WATER PURIFICATION McGraw-Hill book Company 
Puj6^is/ters of75oof for 
Ele c trie al Wor Id The Engineering and Mining Journal 
Engineering Record Engineering News 
Railway Age Gazette American. Machinist 
Signal Engineer American Engineer 
Electric Railway Journal Coal Age 
Metallurgical and Chemical Engineering P o we r (Frontispiece.) 
Cincinnati Filtration Plant. WATER PURIFICATION 
BY 
JOSEPH W. ELLMS, 
•k • 7 
Member American Society Civil Engineers, American Chemical 
Society, American Public Health Association and New 
England Water Works Association. 
First Edition 
McGRAW-HILL BOOK COMPANY, Inc. 
239 WEST 39TH STREET. NEW YORK 
LONDON: HILL PUBLISHING CO., Ltd. 
6 & 8 BOUVERIE ST., E. C. 
1917 Copyright, 1917, by the 
McGraw-Hill Book Company, Inc. 
T II E M A P I, E PRESS YORK PA PREFACE 
In writing this book the object has been to provide the reader 
with a fairly complete account of the development of the art of 
water purification. As a knowledge of the physical, chemical 
and biological characteristics of natural waters is a prerequisite 
to the proper understanding of purification processes, a considera- 
tion of the properties of various classes of waters has been thought 
advisable. The relation of polluted public water supplies to 
water-borne diseases has received especial attention because of 
its importance. The various steps in purification processes, 
such as plain sedimentation, coagulation, filtration and disinfec- 
tion, are described in considerable detail. Special chapters are 
devoted to water softening, and to the removal of iron and 
manganese from ground water supplies. 
The rapid progress made in the art of clarifying and purifying 
turbid waters in the United States during the past quarter of a 
century, has been notable. The evolution of the rapid sand 
filter from its crude beginnings to its present well-developed 
state for purifying waters of this type, is distinctly the result of 
research work undertaken during the early part of this period. 
It was the author's good fortune to have been identified with 
some of the earlier investigations of this problem, and to have 
been able to follow its solution closely in actual practice through- 
out the whole period. 
The experiences of other investigators, as disclosed by their 
published papers, have been drawn upon freely, and while no 
exhaustive examination of the extensive literature of the sub- 
ject has been attempted, it is believed that the writers quoted 
and referred to are sufficiently representative to provide the 
reader with ample information upon the subjects discussed. 
The writer is indebted to a number of his friends who have 
kindly supplied him with original information, illustrations, or 
assistance in writing technical descriptions of apparatus for 
some parts of the book; and to manufacturers of special devices 
VII VIII 
PREFACE 
used in filter plant construction, who have loaned original draw- 
ings and photographs for reproduction. 
The author desires to acknowledge especially the contribution 
of the subject matter in the appendices written by Mr. C. N. 
Miller, Assoc. M. Am. Soc. C. E., dealing with the hydraulics 
of the flow of water through filters, and with the discharge of 
water from waste-water troughs in the operation of rapid sand 
filters. The author is also under special obligation to Mr. 
S. J. Hauser, Chemist and Bacteriologist of the Cincinnati 
Water Purification Plant, for his kindly assistance in reading 
the proof of the book and in preparing the index. 
Cincinnati, Ohio, 
March 1, 1917. CONTENTS 
Page 
Preface vii 
Chapter 
I. Introduction 1 
II. Classification of Natural Waters. . 8 
III. Transmission of Disease Through Drinking Water . 22 
IV. The Effect of Improved Water Supplies upon Health . 34 
V. Objects and Methods of Water Purification .... 43 
VI. Sedimentation 50 
VII. Types of Settling Reservoirs and Coagulation Basins . 60 
VIII. Practical Efficiencies of Settling and Coagulation 
Basins 80 
IX. Filtration of Water 89 
X. Preliminary Treatment of Water for Slow Sand 
Filters 97 
XI. System of Slow Sand Filtration 116 
XII. System of Slow Sand Filtration (Continued) 134 
XIII. Efficiency and Cost of Operation of Slow Sand Filters . 150 
XIV. Rapid Sand Filtration 159 
XV. General Arrangement of Rapid Sand Filter Plants . . 173 
XVI. Details of Rapid Sand Filter Plant Construction. . . 183 
XVII. Details of Rapid Sand Filter Plant Construction 
(Continued) 200 
XVIII. Details of Rapid Sand Filter Plant Construction 
(Continued). 211 
XIX. Regulating, Measuring and Indicating Devices for 
Rapid Sand Filter Plants 220 
XX. Regulating, Measuring and Indicating Devices for 
Rapid Sand Filter Plants (Continued) 238 
XXI. Equipment for the Handling and Storing of Chemicals 
and For the Preparation of Solutions 256 
XXII. Apparatus and Methods for Applying Chemicals and the 
Preparation of Solutions ' 274 
XXIII. Power Plant, Pumping Machinery, Air Compressors, 
Air Tanks, Wash Water Tank and Miscellaneous 
Equipment 290 
XXIV. The Cost of Constructing Rapid Sand Filters. . . . 305 
XXV. Rates of Filtration, Loss of Head and Washing of Rapid 
Sand Filters 312 
IX X 
CONTENTS 
Chapter Page 
XXVI. The Physical and Chemical Changes Produced by the 
Application of Chemical Coagulants, and by the Sub- 
sequent Filtration of the Treated Water 331 
XXVII. Efficiency and Cost of Operation of Rapid Sand Filters 347 
XXVIII. Disinfection of Water Supplies 367 
XXIX. Disinfection of Water Supplies {Continued) 395 
XXX. The Removal of Dissolved Mineral Matter from Water 410 
XXXI. The Removal of Dissolved Mineral Matter from Water 
{Continued) 437 
XXXII. The Control of Water Purification Processes. . . . 452 
Appendix A.-The Flow of Water Through Rapid Sand Filters . 461 
Appendix B.-An Approximate Formula for Calculating the 
Discharging Capacity of Rapid Sand Filter Wash Water 
Troughs 465 
Tables.-Nos. 1 to 12 inclusive 468 
Index 479 WATER PURIFICATION 
CHAPTER I 
INTRODUCTION 
Water purification may be broadly defined as the art of remov- 
ing foreign and polluting substances from solution and suspen- 
sion in water. The character and amount of these impurities, 
and the object for which their removal is undertaken, naturally 
leads to the treatment of this subject under the following heads: 
1. Rendering impure water potable and hygienically safe for 
drinking purposes, and suitable for industrial uses. 
2. Removing putrescible organic matter and disease-produc- 
ing organisms from the fouled water supply or sewage of a 
community. 
In the following pages it is intended to treat only of the first 
division of this subject, since the second covers too broad a 
field to bring it within the scope of this volume.. 
Both the kind and the amount of impurities in a polluted water 
affect the methods which can be employed for its purification, 
and this is so peculiarly true of sewage that its treatment has 
become an art in itself. Those waters which are to be purified 
for drinking purposes demand that a degree of purity shall be 
produced which is not possible and usually unnecessary in the 
case of sewage. The purification of the latter is, as a rule, quite 
incomplete as compared with that required for a safe drinking 
water, and obviously the object to be attained is quite different. 
In the case of waters to be treated for making them suitable for 
steam-boiler purposes, or for general industrial uses, a slightly 
different object is sought. The softening of hard waters or the 
removal of compounds of iron or manganese may be an integral 
part of the purification of a public drinking water supply, in which 
case it is combined with those processes utilized for making a 
water hygienically safe. Of course it is quite possible to carry 
on purification for industrial purposes independently of the more 
refined methods used for drinking water, and frequently this 
course is pursued in connection with manufacturing plants. 
1 2 
WATER PURIFICATION 
HISTORICAL 
Ancient Systems of Water Supply.-Good water for drinking 
purposes has doubtless been appreciated by the human race 
from time immemorial, for it does not take a very high degree 
of intelligence to discriminate between a clear, colorless and 
odorless drinking water, and one that does not possess these 
outward attractive physical properties. Among primitive 
peoples the question of water supply was never of pressing im- 
portance, except in arid and semi-arid regions. In these latter 
countries provision for securing and storing a supply of water 
was usually necessary. Consequently, springs were sought for, 
wells were dug, and cisterns constructed in order that a supply 
of water might at all times be available. 
Wells were common in ancient Egypt, Greece, Assyria, Persia 
and India, and from the sanitary point of view they probably 
furnished a safer drinking water than could be obtained from the 
surface waters in the rivers and lakes. Wells of great antiquity 
may be found today in Egypt and India. Joseph's well at Cairo 
in Egypt is one of the most famous ancient wells, and was 
excavated in solid rock to a depth of 297 ft. The Chinese were 
familiar with the driving of artesian wells and pursued methods 
similar to those now in vogue in sinking them. 
Where turbid surface waters had to be employed for drink- 
ing, domestic filters of unglazed earthenware or of sandstone 
were known to have been used by the ancient Egyptians and 
by the Japanese. Clarification of muddy water by the siphoning 
of the liquid from one vessel to another by the capillary action 
of porous material, such as a strip of cloth, and the consequent 
separation of the water from the suspended matter, was well 
known to the ancients. 
As population became more dense and people began to congre- 
gate in cities, the need for larger volumes of water, than could 
be supplied by springs and wells, became urgent. Works for 
the collection, storage and conveyance of water were built for 
supplying many of the ancient cities, and the ruins of some of 
these works yet remain. 
The ancient water tanks of Aden in Arabia for the collection 
of surface water from the gorges of the volcanic crater, at the 
bottom of which the old city was located, afford an example of 
early impounding reservoirs of an elementary type. These INTRODUCTION 
3 
tanks may have been built by Persian engineers as early as 600 
B.C., or possibly by the Romans; but they without doubt 
antedate the Christian era. 
The infiltration galleries for collecting ground water at Athens 
in Greece were probably constructed over 2,000 years ago, and 
the same method is still pursued to obtain a pure and satisfactory 
water supply. The rain-water cisterns of ancient Carthage 
(150 B.C.) were devised into several storage compartments, two 
of which were apparently used for settling or possibly filtering 
the water. Historians state that the city of Laodicea in Asia 
Minor obtained its supply from the River Caprus. The latter 
had its origin in springs. The water was conveyed to the city 
through a masonry aqueduct over 4 miles in length. A set- 
tling tank with double compartments formed a part of this 
water-works system. In Jerusalem underground cisterns were 
constructed, which were supplied through masonry conduits. 
Probably no more elaborate system of public water supply 
was provided for any ancient city than that for Rome. Until 
312 B.C., the Romans took their supply from the River Tiber 
and from springs and wells in the vicinity of the city. The 
increasing pollution of the river water, as well as the need for 
a greater volume of water, evidently induced the Romans to 
seek elsewhere for a supply. In consequence, three groups of 
springs in the volcanic plain on the left bank of the Tiber were 
made use of, and the water from them conveyed by aqueducts 
to the city. A fourth group of springs was also used which were 
located in the mountains of limestone formation somewhat more 
distant than the three other groups. This latter group furnished 
the best quality of water with which the city was supplied. 
E. H. D'Avigdor, in his "Water Works of Ancient Rome" 
(Engineering, xxi, p. 403), interestingly describes the character 
of the Roman water supply as follows: "The Romans possessed 
three almost independent water services, for which they used 
water of different degrees of purity. The least clear and most 
loaded with sand, such as the Anio aqueduct supplied, was used 
for public baths and the watering of streets; the clearer water 
from Tepula and Alsietina served for tanks, fountains and wash- 
ing troughs; while the very best (Virgo, Marcia and Claudia) 
was confined to drinking purposes, and these springs were 
undefiled even after the heaviest rains." 
Rome was supplied with water from the above-mentioned 4 
WATER PURIFICATION 
four groups of springs through nineteen aqueducts, which were 
built between 321 B.C. and 305 A.D. The aggregate length 
of these aqueducts was 381 miles. 
There can be said to have been no real distribution system 
for the water entering Rome through the system of aqueducts. 
The aqueduct water flowed through small tanks in which the 
heaviest sand and gravel were deposited. To a limited extent 
the small distributing reservoirs (castella) near the city served 
a like purpose. From these reservoirs the water was distributed 
to cisterns, public fountains and private residences. 
During the Middle Ages,1 when disease played such havoc with 
the people of Europe, polluted water undoubtedly assisted in 
conveying infection. At the time of the fall of the Roman Em- 
pire many of the aqueducts built by the Romans at Rome and 
in the provinces were either destroyed or fell into disuse. The 
Moors in Spain during the ninth century constructed some 
important works, as well as repairing in the twelfth century 
some old Roman works. 
Until 1183 A.D. Paris obtained its entire supply of water from 
the River Seine. As late as 1550, Paris used only 1 qt. of water 
per capita per day, and at the end of the seventeenth century 
was using but 2^ qt. per capita per day. With such small 
amounts of water being used one can easily imagine what sanitary 
conditions must have been. 
London was first supplied in small quantities with spring water 
conducted through lead pipes, and masonry conduits. In 1582 
a pump was erected on London Bridge to take water from the 
River Thames, and to deliver it through lead pipes to the city. 
By the invention of the steam engine, pumping machinery of 
adequate capacity and power was made possible. The growth 
and development of water-works plants in reality dates from 
the eighteenth century. However, not until the latter half of 
the nineteenth century was very rapid progress made. The use 
of cast-iron pipe became general in about the year 1800, and 
gradually replaced the wooden mains formerly used. 
Development of Modern Purification Plants.-The methods 
employed in securing and maintaining pure water supplies have 
been in a large measure governed by topographical and geological 
conditions. In different countries, where like physical condi- 
tions prevailed, similar lines of development have not always 
* 1-Turneaure and Russel: "Public Water Supplies." INTRODUCTION 
5 
been followed. Mr. Allen Hazen, in his book on 11 Filtration of 
Public Water Supplies," points out that "it is really marvellous 
how each country has met its problems of water supply from its 
own resources, and often without much regard to the methods 
which had been found most useful elsewhere. England has 
secured a whole series of magnificent supplies by impounding 
the waters of small streams in reservoirs holding enough water 
to last through dry periods, while on Continental Europe such 
supplies are hardly known. Germany has spent millions upon 
millions in purifying turbid and polluted river waters, while 
France and Austria have striven for mountain-spring waters 
and have built hundreds of miles of costly aqueducts to secure 
them. In the United States an abundant supply of some liquid 
has too often been the objective point, and the efforts have been 
most successful, the American works being entirely unrivalled 
in the volumes of their supplies. I do not wish to imply that 
quality has been entirely neglected in our country, for many 
cities and towns have seriously and successfully studied their 
problems, with the result that there are hundreds of water sup- 
plies in the United States which will compare favorably upon 
any basis with supplies in any part of the world; but on the other 
hand, it is equally true that there are hundreds of other cities, 
including some of the largest in the country, which supply their 
citizens with turbid and unhealthy waters which cannot be 
regarded as anything else than a national disgrace and a menace 
to our prosperity." 
Since the above quotation was written, nearly two decades 
ago, many of the cities of the United States have done much to 
redeem the bad reputation which they had because of polluted 
and unwholesome water supplies. There is still much that can 
be done, however, toward improving present conditions, and as 
time goes on and population becomes more dense, the danger 
from polluted water increases. 
Those impurities in water which render it turbid, and which 
for the most part are in suspension, make it unattractive as a 
drinking water. It was probably noted at an early period, that 
a muddy water, if allowed to remain quiescent for a short time, 
lost more or less of its suspended matter by settlement. In 
storing muddy waters for irrigation purposes this phenomenon 
could hardly have failed to have been noticed. 
The appearance of a muddy water was obviously improved by 6 
WATER PURIFICATION 
settlement, and its suitability as a drinking water enhanced. 
If clear spring waters could not be obtained for a public supply, 
resorting to surface waters was a necessity. Hence, we find 
evidence that in some of the ancient water-works systems pro- 
vision was made for settlement of muddy waters in tanks or 
settling basins. The "castellse" and piscinae of the Roman 
aqueduct system evidently performed the function of settling 
tanks, whether originally intended for that purpose or not. In 
other ancient systems the evident purpose of tanks for settling 
purposes is more pronounced. Some of these works have been 
cited in the preceding pages. 
These early devices for improving the quality of a water supply 
were crude and imperfect, and can only be recorded as the 
beginnings from which the really modern art of water purification 
sprang. 
While sedimentation for the clarification of water may be 
looked upon as the earliest step in the art, nevertheless, it can 
only be regarded in most cases as preliminary to more complete 
methods. The straining action of sand and gravel was doubt- 
less also noted by keen observers at an early date, in the same 
way that the clarification effected by settling tanks had been 
observed. The application of this principle to the public water 
supply of London was made in 1829. Filters of sand and gravel 
were also constructed for many of the Continental cities, more 
especially in Germany, where the principles underlying the action 
of filters of this type were carefully studied. 
Mr. Allen Hazen has divided the history of water purification 
in the United States into three epochs, the first beginning with 
James P. Kirkwood's report on the "Filtration of River Waters" 
in 1866, resulting from his study of European practice; the 
second epoch commencing with the work of the Massachusetts 
State Board of Health at its Lawrence Experiment Station in 
1887; and the third with the experiments on very turbid waters 
beginning at Louisville, Ky., in 1896, and continued at Pitts- 
burgh and Cincinnati during the 3 or 4 years following. To 
these three epochs the author would add a fourth, which was 
introduced in 1908 by the experiments at Chicago and at Boon- 
ton, N. J., on the disinfection of water with hypochlorite of lime. 
The use of this latter compound has become widespread in the 
past 5 years throughout the United States, and its usefulness as 
a practical, efficient and economical agent in water purification, INTRODUCTION 
7 
under certain conditions, has been fully demonstrated. The 
sterilization of water with chlorine and its compounds, as well 
as the action of ozone and ultra-violet light, have been carefully 
studied during the past few years, especially in Europe. This 
phase of water purification is well established, and marks an 
important stage in the practical development of the art. 
References 
1. L. F. Vernon-Harcout: "Sanitary Engineering," 1907. 
2. Bolton: "Ancient Methods of Filtration." Pop. Sci. Mo., 16, 495. 
3. Arthur S. Riggs: "Ancient Water Tanks of Aden, Arabia." Eng. 
News, 52, 25, 1904. 
4. "Ancient Water Supply of Athens." Engineer, 101, 215, 1906. 
5. Croes: "The Water Works of Carthage." Eng. Record, 25, 8, 1892. 
6. Edward Wegmann: "The Water Works of Laodicea, Asia Minor." 
Eng. Record, 40, 354. 
7. Edward Wegmann: "Ancient and Modern Water Works." Eng. 
Record, 65, June, 1912. 
8. E. H. D'Avigdor: "Water Works of Ancient Rome." Engineering 
21, 403, 1876. 
9. Geo. Higgins: "The Old Water Supply of Seville." Proc. Inst. 
C. E., 38, 334. 
10. Turneaure and Russel: "Public Water Supplies." 
11. Allen Hazen: "Filtration of Public Water Supplies." 
12. W. P. Mason: "Water Supply." CHAPTER II 
CLASSIFICATION OF NATURAL WATERS 
In order to understand intelligently the methods employed in 
water-purification processes, it is necessary to have a knowledge 
of the composition of the various classes of natural waters which 
require treatment. Although all the water which is now flowing 
on the surface of the ground, as well as that which is found in 
the ground, has a common origin, nevertheless, this water has 
acquired varied characteristics, which have been derived from 
the substances with which it has come into contact. 
Rainfall.-From the water surfaces of the earth the sun is 
constantly evaporating enormous volumes of water, which are 
again condensed and precipitated upon the land and water 
surfaces as rain, snow or ice. That which falls upon the land 
surfaces of the earth is disposed of in different ways. Some of 
the water is evaporated again; some is absorbed by growing 
vegetation; some flows directly over the ground into the rivers 
and lakes; and some sinks into the soil to become the subter- 
ranean water of the earth. The amount of water precipitated 
on the surface of the earth in any particular region depends upon 
a number of factors. Warm air which is saturated with watei 
in the form of vapor will only precipitate this moisture when 
sufficiently cooled. Air currents may be cooled by coming into 
contact with the cold surfaces of the earth, or by being forced 
upward by mountain ranges into colder air currents, or by 
mingling directly with colder air currents at lower levels. Pre- 
cipitation of water will take place under these conditions. 
The water thus precipitated is as near pure water as may be 
found under natural conditions. It probably will contain minute 
quantities of nitrogen compounds, such as ammonium salts, and 
some organic matter consisting of plant spores and bacteria. 
It will also contain in solution small quantities of the gases, 
oxygen and carbon dioxide. 
As the descending rain or snow approaches the earth, it washes 
out of the atmosphere suspended particles of dust and thus puri- 
fies the air. These impurities will, of course, be found in the 
8 CLASSIFICATION OF NATURAL WATERS 
9 
rain water if collected before reaching the ground. Rain water 
caught in or near our large cities is much more impure than that 
collected in the open country. The impurities swept from the 
air are relatively in small amounts as compared with those 
acquired by the water after reaching the earth; but they are the 
first of the dissolved and suspended constituents to be taken 
up by the water, and are soon increased in amount and kind after 
the water reaches the ground. 
SURFACE WATERS 
Flowing Surface Waters.-The character of the land surfaces 
upon which rain or snow is precipitated naturally influences the 
proportion of the volume of water which finds its way over the 
top of the ground into the streams and lakes, and that which 
sinks into the ground to become a part of the underground water 
of the earth. If the land is mountainous, the rain rapidly runs 
off the steep slopes into the valleys, and quickly enters the 
rivers. If the soil is easily washed away, the water will carry 
with it large volumes of sediment, the composition of which will 
depend upon the geological characteristics of the denuded sur- 
faces. On the other hand, if the precipitation occurs in a flat 
level country, a much larger proportion of the water will per- 
colate into the soil. If vegetation is abundant, some of this 
water will be retained by the growing plants, and may sensibly 
retard the discharge of surface water into the natural water 
courses. In water which has been subjected to such conditions 
there will be found those dissolved salts with which the water 
has come into contact, as well as organic matter derived from 
vegetation. 
Flowing surface waters may be characterized, therefore, as 
those which contain relatively small amounts of dissolved mineral 
compounds, more or less of suspended matter, depending upon 
topographical and geological conditions, and variable amounts 
of organic matter, likewise modified by the flora of the region. 
Impounded Surface Waters.-Natural lakes and ponds are 
composed of water which has found its way directly through 
surface streams into the depressions in which these bodies of 
water lay, and of the water which has percolated for some distance 
through the ground and reappears as a surface water at lower 
surface levels. In a like manner the low-water flow of streams 10 
WATER PURIFICATION 
is very largely maintained by the ground-water flow from higher 
levels. 
In consequence, these lake waters are virtually mixtures of 
surface and ground waters, and their composition is, in a large 
measure, intermediate between them. Obviously local topog- 
raphy and the very important factor of sedimentation will 
materially affect their composition. They also have an im- 
portant influence in rendering streams flowing from them more 
uniform in their discharge on account of the storage for water 
which they afford. 
UNDERGROUND WATERS 
The water which sinks into the ground does not all imme- 
diately pass through the subsoil to underground channels. Some 
is held by the surface soil, and in hot countries may soon be 
evaporated into the air; some of the water is absorbed by the 
growing plants; and the remainder sinks slowly or quickly through 
the ground depending upon the porosity of the latter. 
The movement of this ground water is governed exclusively 
by the character of the strata with which it comes into contact. 
Impervious formations will block or divert the flow, while per- 
meable strata will soon become filled and form the channel along 
which the underground stream moves. The velocity of flow is, 
of course, low, and depends upon the porosity of the soil in which 
the water is moving. 
All of these factors have a bearing upon the dissolved constitu- 
ents which will be found in an underground water. The sus- 
pended matter of the surface water will have been filtered out, 
whether in a relatively coarse form like sand and silt, or in a 
finer or colloidal condition like finely divided clay. On the other 
hand, the solution of mineral compounds containing sodium, 
potassium, calcium, magnesium, iron, aluminum, silica, etc., 
will have begun, and will continue as long as the water does not 
become supersaturated, and until "a chemical system of balanced 
values"1 is attained for the various salts with which the water is 
in contact. 
CONTAMINATION OF WATER 
Natural Impurities.-For convenience the impurities in water 
may be divided into classes according to the sources from which 
1 Chase Palmer: "The Geochemical Interpretation of Water Analyses." 
U. S. Geological Survey Bull. 479. CLASSIFICATION OF NATURAL WATERS 
11 
they are derived. As has been previously noted water in the 
course of its flow over the surface or through the ground dis- 
solves more or less material as well as carrying varying amounts 
in suspension. Provided no contact with the wastes of human 
life or activity has occurred, the acquired impurities might be 
designated as natural impurities, in distinction from those 
derived from sewage or manufacturing wastes. Water polluted 
from the latter sources is unfit for human consumption without 
purification as will be shown in a later chapter. Water which 
has not been subjected to such contamination may or may not 
be fit for drinking, depending entirely on the nature of the 
dissolved and suspended impurities which it may contain. 
Dissolved Impurities.-The solvent action of water upon the 
seemingly insoluble constituents of the earth's crust is enormous, 
aside from its erosive action. The surface and subsoil, as well 
as the deeper rock strata, are all subjected to the decomposing 
action of the water with which they come into contact. Erosion 
assists the solution of the mineral compounds composing the 
various strata by reducing them to a finely divided state. 
Disintegration of minerals like orthoclase, for example, is 
able to supply the bases sodium and potassium; dolomitic lime- 
stones may furnish calcium and magnesium; and clay ironstone 
and pyrite may give up iron, aluminum and silica. The acids 
with which the above bases are usually associated are carbonic, 
hydrochloric and sulphuric acids. These latter are derived from 
the disintegration of the minerals in the rock strata the same as 
are the bases. More or less carbonic acid is obtained directly 
from the air and from the oxidation of carbonaceous compounds 
in the breaking down of organic matter. Minute quantities of 
nitric acid may be derived directly from the air, and from oxida- 
tion of nitrogenous compounds. The oxidation of sulphur in 
minerals supplies sulphuric acid in many cases. Sulphates, 
chlorides, nitrates, carbonates and silicates exist as such in the 
earth's crust in enormous amounts, and whenever they come into 
contact with water are dissolved directly, and in amounts 
depending upon the supply available and the solubility of the 
salt being acted upon. 
A classification of natural waters with respect to their dissolved 
constituents, which is based upon the chemical nature and the 
proportional amounts of the radicles present in solution, has been 
suggested by Mr. Chase Palmer in a paper entitled "The Geo- 12 
WATER PURIFICATION 
chemical Interpretation of Water Analyses" (U. S. Geological 
Survey Bull. 479, 1911). He states that: 
"Nearly all terrestrial waters have two general properties, salinity and 
alkalinity, on whose relative proportions their fundamental characters 
depend. Salinity is caused by salts that are not hydrolyzed; alkalinity 
is attributed to free alkaline bases produced by the hydrolytic action 
of water on solutions of bicarbonates and on solutions of salts of other 
weak acids." 
According to his classification waters may possess five special 
properties as follows: 
1. Primary salinity; that is salinity caused by the sulphates 
and chlorides of the alkalies sodium and potassium. 
2. Secondary salinity; that is salinity produced by the sul- 
phates and chlorides of the alkaline earths calcium and mag- 
nesium, or in other words permanent hardness. 
3. Tertiary salinity; that is in reality an "acidity" arising from 
an excess of saline compounds over and above that due to primary 
and secondary salinity. 
4. Primary alkalinity; that is an alkalinity produced by car- 
bonates and bicarbonates of sodium and potassium, or in other 
words permanent alkalinity. 
5. Secondary alkalinity; that is the property caused by alkaline 
earth bicarbonates such as those of calcium and magnesium, or 
in other words temporary hardness. 
By various combinations of these properties five well-defined 
classes of waters are possible and are found in nature. 
Class 1. Waters characterized by properties 1, 4 and 5 as 
stated above. 
Class 2. Those showing properties numbered 1 and 5. 
Class 3. Those exhibiting properties numbered 1, 2 and 5. 
Class 4. Those marked by properties numbered 1 and 2. 
Class 5. Those possessing properties numbered 1, 2 and 3. 
"Surface waters appear to belong chiefly to the first three 
classes, class 4 is represented by sea water and brines, class 5 
is exemplified by mine (acid) waters and waters of volcanic 
origin." 
Suspended Impurities.-Intermediate between a solution, as 
it is commonly understood, and a suspension of finely divided 
particles, like sand for example, there may be so-called colloidal 
suspensions of certain substances in water which possess 
peculiar properties, and which are of considerable importance CLASSIFICATION OF NATURAL WATERS 
13 
in connection with water purification problems. For example, 
silica is found in this state in many natural waters, especially 
those showing primary alkalinity, i.e., waters containing sodium 
and potassium carbonates. The silica compounds characteristic 
of the clays show a marked tendency toward the colloidal state, 
and render the problem of the purification of turbid waters of 
this class almost a problem in itself. 
The enormous amount of the heavier sediment carried by 
many rivers consists largely of sand and clay. Any reduction 
in the velocity of flow of the water laden with such sediment 
causes it to be deposited on the bed of the stream. Gravel and 
sand bars are thus formed in river beds, which may be shifted 
from one point to another by any sudden increase in the velocity 
of the current, such as might be produced by a flood. Lighter 
material like clay is more slowly deposited and more quickly 
moved again by any change in the rate of flow of the water. It 
thus happens that almost all surface waters carry varying 
amounts of suspended material in them which depend upon the 
velocity of the currents of water and on the character of the 
bottom and shores. The action of the wind on large bodies of 
water like the Great Lakes, for example, may stir up the sediment 
on the bottom and render the water, near the shore especially, 
quite turbid. 
Compounds of iron and of manganese are not infrequently met 
with in the colloidal state in natural waters, and offer some of the 
most interesting phases of water-purification work. Organic 
matter found in natural waters is probably always present in this 
form to a greater or less degree. Vegetable stain produced by 
humic substances is a marked characteristic of a very large class 
of natural surface waters, and as such has received considerable 
attention in the study of purification problems. 
How the removal of these impurities, whether in solution or in 
suspension, is effected will be discussed under the description 
of the various methods of purification now employed, rather than 
in this place. It is only desired to emphasize the varied classes 
of impurities which may be found in natural waters, and which 
are virtually "natural impurities," as distinguished from those 
derived from sewage and manufacturing wastes. This waste 
material may furnish similar classes of polluting compounds to 
those derived naturally, but their origin usually justifies their 
consideration separately. 14 
WATER PURIFICATION 
Whether the dissolved salts usually found in natural waters 
are objectionable depends upon the use to which the water is 
to be put. A certain amount of the chlorides, sulphates and 
carbonates of sodium, potassium, calcium and magnesium are 
by no means deleterious, and may possibly be beneficial in a 
drinking water. On the other hand, if too great amounts are 
present these dissolved salts render the water unfit for domestic 
and industrial uses, and actually cause financial losses of no small 
amount to those obliged to use them. Compounds of iron or of 
manganese arc especially objectionable, and not infrequently 
have caused waters containing them to be abandoned as sources 
of supply. Excessive amounts of the fixed alkalies either as 
salts of the strong acids like hydrochloric and sulphuric, or of 
the weak acids like carbonic, make a water unsuitable as a public 
supply. 
Microscopic Plant and Animal Life.-In many of our natural 
waters a luxuriant growth of algae and diatoms is found at certain 
seasons of the year. The character of the mineral and organic 
constituents of the water, as well as the conditions of light and 
temperature materially affect the extent of these growths. They 
occur in both still and running water. Accompanying the growth 
and also the decay of certain of these organisms bad odors and 
tastes are not infrequently developed, and where this occurs 
in public water supplies, they become a nuisance entirely out of 
proportion to their number and size. Certain microscopic 
animal forms may also produce troubles of this same character. 
The odor of growth appears to be due to secretions of an oil- 
like character which they produce, and is usually somewhat 
characteristic of the special organism producing it. When the 
organisms are in large enough numbers they are capable of giving 
an odor to large volumes of water, and not infrequently spoil 
the taste and odor of the whole of a public water supply. Odors 
of decomposition are usually very offensive, and notably so in 
the case of the "blue green algae" or Cyanophyceae. 
It is not probable that impurities of this nature in natural 
waters produce disease in human beings, when swallowed in 
drinking water. They are very objectionable if they produce 
a marked odor, and in such cases are most frequently com- 
plained of in public water supplies. 
Bacteria.-Even lower in the scale of plant life than the 
diatoms and the algse are found the bacteria. They are present CLASSIFICATION OF NATURAL WATERS 
15 
in all natural waters, being the more numerous in surface waters, 
and much less so in ground waters. By far the larger number of 
the various species of bacteria play a beneficent role in the 
economy of nature, and appear absolutely essential to many of 
the normal processes of development of both plants and animals. 
A few species, however, are associated with disease in animals 
and in human beings. So far as the contamination of water is 
concerned, it is only the organisms capable of producing patho- 
logic conditions in man that are of interest. These virulent 
forms usually reach our natural waters through the medium of 
domestic sewage and manufacturing wastes. This class of 
impurities is considered in more detail in the next section. 
IMPURITIES DERIVED FROM WASTE MATERIAL 
Sewage.-The water carriage of waste material of human and 
animal origin has become, in those countries which pay any atten- 
tion to problems of sanitation, the most common method for 
its transfer to some point of ultimate disposal. Wherever 
public water supplies are installed, a system of sewers will of 
necessity follow. Hence the disposal of large volumes of fouled 
water has become a problem of great difficulty, and one that yet 
awaits a completely satisfactory solution. 
It is obvious that some method, even though it is not entirely 
satisfactory, must be used to get rid of this polluted water or 
sewage, and the easiest way has been to turn it into the natural 
water courses. In this manner much of the surface water on 
thickly settled land areas has become polluted with material 
dangerous to the health of human beings, who unwittingly or of 
necessity drink the water thus contaminated. Since disease has 
been found to originate so largely from specific plant and animal 
forms, microscopic in size, which, having produced the disease, 
are discharged from the body chiefly in the excreta and the urine, 
the conveyance of disease through sewage to water has been 
pretty definitely proven. 
Ground waters as well as surface waters may become polluted 
by sewage. The discharge of sewage on the surface of the ground 
or into cesspools, or the leakage or overflow of vaults, may furnish 
the dangerous pollution to well and spring waters by direct 
percolation through fissures in the rock strata, or by more 
indirect routes through the soil itself. 16 
JFA7W PURIFICATION 
Manufacturing Wastes.-In many industries there remains 
after the manufactured product has been completed, a great 
deal of waste material, which for economic reasons it is not worth 
while to work over. Much of this material is in suspension and 
solution in relatively large volumes of water. Its disposal by 
the easiest method is to dump it into the nearest body of water. 
Water fouled with such material is totally unfit for human con- 
sumption. Frequently the material renders even the best 
methods for the purification of domestic sewage inadequate, and 
its proper disposal becomes a special problem in almost every 
case. 
The waste liquids from textile works, dye works, straw board 
factories, paper mills, abattoirs, meat packing establishments, 
dairies, etc., furnish material which is obviously difficult to 
dispose of, and which must pollute in the foulest manner any 
natural water into which they may be turned. 
NATURAL METHODS OF PURIFICATION 
Sedimentation.-Some of the impurities which a natural water 
acquires in the course of its flow, may be lost under certain favor- 
able conditions. For example, a water laden with suspended clay 
or fine sand will deposit this material as soon as the velocity of 
the water is sufficiently retarded as was previously explained. 
This process of sedimentation is one of the most important of 
the natural methods of purification, and plays an important part 
in our artificial methods as well. In flowing streams the deposi- 
tion of sediment is intermittent, being active during low stages 
of the stream when the rate of flow is relatively low, and much 
diminished or practically nil in flood periods. At such times the 
scouring action of the current causes much that has been de- 
posited to be again placed in suspension, and thus carried further 
toward its ultimate disposal in the sea. In this way the immense 
deltas at the mouths of rivers like the Mississippi and the Nile 
are formed. 
Effect of Sunlight.-The purifying action of sunlight on certain 
vegetable compounds in colloidal suspension, such as the brown 
coloring matter in many of the streams and lakes in the north 
central and northeastern parts of the United States, is worth 
mentioning in this connection. A certain amount of bleaching 
out of this coloring matter is apparently effected when this class CLASSIFICATION OF NATURAL WATERS 
17 
of waters are impounded in natural lakes or artificial reservoirs. 
Oxidation of the carbonaceous matter probably occurs, and 
sedimentation in the quiet water undoubtedly assists in the 
clarification. 
Precipitation of Compounds from Solution.-Dissolved salts 
are not usually readily removed once they have gone into solu- 
tion. The chlorides, sulphates and nitrates of either sodium, 
potassium, ammonium, calcium or magnesium will be retained 
on account of their great solubility. Carbonates and bicar- 
bonates of the fixed alkalies, as well as of ammonia, are also 
very soluble. On the other hand, the bicarbonates of the 
alkaline earths have a rather limited solubility and may be 
deposited from solution if the excess of carbon dioxide, which is 
necessary for their retention in solution, is in anyway removed. 
Ground waters in particular may become heavily charged with 
bicarbonates and, on being brought to the surface where the 
pressure is diminished, will lose some of their free carbon dioxide 
and deposit their monocarbonates, which are much less soluble. 
This is particularly true of calcium carbonate. Magnesium 
carbonate, however, is considerably more soluble. 
Oxidizable salts like ferrous sulphate, ferrous carbonate, and 
corresponding salts of manganese occurring in ground waters 
may be deposited from solution on exposure to the air. Such 
purification can be hastened by aeration and thus render some 
unsuitable deep well waters entirely acceptable as a source of 
water supply. 
"Acid mine waters" are usually contaminated with dissolved 
iron compounds, which not infrequently find their way into sur- 
face streams. As much of this iron may be in an unoxidized 
state, the exposure to the oxygen of the air, and to that dissolved 
in the surface water, soon converts the iron to the form of the 
insoluble ferric oxide. Organic matter in colloidal suspension 
may retard the precipitation of the iron, and cause the latter to 
assume a colloidal state itself. 
In waters in which the alkalinity is due to sodium and potas- 
sium carbonates, colloidal solutions of silica and alumina are 
sometimes found, and on account of their slight solubility may 
be deposited, should this "primary alkalinity" be diminished. 
Such a diminution can be effected if waters of this class come 
into contact with chlorides and sulphates of lime and magnesia. 
These latter salts will react with the fixed alkaline carbonates, 18 
WATER PURIFICATION 
forming carbonates of lime and magnesia and the chlorides and 
sulphates of sodium and potassium. The latter salts are without 
power to assist in holding the silica in solution. 
Filtration.-The natural filtration of water through the soil 
effects a high degree of purification provided the ground is of the 
right character. Sand and gravel, when not too coarse, afford 
an excellent purifying medium. Suspended impurities, organic 
matter and oxidizable salts are removed as a result of the strain- 
ing action, and the chemical and biological changes induced 
during filtration. The action of both sedimentation and filtra- 
tion in purifying natural waters is perfectly normal, and one 
which is constantly going on. To these agencies we owe the 
potability of most of our ground waters, and by a study of the 
principles underlying these natural processes we have been able 
to design and operate our modern water-purification plants. 
PURIFICATION BY MEANS OF MINUTE PLANT AND ANIMAL 
ORGANISMS 
Thus far in considering natural methods of purification only 
those agencies have been especially noted which are effective 
without the intervention of organized plant and animal life. 
In the cycle through which inert mineral matter passes into 
organized matter, and then back again into inorganic compounds, 
life in some of its most marvellous forms plays an important and 
essential part. These forms belong both to the animal and 
vegetable kingdom, and for the most part are microscopic in 
size. The borderland between plant and animal, in these almost 
invisible organisms, is extremely ill-defined; but no matter 
how classified their importance in the economy of nature is 
fundamental. 
All living organisms which float about in water between the 
surface and the bottom are designated by biologists as "plank- 
ton." They are moved about by the currents and the wind 
chiefly, although they have slight powers of locomotion. They 
also possess the peculiar ability to remain suspended in the water 
with little effort on their part. The plankton can be divided 
into two general classes: 
"the food producers or plants, which assimilate inorganic matter and 
build up organic compounds by means of their chromophyll coloring 
matter; and the food consumers or animals, such as the microscopic 
protozoa, rotifera, etc., together with the larger ones up to the fishes." CLASSIFICATION OF NATURAL WATERS 
19 
In the lecture from which the above quotation was cited, Dr. 
Marsson1 concisely epitomizes the relations of these two groups 
by stating that the 
"vegetable component of the plankton is the fundamental food supply 
or condition of existence for all aquatic life. It comes from the prod- 
ucts of the decomposition of the albumen which finds its way into the 
water from decaying animals and plants, as well as from sewage. The 
self-purifying power of natural waters is merely the maintenance 
of the proper equilibrium between retrogressive and progressive 
metamorphosis." 
The groups of microscopic plant forms known as the algae, 
diatoms, fungi and bacteria exist in enormous numbers in all 
natural surface waters, and to some extent in ground waters. 
The algae and diatoms through their peculiar cellular structure 
are living laboratories in which light is the energy which tears 
the carbon from carbonic acid, and the nitrogen from its simpler 
compounds and converts them into starch, sugar and albumen. 
Thus oxygen is liberated and becomes available for oxidizing 
organic matter and preventing putrefactive changes. 
The fungi and bacteria find their nutriment in dead organic 
matter, and are the primary agents for its decomposition into 
simpler compounds. Bacterial activity is associated with the 
using up of large amounts of oxygen, where the latter is available; 
and in such cases non-putrefactive disposal of contaminating 
impurities in water is in process in distinction to putrefactive 
changes where the oxygen is not present. 
The bacteria are the natural food for many of the microscopic 
animal organisms. The latter include the protozoa, infusoria 
and metazoa, and where these organisms are found in abundance, 
bacteria and food for bacteria will also be present. They thus 
become indexes of pollution in water, quite as indicative as the 
bacteria themselves. When the food supply is gone they must 
die also. 
Since the smaller plant and animal organisms are the source 
of food for the fishes, and they in turn for human beings, the 
cycle of matter from man through human wastes to mineralized 
1 Max Marsson: "The Significance of Flora and Fauna in Maintaining 
the Purity of Natural Waters, and How They are Affected by Domestic 
Sewage and Industrial Wastes." Eng. News, Aug. 31, 1911. Trans, by 
Emil Kuichling. 20 
WATER PURIFICATION 
compounds and back again to man is complete. The natural 
methods of self-purification of water are going on ceaselessly and 
effectively, but the agencies ordained for this purpose must have 
time and opportunity to do their work. 
References 
1. United States Geological Survey: 
(a) Water Supply Papers: 
"Conservation of Water Resources." No. 234, 1909. 
Herman Stabler: "Some Stream Waters of the Western United 
States." No. 274, 1911. 
R. B. Dole: "The Quality of Surface Waters in the United States." 
No. 236. 
Herman Stabler and Gilbert H. Pratt: "The Purification of Some 
Textile and Other Factory Wastes." No. 235, 1909. 
(6) Water Supply and Irrigation Papers: 
D. D. Jackson: "The Normal Distribution of Chlorine in the Natural 
Waters of New York and New England." No. 144, 1905. 
Herman Stabler: "Prevention of Stream Pollution by Distillery 
Refuse." No. 179, 1906. 
Herman Stabler: "Stream Pollution by Acid Iron Wastes." No. 
186, 1906. 
(c) Bulletins: 
Chase Palmer: " The Geochemical Interpretation of Water Analyses." 
Bull. 479, 1911. 
2. United States Department of Agriculture: 
Office of Public Roads: 
A. S. Cushman: "A Study of Rock Decomposition Under the Action of 
Water." Circular 38. . 
Bulletins: 
A. S. Cushman: "The Effect of Waters on Rock Powders." Bull. 
92, 1905. 
Bureau of Chemistry: 
J. K. Haywood and B. H. Smith: "Mineral Waters of the United 
States." Bull. 91, 1907. 
3. Mass. State Board of Health Report for 1892: 
T. M. Drown: "On the Mineral Constituents of Some Natural Waters 
in Massachusetts." 
4. Engineering News: 
Dr. Max Marsson: "The Significance of Flora and Fauna in Main- 
taining the Purity of Natural Waters, and How They Are Affected by 
Domestic Sewage and Industrial Wastes." Aug. 31, 1911. Trans, by 
Emil Kuichling. J. D. Watson: "Pollution of the River Taine." 
Feb. 8, 1912. 
5. Journal New England Water-works Association: 
G. C. Whipple and D. D. Jackson: "Asterionella: Its Biology, Its 
Chemistry and Its Effect on Water Supplies." Vol. 14, No, 1. CLASSIFICATION OF NATURAL WATERS 
21 
F. S. Hollis and H. N. Parker: "Chlamydomonas in Spot Pond 
(Mass.)." Vol. 14, No. 1. 
R. S. Weston: "The Occurrence of Cristatella in the Storage Reservoirs 
at Henderson, N. C." Vol. 13, No. 1. 
T. M. Drown: "Odor and Color of Surface Waters." Vol. 2, No. 3, 
1888. 
6. Transactions American Microscopical Society: 
D. D. Jackson: "A New Species of Crenthroix (C. Manganifera)." 
Vol. 23, May, 1902. 
G. C. Whipple and H. N. Parker: "On the Amount of Dissolved 
Oxygen and Carbonic Acid Dissolved in Natural Waters, and the 
Effect of These Gases upon the Occurrence of Microscopic Organ- 
isms." Vol. 24, May, 1902. 
H. N. Parker: "Notes on the Growth of Synura in Lake Cochituate, 
Mass." Vol. 30, No. 2, April, 1911. 
7. American Naturalist: 
D. D. Jackson: "Movements of Diatoms and Other Microscopic 
Plants." Vol. 23, No. 461, 1905. 
8. Technology Quarterly: 
D. D. Jackson and J. W. Ellms: "Odors and Tastes of Surface Waters 
with Especial Reference to Anabaena." Vol. 10, No. 4, December, 
1897. 
9. Geo. C. Whipple: "Microscopy of Drinking Water." 
10. Surveyor: 
"Algse and Water Supplies." Aug. 25, 1911. 
James Scott: "The Chara: A Water-purifying Plant." Aug. 25, 1911. 
11. Proceedings Engineers Society of Western Pennsylvania- 
T. P. Roberts: "Acids in the Monongahela River." November, 1911. 
12. Engineering and Contracting: 
Thorndyke Saville: "The Nature of Color in Water." January 
10, 1917. CHAPTER III 
TRANSMISSION OF DISEASE THROUGH DRINKING 
WATER 
The discharge of sewage and waste material of all kinds into 
the streams and lakes obviously affords ample opportunity for 
disease-producing organisms to enter the sources of most of our 
public water supplies. Those diseases peculiar to the intestinal 
tract of the human body are the ones most likely to be dis- 
seminated in consequence of this common practice; and hence, 
typhoid fever, cholera, dysentery and gastro-intestinal disturb- 
ances have come to be regarded as derived in a large measure 
from polluted drinking water, wherever these diseases are endemic 
or even epidemic. Anyone or all of these diseases may be trans- 
mitted in other ways, but where they are widespread, some com- 
mon carrier of infection is generally found to be the source, and 
a common drinking-water supply usually offers the most favorable 
opportunity, for transmitting the disease. 
Probably most of the diseases transmitted by water are of 
bacterial origin. The "spirillum cholerse" of Asiatic cholera; 
the "bacillus typhosus" of typhoid fever, and the "bacillus 
dysenteriae" of dysentery have all been found in contaminated 
drinking water. Pathogenic protozoa may also produce certain 
diseases, and in the case of one form of dysentery, an amoeba 
is known to be the cause. If the theory advanced by Sedgwick 
and MacNutt, that inflammatory diseases of the respiratory 
organs may be also to some extent water-borne, is accepted, then 
polluted water supplies are chargeable with another group of 
diseases particularly prevalent among all classes of people. 
The virility of disease-producing organisms upon their entrance 
to a water is of importance with respect to the real danger which 
they possess. This ability to live and retain their vitality in a 
medium foreign to their natural habitat is also of consequence, 
for if the power to reproduce the disease is soon diminished and 
eventually destroyed, then the length of time before their vitality 
is lost is of the utmost importance. Much experimental work 
has been done to determine the period elapsing before certain 
pathogenic organisms die, when placed in water under varying 
22 TRANSMISSION OF DISEASES 
23 
conditions. Laboratory experiments throw some light on this 
problem, but are not usually conclusive, because of the artificial 
conditions imposed. In almost all the experimental work the 
number of organisms diminish in time, and usually very rapidly. 
This may be the result of a decreasing food supply, or to toxic 
compounds eliminated in the course of growth or decay, which 
kill off rapidly the less resistant organisms. 
The presence of pathogenic forms in a drinking-water supply 
denotes, of course, all the possibilities of dangerous infection. 
Nevertheless, sanitarians have of late regarded the number of 
such organisms, and the length of time which they may have 
been in the supply, as factors of much importance in the epi- 
demiology of disease. This quantitative feature is of consider- 
able significance in connection with the quality of a water 
obtained by the methods commonly employed in the purification 
of polluted waters. 
Spirillum Cholerae and Bacillus Typhosus.-The cholera 
spirillum and the typhoid bacillus are the pathogenic organisms 
which have been most studied in water-borne diseases. It is 
very doubtful whether either of these organisms will multiply 
outside the body, or in impure water. The cholera spirillum 
is not very resistant to adverse conditions outside the human 
body. It is killed in 10 min. by a temperature of 60°C., easily 
destroyed by chemical disinfectants, and does not long retain 
its vitality in association with the ordinary saprophytic bacteria 
in the water. 
The typhoid bacillus is probably more resistant than the 
cholera organism to outside influences. Laboratory experiments 
have demonstrated that the typhoid bacillus will live in sterile 
water in glass vessels for 3 months, and in unsterilized ground and 
surface waters for several weeks. Jordan1 showed by his experi- 
ments with typhoid cultures placed in sacks of collodion and 
parchment and suspended in flowing water, that they would 
retain their vitality under natural conditions for at least 4 or 
5 days. Mr. Geo. A. Johnson's experiments at Columbus, 
Ohio,2 in which he modified Prof. Jordan's technique, showed 
that the ability of the bacteria to pass through the walls of the 
parchment sacks, might indicate that conclusions drawn from 
the disappearance of the bacteria in Jordan's experiments, were 
Jordan, Russel and Zeit: Jour. Infect. Diseases, 1904, 1, p. 641. 
2 Eng. Record, vol. 52, Sept. 23, 1905. 24 
WATER PURIFICATION 
somewhat misleading. If the organisms actually escaped from 
the sacks, failure of samples withdrawn from the latter to 
develop typical cultures, did not necessarily mean that the 
typhoid bacilli had died. Jordan concludes,1 that: 
"It is possible that water may continue to be the vehicle of infection 
during a much longer period (than 4 or 5 days), but the available data 
point to a comparatively short duration of life of the specific germ in 
the water of flowing streams." 
Houston2 has shown that samples of Thames River water 
inoculated with a typhoid emulsion and stored at temperatures 
ranging from 32°F. to 98.6°F., developed negative tests for 
typhoid in 9 weeks at the low temperature, and in 2 weeks at 
the highest temperature. Intermediate temperatures gave nega- 
tive results in conformity with the results stated above, viz., 
the higher the temperature of the water, the shorter the period 
of life of the organism. The history of typhoid epidemics tends 
to confirm in a measure the data obtained in these experiments. 
It emphasizes the protective value of ample periods of sedimenta- 
tion and storage of polluted waters used as public supplies. 
Isolation of Cholera and Typhoid Organisms from Water.- 
The actual isolation of these two organisms from polluted water 
has been accomplished only in a comparatively few well-authenti- 
cated cases. In the case of cholera the organism is discharged 
from the intestines in enormous numbers, but not in the urine. 
Its appearance in sewage and polluted water would, therefore, 
be expected, and has been demonstrated. In 1892 Dunbar 
isolated the spirillum of cholera from the polluted water of the 
Elbe, during the epidemic in Hamburg. Koch3 also reports its 
isolation from the water of two Altona reservoirs supplied also 
from the Elbe. 
The isolation of the typhoid bacillus from natural waters also 
offers a great deal of difficulty, although it probably is more 
virile and capable of living longer in natural water than the 
cholera organism. The bacilli of typhoid fever are discharged 
from the human body both in the urine and the feces. From 
9 to 14 days after infection has taken place are required before 
the disease fully develops. This characteristic feature of typhoid 
fever makes the tracing of infection through natural waters much 
1 E. O. Jordan: "General Bacteriology." 
2 Eng. Record, vol. 65, June 1, 1912, p. 608. 
3 Zeit. fur hygiene und Infect. Krank., 14. TRANSMISSION OF DISEASES 
25 
more difficult, for although the bacilli may have been present 
and caused the disease, they will have probably disappeared 
before suspicion is thoroughly aroused as to their possible presence 
in the water. Comparatively few cases have been recorded, 
therefore, in which the bacillus has been isolated and shown to 
have been the probable cause of a case of typhoid fever. 
The work of Mr. D. D. Jackson and his associates in improving 
methods1 of technique for the differentiation of the colon-typhoid 
group of bacilli offers considerable hope that the isolation of the 
typhoid fever bacillus may yet be accomplished with more cer- 
tainty and ease. Mr. Jackson states that he has isolated B. 
typhosus from a river water used as a source of water supply, 
from a local private water supply and from two points in the 
Hudson River. 
Other Water-borne Diseases.-Intestinal diseases and some 
gastric troubles may be and probably frequently are caused by 
organisms found in water. Among infants this perhaps is truer 
than with adults. Epidemics of diarrhea and dysentery are not 
uncommon and have been traced to impure drinking water. 
The possible infection of a water supply by anthrax (B. anthracis) 
derived from animals sick with the disease produced by this 
organism is rather remote, but not impossible. It is of more 
theoretical interest than practical that the pathogenic organisms 
B. anthracis and B. tetani2 have both been isolated from river 
water by Zeit and Fiitterer; but it goes to show the possibilities 
of water-borne infection. Sewage-polluted waters may contain 
all known pathogens as well as saprophytes, and what role the 
latter forms may play in disease is by no means a settled question. 
The relation between pneumonia, bronchitis and other inflam- 
matory diseases affecting the respiratory organs and polluted 
drinking water has been noted above. The ascertainable facts 
relating to this phase of water-borne diseases are few and difficult 
to satisfactorily classify. The data already collected by Dr. 
W. T. Sedgwick and his associates are extremely valuable, and 
further confirmation of their deductions is hoped for. 
1 D. D. Jackson and T. W. Melia: "Differential Methods for Detecting 
the Typhoid Bacillus in Infected Water and Milk." Jour. Infect. Diseases, 
vol. 6, No. 2, April 1, 1909. 
2 "Report of the Sanitary Investigation of the Illinois River and Its 
Tributaries." Ill. State Board Health, 1900, p. 85. 26 
WATER PURIFICATION 
EPIDEMICS OF WATER-BORNE DISEASES 
Cholera 
Probably no disease is more truly characterized as a "filth 
disease" than is cholera. In certain parts of India it may be 
said to be endemic. Explosive outbreaks are not uncommon, and 
the spreading of the disease by contact is probably constantly 
going on. The insanitary nature of the personal habits of the 
lower classes of natives affords ample opportunity for trans- 
mitting infection, and the streams and lakes frequently serve 
as carriers. 
In 1817 a violent epidemic of cholera broke out in Jessore in 
Bengal, which rapidly spread over a larger part of British India. 
It continued unabated for 3 years, and then began to spread into 
China and Persia. In 1823 the disease had reached Asia Minor 
and Russia. For the next 7 years it did not advance westward 
any further, but a fresh outbreak in 1830 in Russia caused the 
disease to spread all over the latter country and into northern 
Europe and the British Isles. During the next 5 years it spread 
southward, invading northern Africa. 
Another epidemic started in India and China in 1841, reaching 
Europe in 1847; another began in 1850 and entered Europe in 
1853, and was carried across the Atlantic to North and South 
America, where it was particularly severe. The epidemic of 
1865-66 was less extensive than its predecessors. Since 1832 
eight epidemics of cholera have occurred in the United States, 
the last being in 1873. 
With a better idea of the true cause for this disease in par- 
ticular, and with improved methods for combating infection and 
contagious diseases in general, cholera has not been widely 
prevalent in Europe or the United States for a great many years. 
Constant vigilance is required for its suppression, however, and 
only by prompt action, where sporadic cases are discovered, have 
the health authorities prevented epidemics. How many of these 
epidemics have been directly transmitted through drinking water, 
it is impossible to know; but that water acted as a carrier to a 
greater or less extent in many of them is extremely probable. 
In the period from 1831 to 1873, 373,000 people died in Prussia 
of Asiatic cholera, and in 1886 alone 114,000. In 1892, 1,634 
persons died from this disease in Prussia, and from the Hamburg TRANSMISSION OF DISEASES 
27 
epidemic in this same year 8,616 deaths resulted. In 1910 
Germany had but 10 cases of cholera.1 
Circumstantial evidence of a very convincing character has 
been collected, which proves that some cholera epidemics were 
water-borne, and probably no discussion of this subject is com- 
plete without mentioning the disastrous Hamburg epidemic which 
occurred in 1892-93. This city was using unfiltered water from 
the River Elbe, which was contaminated by the sewage of over 
800,000 people. During the fall of 1892, 17,000 cases developed, 
resulting in 8,600 deaths. In fact, wherever the drinking-water 
supply was either filtered or obtained from some source other 
than the river, few or no cases resulted. 
In 1887 the city of Messina, Sicily, suffered from an epidemic 
of cholera, during which 5,000 cases and 2,200 deaths resulted. 
An investigation showed that water purposely diverted from a 
conduit, conveying water to the city, ran into pools, which were 
used for washing soiled clothing by the Messina washerwomen. 
Much of this water found its way back into the open conduit, 
and passed into the city. It was also found that the unglazed 
tile used to distribute the water in the city were broken, and that 
leakage from joints was common. Sewers laid on top and parallel 
with the water mains were in a like condition and offered ex- 
cellent opportunity for further contamination. After a supply 
of pure water, carried in tank ships from the mainland, was 
provided for drinking water, the epidemic ceased at once.2 
A similar outbreak of cholera in 1884 in Cuneo, Italy, which 
resulted in 3,344 cases, was traced to a like cause, viz., washing 
infected linen in a brook emptying into a public water supply.2 
Typhoid Fever 
The prevalence of typhoid fever in civilized countries, where 
no little attention is paid to matters of sanitation, seems at first 
thought surprising. But not until 1880 was the organism which 
causes this disease discovered by Eberth in the spleen of persons 
dying from typhoid fever. Since it seems doubtful that this 
disease, as it develops in human beings, can be reproduced in 
animals, the evidence that the Eberth bacillus is the true cause 
1 Dr. Arthur Lederer: "The Modern Sewage and Water Problem." 
Clinique, August, 1912. 
2 W. P. Mason: "Water Supply." 28 
WATER PURIFICATION 
for the disease has been only slowly accumulating. Another 
factor only recently discovered is that persons showing no clinical 
symptoms of the disease are genuine "culture factories" for 
producing the bacillus and for its dissemination. 
Epidemics resulting from these "typhoid carriers" have been 
satisfactorily traced. The existence of such persons explains in 
some measure the continuance of the disease, and its persistence. 
The transmission by direct contact, by flies, by milk and by water 
has been proven in scores of cases. 
While there may be some extenuating circumstances for the 
continued presence of typhoid fever to the extent to which it still 
exists in the United States, nevertheless it constitutes a national 
disgrace, of which our sanitary authorities and people as a whole 
should be heartily ashamed. Typhoid fever has been practically 
stamped out in Europe as the following table will show.1 
Unit of comparison 
Aggregate 
population 
Deaths per 
100,000 from 
typhoid fever, 
1910 
Thirty-three principal European cities in Russia. 
Sweden, Norway, Austria-Hungary, Germany, 
Denmark, France, Belgium, Holland, England, 
Scotland and Ireland 
31,590,000 
6.5 
Fifty American cities of 100,000 inhabitants or 
over 
20,250,000 
25.0 
Excess of deaths, typhoid fever in American cities 
per 100.000 population 
18.5 
In three-fifths of the population of the United States included 
in the registration area for mortality statistics as compiled by 
the Census Bureau, there occurred 12,673 deaths from typhoid 
fever in 1910, or a death rate of 23.5 per 100,000 of population. 
Assuming the same proportion of deaths in the unregistered 
sections as in the registered, then there were 21,120 deaths from 
this disease, representing probably 200,000 cases. 
The following diagram (Fig. 1) illustrates the prevalence of this 
disease in the United States as compared with certain countries 
in Europe. 
1 Allan J. McLaughlin: "Sewage Pollution of Interstate and Inter- 
national Waters." Public Health and Marine Hospital Service, Hygienic 
Lab. Bull. 83, March, 1912. TRANSMISSION OF DISEASES 
29 
Dr. J. F. Anderson1 concludes from his study of typhoid fever 
epidemics due to contaminated water, that they are characterized 
by: 
(a) A general distribution of cases throughout the area 
supplied by a particular water. 
(6) By the explosive onset of the outbreaks. 
(c) By the trouble occurring in the late winter or spring. 
(d) By the comparative freedom from the disease of persons 
not using the suspected water. 
(e) By evidences of sources of infection found by an inspection 
of the watershed. 
(/) By the outbreak beginning or ending after a change in the 
water supply. 
(g) And by indications of the pollution of the water when 
analyzed. 
TYPHOID FEVER 
DEATH RATE 
PER 
100,000 OF POPULATION 
Fig. 1.-Typhoid fever death rate in various countries. 
As an illustration of one or more of the above features of 
typhoid fever epidemics, the one which came under the author's 
personal observation, may be cited as quite characteristic. A 
town located on the Ohio River was supplied, together with 
two other communities, with water from a tributary of the Ohio 
River. This tributary flows into the Ohio River a short distance 
above the town. The water supply was treated with sulphate 
of iron and lime, and settled in a series of basins, from which 
the water flowed by gravity to the service pipes of the three 
communities. On account of a fire the pumping station was par- 
tially disabled, and the town located on the Ohio River was cut 
off from this supply for a period of about 5 weeks; but the two 
other communities continued to receive water from this same 
1 "A Symposium on Typhoid Fever." Amer. Jour. Public Hygiene, 
May, 1909. 30 
WATER PURIFICATION 
source. In order to obtain a supply of water for this town, a 
pumping station which had been out of service for some time, 
and which drew its supply of water from the Ohio River, was 
started, and pumped Ohio River water into the service pipes for 
35 days, covering parts of the months of October and November. 
Just above the intake of this pumping station is a small creek, 
which drains a ravine. Along this ravine and just above the 
pumping station were located quite a number of houses, about 
which the sanitary conditions were bad. While this pumping 
station was in service, a small artificial pond, located near the 
head of this ravine and connected with a summer amusement 
resort, was emptied. This water was discharged into the Ohio 
River just above the pumping station intake. 
An epidemic of typhoid fever began in the town supplied with 
Ohio River water about the middle of November. The author 
was not called in until about one month later, and at that time 
135 cases had been reported. 
An examination of the milk supply showed insanitary condi- 
tions about many of the dairies, but cases of typhoid were not 
found to be confined to any particular milk route, but were 
general all over the town. The milk supply was evidently 
not the source of the infection. The town had many wells, but 
the distribution of the cases was too uniform to attribute the 
infection to any particular locality in the town. 
At the time the author was making the investigation, the Ohio 
River supply of water had been stopped, and a return to the 
supply formerly used had been made. It was not, therefore, 
possible to obtain much chemical or bacteriological evidence of 
the character of the water which had been taken from the Ohio 
River. It was ascertained, however, that no epidemics of 
typhoid had occurred in the two other communities which had 
continued to receive their regular supply of partially purified 
water. Neither was there known to have been any more than 
the usual number of cases in a town on the opposite side of the 
Ohio River, and taking its supply from the latter stream. 
The logical cause for the epidemic seemed to be, therefore, the 
temporary pumping of a polluted water from the Ohio River, 
which rapidly infected many of the persons who drank it. More- 
over, on account of the explosive character of the outbreak, the 
pollution of the water was probably quite direct. The germs 
may have been washed into the Ohio River, and from thence TRANSMISSION OF DISEASES 
31 
passed into the intake of the pumping station at the time the 
pond in the summer resort was emptied, or following a flushing 
out of the creek by rains. The insanitary conditions along the 
creek, and the negative evidence obtained in investigating the 
milk supply, all led to the conclusion that it was a water-borne 
epidemic of typhoid fever that had occurred. 
A somewhat similar water-borne infection, which caused an 
epidemic of typhoid fever in Columbus, Ohio, in 1903-04, resulting 
in 1,606 cases and 162 deaths in 3 months, probably originated 
from the pollution of the public water supply drawn from the 
Scioto River.1 Cases of typhoid fever at the State Hospital 
were known to have existed 10 days prior to the outbreak in 
Columbus. The sewage from this institution entered the Scioto 
River not far from the Columbus water-works intake. The 
suddenness of the outbreak is shown by the following table: 
Typhoid Fever Epidemic, Columbus, Ohio 
Cases 
Deaths 
Rate per 100,- 
000 of 
population 
December, 1903 
40 
4 
34 
January, 1904 
725 
35 
300 
February, 1904 
798 
94 
805 
March, 1904 
83 
33 
283 
Some of the more important typhoid fever epidemics which 
have been traced to infected water are listed below: 
Place 
Year 
Population 
Number of 
Cases 
Deaths 
Caterham, England 
1879 
5,000 
352 
21 
Plymouth, Pa 
1885 
8^000 
1,104 
114 
Tees River Valley, England 
1890-91 
251,976 
1,330 
100 
Lowell, Mass 
1890-91 
77^696 
2,855 
217 
Lawrence, Mass 
1890-91 
44,654 
1,792 
137 
Worthing, England 
1893 
16^000 
1'411 
168 (Wells) 
Grand Forks, N. D 
1893-94 
6,000 
1,245 
96 
Maidstone, England 
1897 
33^830 
1,928 
150 (Springs) 
Ithaca, N. Y 
1903 
18,000 
1,350 
82 
Butler, Pa 
1903 
13'000 
1^348 
111 
1 Jour. Mass. Assoc. Boards of Health, vol, 
. 14, May, 1904. 32 
WATER PURIFICATION 
Comparatively recent outbreaks of water-borne typhoid fever, 
or a gradually increasing prevalence of this disease, which forced 
the authorities to provide remedial measures, have occurred at 
Erie, Pa., Niagara Falls, N. Y., Evanston, Ill., Coatesville, Pa., 
Ironton, Ohio, Winnipeg, Canada, Rockford, Ill., Memphis, 
Tenn., Council Bluffs, Iowa, and Omaha, Neb. 
The pollution of the Great Lakes in the United States by the 
cities built upon their shores is to a large extent local; but since 
these communities draw their water supply from the same source, 
the problem of preventing the drinking water from becoming 
contaminated, and still obtain a satisfactory disposal of the 
sewage, is a troublesome one. The common remedy of extend- 
ing the water supply intakes out from 3 to 5 miles from the shore 
has been resorted to with fair success, and with marked decreases 
in the typhoid death rate in most cases. The City of Chicago 
has diverted a large part of its sewage through a drainage canal 
into the Illinois River, thus keeping a constantly increasing 
volume of sewage from polluting the lake water farther and 
farther from the shore line. Other lake cities are contemplating 
partial purification of their sewage in order to conserve the purity 
of their water supplies. 
The effect of the wind on these large bodies of water in causing 
currents, the influence which the shore lines may have on these 
currents, and the movement of ice polluted with sewage from the 
shore out into the lake in the spring months are all factors which 
may at times be the cause for water-borne epidemics.1 
Even ground-water supplies, which through carelessness or 
ignorance are not properly protected from pollution after being 
drawn to the surface, are not infrequently the distributors of 
infectious material. A rather remarkable case was recently 
described as occurring at Lincoln, Neb.,2 in which a water from a 
well 60 ft. in depth became contaminated by leakage of sewage 
from a broken sewer. The escaping sewage found its way into 
the ground and into an abandoned pipe which had been forgotten 
and which connected directly with the well. 
An outbreak of typhoid fever occurred during September, 
October and November of 1911, from which six deaths resulted. 
1 D. D. Jackson: "Chlorination at Cleveland, O." Eng. Record, vol. 
65, June 15, 1912. 
2 "A Polluted Well at Lincoln, Neb." Eng. Record, vol. 65, June 15, 1912, 
p. 614. TRANSMISSION OF DISEASES 
33 
About the middle of December a severe outbreak of bowel 
trouble, during which there were several thousand cases, took 
place. This was followed about Dec. 20 by a second epidemic 
of typhoid fever, during which 300 cases were reported. 
References 
1. "Report of the Sanitary Investigations of the Illinois River and Its 
Tributaries." Illinois State Board of Health, 1900, p. 85. 
2. Vitality and Isolation of Cholera and Typhoid Organisms: 
(a) Jour. Infect. Diseases, 1904, 1, p. 641. 
(6) Eng. Record, vol. 52, Sept. 23, 1905. 
(c) Eng. Record, vol. 65, June 1, 1912. 
(d) Zeit. fur Hygiene und Infect. Krank., 14. 
(e) E. O. Jordan: "General Bacteriology." 
3. D. D. Jackson and T. W. Melia: "Differentiation Methods for 
Detecting the Typhoid Bacillus in Infected Water and Milk." 
Jour. Infect. Diseases, vol. 6, No. 2, Apr. 1, 1909. 
4. "A Symposium on Typhoid Fever" (8 papers). Am. Jour. Public 
Hygiene, May, 1909. 
5. "Epidemic of Typhoid Fever at Columbus, Ohio." Jour. Mass. 
Assoc. Boards of Health, vol. 14, May, 1904. 
6. Allan J. McLaughlin: "Sewage Pollution of Interstate and Inter- 
national Waters." Public Health and Marine Hospital Service. Hy- 
gienic Lab. Bull. 83, March, 1912. 
7. D. D. Jackson: "Chlorination at Cleveland, O." Eng. Record, vol. 
65, June, 15, 1912. 
8. "A Polluted Well at Lincoln, Neb." Eng. Record, vol. 65, June 1, 1912. 
9. Typhoid Fever Epidemics and Statistics: 
(a) Eng. Record, vol. 55, February, 1907, p. 131. 
(6) Eng. Record, vol. 58, October, 1908, p. 444. 
(c) Eng. Record, vol. 61, March, 1910, p. 263, 677. 
(d) Eng. Record, vol. 62, December, 1910, p. 630. 
(e) Eng. Record, vol. 63, June, 1911. 
(/) Eng. Record, vol. 65, pp. 254, 300, 591 and 601. 
(^) Eng. Record, vol. 66, July, 1912, p. 95. 
(A) Eng. News, vol. 67, June 13, 1912. 
10. Charles B. Boldrean: "Typhoid Fever in New York City, etc." 
Am. Jour. Public Health, vol. 2, 1912. 
11. H. de B. Parsons: "Our Typhoid Streams." Stevens Inst., Janu- 
ary, 1911. 
12. Nicholas S. Hill, Jr. and Leon R. Whitcomb: "The Relation of a 
Pure Water Supply to Chronic Intestinal Tract Infection." Eng. 
News, vol. 69, No. 5, Jan. 30, 1913. 
13. George A. Johnson: "The Typhoid Toll." Jour. Am. Water-works 
Assn., June, 1916. CHAPTER IV 
THE EFFECT OF IMPROVED WATER SUPPLIES UPON 
HEALTH 
As the relation between impure water and disease becomes 
better understood and appreciated, more attention is being given 
to the quality of public water supplies. Sometimes the method 
pursued is to seek a purer water from some uncontaminated 
source, or to render a polluted supply better by some process of 
purification. By substituting a pure drinking water supply for 
an impure one, the reduction in the death rate from water-borne 
diseases has been notable, and the improved health of the com- 
munity has usually been demonstrated beyond question. 
In the United States the typhoid fever death rates undoubtedly 
furnish the best indicators of the quality of public water supplies. 
The following table taken from Dr. Geo. M. Kober's paper on 
the "Conservation of Life and Health by Improved Water Sup- 
ply" summarizes the statistics of 61 cities in the United States for 
the years 1902-06.1 
Mean Typhoid Fever Death Rate from 1902-06 per 100,000 of 
Population for Cities Using Various Classes of Water 
4 cities using ground water from large wells 18.1 
18 cities using impounded water and conserved rivers 
or streams 18.5 
8 cities using water from small lakes 19.3 
7 cities using water from the Great Lakes 32.8 
5 cities using both surface and underground water.. . 45.7 
19 cities using polluted river water 61.1 
From the same paper is reproduced a diagram (Fig. 2) showing 
in more detail the typhoid fever death rates in different cities 
according to the character and the source of their water supply. 
Spring waters, ground waters from wells, and filtered waters 
evidently furnish the safest supplies. Surface waters, whether 
from streams or lakes, may furnish safe drinking water, but they 
are much more likely to be polluted. 
Probably the most striking effects in reducing typhoid fever 
have come from the purification of polluted supplies. By filter- 
xEng. Record, vol. 57, June, 1908. 
34 IMPROVED WATER SUPPLIES 
35 
ing an impure water supply marked reductions in water-borne 
diseases have almost invariably resulted. Even where the 
' Munich 
L Vienna 
Berlin 
Zurich 
Hamburg 
Paris 
London 
Paterson 
Binghamton 
Albany 
Lawrence 
Watertown 
Richmond Borough 
Queens Borough 
Camden 
Lowell 
Fitchburg 
Cambridge 
Somerville 
Worcester 
Bridgeport 
Hartford 
Malden 
Boston 
Chelsea 
New Bedford 
.Waterbury 
Holyoke 
Bronx Borough 
M anhattan Borough 
Pawtucket 
Newark 
Jersey City 
Baltimore 
Rochester 
Syracuse 
Fall River 
Brockton 
Taunton 
Haverhill 
Portland 
Salem 
Milwaukee 
Detroit 
Chicago 
Buffalo 
Erie 
Cleveland 
Duluth 
St.Paul 
Canton 
Brooklyn Borough 
Columbus 
McKeesport 
Minneapolis 
Seattle 
New Orleans 
Toledo 
Evansville 
Springfield 
Covington 
Grand Rapids 
Wilmington 
Richmond 
Cincinnati 
Louisville 
Philadelphia 
Lancester 
Atlanta 
Harrisburg 
Wheeling 
Allegheny 
Pittsburgh 
Fig. 2.-Typhoid fever death rate according to water supply. 
community has been only partially supplied with purified water, 
the effect on the typhoid death rate has been noticeable. 36 
WATER PURIFICATION 
Typhoid Death Rates per 100,000 of Population for Cities Changing 
from Polluted to Purified Water Supplies 
1907 
1908 
1909 
1910 
Columbus, Ohio 
38 3 
110.5 
20.0 
18.1 
New Orleans, La 
55.5 
33.1 
28.4 
31.5 
Louisville, Ky 
67.9 
44.2 
42.0 
31.7 
Pittsburgh, Pa 
130.8 
46 6 
24.0 
27.8 
Philadelphia, Pa 
60.7 
35.5 
22.3 
17.5 
The filter plant for the City of Columbus, Ohio, was started 
in August, 1908; the high rate for this year was due to an epidemic 
in the early part of the year. The filter plants in New Orleans 
FROM HYGIENIC LABORATORY (U.S.) 
BULLETIN No. 83 
BY 
allan j. mclaughlin 
Fig. 3.-Typhoid death rate by years for city of Pittsburgh, Pa. 
and Louisville were started in 1909. In Pittsburgh and Philadel- 
phia filter plants were placed in operation in 1907 and 1908, 
respectively; but in neither city was the entire population sup- 
plied with the purified water. IMPROVED WATER SUPPLIES 
37 
Fig. 3 from Dr. Allan J. McLaughlin's paper on "Sewage 
Pollution of Interstate and International Waters," which was 
previously referred to, is of especial interest in this connection, in 
showing how pronounced a reduction in the death rate followed 
the introduction of even a limited volume of purer water. 
From the typhoid fever statistics of Cincinnati, Ohio, a most 
convincing argument for the purification of a polluted water 
supply can be presented. 
Unfiltered water from old works 
Filtered water from new works 
Year 
1904 
1905 
1906 
Total 
for 3 
years 
1908 
1909 
1910 
Total 
for 3 
years 
Cases.. 
1,646 
746 
1,940 
4,332 
235 
218 
183 
636 
Deaths 
270 
155 
239 
664 
67 
45 
21 
133 
Number of Cases and Deaths from Typhoid Fever 
The figures for the year 1907 are omitted because water from 
both the old and the new works was supplied to the city. 
If the above figures are expressed as cases and deaths per 
100,000 of population, a better comparison may be made with 
other statistics. 
For 3 years before introducing 
filtered water 
For 3 years after introducing 
filtered water 
Average 
1908 
1909 
1910 
Cases 
417 
67 
62 
50.0 
Deaths 
64 
19 
13 
5.7 
Percentage reduction from the average. 
Cases 
84 
85 
88.0 
Deaths 
70 
80 
91.0 
Number of Cases and Deaths per 100,000 of Population 
Tn a report on the purification of the Montreal water supply 
Messrs. Hering and Fuller present a table showing the effect of 
purification by filtration on the death rate from typhoid fever in 
a number of American citie.s 
The relation between impure water supplies and certain intes- 
tinal diseases other than typhoid fever, is more or less obscure. 38 
WATER PURIFICATION 
Death Rates from Typhoid Fever per 100,000 Population in American 
Cities Using Filtered Water 
City 
Year 
plant was 
com- 
Before 
filtra- 
tion 
After 
filtra- 
tion 
Before 
filtra- 
tion 
After 
filtra- 
tion 
pleted 
Years averaged 
Death rate 
Sand filters 
Albany, N. Y 
1899 
10 
9 
90 
22 
Lawrence, Mass 
1893 
7 
15 
114 
25 
Pittsburgh, Pa 
1907 
8 
1 
133 
471 
Binghamton, N. Y 
1907 
5 
5 
47 
15 
Cincinnati, Ohio 
1908 
4 
1 
50 
16 
Columbus, Ohio 
1908 
11 
1 
78 
20 
Paterson, N. J 
1902 
5 
7 
32 
10 
Watertown, N. Y 
1904 
5 
5 
100 
38 
York, Pa 
1899 
2 
8 
76 
22 
Hoboken, N. J 
1905 
7 
4 
19 
14 
Mechanical filters 
Infant mortality from diarrhea and enteritis is probably both 
directly and indirectly the result of drinking impure water, 
although other causes contribute more frequently to death. 
About 85 per cent, of the deaths listed under "diarrhea and 
enteritis" in the United States census mortality statistics occur 
in children under 2 years of age. Dr. A. J. McLaughlin2 in 
discussing this subject says: 
"Instead of one disease designated under different names we are 
probably considering several diseases with common factors of trans- 
mission. Whatever the real relation between typhoid fever and enteritis 
or diarrhea of children may be, one fact is clear, the same causes operate 
to cause excessive prevalence of both. It is probable that cases of 
typhoid in children under 2 years in many cities are often incorrectly 
diagnosed as enteritis. It must be remembered, however, that the 
causative agent of bacillary dysentery is transmitted in the same way 
and by the same media as that of typhoid. There are too many cases 
of fatal illness in children under 2 years classed as diarrhea and enteritis, 
and an exhaustive investigation should be made to establish the real 
1 Including Allegheny, supplied with unfiltered water. 
2 Public Health and Marine Hospital Service. Hygienic Lab., Bull. 
83, March, 1912. IMPROVED WATER SUPPLIES 
39 
cause of death in enteritis and diarrhea of children. Without such an 
investigation it is impossible to assign the real cause of the excessive 
child mortality from diarrhea and enteritis. In cities of less than 50,000 
population without slums and which are not 'mill' towns an enteritis 
rate in children under 2 years above 100 deaths per 100,000 indicates 
prevalence of an acute intestinal disease preventable by the same 
measures that prevent typhoid fever. It is probable that in such cities 
proper enforcement of prophylactic measures against typhoid fever 
would reduce the enteritis rate below 40 deaths per 100,000. Enforce- 
ment of prophylactic measures would include the installation of pure 
water supplies and proper sewerage systems, coupled with a vigorous 
campaign against the insanitary outdoor privy and the equally dangerous 
shallow well." 
Dr. McLaughlin gives the following table which emphasizes 
the complexity of the problem. 
City 
Typhoid 
death rate 
per 
100,000, 
average for 
10 years, 
1900-1909 
Character 
of water 
supply 
Death rate 
enteritis, 
average 
for 5 years, 
1904-1908 
Remarks 
Rochester, N. Y 
13.7 
Good 
89.5 
Syracuse, N. Y 
14.8 
Good 
105.5 
Sanitary conditions 
Albany, N. Y 
21.9 
Good 
80.0 
good. 
Binghamton, N. Y.. . . 
20.9 
Good 
104.7 
Utica, N. Y 
17.3 
Good 
133.7 
Mill and factory 
towns; bad sanitary 
conditions. 
Schenectady, N. Y... . 
Amsterdam, N. Y 
22.4 
18.6 
Good 
Good 
164.7 
150.7 
Yonkers, N. Y 
9.5 
Good 
207.7 
Cohoes, N. Y 
83.8 
Polluted 
170.9 
Sanitary conditions, 
Niagara Falls, N. Y... 
129.1 
Polluted 
173.2 
exclusive of water, 
Ogdensburg, N. Y 
148.5 
Polluted 
175.0 
good. Mill and fac- 
Buffalo, N. Y 
27.0 
Polluted 
151.6 
tory towns. 
The following mortality statistics taken from reports of the 
Health Department of Cincinnati, Ohio, point pretty plainly 
to the fact that a change from an impure water supply to a 
purified supply made a marked difference in the number of 
deaths from dysentery, diarrhea and enteritis in persons over 2 
years of age. 
The following figures show a reduction in deaths from dysen- 
tery for 3-year periods before and after the introduction of 
filtered water of 64.3 per cent, and for diarrhea and enteritis 
of 55.1 per cent. 40 
WATER PURIFICATION 
Unfiltered Water from Old 
Works. 
Filtered Water from 
New Works. 
Year 
1904 
1905 
1906 
Total 
for 3 
years 
1908 
1909 
1910 
Total 
for 3 
years 
Dysentery 
Diarrhea and enteritis over 2 
27 
21 
22 
70 
9 
11 
5 
25 
years of age 
152 
167 
174 
493 
90 
60 
71 
221 
N. B. The year 1907 is omitted because water was supplied to the 
city both from the old works and from the new works. 
THE MILLS-REINCKE PHENOMENON AND HAZEN'S THEOREM 
The available evidence clearly proves that cholera, typhoid 
fever, dysentery and gastro-intestinal troubles are commonly 
transmitted by infected drinking water, and that there is reason 
to believe that other diseases are also conveyed in the same 
manner. The effect of purifying a polluted water supply, in 
decreasing the deaths from diseases other than cholera and 
typhoid, has been studied recently by W. T. Sedgwick and J. 
Scott MacNutt. They point out that in 1893-94, Messrs. 
Hiram F. Mills, C. E., of Lawrence, Mass., and Dr. J. J. Reincke 
of Hamburg, Germany, respectively, noted independently a 
decline in the general death rate of each of these cities as a result 
of improving their water supplies. Prof. Sedgwick and his asso- 
ciate have collected numerous mortality statistics in a paper' 
on this subject, and have termed the coincidence between a 
lowered death rate and a purified water supply as the " Mills- 
Reincke Phenomenon." In 1904 Mr. Allen Hazen in a paper 
read before the International Engineering Congress held in St. 
Louis, gave a quantitative expression for the phenomenon by 
stating that: "Where one death from typhoid fever has been 
avoided by the use of better water a certain number of deaths, 
probably two or three, from other causes have been avoided." 
Sedgwick and MacNutt have called this expression "Hazen's 
theorem," and have concluded from the studies which they have 
made, that the statement appears sound and conservative, but 
1 W. T. Sedgwick and J. Scott MacNutt: "On the Mills-Reincke 
Phenomenon and Hazen's Theorem Concerning the Decrease in Mortality 
from Diseases Other than Typhoid Fever Following the Purification of 
Public Water Supplies." Jour. Infect. Diseases, vol. 7, 1910. IMPROVED WATER SUPPLIES 
41 
not necessarily precise. The ratios they worked out varied 
widely. For example, at Hamburg, Germany, for every death 
less from typhoid fever after filtration, there were 15.8 deaths 
less from other causes; at Lawrence, Mass., the ratio was 1 to 
4.4, at Lowell, Mass., 1 to 6.0, at Albany, N. Y., 1 to 4.1, and at 
Binghamton, N. Y., 1 to 1.5. 
Millions of People Supplied with Filtered Water 
and Typhoid Fever Death Rates per 100,000 Population 
LEGEND 
Filtered Water 
Population 
■ Typhoid Fever 
Death Rates 
Fig. 4.-Growth of water filtration and decrease in typhoid fever death 
rate in the registration cities of the United States. 
From the data which they compiled with respect to diseases 
of the respiratory organs, the evidence in many cases is striking 
and warrants a more extended study. Factors not specifically 
related to an improved water supply may have been potent in 
reducing the death rate in the cities studied. Greater knowledge 
of methods for the prevention of disease must have played its part WATER PURIFICATION 
42 
in diminishing mortality, and in extending the length of the 
average life. On the other hand, the real causes for many dis- 
eases are still unknown, and the probability that some of them 
can be transmitted through polluted drinking water is by no 
means unreasonable. 
The elimination of disease germs from a water supply would be 
a direct cause for a decrease in the death rate, while an indirect 
cause could come from an increased vital resistance produced by 
the use of a purer drinking water. Probably both of these factors 
are instrumental in bringing about a lowered death rate. 
The accompanying diagram (Fig. 4) from Mr. George A. 
Johnson's paper "The Typhoid Toll," in the Journal of the 
American Water-works Association for June, 1916, is a striking 
graphic presentation of the effect which water purification has had 
in reducing typhoid fever in the United States. 
References 
1. Geo. M. Kober: "Conservation of Life and Health by Improved 
Water Supply." Eng. Record, vol. 57, June, 1908. 
2. Arthur Lederer: "The Modern Sewage and Water Problem." 
Clinique, August, 1912. 
3. E. Bonjean: "Les eaux d'alimentation publique observations gen^rales 
sur leur role epid^miologique; leur choix; etat actuel de 1'epuration." 
Rev. Scient, 49, An. 5, pp. 138-146. 
4. W. T. Sedgwick and J. Scott MacNutt: "On the Mills-Reincke Phe- 
nomenon and Hazen's Theorem Concerning the Decrease in Mortality 
from Diseases Other than Typhoid Fever Following the Purification of 
Public Water Supplies." Jour. Infect. Diseases, vol. 7, 1910. 
5. Geo. A. Johnson: "The Typhoid Toll." Jour. Amer. W. W. Assoc., 
June, 1916. CHAPTER V 
OBJECTS AND METHODS OF WATER PURIFICATION 
The purpose of purifying a drinking water is obviously to make 
it more pleasant and more wholesome to drink. The removal of 
suspended sediment and of vegetable coloring matter, in order to 
convert a dirty water into a clear and colorless one, commends 
itself if for no other reason than that the water becomes more 
acceptable to the senses of taste, smell and sight. If in addition 
to improving its appearance, it is possible to reduce materially 
or eliminate entirely the bacteria, some of which may be capable 
of producing disease, a second important reason for the purifica- 
tion of a drinking water can be advanced. By rendering a water 
hygienically safe for drinking purposes, sickness and death, as 
well as economic losses of no small magnitude are avoided. It is 
not difficult to show that a pure and wholesome water supply is a 
sanitary investment which pays in dollars and cents, as well as in 
health and happiness. 
For industrial purposes many waters are entirely unfitted for 
use without purification. Usually the dissolved salts which pro- 
duce hardness make a water unsuitable for use in steam boilers. 
The softening of hard waters is a method of purification which 
has long been practised, and which could be even more exten- 
sively employed today to the financial advantage of users of 
boiler scale-forming waters. In general, soft waters are more 
desirable for most industrial purposes. In such industries as 
brewing, distilling, starch and sugar manufacturing, the bacterial 
content of the water is of some importance, since certain organ- 
isms may produce fermentations which are undesirable. Waters 
containing iron in solution or suspension are especially objection- 
able for domestic purposes and for such industries as dyeing, 
paper making and bleaching. 
General Methods of Purification.-The impurities in water 
which it is necessary to remove are in suspension and solution, and 
purification methods fall naturally into two classes in consequence. 
For the removal of matter in suspension two processes are em- 
ployed, namely, sedimentation and filtration. These processes 
43 44 
WATER PURIFICATION 
occur normally in nature, but being uncontrolled produce re- 
sults of varying degrees of excellence. When artificially controlled, 
sedimentation and filtration, properly carried out, are able to 
purify waters of very poor quality, and with a degree of certainty 
that have established these methods upon a scientific basis. 
Mineral matter in suspension can be entirely removed, as well as 
practically all of the organic matter, which latter includes the 
bacteria. Chemical reagents may or may not be employed to 
facilitate either or both of these processes. The chemical com- 
pounds used in this connection are mechanical agents exclusively, 
and merely assist and hasten the bringing together of finely divided 
suspended particles into larger masses, which will settle and filter 
out with greater ease. 
For the removal of dissolved impurities, such as salts of lime 
and magnesia, chemical reagents are usually resorted to. A pre- 
cipitation of the bases in an insoluble form, and their subsequent 
removal by sedimentation and filtration, constitute the usual 
methods of purification. In the case of easily oxidizable soluble 
salts, such as those of iron, aeration is effective in breaking down 
the ferrous compounds by the removal of dissolved carbon dioxide, 
and the introduction of sufficient oxygen to bring about the oxida- 
tion of the ferrous to the insoluble ferric compounds. 
Filtration usually follows the process of sedimentation, and is 
carried out in various ways. Sand, gravel, coke, charcoal, coal, 
porous tile and sponges are some of the materials which have been 
and are still used as filtering media. Only the first two media, 
viz., sand and gravel, are commonly employed in large filtration 
plants. 
Purification methods which do not fall under any of the above- 
mentioned classes, are those of distillation and sterilization by 
chemical or physical agents. In the case of distillation all dis- 
solved and suspended impurities are removed, as well as the 
bacteria being killed. It is obviously the most perfect method of 
purification, but not necessarily the most desirable. Its cost 
makes it prohibitive for all practical purposes so far as large 
supplies are concerned. 
Disinfection and sterilization present the most advanced 
phases of water purification, and furnish methods of great prac- 
tical value for supplementing other methods, as well as in some 
cases, furnishing themselves all the purification needed. 
The action of these agents is primarily upon the readily oxidiz- WATER PURIFICATION 
45 
able matter which may be in a water, whether it be of organic or 
of mineral origin. The bacteria being a low form of plant life are 
thus attacked, and are either destroyed or their vitality so im- 
paired that they no longer are able to reproduce themselves. 
The effect of these agents is not selective, except in so far as they 
react first with the more easily oxidized material. If not ex- 
hausted on the latter, they will react slowly upon the more re- 
sistant matter, but may fail to effect a complete oxidation or to 
hinder vital processes. Those bacteria which are capable of form- 
ing spores, and which may be in this stage, are sometimes not 
affected, on account of their well-known resistance to injurious con- 
ditions. The effect of these disinfectants upon the pathogenic 
forms is the same as upon the non-disease-producing bacteria 
which may be present in a water. If practically all the bacteria 
are destroyed, therefore, the water is rendered hygienically safe. 
The agents used for disinfecting water are the hypochlorites of 
calcium and sodium, pure chlorine gas, ozone, copper sulphate, 
and the rays of ultra-violet light. Other chemical compounds 
have been suggested and experimented with, and the use of the 
electric current has frequently been tried. None of these agents 
is of much practical value excepting chlorine and its compounds 
in the form of the hypochlorites, ozone gas and the ultra-violet 
light. All of the chemical disinfectants are used in very minute 
amounts in water-purification work; nevertheless, they are ex- 
tremely effective when properly applied. The relative merits of 
these various agents will be discussed in a later chapter. 
The use of copper sulphate as an agent for killing off growths 
of algae and diatoms in water supplies is of practical importance, 
and of much value. The troublesome character of these growths 
in the operation of water-works, and even of purification plants 
will be considered in detail later. 
Statistics of Purification Plants.-Fifty years ago in the United 
States no plants for the purification of a public water supply were 
in existence. In Europe plants which clarified the water were 
in use, and incidentally effected a certain degree of bacterial puri- 
fication. They were operated upon an empirical basis and with 
no true understanding of the scientific principles underlying the 
art. In fact, the modern theory of the cause for many diseases 
was unknown, and the part played by minute vegetable and 
animal organisms in the dissemination of disease was undreamed 
of. By the wonderful investigations and deductions of Pasteur 46 
WATER PURIFICATION 
and Koch, the relation between polluted waters and the trans- 
mission of certain diseases became apparent and furnished 
a rational basis upon which methods of purification could be 
constructed. 
According to statistics first compiled by Mr. Allen Hazen for 
the International Engineering Congress held in St. Louis in 1905, 
and brought up to 1910 by Mr. Geo. C. Whipple in a paper read 
before the Congress of Technology held in Boston in 1911, over 
10,000,000 people in the United States are now supplied with 
water purified by filtration. Mr. Whipple presents some tables 
which clearly indicate how rapid the growth in the construction 
of purification plants has been. 
Populations Supplied with Filtered Water at Different Dates 
Year 
Total urban popu- 
lation in U. S., 
places of more 
than 2,500 inhabi- 
tants 
Population supplied with filtered water 
Per cent, of 
urban popu- 
lation sup- 
plied 
Sand filters 
Mechanical 
filters 
Total 
1870 
None 
None 
None 
0.00 
1880 
13,300,000 
30,000 
30,000 
0.23 
1890 
21,400,000 
35,000 
275,000 
310,000 
1.45 
1900 
29,500,000 
360,000 
1,500,000 
1,860,000 
6.30 
1910 
38,350,000 
3,883,221 
6,922,361 
10,805,582 
28.20 
In 1870 water supplies in the United States which were purified 
by filtration were practically unknown, while today probably 
30 per cent, at least of communities of more than 2,500 inhabi- 
tants are drinking filtered water. Mr. Whipple estimates that: 
"If the cities of more than 25,000 inhabitants are considered alone, 
it will be found that our 228 cities have a total population of 28,508,000. 
Of these, about 8,098,000 are supplied with water that does not need 
filtration, or at least will not for a long time. This leaves about 
20,311,000 people that are using water from sources subject to contami- 
nation. Of these 8,402,000 or 42 per cent, are adequately protected by 
the filtration of the water. Filters are under construction or have been 
authorized for 648,000 more, thus raising the percentage to 45 per cent. 
Filters have been officially recommended for 3,541,000; 7,720,000 people 
are still using water of questionable quality, although in some of these 
cases filtration has been seriously considered by sanitarians." 
A summary of statistics of population supplied with filtered 
water in the United States is given by Mr. Whipple in another 
table, which is well worth quoting. water purification 
47 
Cities with populations 
Num- 
ber of 
cities 
Total 
population, 
census, 
1910 
Filters i 
ti service 
Filters under con- 
struction and 
authorized 
Filters 
recommended 
officially 
Filters 
needed now 
or probably 
will be 
Filters prob- 
ably not 
Filters not 
needed with 
present water 
supply 
Sand 
Mechan- 
ical 
Sand 
Mechanical 
long time 
Over 1,000,000 
3 
8,501,174 
8,982,345 
1,601,008 
1,452,331 
125,253 
398,103 
176,911 
3,753,606 
129,615 
3,883,221 
104,000 
1,654,495 
893,036 
796,264 
1,300,411 
4,748,206 
2,174,155 
5,922,361 
2,850,600 
2,185,283 
2,971,817 
600,719 
952,484 
1,009,374 
7,719,677 
1,760,283 
1,680,611 
850,550 
1,041,465 
629,588 
5,962,497 
200,000 to 1,000,000 
25 
859,893 
363,198 
237,399 
753,938 
780,768 
2,135,303 
100,000 to 200,000 
50,000 to 100,000 
22 
59 
2,819,528 
4,178,915 
4,026,045 
28,508,007 
9,841,993 
38,350,000 
89,336 
89,336 
112,571 
147,325 
128,993 
3,540,897 
25,000 to 50,000 
Total for cities over 25,000... 
2,500 to 25,000 estimated1.... 
Total for places over 2.5001. . 
119 
228 
... 
558,485 
Summary of Statistics of Population Supplied by Filtered Water in the United States 
1 Census returns not available when table was compiled. 48 
WATER PURIFICATION 
A recent compilation of the rapid sand or mechanical filter 
plants now in operation in America, together with their daily 
capacities is shown in the following table: 
Having daily capacities of 
10 M.G. 
or over 
3 M.G. to 
10 M.G. 
Less than 
3 M.G. 
Total 
Number of plants 
Total daily capacity as grouped.. . . 
29 
892 
112 365 
Millions of gallons 
520 332 
506 
1,844 
Rapid Sand or Mechanical Filter Plants 
Mr. George A. Johnson, in his recent paper on the "Typhoid 
Toll,"1 estimates that the total population of the cities of the 
United States making returns of vital statistics is 34,230,000. 
Filtered water was supplied to 48 per cent, of the population 
of these cities in 1913, or in other words to a population of 
16,500,000 persons. This latter figure represents 17 per cent, of 
the total estimated population of the United States. 
The purification of water supplies by disinfection with calcium 
hypochlorite, either continuously or intermittently, has gone for- 
ward by leaps and bounds within the past 5 years. Today it is 
estimated "that 300 to 350 cities in the United States alone use 
this process."2 The use of disinfecting agents in water supplies is 
also increasing in Europe. 
The investment represented by purification plants mounts into 
many millions of dollars, but the conservation of life effected by 
them, measured in dollars, is very many times more than their 
first cost. Efficiency of these plants is of the utmost importance 
both from the hygienic and economic standpoint, and is becoming 
more and more appreciated by communities owning them. To 
the technical problem of purification much study has been given 
in the past, and is still being given at present; but the art as a 
whole is on a thoroughly scientific and practical basis, as the 
beneficial results of water purification amply testify. 
1 Jour. Am. Water-works Assn., June, 1916. 
2 C. A. Jennings: "Hypochlorite Sterilization of Water Supplies." 
Eng. Record, vol. 66, Sept. 14, 1912. WATER PURIFICATION 
49 
References 
1. Allen Hazen: "Filtration of Public Water Supplies." Appendix IV. 
2. "Water Purification in the United States." Eng. News, 47, p. 310. 
3. Rideal: "Water and Its Purification." 
4. G. C. Whipple: "The Present Status of Water Purification in the 
United States, etc." Proc. Congress of Technology, Boston, 1911. 
5. Engineering Record: 
(a) Geo. W. Fuller: "Importance of Proper Operation of Water 
and Sewage Purification Plants." Vol. 58, p. 498, 1908. 
(6) George C. Whipple: "Policy of Water Filtration." Vol. 60, p. 
718, 1909. 
(c) Rudolph Hering and George W. Fuller: "History of Devel- 
opment of Water Purification-Report on Montreal Supply." Vol. 
62, p. 539, 1910. CHAPTER VI 
SEDIMENTATION 
The removal of suspended mineral and organic matter from 
water supplies is most easily accomplished by settling in reser- 
voirs. The clarification and purification effected by settling in 
this manner is usually spoken of as plain sedimentation. By the 
addition of certain chemical compounds to naturally turbid 
waters, an artificial flocculation or coagulation of the suspended 
particles is produced whereby they settle out more rapidly. 
This process is termed " sedimentation after coagulation" or 
"precipitation with chemicals." 
PLAIN SEDIMENTATION 
Turbid surface waters flowing into impounding reservoirs be- 
come partially clarified in passing slowly through or in standing 
undisturbed in them. This result is produced because the veloc- 
ity of flow of the water, which has maintained these minute par- 
ticles in suspension, is lessened or reduced to practically zero. 
The reduction in the velocity of flow permits the sediment to 
gravitate toward the bottom of the reservoir, and to be finally 
deposited thereon. Provided the wind does not set up surface 
currents, nor temperature changes produce vertical currents, the 
sooner will the deposition of the sediment be effected. For ob- 
vious reasons a perfectly quiescent condition of the water can not 
be produced, and complete clarification, therefore, is not usually 
practicable. 
The ponds and lakes are, as a rule, natural settling reservoirs, 
in so far as the turbid surface waters which may flow into them 
are concerned. The water in them is, therefore, usually quite 
clear. The factors tending to disturb the water and to retard 
settlement are of course active at times. 
Artificial settling reservoirs and coagulation basins form a 
most important part of purification plants treating clay-bearing 
waters. As a preliminary treatment for filtration they are 
absolutely essential for efficiently and economically handling 
these turbid waters. They may be also of much assistance in 
50 SEDIMENTATION 
51 
purifying waters which are only turbid during portions of the 
year. In water softening, settling basins are essential to obtain- 
ing a well-clarified effluent. 
Theory of Sedimentation.-The very practical importance of 
sedimentation in water purification has caused the conditions 
which influence the phenomena to be carefully studied. Much of 
the experimental work has of necessity been empirical in charac- 
ter, because of the number of factors which are involved. Some 
investigators, however, have approached the problem from a 
theoretical standpoint, and many of their deductions are valuable 
and suggestive. The early work of Brewer, Durham, Hunt, 
Barus and Seddon is of interest in connection with this general 
subject. The more recent paper of Mr. Allen Hazen,1 however, 
has a more practical bearing on sedimentation, although discuss- 
ing in a theoretical manner certain important factors. 
From the simple assumption, "that whenever a particle of sus- 
pended matter hits the bottom, it remains where it strikes, and 
is never carried forward on the bottom, or picked up again; 
second, that all the sediment in the water is of the same hydraulic 
value, that is to say that every particle settles through the water 
at the same rate as every other particle," Mr. Hazen proceeds to 
develop mathematically fifteen special propositions. He dis- 
cusses the velocity with which particles of sediment settle 
through still water, and the effect of temperature, flocculation 
and coagulation upon the rate of settlement. 
The minute size of the particles, their very slow subsiding 
velocities, and the mixing action produced by surface and vertical 
currents, all tend to maintain the finest particles in practically 
constant suspension. Mr. Hazen gives a table of the subsiding 
velocities of various-sized particles which is reproduced on the 
following page. 
Mr. Hazen's resume of the deductions and conclusions at the 
close of the paper is clearly and concisely stated as follows: 
Resume 
"The fundamental proposition, in clearing water by sedimentation, 
seems to be that every particle of sediment moves downward through 
the water at a velocity depending upon its size and weight and upon 
the viscosity of the water. Particles of sediment are generally so far 
apart that they do not influence each other; and, while there is no doubt 
1 Trans. Am. Soc. C. E., vol. 53, p. 45, 1904. 52 
WATER PURIFICATION 
Velocities at which Particles of Sediment Fall in Still Water 
Diameter of particles 
in millimeters 
Hydraulic value in 
millimeters per second 
10°C. = 50°Fahr. 
Remarks 
1.0000 
100.0 
Experiments by the writer. 
0.8000 
83.0 
Experiments by the writer. 
0.6000 
63.0 
Experiments by the writer. 
0.5000 
53.0 
Experiments by the writer. 
0.4000 
42.0 
Experiments by the writer. 
0.3000 
32.0 
Experiments by the writer. 
0.2000 
21.0 
Experiments by the writer. 
0.1500 
15.0 
Experiments by the writer. 
0.1000 
8.0 
Experiments by the writer. 
0.0800 
6.0 
Interpolated from connecting curve. 
0.0600 
3.8 
Interpolated from connecting curve. 
0.0500 
2.9 
Interpolated from connecting curve. 
0.0400 
2.1 
Interpolated from connecting curve. 
0.0300 
1.3 
Interpolated from connecting curve. 
0.0200 
0.62 
Wiley's formula. 
0.0150 
0.35 
Wiley's formula. 
0.0100 
0.154 
Wiley's formula. 
0.0080 
0.098 
Wiley's formula. 
0.0060 
0.055 
Wiley's formula. 
0.0050 
0.0385 
Wiley's formula. 
0.0040 
0.0247 
Wiley's formula. 
0.0030 
0.0138 
Wiley's formula. 
0.0020 
0.0062 
Wiley's formula. 
0.0015 
0.0035 
Wiley's formula. 
0.0010 
0.00154 
Wiley's formula. 
0.0001 
0.0000154 
Wiley's formula. 
Note.-These values are not given as being precise, but they are believed 
to be sufficiently accurate for the purpose of this discussion. 
that they do sometimes collect in groups and thus change the condi- 
tions, it seems to be generally true that each particle will settle as if no 
other particle were present. 
"If the water in a basin were absolutely quiet there would be a 
regular sequence of clearing beginning at the top. The coarsest particles 
would go down fastest, but at any given point there would be a gradual 
clearing, and this clearing would take place most rapidly at the top, 
and, after longer intervals, at lower points in the basin. 
"Seddon started out with this theory, but found it to be not in 
accordance with the facts. His observation showed that while the 
amount of sediment in the water in the top was a little less than in the 
water in the bottom, the distribution was nearly equal throughout 
the mass, a condition of affairs inconsistent with the theory. He SEDIMENTATION 
53 
accounted for this distribution of sediment by the constant mixing 
of the water from top to bottom, and to the sustaining power of vortex 
motions in the water. These motions he thought arose from the internal 
motion of the water at the time of entrance, and from wind, and from 
temperature changes. 
"The writer has taken Seddon's development of the case as his 
starting point, and has carried the discussion further. He believes that 
while the internal motions keep the water mixed, and with nearly the 
same density of sediment from top to bottom, the tendency of the 
particles of sediment to settle is nevertheless an unbalanced force always 
acting to take the particles to the bottom, and the number of particles 
that hit the bottom in a given time is proportional, first, to the velocity 
at which the individual particles settle, and second, to the density of 
sediment in the water immediately above the bottom. 
"With these fundamental relations in mind, it is easy to compute 
and to express by simple formulas the proportions of particles of sedi- 
ment of a given hydraulic value which will hit the bottom under given 
conditions and which, therefore, presumably will be removed. 
"The fundamental propositions may be very concisely expressed. 
They are: First, that the results obtained are dependent upon the 
area of bottom surface exposed to receive sediment, and that they 
are entirely independent of the depth of the basin; and second, that the 
best results are obtained when the basins are arranged so that the 
incoming water containing the maximum quantity of sediment is kept 
from mixing with water which is partially clarified. In other words, 
the best results are obtained where any given lot of water goes through 
the basin with the least mixing with the water which enters after it. 
This is practically accomplished by dividing the basins into consecutive 
apartments by baffles or otherwise. 
"Thus far, the discussion is easy and apparently certain. The next 
step is a more difficult one. It relates to bottom velocities, and has to 
do with the question whether these velocities are such as to allow the 
particles to remain on the bottom when they get there, or whether 
they will be taken up again and be kept in motion with the body of the 
water. This is a point upon which further experimental data are 
needed. The problem of securing such data seems to be difficult. 
The observations must be made at the bottom of a layer of liquid of 
considerable thickness, where the conditions of observation are not 
favorable. The observations, further, must be made on very low 
velocities and on particles so small as to be practically microscopic. 
"Whatever view may be taken of the second part of the problem, or 
whatever researches upon it may show, the arrangements of basins 
most favorable to taking particles to the bottom should stand." 54 
WATER PURIFICATION 
Coagulation of Sediment.-The collection of minute particles 
of suspended matter into larger aggregates is a process which 
appears to occur only to a limited extent in natural waters, al- 
though it may be artificially produced by the addition of certain 
chemical compounds. Since the deposition of these finely divided 
particles greatly facilitates sedimentation and subsequent filtra- 
tion, when the latter forms a part of the purification process, the 
phenomenon is of much practical interest and importance in 
water purification. 
Colloidal Character of Sediment.'-The analogy between the 
physical properties of liquid suspensions of certain chemical com- 
pounds, and those of turbid clay-bearing waters is so marked that 
some description of their properties will be of interest. In a paper1 
published some years ago by the author attention was called to 
Dr. A. A. Noyes'2 classification of colloidal mixtures, in which he 
defined non-viscous, non-gelatinizing, but readily coagulable 
mixtures as "colloidal suspensions." He regarded them as 
really suspensions of minute particles and not true solutions. He 
differentiates further by designating mixtures in which the par- 
ticles may be visible under the microscope as "microscopic sus- 
pensions," reserving the term "colloidal suspensions" for those 
containing particles beyond the limit of microscopic visibility. 
Physical and Electrical Properties.-Natural waters furnish 
many examples of colloidal suspensions. Most turbid waters in 
their natural condition may be considered as mixtures of "mi- 
croscopic and colloidal suspensions." A beam of light passed 
through Ohio River water, even after the water has stood for 
many weeks, is plainly visible, in the same manner as when a sun- 
beam passes through dusty air. This is a fairly good proof of the 
presence of minute particles in suspension, which reflect the light 
from their surfaces. Colloidal suspensions of gold and arsenious 
sulphide artificially prepared act in a similar manner to a ray of 
light. 
Another property of colloidal and microscopic suspensions, 
apparently depending on the presence of electric charges upon 
them, is seen in the migration of the colloidal particles by the 
passage of an electric current through the mixture. Thus the 
particles of a colloidal suspension of ferric hydroxide or aluminum 
1<'The Coagulation and Precipitation of Impurities in Water Purifica- 
tion." Eng. Record, vol. 51, May 13, 1905. 
2 Jour. Am. Chern. Soc., vol. 27, No. 2. SEDIMENTATION 
55 
hydroxide migrate with the positive current toward the cathode, 
while kaolin and other similar colloidal or microscopic particles 
migrate with the negative current toward the anode. The col- 
loidal suspension of clay particles characteristic of turbid Ohio 
River water act in a precisely similar manner to those of kaolin. 
Experiments by the author with slightly turbid Ohio River water 
produced an average rate of travel of the particles toward the 
anode of 0.29 cm. per hour for a potential gradient of 4 volts per 
centimeter. These results are of the same order of magnitude as 
were obtained by Whitney and Blake for the migration of par- 
ticles in a colloidal suspension of silicic acid.1 Silica doubtless 
formed the larger proportion of the particles in the sample of 
Ohio River water with which the above experiments were made. 
The clearing of the water at the cathode and the movement of the 
particles toward the anode indicated them to be negatively 
charged. 
Coagulation with Chemicals.-Turning now to another impor- 
tant property of colloidal suspensions in general, viz., their coagula- 
tion, we find that turbid natural waters possess similar properties 
to colloidal suspensions artificially prepared. Non-electrolytes 
do not have the power of coagulating colloidal suspensions; 
but all electrolytes will do so with a proper degree of concentration 
of the latter, and a sufficient length of time. Acids, bases and 
salts, therefore, will coagulate colloidal suspensions, all being 
more or less dissociated in aqueous solution and capable of con- 
veying an electric current. In a similar manner hydrochloric acid, 
caustic soda, caustic lime or an ordinary salt solution will each, if 
of the proper concentration, coagulate the colloidal clay of a natu- 
rally turbid water like that in the Ohio River. It should always 
be borne in mind in dealing with natural waters that we are work- 
ing with extremely dilute solutions of salts, even in those waters 
which we commonly speak of as high in soluble compounds. In 
other words the concentration of the electrolytes is very low. 
In a number of modern purification plants in which lime is used 
to soften the water, in addition to the employment of sulphate of 
iron or aluminum sulphate to effect clarification, the action of the 
caustic alkali is virtually that of a coagulant, especially if the lime 
is added in sufficient amounts to produce a slightly caustic con- 
dition. Some quite extensive experiments undertaken on a large 
scale some years ago, clearly proved that by rendering the Ohio 
1 Jour. Amer. Chern. Soc., October, 1904. 56 
WATER PURIFICATION 
River water caustically alkaline with lime, the suspended clay 
could be coagulated, and could be subsequently removed by 
rapid filtration through sand. The explanation of the phenom- 
enon seems to be that the introduction of the base calcium hy- 
drate in excess, furnished an electrolyte of sufficient concentration 
in the water to effect coagulation of the colloidal clay. 
The similarity between artificially prepared colloidal suspen- 
sions and the very small suspended particles characteristic of tur- 
bid waters as above described, seems to warrant the classification 
of the latter mixtures with the former. If we seek further analo- 
gies in the co-precipitation or absorption by colloids of other sub- 
stances in solution and suspension with them when these colloids 
are coagulated, the true character of turbid waters seems even 
more apparent. 
When a turbid water containing carbonates of lime and mag- 
nesia in solution is treated with sulphate of alumina, for example, 
a reaction results setting free aluminum hydroxide. The liber- 
ated aluminum hydroxide absorbs or mechanically traps the 
finely divided material in suspension, and in the course of time 
the finer particles come together to form larger particles which 
settle out readily. This process of coagulation of the colloidal 
clay suspension and co-precipitation of the larger suspended par- 
ticles, appear to be started by the coagulation of the aluminum 
hydroxide. It is evident, as Dr. Noyes points out in the paper 
referred to, "that the mechanism of this coagulation is not yet 
understood," although it appears, "that it is the ion with a charge 
opposite to that of the colloid particles that is mainly responsible 
for their coagulation." It is also probable that it is the electric 
charge upon the particle which tends to hold it in suspension. 
Time Required for Coagulation.-The time factor in the 
coagulation of turbid waters is an important one. The length of 
time required to bring about the proper degree of coagulation 
varies with different waters. It is apparently influenced by the 
character of the suspended colloids, by the kind and amount of 
salts in solution, by the quantity of the coagulating chemical 
applied, by the temperature of the water, and by the agitation to 
which the mixture is subjected. The presence of much organic 
matter may retard the formation of the floc. An ordinary clay 
suspension is usually quickly coagulated. Complete flocculation 
and partial clarification may take place under favorable condi- 
tions in as short a time as 10 or 15 min.; on the other hand, the SEDIMENTATION 
57 
author has seen a water which was not affected at the end of 4 
hrs. This water had only a slight turbidity. Two hours after 
applying the coagulating chemical to this water it could be passed 
through a sand filter without removing the suspended matter or 
all of the coagulant. The large amount of soluble organic 
matter present in the water evidently produced a colloidal sus- 
pension with the aluminum hydroxide, formed by the decomposi- 
tion of the aluminum sulphate which was added, and was only 
slowly precipitated. 
Under ordinary conditions a period of 3 to 5 hrs. is sufficient to 
properly coagulate and settle a turbid water. Of course a proper 
arrangement of basins, inlets, outlets and baffles is requisite for 
bringing about effective coagulation and clarification. The con- 
struction of typical coagulation and settling basins will be de- 
scribed in a succeeding chapter. 
Loss of Chemical by Adsorption.- Another phenomenon 
common to the coagulation of turbid waters is the adsorption of 
the applied chemical compound by the flocculent material pro- 
duced by the reaction and the associated suspended matter. The 
particular significance of this fact in water purification was pointed 
out by Mr. George W. Fuller in his report on the experiments 
on Ohio River water made at Louisville, Ky., some years ago. 
While adsorption or concentration of a liquid on the contact- 
surface of the particles is not alone peculiar to colloids, but is 
characteristic of all solid precipitates, it has a practical bearing 
on coagulation in that there is some loss of chemical on account 
of this property. The amount of chemical thus lost is probably 
proportional to the surface area of the particles of the precipitate, 
and is a function of the nature of the solid and dissolved bodies, 
and of the concentration of the latter. 
Natural Colloids.-The "schmutzdecke, " which forms on the 
top of slow sand filters, is doubtless a true colloid, and by its 
adsorptive power removes the organic and inorganic suspended 
matter. The difference between the appearance of the sand in 
the beds of slow sand filters receiving water carrying considerable 
dissolved and suspended organic matter, and those which receive 
water holding more or less clay in suspension, as pointed out by 
Mr. George W. Fuller in his discussion of Mr. Allen Hazen's 
paper " On Sedimentation, " previously quoted from, may possibly 
be explained by a difference in the character of the colloids pro- 
duced or carried by the two types of water. The colloid formed 58 
WATER PURIFICATION 
by the soluble and suspended organic matter in a water has prob- 
ably the characteristic of a true colloid, i.e., a viscous, gelatinizing 
compound not coagulated by salts, like gelatine for example. On 
the other hand clay-bearing waters produce or carry a colloid 
which is non-viscous and non-gelatinizing, but which^may be 
readily coagulated by compounds like aluminum sulphate or 
sulphate of iron. Bacterial activity very likely aids in the forma- 
tion of the natural scums found on slow sand filters. 
In the absence of any artificially applied coagulant like alumi- 
num hydrate the 11 granular appearance" of sand beds receiving 
clay-bearing waters indicates the formation of a colloid whose 
ability to prevent the passage of finely divided suspended matter 
is limited. The colloidal particles pass deeper into the sand than 
would a true gelatinous colloid, and their power of adsorption is 
soon exhausted. The "ripening" of the sand beds of both slow 
sand and rapid sand filters, is, therefore, probably a process of the 
formation and deposition of colloidal particles upon the sand 
grains, by which their adsorptive power is slowly increased. 
It is, therefore, in the character of the colloid formed that one 
must seek for the explanation of the phenomena connected with 
the coagulation and filtration of natural waters. It is evident 
that much more light is needed on a great many of these prob- 
lems, but that some of them are associated in some way with the 
colloidal state of the clay, of the organic matter and of the artifi- 
cially applied coagulants, is self-evident. While the above points 
may at first sight appear to be only of scientific interest, they 
are in reality intimately associated with the most efficient de- 
sign of purification plants. The size of coagulation basins, the 
mixing and the agitation of the raw water with the coagulant, the 
ability to vary the period of coagulation for any given water, the 
best chemical compounds to use for the proper coagulation of the 
suspended matter in the water, the advisability of providing 
for plain sedimentation before attempting coagulation and the 
size and character of the sand grains composing the filter bed are 
all factors entering into the practical design of water-purifica- 
tion plants. The poor results often obtained in the operation 
of plants is quite as frequently the result of improper design as 
of faulty methods of handling the plant. SEDIMENTATION 
59 
Theories of Sedimentation: 
1. Brewer: Proc. Nat. Acad. Sciences, November, 1883. 
2. Durham: Chem. News, 30, p. 57, 1874; ibid., 37, p. 47, 1878. 
3. Hunt: Proc. Boston Soc. Natural History, February, 1874. 
4. Barus: Bull. U. S. Geological Survey, No. 36, 1886. 
5. Seddon: Jour. Assoc. Eng. Soc., p. 477, 1889. 
6. Hazen: Trans. Amer. Soc. C. E., vol. 53, p. 45, 1904. 
7. Longley: Eng. Record, vol. 57, pp. 793, 813. 
Coagulation : 
1. Ellms: Eng. Record, vol. 51, p. 552, 1905. 
2. Eng. Record, vol. 54, pp. 439, 475, 1906. 
3. Eng. Record, vol. 58, p. 292, 1908. 
4. Eng. Record, vol. 61, pp. 176, 215, 247, 599, 1910. 
5. Eng. Record, vol. 64, p. 476, 1911. 
General Experimental Data: 
1. Fuller: "Report on Water Purification at Louisville, Ky." 
2. Fuller: "Report on Water Purification at Cincinnati, Ohio." 
3. Weston: "Report on Water Purification at New Orleans, La." 
4. "Report of Filtration Commission of Pittsburgh, Pa." 
5. George F. Catlett: "Colloidal Theories Applied to Colored Water." 
Eng. Record, June 3, 1916. 
6. Thorndyke Saville: "The Nature of Color in Water." Engineering 
and Contracting, January 16, 1917. 
References CHAPTER VII 
TYPES OF SETTLING RESERVOIRS AND COAGULATION 
BASINS 
The design of reservoirs and basins for efficient sedimentation 
has received considerable attention in water-purification work, 
but not as much as the subject deserves. The complexity of the 
problem has not always been appreciated, and too frequently this 
portion of the plant has been made to accommodate itself to other 
features of the design instead of being properly coordinated with 
them. 
A certain amount of flexibility in the design of these reservoirs 
is quite possible without materially sacrificing efficiency. Never- 
theless, it is the author's opinion that much more of the work of 
clarification should be thrown on this part of the purification 
plant than is now the practice. They are by far the least com- 
plicated portions of the works, and the least expensive to operate. 
Their tendency to make more uniform the operation of the filters 
strongly commends them to the operators of such plants. It is 
believed that in designing settling reservoirs of all kinds more con- 
sideration should be given to obtaining efficient sedimentation 
and to making them as large as good engineering practice will 
permit, due regard being had for other portions of the plant, in 
order to produce a well-balanced design. 
Impounding Reservoirs.-Impounding reservoirs built pri- 
marily for storing water are, of course, to be considered from a 
somewhat different standpoint than those especially designed 
for sedimentation purposes. The depositing of suspended sedi- 
ment in storage reservoirs is an incident rather than the object of 
their operation. However, considerable purification is effected 
in them, and the influences which lead to such results will be 
discussed in another section. 
Settling Reservoirs.-Storage of turbid waters for purposes of 
plain sedimentation, in reservoirs especially constructed for this 
purpose, is undoubtedly good practice, but is not always provided 
for. Usually such supplies are drawn from turbid streams in 
which the water may at 'times be loaded with sediment. Pro- 
vided the reservoirs are relatively large in proportion to the con- 
60 RESERVOIRS AND COAGULATION BASINS 
61 
sumption of water, quite a number of days of settlement may be 
possible before the water is drawn off. The relative positions of 
the inlet and outlet of these reservoirs is obviously of great im- 
portance in even approximating theoretical displacement of the 
water if they are used continuously. The tendency of the 
water currents to seek the shortest path between the inlet and 
outlet, and thereby render more or less ineffective certain parts 
of the reservoir, is a commonly observed condition. Baffling 
undoubtedly has the effect of breaking up "short-circuiting 
currents of water, " but is not always employed in plain sedi- 
mentation basins. 
Cleaning out the sediment deposited in this class of reservoirs 
is greatly facilitated by smooth and hard linings to the reservoirs, 
by sufficient slope to the sides and bottom and by an adequate 
system of gutters and drains. Proper facilities for flushing out 
the mud by streams of water under pressure, or by scraping the 
mud to the gutters assisted by a flow of water which is not under 
pressure, are necessary in all well-designed basins. 
In all settling basins the sand and heavier portions of the silt 
will be found deposited close to the inlet. In some cases the 
amount of sediment dropped near the inlet is very great, especially 
where the turbid stream supplying the water carries a good deal of 
sand and silt, and where the pumping from the stream is practi- 
cally continuous. 
Cincinnati Storage and Settling Reservoirs.-Combined stor- 
age and settling reservoirs are in operation at Cincinnati, Ohio, 
(Fig. 5) in which Ohio River water undergoes plain sedimentation 
before being coagulated with chemicals and filtered. These two 
reservoirs hold approximately 392,000,000 gal. of water, and 
were designed with the idea of their being operated on the fill-and- 
draw plan. This method of operation, however, has never been 
followed. The water flows continuously through them in parallel. 
The inlets to the reservoirs are between 500 and 600 ft. from the 
outlets. The latter consist of movable pipes (Figs. 6 and 7) 
with their mouths held about 4 ft. under the surface of the water 
by means of floats. The water is thus continuously skimmed 
from the surface. The depth of water in the reservoirs varies 
from 35 to 50 ft. 
The irregular shape of these reservoirs (Fig. 8) is accounted for 
by the desire in construction to make the excavations equal the 
embankments as nearly as possible. As it was intended that 62 
WATER PURIFICATION 
they should be operated by first filling with water, then allowing 
the latter to stand and deposit its sediment for a day or so, and 
finally drawing off the settled water, it was of no particular conse- 
quence if the inlet was close to the outlet. Since they are not 
Fig. 5.-Cincinnati settling reservoirs. 
Fig. 6.-Cincinnati settling reservoirs, effluent float tubes. 
utilized in this manner, but are operated continuously, their 
efficiency is undoubtedly somewhat reduced because of their 
irregular outline, and the short distance between the inlet and 
outlet. The apparently dead spaces in these reservoirs, however, RESERVOIRS AND COAGULATION BASINS 
63 
are by no means useless, since experience shows that diffusion of 
the sediment causes a much more uniform deposition of the latter 
than would be supposed. The depth of mud in the lobes of these 
COMPRESSED EARTH 
Fig. 7.-Cincinnati settling reservoirs, profile showing float tubes. 
Reservoir No. 2 
Reservoir No.l 
PLAN 
SECTION ON LINE A-B 
Fig. 8.-Cincinnati settling reservoirs, plan and cross section. 
reservoirs will probably run from 35 to 40 per cent, of the average 
depth in the portions of the reservoirs where more active sedimen- 
tation is in progress. 64 
WATER PURIFICATION 
Louisville Settling Reservoirs.-The two Crescent Hill reser- 
voirs at Louisville, Ky., perform a similar service to those at 
Cincinnati. They cover an area of 750,000 sq. ft., and hold, when 
filled to a depth of 20 ft., a little over 100,000,000 gal. of water. 
Originally the water as it was pumped from the Ohio River was 
delivered into the gate house at the end of the division wall sepa- 
rating these two basins. It passed over a weir above the 20-ft. 
level into the first basin, and was drawn off at a point diagonally 
opposite at the bottom through a conduit 5 ft. in diameter. 
Through this conduit the water passed to the corresponding cor- 
ner of the second basin, where it entered at the bottom. It was 
drawn off at the diagnonally opposite corner through the gate 
house. 
Delivering the muddy water at the surface of these reservoirs 
and drawing off at the bottom has been recently changed, so that 
the water now enters from the gate house at the bottom of the 
first basin, and is withdrawn over a weir tower built to enclose the 
inlet end of the conduit. After passsing through the latter to the 
bottom of the second basin, it is skimmed off at the top of this 
basin at the gate house from which it flows to the coagulation 
basins. 
Mr. George W. Fuller observed, when these reservoirs as origi- 
nally arranged, were drawn off to be cleaned, that with the ex- 
ception of the coarser material piled up near the inlet, the depth 
of sediment was substantially uniform over the entire bottom. 
Mr. Alien Hazen has noted the same condition in other reservoirs, 
and accounts for it by the mixing action by which water in all 
parts of the reservoir is made substantially of the same quality. 
New Orleans Grit Reservoirs.--For removing the heavy silt 
and sand of a normally turbid water, relatively small grit reser- 
voirs have been used. In the New Orleans purification plant two 
such reservoirs are in service. Each reservoir is 75 by 150 ft. in 
area, and has outer walls 20 ft. high. These reservoirs are pro- 
vided with a center baffle which causes the water to travel up one 
side of the reservoir and down the opposite side to the outlet. At 
the normal capacity of the plant a sedimentation period of about 
1 hr. is obtained with one basin in service. The heaviest sedi- 
ment settles nearest the inlet, and grows much lighter as the out- 
let is approached (Fig. 9). 
Albany Settling Reservoir.-The settling basin designed to be 
used in conjunction with the slow sand filtration plant at Albany, RESERVOIRS AND COAGULATION BASINS 
65 
Future Clear Water Reservoir 
Clear Water 
Reservoir No.2 
Clear Water 
Reservoir No.l 
Fig. 9.-New Orleans water purification plant, general plan. 
Future Reservoirs 
Future Reservoirs 
Reservoirs 
Future 66 
WATER PURIFICATION 
N. Y., was intended to be operated continuously. The basin has 
an area of 228,000 sq. ft. and a depth of 9 ft. It holds 14,600,000 
gal. of water. Provision was made to allow the water entering 
the reservoir to discharge through 11 inlet pipes, which rise 4 ft. 
above the water line. By discharging into the basin in this 
manner aeration of the water was produced. The water was with- 
drawn through 11 outlets. 
At the present time the water is not aerated as it enters the 
reservoir through two 18-in. perpendicular and one 36-in. hori- 
zontal inlets. At the present rate of consumption a 17-hr. period 
of subsidence is obtained. 
COAGULATION BASINS 
The bringing together of finely divided particles of suspended 
sediment in natural waters by means of the action of certain 
chemical compounds is termed coagulation. Basins, in which 
this action may take place and the flocculated sediment settle out, 
are commonly provided in connection with rapid sand filter 
plants. They may also form a part of slow sand filter plants 
where the sediment in the water settles out slowly and is liable to 
clog the sand in the filters. 
Mixing Channels.-The introduction and uniform distribution 
in the water to be treated of chemical solutions of the strengths 
usually employed in water purfication offers some mechanical 
difficulties. Since the efficiency of the coagulating compound 
depends upon the formation of a large and well-defined floc, the 
size and character of which is directly influenced by the tempera- 
ture of the water and the nature of the salts and colloids in solu- 
tion and suspension, a thorough mixing of the water with the 
solution is of the utmost importance. 
By sending the water through relatively narrow channels at a 
velocity of 1.5 to 2.0 ft. per second, and allowing the travel 
through these channels to take about 1 hr., a thorough mixing and 
a complete coagulation is effected, which produces rapid sedimen- 
tation in the coagulation basins into which the water discharges. 
The eddies formed in the channels by reversing the direction of 
flow or by baffles or by projecting portions of the concrete con- 
struction in the channels, assist materially in bringing about a 
thorough mixing. 
In New Orleans mixing channels are provided for applying a 
5 per cent, milk of lime in the manner described above. In this RESERVOIRS AND COAGULATION BASINS 
67 
plant duplicate reservoirs 75 by 320 ft. each with outer walls 19 
ft. high, are divided into 16 rectangular double-decked passages 
with a cross-sectional area of over 60 sq. ft. each, and an aggre- 
gate length of about 5,120 ft. When passing 40,000,000 gal. of 
water in 24 hrs., the flow of the water through the channels 
requires about 1 hr. 
Iron sulphate for coagulation purposes, and soda ash for re- 
ducing the permanent hardness of the water, may be both intro- 
duced in these channels, but provision is also made for applying 
a coagulant at various points in the flow of the water through the 
coagulation basins proper. 
Obviously, mixing channels may become settling chambers un- 
less the velocity of flow is sufficient to retain the flocculated sedi- 
ment in suspension. This does take place, but corrects itself by 
reducing the cross-section of the channel by deposition of sedi- 
ment until the velocity is sufficient to maintain a scouring action 
through the passage. On the other hand, too great a velocity 
may break up the floc and thereby diminish its ability to settle 
rapidly, as well as failing to entrap the very fine suspended sedi- 
ment for which the coagulating chemical was added. Provided 
a very thorough mixing action is obtained in the channels, it is 
evident that the coagulation basins themselves may be made 
smaller than they otherwise could be. 
A more recent design is found in the Grand Rapids, Mich., 
plant, where the water after passing through a grit chamber hold- 
ing about a 26.5-min. supply at a normal rating of 20,000,000 gal. 
per day, enters a mixing chamber 44 ft. wide by 160 ft. long. The 
chamber holds 732,000 gal., or about a 53-min. supply at normal 
rating. The basin is provided with wooden baffles of " around the 
end type, " spaced 3 ft. apart for the full length of the chamber. 
Mr. J. W. Armstrong who designed this plant, as well as the 
plant at New Orleans, states this type of baffle permits the opera- 
tion of the plant with varying heads of water, and offers reason- 
ably good facilities for cleaning and inspection. 
Settlement After Coagulation.-The deposition of coagulated 
sediment should take place rapidly if the proper amount of 
chemicals have been added to the water, a thorough mixing of the 
watev with the chemicals has been effected, and temperature con- 
ditions are favorable. The same general principles governing the 
deposition of sediment in plain sedimentation basins, as previously 
discussed, apply here, except that the suspended particles are 68 
WA TER P U RIF IC A TION 
larger and should gravitate toward the bottom more rapidly. 
A velocity of flow of 1.0 to 1.5 ft. per minute can probably be 
safely maintained in most cases without carrying too much of 
the finest material forward to the outlets of the basins. Baffleswill 
probably be found advantageous in most basins of this type. A 
uniform and even distribution of flow across the entire cross- 
section of the basin is desirable between the inlets and outlets. 
Skimming from the surface at the outlets is the proper method 
for the withdrawal of the settled water. Provision for secondary 
application of the coagulating chemicals as the water flows 
through the basins for the purpose of correcting the original 
treatment, should never be omitted in any well-designed plant. 
If possible, an auxiliary basin for getting the benefit of this 
secondary treatment ought to be installed, otherwise too much 
coagulated material is liable to be carried through the outlets to 
the filters. 
The primary coagulation basins are usually constructed in 
duplicate in order to be able to clean one of them at a time 
without interrupting the operation of the whole plant. An ade- 
quate number of sumps and drains ought to be provided, so 
that the deposit when ready to be cleaned out does not have to 
be moved too far. The size of the drains should be ample to 
prevent clogging, and plenty of water should be available for 
flushing purposes. 
St. Louis Settling Basins.-The six original settling basins of 
this water-works plant (Fig. 10) were constructed as plain sedi- 
mentation basins. For a number of years since the chemical 
coagulants sulphate of iron and caustic lime have been used to 
clarify the water, they have become coagulation basins. They 
originally consisted of six rectangular masonry tanks 670 by 400 
ft. in plan, and placed side by side. The walls are of masonry but 
the bottoms are of concrete. They were intended to be operated 
in parallel, but since coagulants have been used the walls dividing 
adjoining basins have been cut down, so that the water now 
passes through them in series, entering through four 3 by 3 ft. 
openings at the end of one of the basins. The water is withdrawn 
through a masonry conduit at the end of the sixth basin after 
having passed through the preceding five basins. Sluice gates in 
each basin, connected with both the inlet and outlet conduits, 
make it possible to withdraw basins from service for cleaning, 
and to provide for emergency conditions. RESERVOIRS AND COAGULATION BASINS 
69 
Two additional basins, covering together an area 400 ft. wide 
by 1,660 ft. long have been constructed of concrete. The two 
basins are separated by a division wall, and are each 400 by 800 
ft. in plan, and have an average depth of about 21 ft. Their 
combined capacity is 75,000,000 gal. This makes the total 
capacity of the eight basins 255,000,000 gal. These are the 
Fig. 10.-St. Louis settling basins. 
largest coagulation basins known to the author, and they handle 
effectively a very turbid water. The effluent from these basins 
is now filtered. 
Fig. 10a.-St. Louis filtration plant. 
Cincinnati Coagulation Basins.-A good example of coagulation 
basins which are effective, but which are unprovided with baffles, 
are the three basins of the Cincinnati plant (Figs. 11 and 12). A 
lined basin or reservoir 400 by 400 ft. in plan is divided by a con- 
crete wall into two basins, each of which may be operated independ- 
ently, but which are usually operated together in parallel. The 
depth of these basins is about 21 ft. The water enters the two 70 
WA TER P U RIF IC A TION 
basins through sluice gates at the bottom of a gate house at one 
end of the division wall. It flows into the bottom of the basins 
through two reinforced-concrete conduits on either side of the 
above-mentioned gate house. These conduits are 7 ft. in diame- 
Fig. 11.-Cincinnati coagulation basins 
Coagulation 
Basin, No.l 
Clear Water Reservoir 
Coagulation 
Basin No.2 
Coagulation 
Basin No.3 
Chemical 
House 
Filter House 
Head House I 
PLAN 
Slope 1^-1 
Slope 1^-1 
Fig. 12.-Cincinnati coagulation basins, plan and cross section. 
SECTION ON LINE A-B 
ter and have set in the top of each of them twenty-one 20-in. tees. 
Each of these tecs discharges through the two openings in line 
with each other, and in parallel with the line of flow through the 
body of the conduit. In this manner the velocity of the counter- RESERVOIRS AND COAGULATION BASINS 
71 
currents of water tend to oppose each other, and to produce a 
uniform and quiet inflow. The water then passes across to the 
opposite side of the basins, and is skimmed off at the surface by 
openings into a steel conduit which conveys the settled water to 
either a third basin for secondary treatment, or directly to the 
filters. Two mud valves hydraulically operated are placed in the 
bottom of each basin. The slope of the bottom toward the 
sumps of these valves is approximately 2 per cent. The sides 
have a slope of 1.75 to 1. These basins have a concrete lining 
covered with hard-burned brick which are grouted in place so as 
to form a smooth hard surface for flushing out the deposited mud. 
The third basin is 400 by 80 ft. and about 16 ft. deep. It is 
used in series with the two larger basins, and although operated 
Fig. 12a.-Cincinnati filtered water reservoir. 
continuously to obtain the maximum period of sedimentation, was 
originally intended for settling the water after a secondary appli- 
cation of chemicals, when the first application had been found 
insufficient. Whenever it is necessary to make a second applica- 
tion of chemicals, which is rarely the case, the third basin is used 
as originally intended. 
The two large basins have a capacity of 10,000,000 gal. each, and 
the small basin a capacity of 2,000,000 gal. At the maximum 
output of this plant a period of 4 to 5 hr. for settlement is possible 
in these basins. 
Grand Rapids Coagulation Basins (Fig. 13).-As an example 
of well-baffled coagulation basins those at Grand Rapids, Mich.,1 
may be cited. There are two covered coagulation basins in this 
plant. The smaller basin is 88 ft. 6 in. by 118 ft. 9 in., and holds 
1 "The Municipal Water Purification Plant at Grand Rapids, Mich." Eng. 
Record, vol. 64, p. 379, 1911. 72 
WATER PURIFICATION 
1,134,000 gal. of water, which is about 1 hr. and 22 min. supply 
at the normal rating of 20,000,000 gal. per day. The larger basin 
is 118 ft. 6 in. by 118 ft. 9 in., and holds 1,452,000 gal., or about 
1 hr. and 44 min. supply at the normal rating. The basins may 
be operated in series or in parallel. 
Mr. J. W. Armstrong in describing this plant notes that in 
basins having but few baffles, that there is a tendency for the 
WaterConduit 
Clear WaterBasm 
GritChamber 
New Curb 
J MixingChamber 
Center Passage' 
Coagulating 
g Basin 
Coagulating 
Basin 
Fig. 13.-Grand Rapids water purification plant, general plan. 
Plan of Works. 
water to short-circuit, and for the floc to settle out unevenly in 
different parts of the reservoir. In order to overcome this diffi- 
culty and to maintain a more even distribution of the floc, the 
baffles in these basins are placed closer than usual, being 15 ft. 
apart on centers. 
CLEANING SETTLING AND COAGULATION BASINS 
The method of removing the sediment deposited in settling and 
coagulation basins should always receive careful consideration in 
design. Where large amounts of sediment must be removed, RESERVOIRS AND COAGULATION BASINS 
73 
adequate means for handling the accumulated mud are necessary 
in order not to keep a basin out of service for too long a period. 
In small plants the time factor is not of so much importance, 
although even here great inconvenience may arise from limited 
facilities for handling the accumulated sediment. Basins ought 
always to be in duplicate, in order that one at a time may be with- 
drawn from service for cleaning. Where the sediment accumu- 
lates rapidly, frequent cleaning appears preferable to allowing a 
large amount to accumulate, which would require a considerable 
time to remove. 
Concrete or concrete- and brick-lined basins form the best sur- 
faces from which to wash off the mud. Drains should he of 
ample size, and there should be a good slope toward them. The 
general method of cleaning employed will depend on the slope of 
the sides and bottom toward the sumps, and on the distance the 
mud must be moved. Where the deposit is washed out by 
streams of water under pressure, the author has found that slopes 
of less than 1 in 20 greatly retarded the process of cleaning. One 
and 2 per cent, grades are not enough when cleaning by this 
method. On such grades the semi-fluid deposit has to be re- 
peatedly pushed forward toward the sumps, thereby losing much 
time and requiring large volumes of water. 
Where scraping is employed to assist the flushing with water, 
the latter need not be under much pressure, but a considerable 
volume of water is needed to give a semi-fluid consistency to the 
mud. At the St. Louis plant horse-drawn scrapers are employed 
made from 2-in plank 10 in. wide and 12 ft. long. Men follow 
the teams with hand scrapers, cleaning the bottom as they go and 
arranging gutters so as to confine the flushing water used to a 
narrow strip nearest the teams and to prevent its being wasted 
over the cleaned bottom. 
The St. Louis basins have a comparatively flat bottom with a 
gutter running through the center. This gutter runs lengthwise 
of the basin and has a 1 per cent, slope to the sewer into which it 
discharges. No trouble is experienced in moving the mud along 
the gutter. More difficulty is encountered with the heavy silt 
and sand which settles near the inlets, and which requires teaming 
and much water to move. 
It has been the experience of the author that sand and silt 
deposits are easily moved on 8 and 10 per cent, slopes, provided 
the deposit is first disintegrated by a stream of water under a 74 
WATER PURIFICATION 
pressure of 70 to 75 lb. per square inch. The mud banks in such 
cases need not be entirely liquefied; but if they are undermined 
by a stream of water (Figs. 14 and 15) from a nozzle 1 in. or even 
Fig. 14.-Cleaning Cincinnati settling reservoirs. 
Fig. 15.-Cleaning Cincinnati settling reservoirs. 
% in. in diameter and under the above-mentioned pressure, they 
can be readily floated away to the sump. On the other hand, 
the gelatinous clay deposits usually have to be well liquefied be- 
fore they can be moved, although on slopes of 10 or 12 per cent. RESERVOIRS AND COAGULATION BASINS 
75 
the mud can be floated to the sump in larger pieces than when the 
grades are less. These observations apply particularly to the 
kind of silt and clay carried by the water of the Ohio River, and 
to handling large amounts of sediment in large settling reservoirs. 
Nevertheless, they are believed to be generally true, although 
the original character of the sediment must necessarily influence 
the general properties of the deposit. In coagulation basins the 
applied chemicals or the products of the reactions, such as hy- 
drates of iron, aluminum, magnesium, and carbonate of lime, 
must of necessity modify the nature of the deposit. 
Cost of Cleaning Settling and Coagulation Basins.-The cost 
of cleaning basins will necessarily vary with the kind and amount 
of deposit to be removed. This has been found to be the case 
both at St. Louis and at Cincinnati. At St. Louis the cost seems 
more dependent on the amount of sand in the deposit than in 
the amount to be moved. The following table compiled from 
the records of the Cincinnati Water Department shows the cost 
of cleaning one of its large plain sedimentation reservoirs. 
Total cost 
Estimated cost 
per cubic yard of 
sediment 
removed 
T , 
Labor 
81,100.45 
$0.0267 
Supplies 
440.23 
0.0107 
Power 
154.44 
0.0037 
Value of water used in cleaning 
62.82 
0.0015 
Value of water wasted in draining 
183.13 
0.0044 
Total 
$1,941.07 
$0.0470 
N.B. The cost of cleaning per million gallons of water settled was 80.056- 
The cost of cleaning settling basins at St. Louis, Mo., in 
which coagulated sediment is precipitated is shown by the fol- 
lowing table compiled from the annual reports of the Water 
Commissioner. 
Cost of Cleaning Chain of Rocks' Basins 
1908 
1909 
1910 
1911 
Cubic yards sediment 
removed 
Total cost 
Cost per cubic yard.. . 
135,108 
$2,631.85 
$0.0195 
129,035 
$2,947.26 
$0.0228 
182,500 
$3,159.11 
$0.0173 
144,200 
$2,158.39 
$0.0150 76 
WA TEH PUHIFICA TION 
The costs given for cleaning the Chain of Rocks' basins are 
for labor and teams, for furnishing and keeping in repair all tools, 
boots, etc., used in cleaning, and for moving such apparatus to 
and from the basins before and after cleaning. It does not in- 
clude the cost of the water lost by draining or used in cleaning 
the basins. 
In April, 1908, 700 cu. yd. of mud were removed from the 
Baden storage reservoir at St. Louis, at a cost of $149, or $0,213 
per cubic yard. The basins at Bissell's Point were cleaned in 
August of the same year at a cost of $0,133. 
Cost of Cleaning Cincinnati Coagulation Basins in 1911 
Labor $89.67 
Power 24.75 
Value of water lost in draining and used in cleaning. 76.52 
$190.94 
N.B. Cost per cubic yard of mud estimated to have been removed, 
$0,024. 
For cleaning these basins twice in 1910, when 15,600 cu. yd. of 
mud were estimated to have been removed, the total cost was 
found to be $468, or a cost of $0.03 per cubic yard. If the value 
of the water lost in draining, and that used in cleaning is deducted 
from this total cost, the cost per cubic yard is practically one- 
half that given above or $0,015. This latter figure is the one to 
be compared with those costs obtained in cleaning the Chain of 
Rocks' basins in St. Louis, since there no charge for the value of 
the water lost by draining or that used in cleaning is included. 
The cost of cleaning reservoirs, passages, and chambers of the 
Carrollton plant of the New Orleans water purification plant is 
of interest, on account of the completeness of the detailed list of 
cost items. The following table is taken from the Report of the 
Sewerage and Water Board for the year 1911: 
Total amount of wet mud removed from reservoirs 45,000 cu. yd. 
Total amount of dry material removed from reservoirs 18,000 cu. yd. 
Total amount of water treated during year 5,274 M. gal. 
Total amount of filtered water required, 4 M. gal., value... $62.40 
Total amount of treated water used and wasted, 17 M. gal., 
value 174.59 
Total amount of raw water used and wasted, 12 M. gal., 
value 47.52 
Total amount of labor required, value 302.05 RESERVOIRS AND COAGULATION BASINS 
77 
Cost of labor for cleaning per million gallons water treated. 0.053 
Cost of water for cleaning per million gallons water treated. 0.049 
$0,102 
Total estimated cost of cleaning per cubic yard of dry 
material removed 0.033 
Value of raw water taken at $3.96 per million gallons. 
Value of treated water taken at $10.27 per million gallons. 
Value of filtered water taken at $15.55 per million gallons. 
The cost of cleaning based on the wet mud removed is $0,013 
per cubic yard, and is comparable with the figures given for clean- 
ing the Chain of Rocks, basins at St. Louis or the coagulation 
basins at Cincinnati. 
Cost of Clearing Water by Settling.-Mr. S. Bent Russell in a 
paper1 discussing the cost of clearing water by sedimentation in 
reservoirs gives some interesting data on the costs obtained at St. 
Louis in operating the Bissell's Point basins between 1881 and 
1894. These basins were operated as plain sedimentation basins 
on the intermittent-flow plan. The head lost by this method of 
operation was 14 ft., and the cost of the increased lift is figured 
in the following table as one of the six items of cost chargeable 
against the clarification of the water. 
Analysis of Cost of Sedimentation in Bissell's Point Basins at St. 
Louis, Mo., between 1881 and 1894. 
Items 
Cost per million 
gallons settled 
Cost per year in per 
cent, of first cost 
1. 
Interest 
$2.820 
5.000 
2. 
Depreciation 
0.790 
1.400 
3. 
Repairs 
0.054 
0.068 
4. 
Operating 
0.281 
0.340 
5. 
Cleaning 
0.198 
0.230 
6. 
Increased lift 
0.329 
0.585 
Total 
$4,472 
7.623 
Mr. Russell concludes that: 
"The cost of clearing is dependent upon the quantity of water handled 
and the proportion of sediment removed. The area or dimensions of 
the floors will influence the cost. The inclination of floor and drains, 
etc., are important factors, and the character of the sediment must be 
considered. This item is of some importance, where there is much 
1 Eng. Record, vol. 60, Oct. 16, 1909. 78 
WATER PURIFICATION 
sediment, and to keep the cost within proper limits we are justified in 
adding considerably to the first cost of the plant." 
The cost of plain sedimentation, as shown by the operation 
costs of the Cincinnati, Ohio, settling reservoirs, was $0.20 per mil- 
lion gallons of water settled in both 1909 and 1910; in 1911 the 
cost was $0.19, and in 1912 $0.42 per million gallons of water 
settled. These costs do not include fixed charges, but are the 
entire cost of operating and maintaining the reservoirs and the 
grounds about them which are quite extensive. They include 
in the years 1909 and in 1912, respectively, the cost of cleaning 
one reservoir. Probably on an average 50 per cent, of the fore- 
going costs at Cincinnati are chargeable to the upkeep of grounds 
about the reservoirs, and should not properly be made a part of 
the cost of sedimentation. 
In 1912 labor costs were greater than in 1909, and this to- 
gether with certain changes made in drainage valves during the 
cleaning of the reservoir in that year accounts in part for the 
increased cost. The cost of upkeep of the grounds about the re- 
servoirs in 1912 was over 40 per cent, of the total. 
First Cost of Settling Reservoirs.-The first cost of construct- 
ing settling reservoirs naturally varies with their size, manner of 
construction and the general topography of the locality. Reser- 
voirs formed by the damming of a valley and having the land to 
be flowed stripped of its surface soil, have been found to cost 
from $135 to $600 per million gallons of capacity. Reservoirs 
formed by an earth embankment damming a natural ravine, and 
lined with concrete and brick, like those at Cincinnati, cost nearly 
$4,000 per million gallons of capacity; while the coagulation 
basins built principally in embankment and lined with concrete 
and brick cost nearly $14,000 per million gallons of their hold- 
ing capacity. On the other hand, masonry-walled basins like 
those at St. Louis cost from $6,000 to $6,500 per million gallons 
capacity. 
The cost of the coagulation basins of the Toledo, Ohio, water 
purification plant was $2,500 per million gallons of the daily 
rated capacity of the plant. The coagulation basins of the Cin- 
cinnati plant mentioned above when estimated in a similar man- 
ner cost about $2,700 per million gallons of daily capacity. The 
settling basins of the Columbus, Ohio, purification plant cost 
$5,630 per million gallons of daily capacity, while the large mix- 
ing tanks cost $1,470 per million gallons of daily capacity. RESERVOIRS AND COAGULATION BASINS 
79 
These figures are given merely to show ranges of cost rather 
than to make any comparisons, since the local topographical 
conditions, general type of plant and style of construction are so 
varied that a strict comparison of costs is improper and may be 
misleading. 
References 
1. F. B. Leopold: "The Water Filtration Works at Anderson, Ind." 
Eng. Record, vol. 51, p. 125, 1905. 
2. "The Revised Plans for the Purification of the Pittsburgh Water 
Supply." Eng. Record, vol. 51, p. 133, 1905. 
3. "A Concrete Settling Reservoir at McKeesport, Pa." Eng. Record, 
vol. 51, p. 597, 1905. 
4. "The Water Filtering and Softening Works at Columbus, Ohio." 
Eng. Record, vol. 53, p. 202, 1906. 
5. J. H. Gregory: "The Improved Water and Sewage Works at Colum- 
bus, Ohio." Discussion of paper in Proc. Amer. Soc. C. E., vol. 36, 
No. 3, 1910. 
6. "The Water Filtration Plant at Moline, Ill." Eng. Record, vol. 55, 
p. 705, 1907. 
7. "The New Settling Basins and Other Improvements to the St. Louis, 
Mo., Water Supply System." Eng. Record, vol. 56, p. 13, 1907. 
8. "Water Filtration Plant at Sandusky, Ohio." Eng. Record, vol. 60, 
p. 431, 1909. 
9. "Sedimentation Basin Cleaning at Poughkeepsie, N. Y." Eng. Record, 
vol. 63, p. 329, 1911. 
10. "The Use of Coagulants with Slow Sand Filtration." Eng. Record, 
vol. 64, p. 476, 1911. 
11. S. Bent Russell: "The Cost of Clearing Water by Sedimentation." 
Eng. Record, vol. 60, Oct. 16, 1909. 
12. "New Reinforced Concrete Reservoir at Council Bluffs, la." Eng. 
Record, vol. 67, p. 39, 1913. 
13. Annual reports of the Water Departments of New Orleans, La., Louis- 
ville, Ky., St. Louis, Mo., and Cincinnati, Ohio, for the years 1911 
and 1912. CHAPTER VIII 
PRACTICAL EFFICIENCIES OF SETTLING AND COAGU- 
LATION BASINS 
The practical efficiency of settling basins in the removal of 
sediment from turbid waters is variable, and depends upon a 
number of factors which have been discussed in the preceding 
two chapters. The manner in which the basins are operated, 
either as a result of unintelligent methods of handling, or because 
of the demands upon the plant which the operator must meet, 
are conditions contributing to their inefficiency. How great a 
percentage of the sediment is removed in the different types of 
basins is best illustrated by actual examples. 
The very finely divided clay sediment in the Allegheny River 
at Pittsburgh, Pa., is difficult to settle out. The sedimentation 
basins hold approximately 120,000,000 gal. of water. On the 
basis of 90,000,000 gal. daily consumption there is a theoretical 
storage of 32 hr. It has been concluded, however, that the water 
frequently passes through the basins in as short a time as 11 or 
12 hr. The following table compiled from the 1910 report of 
the Bureau of Water of Pittsburgh indicated a yearly average 
removal of but 28.2 per cent., and ranged from zero to nearly 49 
per cent. 
At Cincinnati, Ohio, the sediment in the Ohio River water is 
possibly somewhat more easily removed by settling. In the set- 
tling reservoirs there is provided a considerably longer period of 
storage, than at Pittsburgh, which probably accounts for the 
larger percentage of sediment removed. 
In the Cincinnati reservoirs there is a theoretical storage for 
6.6 days if based on a consumption of 50,000,000 gal. per day and 
an available storage capacity of 330,000,000 gal. From obser- 
vations made it has been concluded that the water actually 
passes through these basins in as short a time as 40 to 48 hr. The 
relative positions of the inlets and outlets in these reservoirs, 
as previously described, obviously make complete displacement 
impossible. 
80 SETTLING AND COAGULATION BASINS 
81 
Turbidity of Allegheny River Water at Ross Pumping Station and 
of Water after Settling 
Averages of 2-hr. readings in 1910-1911 
Year, 
1910 
River water 
Settled water 
Average 
Maxi- 
mum 
Mini- 
mum 
Average 
Maxi- 
mum 
Mini- 
mum 
Percent- 
age 
reduc- 
tion 
February 
38 
388 
7 
22 
110 
8 
42.1 
March 
65 
366 
14 
58 
200 
14 
10.8 
April 
44 
300 
18 
29 
110 
11 
34.1 
May 
24 
53 
10 
18 
38 
8 
25.0 
June 
27 
147 
13 
21 
110 
10 
22.3 
July 
17 
27 
8 
15 
25 
8 
11.7 
August 
18 
29 
12 
17 
23 
12 
5.5 
September 
50 
158 
6 
40 
140 
6 
20.0 
October 
19 
56 
9 
20 
80 
11 
November 
35 
105 
13 
29 
60 
13 
17.1 
December 
1911 
54 
833 
8 
28 
270 
7 
48.1 
January 
78 
328 
16 
40 
100 
16 
48.7 
Average 
39 
28 
28.2 
Average Percentage Removal of Sediment by Plain Sedimentation 
in Cincinnati Settling Reservoirs 
Month 
Average turbidity 
Percentage removal 
River water 
Settled water 
1911 
1912 
1911 
1912 
1911 
1912 
Parts per million 
January 
240 
122 
85 
54 
64.6 
55.7 
February 
190 
226 
105 
83 
44.7 
63.2 
March 
140 
360 
• 62 
190 
55.7 
47.5 
April 
160 
291 
57 
105 
64.4 
64.0 
May 
55 
202 
18 
97 
67.3 
52.1 
June 
76 
100 
24 
20 
68.4 
79.9 
July 
50 
712 
24 
190 
52.0 
73.4 
August 
64 
385 
27 
170 
57.8 
55.8 
September 
410 
328 
119 
140 
71.0 
57.3 
October 
257 
60 
102 
13 
60.3 
78.3 
November 
148 
76 
40 
16 
73.0 
79.0 
December 
128 
87 
36 
25 
71.9 
71.3 
Average 
159 
245 
58 
92 
63.5 
62.5 82 
WA TER P U RIF IC A TION 
At the New Orleans purification plant the turbid Mississippi 
River water is pumped directly into so-called grit reservoirs, and 
flows from the latter into mixing channels, where the water re- 
ceives the lime and such small amounts of coagulant as are used 
at this plant. It then flows to the settling basins, and from 
thence to the filters. The following table taken from the 1911 
report of the Sewerage and Water Board of New Orleans indi- 
cates the reduction in turbidity effected at various stages of the 
process. 
Turbidities of Mississippi River Water and Effluents of Grit Reser- 
voirs and Coagulation Reservoirs at New Orleans 
Purification Plant 
River water 
Effluent grit res. 
Effluent coag. res. 
1909 
1910 
1911 
1909 
1910 
1911 
1909 
1910 
1911 
Parts per million 
Maximum. . . 
1,600 
1,700 
1,400 
1,550 
1,450 
1,250 
340 
525 
280 
Minimum . . . 
80 
55 
150 
75 
55 
130 
1 
2 
5 
Average 
525 
550 
500 
475 
450 
425 
44 
32 
32 
It will be noted that the reduction in the turbidity of the river 
water after passing through the grit reservoirs ranges from 8 to 
18 per cent, on an average, while the reduction after passage 
through the coagulation reservoirs averages over 90 per cent. 
In all probability the grit reservoirs really remove considerably 
more sediment by actual weight than is indicated by the turbid- 
ity readings. They are to be regarded as settling chambers for 
the removal of the coarse silt and sand only, and not for deposit- 
ing the finer clay particles. 
The purification plant at St. Louis, Mo., offers some interest- 
ing data on the removal of sediment by coagulation with lime 
and sulphate of iron, and the effect of the series of settling basins 
through which the treated water is passed. Mr. W. F. Monfort, 
chemist at this plant, states in the report of the Water Commis- 
sioner issued in January, 1913, that: 
"When a water properly treated is passed through the basins 97 
to 99 per cent, of the suspended matter is precipitated in the first basin, 
the percentage removal depending upon the character and quantity 
of solids contained, the temperature, wind velocity, and direction, and 
the amount of sludge in the filling basin. In passing succeeding basins 
the remaining suspended matter undergoes a further reduction to one- SETTLING AND COAGULATION BASINS 
83 
half or one-sixth, likewise dependent upon velocity, size of particles, 
wind, and temperature. The major portion of clarification is, however, 
accomplished in the filling basin. There is no provision for applying 
chemicals after water enters the basins. Efficiency of the plant, there- 
fore, depends primarily upon the volume of water which can be 
satisfactorily clarified in the filling basin (Fig. 16). 
"The weight of suspended matter in the effluent from successive 
basins varies with the weight of solids carried by the raw water. Some 
illustrations are given: 
Parts 
per 
million 
Per cent, 
removal 
Parts 
per 
million 
Per cent, 
removal 
Parts 
per 
million 
Per cent, 
removal 
Parts 
per 
million 
Per cent, 
removal 
River 
1,444 
3,000 
4,500 
1,000 
Basin 1. 
14.0 
99.01 
47 
98.5 
95 
98.0 
25 
99.75 
Basin 2. 
12.1 
99.16 
21 
99.3 
50 
98.8 
20 
99.80 
Basin 3. 
8.4 
99.40 
14 
99.5 
35 
99.2 
10 
99.90 
Basin 4. 
7.1 
99.50 
12 
99.6 
20 
99.5 
10 
99.90 
Basin 5. 
5.8 
99.60 
10 
99.7 
15 
99.7 
5 
99.95 
Basin 6. 
5.6 
99.60 
10 
99.7 
15 
99.7 
5 
99.95 
Basin 7. 
5 
99.95 
Basin 8. 
5 
99.95 
"A change of pumping from a rate of 60,000,000 gal. per day to 
90,000,000 gal. has increased the suspended matter in the treated water 
from 2 to 7 parts per million, and a further sudden increase to 120,000,000 
gal. per day has caused a further rise of 10 to 13 parts. 
"In the spring when the temperature of the water in the river and 
basins is rising the sludge is less subject to disturbance than when, as 
in the fall and early winter, with falling temperature, the influent 
water, being more dense than the warmer water in the basins, passes 
downward over the sludge, causing it to carry over the weir from the 
filling basin. 
"In the fall with lowering atmospheric temperatures, the sludge and 
water in the bottom of the basin are sometimes 1°F. or more warmer 
than the surface water of the basins and river. Circulation in the 
basins, therefore, is effective in changing the course of the influent water 
currents, making them deeper, and increasing the scour. 
"The sludge is further subject to disturbance by wave action when 
high winds prevail for a day or two, as frequently occurs in March. 
In such cases the amount of suspended matter carrying over the weir 
from the filling basin shows a marked increase. 
"The basins at the Chain of Rocks (52 acres) are all uncovered, all 
used in series, and, therefore, subject to disturbance by each of the 
agencies affecting their successful working." WATER PURIFICATION 
SETTLING AND COAGULATION BASINS 
MAXIMA AND MINIMA OF ALKALINITY 
AND COLOR IN RAW WATER 
Diagram No. 1 
MAXIMA AND MINIMA OF TURBIDITY 
AND SUSPENDED MATTER, PARTS 
PER MILLIONS, IN RAW WATER 
Diagram No. 2 
COEFFICIENT FINENESS OF SOLIDS 
IN RAW WATER 
Diagram No. 4 
TYPICAL FLUCTATIONS IN CHARACTER 
OF RAW WATER 
Bacteria 
Suspended 
Solids 
Bicarb. 
Alkalinity 
B.Coli. 
Diagram No. 5 
REDUCTION THROUGH THE BASINS 
Diagram No. 7 
COMPARISON OF CALCIUM AND 
MAGNESIUM IN RAW AND TREATED WATER. 
Diagram No. 8 
CHARACTERISTICS OF TREATED WATER 
Diagram No. 6 
COMPARISON OF ALKALINITY, SUSPENDED 
SOLIDS AND COLOR IN RAW AND 
TREATED WATER 
Fig. 16.-Diagrams showing characteristics , t . 
or raw and treated water at bt. Louis, Mo. 86 
WATER PURIFICATION 
The author's experience with the settling and coagulation 
basins of the Cincinnati purification plant confirms the observa- 
tions of Mr. Monfort on the effect in change in rate of flow, tem- 
perature, velocity of wind and amount of sludge in the reservoirs 
and basins. 
Summaries of the percentage removal of sediment by the proc- 
ess of plain sedimentation followed by coagulation and filtra- 
tion at the Cincinnati plant are given in the following table: 
Turbidity, parts per million 
1908 
1909 
1910 
1911 
1912 
River water 
190 
225 
190 
159 
245 
Settled water 
100 
85 
80 
58 
92 
Coagulated water 
32 
23 
21 
15 
16 
Percentage removal 
By settling reservoirs.... 
48.4 
62.2 
57.9 
63.5 
62.5 
By coagulation basins.. . . 
34.6 
26.7 
31.1 
27.0 
31.0 
By filters 
17.0 
10 2 
11.0 
9.5 
6.5 
Bacteria in River, Settled, Coagulated and Filtered Water 
1908 
1909 
1910 
1911 
1912 
Bacteria per cubic centimeter 
River water 
Settled water 
Coagulated water 
Filtered water 
7,000 
3,400 
700 
100 
9,300 
2,500 
475 
75 
8,900 
3,200 
750 
75 
13,790 
3,140 
776 
39 
11,130 
3,945 
898 
26 
Average percentage re- 
moval 
98.5 
99.2 
99.2 
99.7 
99.8 
The effect of coagulation and settlement upon suspended sedi- 
ment and the bacteria in treating the Scioto River water at the 
Columbus, Ohio, water-purification plant is shown in the follow- 
ing tables. In this plant lime is used to soften the water and 
only small amounts of chemical coagulants are applied. SETTLING AND COAGULATION BASINS 
87 
Month 
Turbidity, parts per million 
1909 
1910 
1911 
River [ 
water । 
Settled 
water 
River 
water 
Settled 
water 
River 
water 
Settled 
water 
January 
7 
94 
21.0 
137 
7.0 
February 
256 
77 
12.0 
146 
8.0 
March 
109 
17.0 
66 
12.0 
28 
1.0 
April 
102 
21.0 
22 
4.0 
117 
5.9 
May 
196 
28.0 
41 
8.0 
61 
3.8 
June 
92 
14.0 
15 
3.0 
19 
1.5 
July 
86 
9.6 
24 
4.0 
14 
0.6 
August 
24 
4.0 
19 
4.0 
18 
0.0 
September 
20 
4.0 
16 
1.3 
39 
0.1 
October 
13 
2.0 
44 
3.6 
72 
3.4 
November 
17 
1.4 
6 
1.6 
98 
5.0 
December 
107 
13.0 
20 
1.0 
64 
2.5 
Month 
Bacteria per cubic centimeter 
1909 
1910 
1911 
River 
water 
Settled 
water 
River 
water 
Settled 
water 
River 
water 
Settled 
water 
January 
435 
12 
52,000 
1,600 
36,000 
43 
February 
32,825 
2,146 
15,500 
190 
8,000 
9 
March 
18,228 
828 
20,000 
190 
1,500 
7 
April 
8,500 
533 
2,600 
28 
12,400 
27 
May 
16,309 
3,523 
3,300 
24 
3,546 
383 
June. 
3,952 
180 
1,100 
12 
800 
9 
July 
4,009 
42 
1,900 
7 
1,700 
40 
August 
1,888 
83 
550 
4 
1,800 
190 
September 
1,171 
65 
700 
2 
2,400 
800 
October 
707 
100 
2,200 
42 
6,500 
160 
November 
1,300 
60 
300 
10 
44,000 
275 
December 
21,500 
65 
12,000 
14 
17,000 
120 
Averages 
9,235 
636 
9,338 
177 
11,470 
172 
The bacterial reduction produced by settling the Susquehanna 
River water with sulphate of alumina is shown in the following 
table of averages: 88 
WATER PURIFICATION 
Bacteria per Cubic Centimeter in River Water and Settled Water 
and Percentage Reduction Effected 
. Year 
Bacteria per cubic centimeter 
Percentage 
reduction 
River water 
Settled water 
1906 
12,372 
3,228 
73.91 
1907 
10^712 
4,504 
57.96 
1908 
4,949 
1,662 
66.43 
Average 
9,344 
3,131 
66.49 
It is obvious that the amount of coagulant used in treating the 
water, which contains suspended sediment, will have a marked 
influence on the apparent efficiency of settling basins. The more 
coagulant used the greater the precipitation of both the clay par- 
ticles and the bacteria. 
References 
1. Report of Sewerage and Water Board of New Orleans, La., for 1911. 
2. Report of Bureau of Water of Pittsburgh, Pa., for 1910. 
3. Report of Water Department of Cincinnati, Ohio, for 1911-1912. 
4. Report of Division of Water of the City of Columbus, Ohio, for 1911. 
5. Report of Board of Commissioners of Water and Lighting Department, 
Harrisburg, Pa., for 1908. 
6. W. F. Monfort: "The Water Supply of St. Louis, Mo." Reprint 
from Report of Water Commissioner for 1913. CHAPTER IX 
FILTRATION OF WATER 
Neither plain sedimentation nor that assisted by the use of 
coagulating chemicals, or both processes combined can, as a rule, 
produce complete clarification of turbid waters. Excellent re- 
sults may be obtained, however, provided the periods of sedi- 
mentation are sufficiently prolonged and enough coagulating 
chemicals are used. As a method of treatment preceding filtra- 
tion, however, they are very effective, since they naturally 
reduce the work to be done by the filters. The efficiency of 
settling as a method for the removal of the bacteria has been 
illustrated by practical examples in the preceding chapter. In 
this chapter the part played by filtration in the purification of 
water will be considered. 
THEORIES OF FILTRATION 
Various kinds of material may be used for filtering water, but 
sand and gravel are the most practical materials for this purpose 
and are almost universally employed. The passage of a natural 
water through a layer of sand and gravel of the proper character 
and thickness, has the effect of removing material in suspension 
in the water. The effluent from the filter in usually clear, having 
been freed from visible suspended particles, and is also found to 
have had removed from it a large proportion of the bacteria. 
Even vegetable coloring matter and colorless organic matter, 
which are probably in suspension in a colloidal form, are to a 
certain extent removed from water containing them by filtering 
through sand. 
The several factors directly affecting the efficiency of sand fil- 
ters are the thickness of the bed, the size and uniformity of the 
sand grains, the amount of organic matter in the water and ad- 
hering to the sand grains, the kind and amount of coagulating 
compounds which may have been employed, the temperature of 
the water, and the rate of flow. Why these factors influence the 
process of filtration requires a more extended discussion of the 
probable structure and action of a filter when in operation. 
89 90 
WATER PURIFICATION 
Function of a Filter.-The primary function of a filter is that 
of mechanically straining out suspended particles. These par- 
ticles may and usually do vary greatly in size. It can be readily 
understood how particles larger in size than the interstices of the 
sand bed will be strained out. It is not so easy to understand 
how particles with diameters of perhaps 0.0001 mm., and even 
much smaller are removed. The submicroscopic particles are 
much smaller than the bacteria, and yet a filter is able to strain 
them out of the water in which they are suspended. Evidently 
agencies are operating within the filter bed which are remark- 
ably effective in holding back these fine particles, and which do 
not depend alone upon the sand grains. 
For a long time it has been known that a sand filter was more 
effective after it had been in service awhile than when first 
started. Rapid sand filters as well as slow sand filters appear 
to undergo this improvement. The more organic matter contained 
in the water being filtered, the quicker this so-called "ripening" 
is effected. Apparently it is the organic matter which plays the 
chief part in the straining process. This organic matter may be 
either living or dead, that is, it may be organized in the forms of 
plant life such as the bacteria, algae, diatoms and lower forms of 
animal life, or it may be unorganized like the vegetable coloring 
matter and the colorless nitrogenous and carbonaceous com- 
pounds characteristic of natural waters. 
Real Structure of a Filter.-A close examination of a well-" rip- 
ened" sand filter reveals the sand grains covered with a very 
thin film of a gelatinous character, which is most abundant near 
the surface of the bed, but appears to permeate it to some depth. 
This material is evidently in part composed of organic matter 
both living and dead. It is in the colloidal form and seems to 
fill partially or wholly the interstices of the bed, and to be sup- 
ported by the sand grains. Not all of this material between the 
sand particles is of an organic character. Some of it is inorganic, 
but it too is apparently in the colloidal form. 
It will be easier now to conceive of the sand bed as a frame- 
work or skeleton of sand grains supporting and carrying the deli- 
cate structure of the filter proper, which latter consists of the 
colloidal matter deposited and developed within the bed. In a 
rapid sand filter, which is frequently disturbed by washing, the 
colloidal matter has less chance to develop and less opportunity 
for lodgement in the openings between the sand grains than in a FILTRATION OF WATER 
91 
slow sand filter. Nevertheless, when organic matter is plenti- 
ful in the water being treated, the sand grains become coated, 
and not even violent washing of the bed will entirely remove it. 
Moreover, in this class of filters colloids produced by the reaction 
of certain chemical compounds which are added to the water, are 
employed to assist the natural colloids in their straining action. 
Properties of the Colloidal Form of Matter with Reference to 
Filtration.-The filtering action of a sand bed must, therefore, be 
attributed to the properties of the colloids deposited or formed 
therein. A consideration of some of the properties of this form 
of matter will perhaps explain the ability of filters to perform the 
work which is so characteristic of them. The colloidal state of 
matter can be considered as a mesh structure, or a sort of porous 
framework. Through it water diffuses readily, but suspended 
colloids contained in the water will be held back. Thus the bac- 
teria, which are of a colloidal nature, are unable to pass through 
the mesh-like structure of the colloids attached to the sand grains, 
or are held back by adhering to the surface of the colloidal matter 
upon the sand and in the passages between the sand particles. 
Inorganic suspended colloids are likewise prevented from passing 
through the jelly-like colloids within the bed, either because they 
have been previously coagulated, whereby the particles have 
become too large to pass through the openings, or because they 
stick to the surfaces of the colloidal matter within the bed, the 
same as do the bacteria and other matter too large to pass through 
the capillary passages. 
The adhesion of colloidal particles to each other may be and 
probably is more than a mechanical attachment, that is, it may 
be an electrical attraction. It is known that coagulation of sus- 
pended colloidal particles is effected by the ions of an electrolyte 
having an electric charge of an opposite sign from that of the 
colloid. By neutralization of the electric charges of opposite 
signs coagulation is effected. It is also known that absorption 
of ions occurs on the surfaces of colloid particles, and it may be 
that coagulation by electrolytes is a phenomenon of adsorption. 
It thus becomes conceivable that within the mesh-like structure 
of the colloidal matter permeating the sand bed minute coagula- 
tion and sedimentation chambers exist in which the suspended 
colloidal matter in the water being applied to the bed is deposited 
and held. If such an explanation is accepted, then there must 
come a time in the operation of any filter when it can no longer WATER PURIFICATION 
92 
remove matter in suspension in the applied water, i.e., these co- 
agulation and sedimentation chambers and passages become 
overloaded. When this occurs the filter becomes ineffective. 
This is a well-known condition in filter operation. 
A sand filter, through which a colloidal solution of ferric hy- 
droxide is passed, retains the hydroxide and delivers a clear 
effluent for a certain length of time. Only a quite definite quan- 
tity is thus held back, after which the ferric hydroxide passes 
through unchanged. "It is reasonable to assume that the posi- 
tive (electrically charged) colloidal particles (of ferric hydroxide) 
are discharged and retained by the negatively charged sand 
grains."1 This explanation seems reasonable, and probably very 
similar conditions are possible within a rapid sand filter when it is 
filtering water treated with a chemical coagulant. Overloading 
or inability of a filter to pass clear water through it, after it has 
become clogged with sediment, is characteristic of both slow and 
rapid sand filters. 
Explanation of Some Well-known Filtration Phenomena.-In 
the experiments on filtration undertaken at Louisville, Ky., in 
1896-97, it was shown that by the addition of sulphate of alumina 
to the turbid Ohio River water a theoretical reduction in the 
alkalinity did not result, and that the kind and amount of the 
suspended clay present materially affected the proportion of the 
carbonate alkalinity used up. 
Suspended matter,2 parts per 
million 
200 
400 
800 
1,200 
Percentage which the actual 
reductions in alkalinity were of 
the theoretical 
85 
80 
75 
65 
The influence upon the amount of sulphate of alumina unac- 
counted for by reaction with the alkaline carbonates in the water, 
when the amount of suspended matter by weight was practically 
the same, is shown by the following table taken from this same 
report. 
1 Emil Hatshek: "Introduction to the Physics and Chemistry of 
Colloids." 
2 George W. Fuller: "Water Purification at Louisville." FILTRATION OF WATER 
93 
Sample 
Nos. 
Suspended solids, parts 
per million 
Percentage which the actual 
reduction in alkalinity was of 
the theoretical 
1 
500 
57 
2 
534 
74 
3 
584 
77 
4 
516 
80 
5 
558 
84 
Average 
550 
74 
Mr. George W. Fuller in this work also found "that with suc- 
cessive equal amounts of applied sulphate of alumina the work 
accomplished is not regularly progressive, and that for some rea- 
son in the ordinary river (Ohio) water the specific efficiency of 
the first portion of chemicals is very low, and less than that of 
subsequent ones." 
All of these facts find an explanation in our present knowledge 
of the properties of colloids. Loss of chemical coagulant is due 
to adsorption upon the surfaces of the colloidal particles. The 
ineffectiveness of the first portions of chemicals added is probably 
due to the fact that a definite minimum concentration of the 
electrolyte in every case is necessary to produce coagulation. 
This feature of coagulation has been very extensively investi- 
gated for a large number of electrolytes, and the minimum 
amounts necessary to produce coagulation have been accurately 
determined. 
For example, in a colloidal solution of arsenious sulphide the 
minimum concentration of a large number of electrolytes, which 
would bring about coagulation of this colloid, has been deter- 
mined. The striking feature of the results obtained was that 
the amount of the salts containing monovalent cations re- 
quired to produce coagulation was roughly 70 or 80 times that 
needed of salts containing divalent cations, and 600 times that 
necessary to be used in the case of salts with trivalent cations. 
It is, therefore, evident why the salts with trivalent elements 
like aluminum and iron are the best chemicals to be used in coagu- 
lating the suspended colloidal matter in natural waters, since the 
weight of the electrolyte needed is relatively small. 
In water softening the precipitates formed are of a colloidal 
character, and assist materially in clarifying the water, especially 
if other suspended material is present. Certain of the precipi- 94 
WA TEH P U RIF IC A TION 
tates, however, are very slowly deposited, as they appear to be 
held in colloidal solution for long periods. The so-called "after- 
deposits" of softened waters may be ascribed to this form of 
matter. 
The effectiveness of "roughing filters" must be due very 
largely to the action of the colloids formed and deposited within 
the filter bed. The relatively large size of the gravel and coarse 
sand of which such filters are usually constructed, as compared 
with the fine sand used in slow and rapid sand filters, appears 
to confirm the theory that the real filtering medium of any filter 
is the colloidal matter within the filter bed. 
If a sand filter bed is composed of fairly uniform sized grains, 
or if those of like size are stratified, as is the case in those rapid 
sand filters in which wash water is applied at high velocity, it 
is evident that a more even distribution of the colloidal matter 
in the bed is produced, and a more uniform rate of flow through 
all parts of the bed is possible. Too great a depth of bed im- 
poses a needless amount of friction to the flow of water through 
sand filters, although too shallow a bed may not produce enough 
resistance with a consequent tendency to uneven rates of 
filtration. 
The temperature of the water necessarily affects its viscosity. 
The more viscous it is, which is the case at low temperatures, 
the more frictional resistance it offers to passage through the 
bed. A low temperature of the water also appears to affect 
adversely the colloidal content of the bed, since the efficiency 
of sand filters is always lower during the winter months. 
A possible explanation for this fact may be that during the 
warmer months of the year, when the temperature of natural 
surface waters is the highest, the microscopic plant and animal 
life is most abundant. In other words, the quantity of colloidal 
matter brought to the sand bed is much greater when the water 
is warmer, and renders the filter relatively more efficient than 
when the water is at its Iqwest temperature. If high turbidities 
due to suspended clay particles occur during this period of low 
temperature, the ratio of. this class of colloids to the organic 
jelly-like colloidal matter of the filter is much greater. Hence, 
it may be that a filter under such circumstances becomes less 
efficient, because of an actual diminution in the quantity of the 
real filtering medium within the sand bed. In actual practice 
in operating rapid sand filters such difficulties are overcome by FILTRATION OF WATER 
95 
increasing the amount of coagulating chemicals, i.e., by provid- 
ing more colloidal matter in the form of aluminum or ferric 
hydroxide. 
The author has personally noted in operating rapid sand filters, 
when a sudden increase in the clay particles producing turbidity 
took place in the warm months of the year, that the filters were 
able to successfully clarify a water with four and five times the 
turbidity that they did during the winter months, and that a 
relatively small increase in the quantity of applied chemical 
coagulants sufficed to maintain a clear effluent. If the greater 
content of organic colloids in the bed at such a time was not the 
cause for the better efficiency of the filter, it is rather difficult 
to conceive of an explanation. 
The rate of flow of water through a filter must be properly 
regulated if a well-clarified and purified effluent is desired; but 
the rate can be varied widely provided a change from a lower to 
a higher rate is not made too rapidly, and certain maximum limits 
for a given depth and kind of bed are not exceeded. Uniformity 
of flow through all parts of the sand bed is absolutely essential 
for obtaining good results, and it is easy to see why this is the 
case if the colloids are regarded as the real filtering medium. 
Too rapid a flow must have the effect of washing out or breaking 
down the colloidal structure, and of subjecting it to rapid 
overloading with suspended material in the water applied to the 
filter. 
Enough illustrations of the phenomena associated with filter op- 
eration have been given to show that we must look to the physics 
and chemistry of the colloidal form in which both organic and in- 
organic matter may exist to explain in a rational manner the facts 
of the filtration of water. Biology undoubtedly plays an impor- 
tant part in producing and developing colloidal matter within the 
filter; and chemical changes which are incidental to the activity 
of the life and death of the organisms, are a part of the derived 
phenomena. However, the principles of filtration are in reality 
based upon the physical properties of the colloids, and upon 
them must be founded any rational theory of filtration. 
References 
1. Emil Hatschek: "An Introduction to the Physics and Chemistry of 
Colloids." 
2. Paul Rohland: "The Colloidal and Crystalloidal State of Matter." 96 
W A TEH P U RIF IC A TION 
3. A. A. Noyes: Jour. Am. Chern. Soc., vol. 27, No. 2. 
4. Whitney and Blake: Jour. Am. Chern. Soc., October, 1904. 
5. Ellms: "The Coagulation and Precipitation of Impurities in Water 
Purification." Eng. Record, May 13, 1905. 
6. Ellms and Snell: "Colloidal Suspensions and Their Relation to 
Problems in Water Purification." Proc. Am. Chem. Soc., June 22,1905. 
7. George W. Fuller: "Water Purification at Louisville, Ky." CHAPTER X 
PRELIMINARY TREATMENT OF WATER FOR SLOW 
SAND FILTERS 
Many waters require but little clarification in order to make 
them acceptable for domestic purposes, although from a sanitary 
standpoint constant purification by means of filtration is essen- 
tial. For this class of waters filtration through sand beds of 
large area at a comparatively low rate of flow may be practised 
for the larger part of the year with excellent results. However, 
at certain periods, such as follow a heavy rainfall and subse- 
quent flooding of the streams, these waters may become quite 
turbid with finely divided suspended matter. Slow sand filters 
will not continuously clarify such waters, and in consequence 
preparatory treatment has become quite general. 
This preliminary treatment may consist of the coagulation of 
the suspended matter with chemicals and settling out as much 
of it as possible in reservoirs before passing the water to the 
filters. In so far as this method of procedure is followed, it 
differs in no way from that described under coagulation and 
sedimentation in a preceding chapter. 
The efficiency of this type of preliminary treatment for the 
preparation of turbid waters for filtration through slow sand 
filters, is strikingly shown in the case of the Washington, D. C. 
filter plant. In 1908, Mr. Francis F. Longley, assistant super- 
intendent of the plant, conducted1 a series of experiments on 
Potomac River water for the purpose of studying the effect 
of varying periods of sedimentation (Figs. 17 and 18) with and 
without coagulants, and of the subsequent passage of the water 
through various types of preliminary filters, followed by filtration 
through slow sand filters. 
His experiments showed, that with one day of sedimentation 
followed by filtration through either one of two different types 
of pre-filters, and then through a slow sand filter, that the 
1 Francis F. Longley: "Experiments on Preliminary Treatment of 
Potomac River Water at Washington, D. C." Eng. Record, vol. 57, June 
27, 1908. 
97 98 
WA TER P U RIF IC A TION 
process failed to remove all of the turbidity. Even after 3 
or 4 days of sedimentation followed by triple filtration through 
slow sand filters, a perfectly clear effluent could not be obtained. 
WITHOUT COAGULATION 
YEARS 1906 TO 1911 
Perfect effluent 
below 'SO; 
WITH COAGULATION 
YEAR 1912 
Fig. 17.-Diagram of Potomac river water turbidities and effect of coagu- 
lation with alum. 
Readings were 300,360. 
100,500 and 300 
Turbidity Parts per Million 
Alum,brains 
per ballon 
Fig. 18.-Diagram of Potomac river water turbidities and effect of coagu- 
lation with alum. 
Mr. Longley, therefore, concluded that the complete removal 
of turbidity 
"can not be effected by any device applicable to large quantities of 
water, that involves purely mechanical principles in its operations." PRELIMINARY TREATMENT OF WATER 
99 
The clarification of the water was effected, however, by the 
use of sulphate of alumina, which produced coagulation of the 
minute clay particles in suspension. The coagulated water was 
next settled in sedimentation basins and then filtered through 
a slow sand filter. He, therefore, concluded that: 
"the desired improvement in the water (supply of Washington) can 
be effected by occasional coagulation with subsequent thorough sedi- 
mentation in the two existing reservoirs. This process is so entirely 
flexible that with its use the final product of the filters so much desired, 
is assured." 
The soundness of Mr. Longley's deductions were fully verified 
4 years later by results obtained in treating the water supply of 
Washington as he had outlined. These results are described 
by Mr. W. F. Wells in a recent paper,1 and are summarized 
below. 
"1. Seven days' storage is not sufficient to prepare the Potomac 
water, when the turbidity is above 100, for satisfactory slow sand 
filtration. 
"2. Preliminary treatment with alum introduces an ideal flexibility 
into the system, whereby the turbidity of water flowing on to the 
filters may be kept uniformly below 20 (or lower if desired). 
"3. Filtration of this water yields a water of constant purity, per- 
fectly clear, and with less than 10 bacteria per cubic centimeter. 
"4. The few bacteria surviving filtration are mostly harmless hardy 
soil forms, making it inadvisable to add hypochlorite of lime for 
sterilizing purposes. 
"5. Cutting down the peak load greatly reduces the cost of the filter 
operation, and the treated water keeps the beds in better condition. 
"6. The rate of filtration following alum treatment can be more 
than doubled to advantage, and an economical balance struck between 
the rate of filtration and the quantity of alum." 
Pittsburgh's Contact Roughing Filters and A-frame Baffles.- 
An interesting method for the preliminary treatment of the 
Allegheny River water prior to its passage through slow sand 
filters at Pittsburgh, Pa., has been worked out by Messrs. 
Johnson, Weston and Drake.2 The necessity for increasing 
1 Wm. F. Wells: "Some Notes on the Use of Alum in Connection with 
Slow Sand Filtration at Washington, D. C." Proc. Am. Water-works 
Assn., June, 1913. 
2 G. A. Johnson: "Preliminary Treatment of Water for Slow Sand Filters 
at Pittsburgh, Pa." Eng. News, vol. 68, Oct. 3, 1912. 100 
WATER PURIFICATION 
the yield of the slow sand filters of this large plant arose because 
of the peculiar character of the water which it is obliged to handle. 
This water is a mixture of the flow of the Allegheny and Kis- 
kiminetas Rivers. The water of the former river is polluted by 
manufacturing wastes from oil refineries, tanneries and dis- 
tilleries. In consequence it contains large quantities of organic 
matter and considerable coloring matter. The Kiskiminetas 
River water is fouled by the acid wastes of a coal mining region. 
These two streams unite about 15 miles above the filtration 
Fig. 19.-Plan of contact baffles and details of A-frame baffles at Pittsburgh 
filtration plant. 
Contact Baffles 
Cop Detail 
Section and Elevation of A-Frame Baffle 
plant, and a reaction between the acid and alkaline waters of 
the two streams results, which causes coagulation and the pro- 
duction of much matter in the colloidal form. This colloidal 
matter is formed partly because of the irregular mixing and 
varying proportions of the waters of the two streams containing 
these different kinds of suspended and dissolved matters. The 
petroleum oil wastes further complicate conditions especially 
in cold weather, when the paraffin of the oils deposit in the sand 
bed and clog it rapidly. The yield of these filters between 
scrapings has been cut down at times as much as 70 or 80 per 
cent, on account of these adverse conditions. PRELIMINARY TREATMENT OF WATER 
101 
In the experimental investigation which preceded the design 
and construction of the preliminary treating plant, it was found 
that the addition of lime to neutralize the acid condition of the 
water availed little toward preventing the sand beds becoming 
clogged and caked. It was also found that the use of bleach- 
ing powder assisted very materially in lengthening the periods 
of service of the filters, but for what reason it was not clearly 
understood. 
The plan devised to remedy these difficulties consisted in 
fitting one of the two 54,000,000-gal. concrete sedimentation 
basins (Fig. 19) with 24 contact baffles or roughing filters of very 
coarse stone, and with two hollow A-frame baffles. The second 
basin was also to be fitted up later on. The object sought in 
these devices was less the removal of suspended colloidal matter 
by actual straining, than it was to effect such a mixing and react- 
ing of the various constituents of the water as would permit of 
their subsequent settling out in the sedimentation basins. These 
baffles would also be serviceable when coagulants were applied 
to the water by hastening the chemical reactions. 
The contact baffles (Fig. 20) or roughing filters were arranged 
in four batteries of six units each in one end of the large sedi- 
mentation basin. Each unit was a concrete box 61.5 by 41.6 
ft. inside dimensions, and about 17 ft. deep. They were each 
provided with a false bottom composed of four rows of concrete 
longitudinal beams of arched construction resting on the floor 
of the tank, and supporting a series of 3 by 8-in. reinforced-con- 
crete beams 7 ft. 4 in. in the clear, and spaced 2 in. apart. On 
this false bottom 8 ft. of gravel and coarse stone were placed, 
ranging in size from 3 in. in the bottom layer to in. on top. 
The water is admitted to each contact baffle through a 24-in. 
inlet pipe with its invert at the level of the surface of the gravel 
filling. Under the usual conditions about 4 ft. of water stands 
on top of the gravel. A 20-in. pipe at the end of the tank oppo- 
site the inlet serves as the outlet or effluent pipe. The cleaning 
of the contact filters is effected by rapid draining through the 
coarse gravel and stone to appropriate drain piping by means 
of quick-opening valves. The roughing filters are operated at 
the rate of 75,000,000 gal. per acre daily. 
From the 24 contact or roughing filters the water passes 
through openings in a concrete wall to the main body of the 
sedimentation basin. In this part of the basin are provided two 102 
WA TER P U RIF IC A TION 
Fig. 20.-Isometric view of contact baffle unit, Pittsburgh filtration plant. PRELIMINARY TREATMENT OF WATER 
103 
A-frame baffles extending the full width of the basin and keyed 
into the walls of the latter on either side. These A-frames are 
hollow and are 15 ft. in height. The water enters near the top 
of the A-frame through openings provided, and after passing down 
inside the baffle flows out at the bottom into the next section of 
the sedimentation basin. A sludge drain extends along the 
bottom of the basin and through the baffles to be used in cleaning 
out the deposited sediment. 
PRELIMINARY RAPID SAND FILTERS 
To avoid the use of coagulating chemicals preliminary filters 
have been designed and installed in different plants for the pur- 
pose of removing a portion of the suspended matter, so that the 
slow sand filters may not become too quickly overloaded. These 
filters are designed with the idea of mechanically straining out 
as rapidly as possible the coarser particles, and are, therefore, con- 
structed of rather coarse gravel and sand, and are operated 
at relatively high rates of flow, viz., from 28,000,000 to 100,- 
000,000 gal. per acre per day. The effluent from these filters 
is passed through slow sand filters for a second and final filtra- 
tion. A water which has received such preliminary treatment 
may be filtered through slow sand filters at rates from 4,000,000 to 
8,000,000 gal. per acre per day, or considerably higher than the 
2,500,000 to 3,000,000 gal. per acre daily at which they are 
usually operated. 
In some cases these preliminary filters have been upward-flow 
filters, i.e., those in which the water passed upward through 
layers of gravel, slag or coke and compressed sponge clippings. 
This type of filter, known as the Maignen filter was installed in a 
plant (Fig. 21) at South Bethlehem, Pa., in connection with slow 
sand filters for the treatment of water from the Lehigh River. 
The latter stream carries at times of freshets a considerable 
amount of fine culm from coal mines, which gives it a deep black 
color. To prepare this water for slow sand filtration a scrubber 
or preliminary filter was placed at the rear of each slow sand fil- 
ter tank, and was separated from the latter by a low division wall. 
The water passed upward through the scrubbers and across the 
division wall on to the slow sand filters. 
The supply pipe to the scrubbers is an 8-in. cast-iron pipe con- 
nected to a Y in each scrubber. From the Y extend two lines of 104 
WATER PURIFICATION 
6-in. terra-cotta pipe, which are laid on the floor of the filter 
and at equal distances from the center line and side walls. The 
terra-cotta pipe is laid with tight joints, but each 3-ft. length is 
perforated with nine 1-in. holes in three rows, one row on the bot- 
tom and one row on each side of it at an angle of 45° with the bot- 
tom center-line. A wash-water drain of reinforced concrete of 
6 Perforated 
.Vitrified Pipe 
Sponge 
W Coke 
Rows of Slate 
Imbedded in Coke 
2 to 3 Pebble 
and Coke 
Puddle 
Concrete and 
Expanded Metal 
Underdrain 
PART TRANSVERSE SECTION 
Cedar Slats^ 
over Sponge 
Scrubber 
Filter 
J Raw Water Inlet from 
Storage Reservoir 
PART PLAN 
Filtering Material 
Concrete and Expanded 
Metal Underdrain 
LONGITUDINAL SECTION 
Fig. 21.-Details of filtering materials and piping connections of scrubbers 
at So. Bethlehem plant. 
half elliptical cross-section with its convex side upward is laid 
along the center line of the filter and connects with a 12-in. drain 
at the center. Two by 3-in. openings on 7 ^-in. centers along 
each edge, where the drain rests on the floor of the filter, provide 
inlets to the drain. 
"The space between the supply pipe and the waste water drain is 
filled with selected 3-in. gravel. The remainder of the lower foot of 
filtering material is made up of 3-in. coke. Four layers of IJ^-in. coke, PRELIMINARY TREATMENT OF WATER 
105 
with an aggregate thickness of 2 ft., are placed over the 3-in. coke. 
In each layer of the lJ4-in. coke are placed regular rows of slates about 
the size of ordinary roof slates. The rows in the lower layer are 
placed longitudinally in the scrubber, and those in the layers above 
it alternately, transversely and longitudinally. The slates in the 
lower and upper longitudinal rows are inclined about 30° from zero 
and 180°, respectively; the same amount and relation of inclination is 
maintained between the two sets of transverse rows. Over the coke 
containing the rows of slates is an additional 10 in. of IJ^-in. coke, and 
over that a layer of sponge, originally 18 in. thick, but compressed to 
make the total depth of the filtering material in the scrubber 4 ft. 6 
in. The mass of filtering material is prevented from floating, due to 
its buoyancy and the upward flow of water through it, by a grillage of 
2 by 3-in. cedar slats placed transversely in the scrubber with J^-in. 
spaces between them. The cedar slats are held in place by longitudinal 
6 by 8-in. yellow-pine stringers wedged against transverse 12-in. I- 
beams, imbedded at their ends in the concrete of the division and side 
walls, as shown in one of the illustrations."1 
The Maignen type of filter was installed in the Lower Rox- 
borough filter plant in Philadelphia, and was constructed essen- 
tially as described above for those built at the South Bethlehem 
plant. At the Belmont plant in Philadelphia there are nine 
separate filters divided into three compartments each. The 
first compartment contains ordinary coke, and the water enter- 
ing at one end at the bottom flows through it horizontally to the 
bottom of the next compartment. This latter compartment is 
filled with a layer of sponges about 6 ft. deep. The water flowing 
upward through this sponge layer next passes to the third com- 
partment, which is filled with coke breeze ranging from % to 
in. in diameter. The water filters downward through this coke 
breeze at the rate of 40,000,000 gal. per acre per day. 
It was found that the first two compartments of these scrub- 
bers were not effective in removing turbidity, and the third com- 
partment is now used only for preparing the water for the final 
filtration through slow sand filters. 
Torresdale Preliminary Filters.-The preliminary filters of 
the Torresdale plant at Philadelphia (Fig. 22) are a good example 
of 11 roughing filters" in which no coagulants are used. They are 
used to treat the water of the Delaware River preparatory to its 
filtration through slow sand filters of the usual type. They are 
'"Water Purification at South Bethlehem, Pa." Eng. Record, vol. 52. 
No. 3. 106 
WATER PURIFICATION 
CROSS SECTION THROUGH FILTER HOUSE 
20"Hydraulic Valves 
Fig. 22.-Torresdale preliminary filters, Philadelphia, Pa. 
PART SECTIONAL PLAN OF PLANT 
- Elevated Tank 
Influent Gullet 
6 "Clear Water 
Filter House 
Effluent Gullet PRELIMINARY TREATMENT OF WATER 
107 
in fact mechanical filters of the rapid sand type, differing princi- 
pally in the character of the filtering layers of sand and gravel. 
They consist of a series of 120 beds arranged in eight rows or 
batteries of 15 beds each. There are four filter houses each run- 
ning across the entire width of the plant and controlling two bat- 
teries of 30 filters each. Each bed measures 20 ft. 3 in. by 60 ft. 
in the clear, and is controlled from its own individual operating 
table. The filtering material consists of 15 in. of gravel varying 
in size from 2 to 3 in. in diameter, 4 in. of gravel varying from % 
to in., 3 in. of gravel varying from to in., 8 in. of gravel 
varying from % to in., and 12 in. of coarse sand varying from 
0.03 to 0.04 in. 
The influent gullets1 for these filters run the full length of the 
plant, and serve one or two batteries of beds. They are formed 
by longitudinal walls at the backs of the beds of adjacent 
batteries, and are roofed with 6-in. reinforced-concrete slabs. 
The wash-water gullet is constructed of reinforced concrete 12 in. 
wide and 4 ft. 3 in. deep. Extending out laterally on each side of 
this gullet, and terminating at the side walls of a filter are sheet- 
steel wash-water troughs 18 in. wide, increasing in depth from 6 
in. at the wall line to 9 in. at the gullet. Four feet of water is 
carried over the filter bed when in operation. 
Two collectors extend the full length of each bed and on each 
side of the wash-water gullet; they are constructed of reinforced 
concrete rectangular in section, 30 in. wide and 8 in. deep, inside 
measurements. After passing through a hydraulically operated 
valve and an effluent controller, the water passes to the effluent 
gullet, which is of reinforced concrete rectangular in shape and is 
located under the operating gallery of each house. Thirty-inch 
wash-water mains are laid on top of the effluent gullets, and from 
these the wash-water piping is led into each filter through 20-in. 
spiral-riveted galvanized pipe suspended from the roof and con- 
nected at the center of the bed and directly above the wash-water 
gullet with four 12-in. pipes diverging toward the corners of the 
bed. These diverging pipes in turn each connect with two 8-in. 
pipes directly above the filtered-water collectors, and from these 
8-in. pipes vertical downtakes of the same diameter connect with 
a manifold. There are eight manifolds in each filter which have 
l|^-in. galvanized laterals with %g-in. holes in the bottom ex- 
1 "Description of Filtration Works and Pumping Stations of Philadel- 
phia." Bureau of Water, Philadelphia, Pa. 108 
WATER PURIFICATION 
Fig. 23.-New sedimentation basin, Tbrresdale filter plant, Philadelphia. PRELIMINARY TREATMENT OF WATER 
109 
Cross-Section 
Fig. 24.-Arrangements for treating water with chemicals at Torresdale filter plant, Philadelphia. 
Longitudinal Section 110 
WA TER P URIE IC A TION 
tending on 5^-in. centers. The holes are likewise placed on 
5^-in. centers. These holes are brass-bushed. 
The wash water is drained from the wash-water gullet through 
a 12 by 36-in. hydraulically operated sluice gate discharging 
into the wash-water drain, which simply consists of an open space 
between the wall of the filter bed and the side of the effluent gullet. 
The main air supply consists of a 20-in. pipe running the full 
length of each filter house and is suspended from the roof. The 
air system is connected to the wash-water piping and air is 
introduced through the manifold in the bottom of each filter. 
Fig. 25.-Arrangements for treating water with chemicals at Torresdale 
filter plant, Philadelphia. 
The agitation of the water in the filter bed by compressed air 
precedes the washing with water. Periods of service between 
washings may vary from 12 hr. to 4 days. The usual rate of 
filtration is 80,000,000 gal. per acre per day. The rated capacity 
of the plant is 240,000,000 gal. per day. 
The difficulty encountered at the Torresdale plant in obtain- 
ing at all times sufficient clarification of the raw water with the 
roughing filters alone finally led to the construction of a sedi- 
mentation basin in which coagulants might be used if desired. 
The Puech-Chabal System.-Water-purification plants of 
this type make extensive use of preliminary filters prior to apply- 
ing the water to slow sand filters. A number of plants have 
been installed near Paris and in other European cities. PRELIMINARY TREATMENT OF WATER 
111 
In this system the raw water first passes through a set of three 
or more roughing filters of coarse-graded gravel, which varies 
Raw Water 
Canal 
Inlet Canal Supported on Concrete Columns 
SECTIONAL PLAN 
Fig. 26.-Puech-Chabal system of filtration. 
in size from 1 in. in the first basin to smaller sizes in each suc- 
ceeding basin. By graduating the sizes of the material in these 
Fig. 27.-Puech-Chabal system of filtration. 
roughing filters the velocity of the water through them is greater 
in the first basins than in the last. 112 
WATER PURIFICATION 
The partially clarified effluent from the roughing filters passes 
through a so-called pre-filter and from thence to a slow sand 
filter of the usual type. 
This system has been described in some detail by Mr. William 
F. Johnston in the Engineering Record, May, 6, 1911. The 
accompanying cuts are illustrative of the process. 
Operating Results and Costs of Preliminary Treatment.- 
The results obtained in the operation of plants for the pre- 
liminary treatment of water for slow sand filters depend upon the 
adequacy of the particular method employed. Variations in the 
Laboratory 
Machinery 
Room 
Clear Water 
F Reservoir 
Rough 
Filters 
Submerged 
Filter 
Distributing 
Ultra Violet Ray 
Sterilized 
Fig. 28.-Puech-Chabal system of filtration. 
amount and kind of sediment to be handled often tax such plants 
severely, and naturally the slow sand filters have to bear some- 
times more and sometimes less than their proportion of the 
burden. 
At the Washington purification plant the benefits obtained by 
preliminary treatment with aluminum sulphate have been partly 
indicated in the preceding pages. The following table shows 
quite clearly the great aid this preliminary treatment is to the 
slow sand filters in reducing both the bacteria and sediment which 
they are called upon to remove. PRELIMINARY TREATMENT OF WATER 
113 
Bacteria and Turbidity Averages-January to July 
1906-1911 
1911 
1912 
Days alum 
was added 
Tur- 
bidity 
Bac- 
teria 
Tur- 
bidity 
Bac- 
teria 
Tur- 
bidity 
Bac- 
teria 
Tur- 
bidity 
Bac- 
teria 
Great Falls 
128 
9,000 
7,043 
3,242 
73 
70 
39 
26 
14 
0 + 
8,735 
6,387 
2,357 
46 
302 
135 
65 
11 
0 + 
12,707 
9,367 
1,769 
60 
678 
302 
120 
10 
0 
22,534 
15,335 
1,310 
28 
Dalecarlia inlet 
Dalecarlia outlet1 
62 
30 
3 
McMillan outlet 
Filtered water 
1 Water flows through Dalecarlia Reservoir to its outlet, at which point it receives the 
solution of aluminum sulphate. Some alum was used in winter months of 1911, but 
regular plant for applying coagulant was not in operation until 1912. 
The amount of aluminum sulphate used between the first of 
July, 1911, and June 30, 1912, was 534.49 tons, and was applied 
on 110 days in this period. The average quantity of aluminum 
sulphate added to each gallon of water was approximately 1 
grain. The cost of this preliminary treatment was $0.53 per 
million gallons filtered, of which $0.48 was for the chemical and 
$0.05 for labor. 
The Torresdale preliminary filters in Philadelphia were placed 
in operation on Jan. 21, 1909. The daily average reductions 
in turbidity and bacteria for the year 1910 were 67.4 and 68.5 
per cent., respectively. The average turbidity of the water 
applied to the pre-filters was 25 parts per million, and the 
maximum 260 parts per million. The percentage of wash water 
used averaged 1 per cent, of the amount filtered. Each pre- 
filter was cleaned on an average of 334 times during the year, 
or the period of service between cleanings averaged about 1.1 
days. 
The cost of operating the Torresdale preliminary filters was 
$0.30 per million gallons. This does not include any charge for 
laboratory supervision. The operating costs per million gallons 
for preliminary filtration through the Maignen type of filters 
at the Lower Roxborough and Belmont plants in Philadelphia 
were $1.23 and $0.53, respectively. 
The Lower Roxborough filters effected an average removal of 
turbidity of 51.4 per cent., and of bacteria 41 per cent. The Bel- 
mont pre-filters reduced the turbidity 50.1 per cent., and the 
bacteria 44.5 per cent. 
At Wilmington, Del., preliminary filters of the Maignen type 114 
WA TER P U RIF IC A TION 
showed the following efficiencies in operation for a series of years 
ending June 30, 1912:' 
Year 
Unfiltered water 
Effluent from pre- 
liminary filters 
Per cent, removed 
Parts pe 
million 
1907-08 
24.8 
10.6 
57.26 
1908-09 
33.2 
11.8 
64.46 
1909-10 
30.6 
14.6 
52.25 
1910-11 
49.0 
20.0 
59.18 
1911-12 
68.0 
36.0 
47.06 
Average 
41.3 
18.6 
54.96 
The average reduction in color by the pre-filters was 2.3 parts 
per million of the standard scale in 1912, or from 24.5 parts to 
22.2 parts per million. This is a removal of 9.38 per cent, of the 
total color. The range in color in the raw water was from 70 
parts per million to 17 parts of the standard scale. 
The efficiency of the pre-filters in removing bacteria is shown 
in the following table: 
Year 
Unfiltered water 
Effluent from prelimi- 
nary filters 
Per cent, removed 
Bacteria per cu 
bic centimeter 
1907-08 
5,116 
3,545 
30.712 
1908-09 
4,142 
2,171 
47.792 
1909-10 
5,248 
1,250 
76.19 
1910-11 
31,955 
16,256 
49.13 
1911-12 
49,533 
29,816 
39.06 
Average 
15,999 
8,840 
44.75 
N.B. The average efficiency of these filters judged from counts made 
from cultures on agar as a medium was 39.29 per cent., while with gelatin 
the percentage removal indicated was 54.79 per cent. 
The preliminary filters showed a reduction of 4.94 per cent, 
in the number of positive tests for the Bacillus coli during the 
year 1911-12. 
1 Report Board Water Commissioners of Wilmington, Del., 1911-12. 
2 Agar counts. PRELIMINARY TREATMENT OF WATER 
115 
The cost of operating the Wilmington preliminary filters is 
given in detail in the following table: 
Total cost 
Cost per million 
gallons 
Salaries 
$863 25 
$0.21 
0 005 
Additional labor 
10.63 
Supplies 
2 34 
0 001 
Repairs and renewals 
96.41 
0 023 
Light and power 
511.55 
0.124 
Other operations 
51.21 
0 012 
Special 
688.30 
0.166 
Total 
$2,223.69 
$0,541 
The above figures are gross operating costs but do not include 
interest or depreciation on plant investment. The item "Special" 
refers to improvements on the filter beds, and if omitted would 
bring the cost down to $0,375 per million gallons. 
References 
1. George W. Fuller: "Report on Water Purification at Cincinnati, 
Ohio." 
2. F. F. Longley: "Experiments on Preliminary Treatment of Potomac 
River Water at Washington, D. C." Eng. Record, vol. 57, June 27, 
1908. 
3. Wm. F. Wells: "Some Notes on the Use of Alum in Connection with 
Slow Sand Filtration at Washington, D. C." Proc. Am. Water- 
works Assn., June, 1913. 
4. G. A. Johnson: "Preliminary Treatment of Water for Slow Sand Fil- 
ters at Pittsburgh, Pa." Eng. News, vol. 68, Oct. 3, 1912. 
5. "Water Purification at South Bethlehem, Pa." Eng. Record, vol. 
52, No. 3. 
6. "Description of Filtration Works and Pumping Stations of Philadel- 
phia." Bureau of Water, Philadelphia, Pa. 
7. Report Board of Water Commissioners of Wilmington, Del., 1911-12. 
8. W. F. Johnston: "Water Purification by the Puech-Chabal System." 
Eng. Record, vol. 63, May 6, 1911. CHAPTER XI 
SYSTEM OF SLOW SAND FILTRATION 
The most efficient method for the final treatment of a water, 
which contains finely divided material in suspension and one 
which has proven entirely satisfactorily in practice, is to subject 
it to filtration through a bed of sand. Other filtering mediums 
may be used; but the small size of the particles of which the sand 
is usually composed, which makes it possible to prepare a com- 
pact yet porous bed, its slight deterioration by usage and the 
relative ease with which it may be cleaned, its wide distribution, 
and its low cost, are the reasons for its general employment. 
Sand filtration, as commonly practised, may be divided into 
two general methods, depending on the rate at which the water 
flows through the sand bed. These rates of filtration are so 
widely different that a classification as slow sand filters and rapid 
sand filters is sufficient to characterize the two systems. Where 
the rate of filtration does not exceed 6,000,000 to 8,000,000 gal. 
per acre per day, the filters are called slow sand filters. Rates 
as low as 1,500,000 gal. per acre per day are used, but the 
average rate is probably between 2,500,000 and 3,000,000 gal. 
per acre per day. Under some very favorable circumstances 
the yield may be as high as 6,000,000 or 8,000,000 gal. per acre 
per day. In such cases the unpurified water is usually quite 
free from suspended matter, or preliminary treatment of some 
sort has been employed. 
The rates at which rapid sand filters are operated vary widely, 
the usual rate being about 125,000,000 gal. per acre daily. The 
rate may be 40 per cent, lower or 20 per cent, higher than the 
above-named rate and still be regarded as a rapid sand filter. 
Rapid sand filters almost invariably require the water to be 
prepared for such high rates of filtration by being treated with 
some agent which will coagulate the impurities suspended in the 
water, and which will furnish material in a colloidal condition 
that will cover the surface of the sand as soon as the water begins 
to pass through the bed. In this class of filters the large volume 
116 GENERAL PLAN OF 
WASHINGTON, FILTRATION PLANT 
SHOWING MAIN PIPE LINES 
SCALE OF FE£T 
0 50 100 150 200 250 300 
NORTH CAPITOL STREET 
24 Tile Waste Drain 
FIRST STREET WEST 
N>Cap Street Sewer 
24 Tile Drain to Reservoir 
Fig. 29.-General plan of Washington filtration plant showing main pipe lines. 
(.Facing page 116.) SYSTEM OF SLOW SAND FILTRATION 
117 
of water, which passes through the bed in a comparatively short 
period of time, makes it possible to build them of relatively small 
area as compared with slow sand filters, and also necessitates 
efficient and rapid methods of cleaning. These filters will be 
considered in detail in subsequent chapters. 
General Form and Construction of Slow Sand Filters 
(Figs. 29, 30, 31 and 32).-If the rate of filtration is as low as 
2,500,000 or 3,000,000 gal. per acre daily, it is obvious that in 
order to obtain a sufficient quantity of water for a public supply, 
a considerable area of sand must be used for filtering the water. 
In consequence the slow sand filter plants consist of beds of sand 
varying in size from to 1 acre in extent. As many beds are 
provided as may be necessary to supply the volume of water 
needed at times of maximum draft, with proper allowances made 
for those beds which may be withdrawn from service for the 
purpose of cleaning or repairing them. 
The sand beds are built both with and without covers. Un- 
covered beds during the winter months may be operated with 
difficulty, if the temperature is low enough to form ice in the 
water on the top of the filter. According to Mr. Allen Hazen, 
where the normal mean January temperatures are lower than 
32°F., it is probably safer to provide covers. 
The form and general method of construction of the filter beds 
depend upon the particular conditions which may exist at the 
site selected. In form they are usually rectangular, but may 
take other shapes best adapted to the land available. As these 
filters are virtually shallow reservoirs and cover considerable 
area, it is necessary that the embankments and bottom below the 
underdrains be made water-tight. Paved earth embankments, 
masonry or reinforced-concrete walls are used for the sides of 
the filter. The bottom is constructed of either a thin layer of 
concrete, or of a pavement over a puddle layer, or of inverted 
groined arches of concrete carrying piers to support a cover for 
the filter. If the bottom of the filter is not water-tight, troubles 
may arise from the infiltration of ground water, when the level 
of the latter is higher than the water on the filters. Loss of 
water may also result by leakage either to adjoining filters or 
to some point outside the filters should the water level in the 
latter be higher than the ground water. According to Mr. 
Allen Hazen, leaky filter bottoms have probably caused more 
annoyance in operation than perhaps any other single defect. 118 
WATER PURIFICATION 
Fig. 30.-Belmont slow sand filters, Philadelphia SYSTEM OF SLOW SAND FILTRATION 
119 
Cracks in division walls and side walls may also cause loss of 
water and become troublesome in operation. 
Fig. 31.-Plan of Queen Lane slow sand filters, Philadelphia. 
The covers for slow sand filters are constructed of brick or con- 
crete supported by pillars. The groined-arch construction is the 120 
WATER PURIFICATION 
Fig. 32.-Sections through Queen Lane slow sand filters, Philadelphia. 
SECTION ON B-B 
TRANSVERSE SECTION 
LONGITUDINAL SECTION 
SECTION ON A-A 
12*Throngh Outlet- i 
lO'Fllter Drain-- 
2O"Casi Iron Sewer ' SYSTEM OF SLOW SAND FILTRATION 
121 
more common form of building the cover, although flat rein- 
forced-concrete covers are used as well. Over the vaulting is 
spread a layer of earth from 2 to 3 ft. in thickness. The top is 
provided with openings for the admission of light and air when 
needed during cleaning, and with entrance galleries for the re- 
moval or introduction of sand. 
Underdrain System.-The underdrain system which supports 
the gravel and sand of the filter bed is usually constructed of 
brick and tile, and is so arranged as to collect the water from all 
parts of the bed as uniformly as possible. The main drain pass- 
ing through the center of the bed is fed by laterals which usually 
enter at right angles to the former. The main drain is often 
constructed of brick or concrete, but the laterals are usually of 
tile. The drains are built either above, partly above, or flush 
with the floor of the filter. If placed on the floor of the filter, 
their tops do not extend above the coarsest gravel in order to 
prevent the open joints becoming clogged with small gravel or 
sand. Some double bottoms of brick have been used in Euro- 
pean filters, in which small arches or other convenient method of 
supporting a false bottom of open brickwork are employed. On 
this false bottom rest the gravel and sand. 
The frictional resistance to the flow of water in the under- 
drains should be very low, and so distributed as to avoid as much 
as possible unequal rates of filtration in different portions of the 
bed. Mr. Allen Hazen suggests that velocities of flow in lateral 
tile drains between 4 and 12 in. in diameter should be from 0.3 
to 0.5 ft. per second, and that in the main drains 0.55 ft. per 
second should not be exceeded, where the rate of filtration is 
about 2,570,000 gal. per acre per day. In order to reduce the 
size of the center drain, and to maintain more uniform rates of 
filtration throughout the bed, compensating orifices were used 
by Messrs. Hazen and Hardy in constructing the Washington, 
D. C., slow sand filters. These orifices consisted of circular 
openings cut in brass disks, and which were placed at the 
entrance of the laterals to the main drain. The object sought 
was to introduce frictional resistance at these points "equal to 
the combined frictional resistance and velocity head in the main 
drain from the most remote point to the point in question."1 
Gravel Layers.-The gravel to be used in slow sand filters 
must be carefully selected and placed. Only gravel composed of 
1 Trans. Amer. Soc. C. E., vol. 57, p. 307. 122 
WATER PURIFICATION 
the harder rocks should be used, i.e., limestone or other easily 
crushed rock should be avoided. The manner in which the 
underdrains are laid, and the nature of the floor of the filter 
determine in a large measure the thickness and way in which the 
layers of gravel of various sizes are distributed. Around the 
drains the largest stones are placed, and above them successive 
layers of intermediate sizes of gravel until the sand is reached. 
At the Albany, N. Y., filter plant, three grades of gravel were 
used, the largest size being material which passed a screen having 
holes 3 in. in diameter, and which was held by a screen having 
holes 1 in. in diameter. The next size of gravel was that which 
passed the 1-in. holes and was held back by a screen with %-in. 
holes; and the smallest size of gravel that which passed the %-in. 
holes and was retained by a screen with ^g-in. holes. The 
largest size of gravel was placed entirely around the 6-in. pipe 
drains, and slightly above their tops. In some cases it even 
covered the entire floor of the filter, but this was not uniformly 
carried out. The next smaller size of gravel was so laid as to 
bring the gravel layer to the same level all over the filter, and 
averaged about 2 in. in thickness immediately above the drains. 
At other points it was somewhat thicker, depending upon the 
amount of coarse gravel underneath. On top of this second 
layer was placed a third layer of the finest gravel to an average 
depth of 2% in. 
In many of the European plants much deeper gravel layers 
have been used than that described above. Deeper layers are 
of no particular advantage, since there is little likelihood of the 
finer gravel and sand getting down into the drains, if the layers 
are properly graded. In general, if any supporting layer of 
gravel does not exceed by more than three or four times the 
average size of the stones of the next layer above it, there is 
little danger of the finer material penetrating to the lower layers. 
Each layer must be complete in itself. The gravel should be 
washed free from sand, earth or other fine material, as a very 
little matter of this character may clog the gravel layer and pro- 
duce unequal rates of flow through the bed. 
The total depth of gravel layers varies considerably in prac- 
tice, but is usually from 2 to 3 ft. in most of the European filters. 
Recent American practice has cut this depth down to 12 and 16 
in. Even 6-in. gravel layers, when carefully graded and placed, 
have proven entirely satisfactory. A gravel bed composed of SYSTEM OF SLOW SAND FILTRATION 
123 
6 in. of stones from 1 to 3 in. in diameter, 2 in. of smaller stones 
from y to 1 in. in diameter, 2 in. of coarse gravel with stones 
from to y in. in diameter, and 2 in. of a very coarse sand with 
particles from ^2 to y& in. in diameter, should, if carefully laid, 
provide an excellent foundation for the sand bed above it. 
In the Washington, D. C., filters, the gravel was placed on 
the inverted groined arch floor in a manner similar to that at 
Albany, N. Y., and approximated 12 in. in depth midway between 
the piers, and decreasing to 3 in. at the piers. Crushed trap rock 
or granite was used in three sizes, the lower layer of coarse gravel 
being about 7 in. thick, and the two finer layers each 2^2 in- 
thick. The gravel was kept 2 ft. away from the side walls. 
Somewhat deeper layers of gravel were used in several of the 
Philadelphia filter plants. In these plants gravel ranging in size 
from 1% to 3 in. in diameter was placed around the collectors 
and for a height of 6 in. from the floor of the filter. Above this 
layer was placed 4 in. of gravel ranging from % to 1% in- in 
diameter. On top of this was a 3-in. layer of gravel with stones 
ranging from y to % in. in diameter, and above the latter a 
layer of 2 in. of fine gravel of a size from y in. in diameter to 
material retained on a sieve with 14 meshes to the linear inch. 
On top of the 2-in. layer was a 1-in. layer of coarse sand, which 
passed a 14-mesh sieve and was retained on a 20-mesh sieve. 
The total depth of the gravel and coarse sand was 16 in. 
Sand.-The large amount of sand required for slow sand filters 
necessitates the use of the sands usually found in river beds, or on 
the shores of the ocean or great lakes, or in deposits at or near the 
earth's surface. Not all of this material is suitable in its compo- 
sition, and frequently not so as regards the size of the particles. 
It usually is composed of quartz and hard silicates whose grains 
are sometimes quite irregular in shape, and with sharp angles. 
On the seashore, and wherever erosive action of the water has 
rounded the particles, a much less angular grain is obtained, and 
one which may be considerably finer than the bank sands. Natu- 
ral sands are usually mixed with more or less finer material like 
clay, which should be removed by washing before being used. 
Sands for filter purposes should be practically all quartz, or 
silicates which do not break down easily, and should not contain 
carbonates of lime or magnesia, except in very small amounts. 
The presence of aluminous and calcareous material tends to 
increase the frictional resistance of the sand to the flow of the 124 
WATER PURIFICATION 
water, and salts of lime and magnesia to harden it. Hence speci- 
fications for providing filter sand usually set limits for the 
amounts of these materials, which if exceeded, renders the sand 
unacceptable. 
Size of Sand.-The sizes of the sand grains are an important 
factor in the operation of a filter, since they materially influence 
the rate of filtration. In natural sands the relative amounts of 
the various-sized particles differ considerably. As far as the 
transmission of the water is concerned, the influence of the finest 
sand grains is far greater in proportion to their number than are 
the coarser particles. By filling the voids between the larger 
grains the finer particles may render an otherwise coarse sand so 
compact as to offer a high resistance to the flow of the water. 
It is evident, therefore, that it is the size and number of the 
pores between the grains of sand, or in other words, the porosity 
of the sand bed which governs primarily the rate of flow. 
As the rate of flow is a function of the size of the sand grains 
and of their compactness, it is customary to express the velocity 
of flow in terms involving the so-called "effective size" of the 
sand grains, together with other factors, such as the head of 
water, thickness of the sand layer, and temperature of the water. 
Mr. Allen Hazen's1 experiments on filter sands led to the develop- 
ment of the following formula: 
1 = cd I \60° / 
where V = velocity of water in meters daily in a solid column; 
c = a coefficient varying from 700 to 1,000 for new and 
clean sand of fairly uniform size, to as low as 400 
for old compacted sand; 
d = effective size of sand; 
h = head of water; 
I = depth of sand layer; and 
t = temperature of water in degrees Fahrenheit. 
The effect of the temperature on the rate of flow is quite 
marked, the rate decreasing or increasing with the fall or rise of 
the temperature. For example, twice as much water- will 
pass through a given sand at 74°F. as will at 32°F. 
Effective Size of Sand.-For a uniformly sized sand of a certain 
compactness, it has been found that the velocity of flow is nearly 
1 Allen Hazen:-"The Filtration of Public Water Supplies." SYSTEM OF SLOW SAND FILTRATION 
125 
proportional to the square of the diameter of the sand grain, but 
as natural sands are not uniform, it is necessary to introduce a 
coefficient into the formula. Moreover, since the finer particles 
of a natural sand appear to be the sizes of grains which most 
materially influence the rate of flow, it becomes a question as 
to what size shall be chosen to represent the sand as a whole in 
calculating its frictional resistance. Hazen decided from his 
experiments that the size of grain such that 10 per cent, of the 
particles are smaller and 90 per cent, are larger than this size, 
should be regarded as the "effective size." The ratio of the size 
of the grains, such that 60 per cent, of the sand is finer than this 
size to the "effective size," is termed the "uniformity coefficient," 
or in other words a ratio of the "large particles" to the "small 
particles" in a sand as above defined. 
A very fine sand for use in a filter would be one that had an 
effective size as low as 0.17 mm., and a very coarse sand one 
with an effective size as high as 0.50 mm. The more usual sizes 
for slow sand filters range between 0.20 mm. and 0.40 mm. in 
effective size. Natural sands will range from 1.5 to 2.5 in their 
coefficients of uniformity, but the best sands for filter purposes 
should be less than 2.0. 
Mechanical Analysis of Sand.-Sand may be separated into 
its different-sized particles by means of a series of sieves having 
meshes varying in number from 10 to 200 per linear inch. 
Coarser sieves are used for the very coarse sands and the finer 
gravels. For sands finer than 0.10 mm. (sieve having about 
200 meshes per linear inch would separate this size), elutriation 
methods must be used, as it is difficult to make a much finer 
sieve than one with 200 meshes per inch with which results of 
any practical value can be obtained. 
By sifting a known weight of a dried sample of sand through 
a series of graded sieves, it is possible to calculate from the 
weights of sand left on the different sieves the proportion retained 
to that passed by any given size of mesh. Because of the varia- 
tion in the sizes of the openings in the wire cloth of any given 
mesh, Mr. Allen Hazen derives the size of separation of a sieve 
by taking the last particles of sand passing through it, and de- 
termining their weight and specific gravity. Assuming each 
particle to have a diameter of a sphere of equal volume, and taking 
the specific gravity of a quartz sand as 2.65, it is possible to 
calculate the mean diameters of the particles from the above 126 
WATER PURIFICATION 
data. Where d represents the diameter of the particle in milli- 
meters, and w its weight, then the size of separation may be 
obtained from the formula: 
d = 0.9^/ w. 
From actual measurements of the openings in the mesh by 
means of a microscope, using stage and ocular micrometers to 
obtain accurate readings, Mr. Philip Burgess obtains the size 
of separation of a sieve by multiplying the width of the space by 
a factor derived from a comparison with results obtained by 
Hazen's method. This factor is 1.095 for woven-wire cloth, 
and 1.19 for twilled-wire cloth. These factors were obtained 
after counting over 250,000 sand grains, which had been separated 
by Hazen's method of sifting, and then determining their weight 
and specific gravity. 
By a sufficient number of separations the percentages by weight 
of the sand smaller than the various sizes may be plotted to a 
uniform or to a logarithmic scale. From this curve intermediate 
points may be determined with a fair degree of accuracy, and 
the relative proportions of the variously sized particles be 
seen at a glance. From such a curve the effective size of the 
sand and the uniformity coefficient are most conveniently 
determined. 
Specifications for filter sand may be and usually are quite 
explicit, as the following specifications for supplying sand to the 
slow sand filters at Providence, R. I., will show: 
"The filter sand shall be clean bank sand, or equal, screened and 
washed, as may be necessary to secure the requisite cleanliness and 
grain sizes, and shall be entirely free from clay, dust, roots and other 
impurities. The grains shall, all of them, be of hard material which 
will not disintegrate, and shall be of the following diameters: Not 
more than one (1) per cent, less than thirteen-hundredths (0.13) of a 
millimeter, nor more than ten (10) per cent, less than twenty-six 
hundredths (0.26) of a millimeter. At least ten (10) per cent, by weight 
shall be less than thirty-four hundredths (0.34) of a millimeter; at least 
sixty (60) per cent, less than eighty-three hundredths (0.83) of a milli- 
meter, and at least eighty-five (85) per cent, less than two and ten 
hundredths (2.10) millimeters. No particle shall be more than five (5) 
millimeters in diameter, and the sand shall be passed through screens 
and sieves of such mesh as to stop all such particles, and no screen 
shall be used containing at any point holes or passages allowing grains SYSTEM OF SLOW SAND FILTRATION 
127 
five (5) millimeters to pass. The diameter of sand grains shall be com- 
puted as the diameters of spheres of equal volumes."1 
When sand is required to be washed, because of the clay or 
other fine soil it may contain, specifications have sometimes 
required that the washing shall be continued until a definite 
weight of the washed sand when shaken with a certain volume 
of clear water, does not produce more than a certain turbidity, 
measured according to the accepted silica standard scale. At 
Pittsburgh, Pa., it was required that 100 grams of the washed 
sand when shaken with 1 liter of water should not cause a tur- 
bidity greater than 200 parts per million. At Washington, D. 
C., a turbidity was not permitted which exceeded about 0.2 
per cent, of clay in the washed sand, or about 4,000 parts per 
million of turbidity. 
At Providence the sand specifications stated: "the filter sand 
shall not contain more than 1 per cent, of lime and magnesia 
soluble in dilute hydrochloric acid in 24 hr. at 70°F., taken 
together and calculated as carbonates." At Albany, N. Y., 
2 per cent, was made the limit for lime and magnesia and 
calculated as at Providence. 
Depth of Sand.-Since the sand bed offers the greatest resist- 
ance to the flow of the water, its depth should not be greater 
than will provide good purification of the water being filtered. 
Beds of fine sand may be of less depth than those of coarse sand, 
but the former produce greater losses of head than do the latter, 
and are likely to clog more quickly. With the average sand of 
0.35 mm. effective size, and a uniformity coefficient of 2.0, the 
depth may be from 30 to 40 in. This depth will allow for 
removals of more or less sand by scraping for the purpose of 
cleaning the filter bed. Deep beds of sand are likely to be less 
adversely affected by changing rates of flow or by scraping than 
shallow beds, and are, therefore, safer and easier to operate. 
Fine sands permit much less water to pass through them between 
cleanings of the bed by scraping than do coarse sands. The 
penetration of the impurities into the sand is also much less in 
fine sands than in coarse, and for that reason are somewhat more 
quickly and economically cleaned, since it is necessary to handle 
less sand when scraping the beds. 
In Germany the law does not permit a sand bed of a filter to 
1 Otis F. Clapp: " Description of Slow Sand Filtration Plant,Providence, 
R. I." 128 
WATER PURIFICATION 
!Fig. 33.-Detailed plan of piping in sand court, Washington, D. C., filtration plant. 
FILTER NO. 1 1 
FILTER NO. 16 
FILTER NO. 10 
FILTER. NO. 1 5 SYSTEM OF SLOW SAND FILTRATION 
129 
be used after the sand has been reduced by scraping to less than 
12 in. In actual practice, however, both in this country, and in 
Europe, the sand beds are usually considerably more than 12 in. 
in depth after the final scraping and before being refilled with 
sand to their original depth. 
Arrangement of Sand Courts and Controlling Chambers.- 
Slow sand filter (Fig. 33) beds are usually built side by side in 
two rows facing a common open area between them. In large 
plants a duplication of this arrangement is. carried out where 
possible, as it affords a convenient place for storing and handling 
the large amounts of sand which are employed in this process. 
Economy of construction is also obtained by this arrangement, 
since a common division wall between adjoining filters reduces 
the cost of masonry and embankments. Water to be filtered is 
readily distributed to the sand beds on either side of the court, 
and effluent piping for the filtered water is also most conveniently 
located in this same area. 
Controlling chambers, in which the piping from two or more 
filters are brought together, are usually placed on either side or 
in the center of the court as may be most convenient. Drain 
piping, sand bins, sand-washing devices and conveying apparatus 
are necessarily located within this space between the rows of 
filters. 
The use of this space for handling and storing sand has not 
always proven satisfactory, owing to the insufficient area pro- 
vided. At the filter plant in Albany, N. Y., the use of the court 
between the filters has been abandoned for washing or storing 
sand for this reason. Methods of washing the sand in place, or 
of immediate transfer of the washed sand to another filter, have 
obviated to some extent the necessity for sand courts. 
A somewhat unusual arrangement of filters exists in the Lower 
Roxborough filter plant at Philadelphia, where filters were placed 
in a series of steps owing to the topography of the site. Adjacent 
filters differ in level by 2 ft. 9 in. 
Depth of Water on Sand.-The depth of water over the sand 
layer varies considerably, but in modern plants ranges from 3 
to 4 ft., when the sand is at its maximum depth before scraping 
of the bed has commenced. The older plants sometimes carry 
a greater depth of water than 4 ft., but in such cases the method 
of controlling the outflow from the filter is crude and unreliable. 
From 5 to 7 ft., of water are carried over the sand at the Pitts- 130 
WATER PURIFICATION 
burgh slow sand filter plant, which seems to be considerably more 
water on top of the sand than is the usual custom (Fig. 34). 
Comparatively simple devices for regulating the inflow of 
water to the bed are now generally used and, when employed 
STANDARD SECTIONS 
OF 
WALLS AND VAULTING 
PltR 
SECTIONS OF WALLS USED FOR DIFFERENT WIDTHS OF EMBANKMENTS. 
DEEP FOUNDATIONS USED IN SHALLOW CUT OR ON FILL. RAISED FOUNDATIONS 
(DENOTED BY PRIMED LETTERS) USED IN DEEP CUT. 
Fig. 34.-Cross-section of filters, Washington filtration plant. 
DIVIDING WALL. 
SECTION G and G' 
COURT WALLS. 
SECTION A 
in addition to regulating devices for controlling the rate of dis- 
charge from the filter, make it possible to maintain a practically 
constant depth of water over the sand. With reliable controlling SYSTEM OF SLOW SAND FILTRATION 
131 
apparatus there is no reason why in most cases one-half to one- 
third the prevailing depths of water on the sand may not be 
used with results equally as good as are now obtained. 
Loss of Head.-The resistance offered to the passage of the 
water through the filter bed varies, being slight at first when the 
sand is clean, and gradually increasing as it becomes clogged 
with material strained out of the water and deposited on the 
surface of the bed and in the passages between the sand grains. 
The difference in level between the water on the top of the filter 
bed and the height to which the water will rise after issuing from 
the underdrains of the filter is the measure of this frictional resist- 
ance or the loss of head. The loss of head is measured by means 
of floats located in chambers connected with the water on the 
top of the filter, and the water discharging from the underdrains. 
These floats in their separate chambers are connected to indicat- 
ing apparatus, which may be placed in any convenient position 
for easy observation. 
Limits to the Loss of Head in Operation.-As the loss of head 
indicates the extent of clogging in the filter, it furnishes a method 
of determining how long the filter may be safely operated, pro- 
vided experience has shown that with any given water an increase 
in the loss of head beyond certain limits accompanies a deteriora- 
tion in the quality of the filtered water. The relation between 
loss of head and quality of water is by no means constant for 
any given water, and depends upon several factors, such as the 
character of the material being strained out of the water, the 
rate and fluctuations in the rate of filtration, the fineness and 
depth of the sand bed, and the disturbance of the filtering medium 
at and near the surface of the sand by the escape of air. Carry- 
ing the loss of head to a point where the mouth of the effluent 
pipe is uncovered, will permit air to enter the underdrains and 
work its way up through the bed. This is not of course per- 
mitted in well-operated plants. Air dissolved in the water 
may also be liberated in any zone of diminished pressure within 
the sand bed, and in moving upward through the sand bed 
produces similar disturbances to those caused by air entering 
the underdrains. 
In practice the loss of head permitted, before filters are with- 
drawn from service for cleaning, varies widely. In some cases 
only 18 to 24 in. are allowed, while in others 6 or 7 ft. may be 
reached. The average loss of head is probably from 4 to 5 ft., 132 
WATER PURIFICATION 
at least in American practice with modern regulating apparatus. 
In Europe, however, Mr. Allen Hazen noted a strong tendency 
to limit the loss of head to about one-half the average figures 
stated above. Only actual tests will in most cases show the loss 
of head which may be permitted, consistent with maintaining 
a high grade of filtered water and economical operation of the 
filters. Too frequent cleanings of the sand beds mean increased 
cost of maintenance, which can easily be brought about by arbi- 
trarily fixing a limit to the loss of head, above which the filter 
shall not be operated, without knowing whether a greater loss 
of head might not be utilized and still obtain a good quality of 
effluent. 
Causes for Deterioration in Quality of Water with High 
Loss of Head.-In some of the older slow sand filter plants, the 
filters were connected with the filtered water reservoir in such 
a way that fluctuations in the water level of the latter, due to 
increased or decreased consumption of water, produced corre- 
sponding changes in the rates of filtration. With clean filter 
beds fluctuations in the rate of filtering were not especially 
harmful, but as the loss of head increased the sudden changes in 
the rate were liable to cause the filters to discharge a poorer 
quality of water. The theory that the surface films on the sand 
layer were broken through by the increased pressure due to a 
high loss of head has been commonly held, especially in Europe. 
Mr. George W. Fuller1 in his report upon the operation of the 
experimental slow sand filters at Cincinnati, Ohio, makes a 
statement which seems to fit into modern theories of the action 
of filters as set forth in a preceding chapter. He states that 
"it may be noted that the increased losses of head occur at times 
when the retentive capacity of the sand layer is more or less ex- 
hausted; and, upon the application of very turbid water, the 
applied clay passes through the portion of the filter where the 
frictional resistance is the greatest" 
If the colloidal matter which is supported by the sand grains 
loses its "retentive capacity" by becoming surcharged with 
suspended particles, and is no longer able to hold these particles, 
whether it be by electrical attraction or by purely mechanical 
means, it necessarily follows that further additions of suspended 
particles will result in their passing through the minute capillary 
passages of the bed and their appearance in the effluent of the 
1 Report on Water Purification at Cincinnati, Ohio, p. 189. SYSTEM OF SLOW SAND FILTRATION 
133 
filter. This explanation holds as well for the minute clay par- 
ticles, which may be even smaller than the bacteria, as it does 
for the bacteria themselves. If a filter has been recently scraped, 
the colloidal matter, or real filtering medium, must be more or 
less disturbed and broken up. In this case there are very prob- 
ably "holes" in the filter, which must become sealed by the 
formation and deposition of new colloidal matter before the 
filter regains its original efficiency. In fact, a recently scraped 
filter exhibits in a limited way the same inability to strain out 
suspended matter, as does a filter filled with new sand which 
has never been used for filtering purposes. 
Viewed in the light of the above theory, the height to which 
the loss of head may be safely carried is only relative, as the 
quality of the effluent depends upon several factors, which may 
not be even constant for the same water at different periods of 
the year, and which certainly vary with different classes of waters. 
The inconstant relation between loss of head and the quality 
of the effluent is a fact well known to all operators of filters, 
and would seem to find a rational explanation in the theory 
suggested above. 
References 
1. Allen Hazen: "Filtration of Public Water Supplies." 
2. Turneaure and Russell: "Public Water Supplies." 
3. Slighter: "Water Supply Paper of U. S. Geological Survey," 1902. 
4. Annual Report U. S. Geological Survey, Pt. II, p. 295, 1899. 
5. Otis Clapp: "Slow Sand Filtration Plant at Providence, R. I." 
(Pamphlet). 
6. George W. Fuller: "Report on Water Purification at Cincinnati, 
Ohio." 
7. "Description of Filtration Works and Pumping Stations at Phila- 
delphia." Bureau of Water (Pamphlet, 1909). 
8. Allen Hazen and E. D. Hardy: "Works for the Purification of the 
Water Supply of Washington, D. C." Trans. Am. Soc. C. E., vol. 57, 
p. 307, 1906. 
9. George A. Johnson: "The Purification of Public Water Supplies." 
Water Supply Paper No. 315. U. S. Geological Survey. CHAPTER XII 
SYSTEM OF SLOW SAND FILTRATION (Continued) 
Influent Controlling Devices.-The water to be applied to a 
filter must enter at low velocity in order not to disturb the sand 
APPARATUS FOR CONTROLLING RATE OF FLOW 
Eler. C.L.(157.3)- 
16* Refill Eler. (156.3) 
.Venturi Meter 
20* Filtered-Water Effluent 
Eler. C.L.l 153.7) 
<Eler. (152.2) 
SECTION ON C-C 
'C.l. Drain Eler. of C.L. (151) 
Drj Chamber for Indicator Apparatus - 
Spec. Fa~ 
Main Effluent for Filtered-Water 
20*Filtered-Water Effluent 
NOTE: 
Detail arrangements differ slightly 
in the different houses. 
Venturi Meter - 
PLAN AND SECTIONS 
REGULATOR HOUSES 
~ 20 Flltered-Water Effluent 
20* Outlet for Filtered Water 
FLOOR PLAN 
Main Effluent for Filtered-Water 
Venturi Meter 
20* Flltered-Water 
Effluent। 
-20"Tlle Drain 
Fig. 35.-Regulator houses, Washington filtration plant. 
SECTION ON D-D 
C.l. Drain Eler. of O.L. (151) 
bed. If the pressure at the point of discharge into the filter 
is very low, no method of control may be necessary. If the 
pressure is too great, however, floats on the surface of the water 
in the filter are used to operate a simple butterfly valve, or a 
balanced valve of more elaborate design, located in the supply- 
134 SYSTEM OF SLOW SAND FILTRATION 
135 
pipe line. When the influent pipe discharged the water just 
above the surface of the sand, as was the case in some of the older 
plants, paving about the mouth of the pipe was utilized to pre- 
vent disturbance of the sand. In modern plants separate 
chambers and weirs from these chambers over which the water 
enters the filter proper at a low velocity, are usually provided. 
In the Pittsburgh1 slow sand filters the water is admitted 
through a 20-in. pipe line to an inlet chamber 6 ft. 9 in. by 9 ft. 
6 in. at the inlet corner of the bed. The tops of the walls of this 
well are at the minimum sand surface level in the bed, and when 
the sand is above that level, they are carried up to the surface 
of the sand with bricks laid dry. The controlling valve on the 
influent pipe is of the disk type, the water entering it at the side 
and flowing out both top and bottom. The float on the stem 
of this valve rests on the surface of the water over the sand in 
the filter, and within certain limits may be set at any desired 
height, which height determines the depth of water over the 
sand. 
Effluent Controllers.-In order to obtain a proper degree of 
purification of the water as it passes through the sand bed, it is 
necessary to permit it to flow at only relatively low rates. With- 
out artificially interposing a resistance to this flow at the outlet 
of the filter, a clean bed of sand would pass the water at too high 
a rate. A gradually increasing resistance to flow is produced by 
the material deposited on and in the sand bed, as explained in 
the preceding chapter. Hence, as this latter frictional resist- 
ance increases, the artificial resistance should be proportionally 
decreased, if a uniform rate of discharge is to be maintained. 
Numerous devices for maintaining a uniform rate of flow are 
in use, of which only one or two can be described. The principle 
involved in many of them is that of throttling the flow of water 
from the effluent main, so that a constant difference in head is 
maintained on the two sides of a fixed orifice or weir in an effluent 
chamber. This regulation may be effected by hand or be done 
automatically. In some devices there is no attempt to throttle 
the discharge, but in place of the stationary orifice a movable 
one is employed on which a constant head is kept by means of 
floats in an effluent chamber. 
At the Albany, N. Y., slow sand filter plant, a fixed orifice, 
on which a constant head is maintained by hand regulation of 
1 "Pittsburgh Filtration Plant." Eng. Record, vol. 54, No. 24, 1906. 136 
WATER PURIFICATION 
the effluent valve, is employed. At Hamburg, Germany, a 
movable weir is used, which can be adjusted by hand. Auto- 
matic regulation of the rate by means of movable weirs were 
provided at the Pittsburgh and Philadelphia plants. These 
weirs consist of telescopic tubes carried by floats. The upper 
end or mouth of the sliding tube can be carried at any fixed level 
below the surface of the water, and thus produce a constant dis- 
charge from the effluent chamber irrespective of the height of the 
water in the chamber, and indirectly independent of the loss 
of head. At the Pittsburgh plant, the float operates a valve in 
the effluent line, and thus controls the loss of head directly, so 
far as the introduction of frictional resistance to the flow of 
water leaving the effluent pipe could be effected. This auto- 
matic control at the Pittsburgh plant has been abandoned, as 
hand regulation has been found more satisfactory for the reason 
that the filters can be operated to a greater loss of head, and 
consequently with a greater yield of water between cleanings. 
In some of the more modern plants the use of the Venturi 
meter to measure the rate of flow, has become quite common. 
These meters are employed at the slow sand filter plant in Wash- 
ington, D. C. Regulation is effected by hand and not automatic- 
ally. By utilizing the difference in pressure between that on the 
upstream side of the throat of the Venturi tube, and that at the 
throat, automatic regulation may be effected through auxiliary 
apparatus. This is done in several of the Venturi type of rate 
controllers now in use in many of the rapid sand filter plants, 
where the use of rate controlling apparatus of an automatic 
character is quite essential to the efficient operation of the plant. 
These rate controllers will be more fully described in a succeed- 
ing chapter. 
Sand Cleaning Methods and Apparatus.-When a slow sand 
filter has been in service long enough to become clogged with 
material removed from the water during filtration, it becomes 
necessary to remove this material in order to maintain the quan- 
tity and quality of the effluent. Various methods are employed 
for cleaning the surface layers of sand, all of which are based 
on washing a thin surface layer of sand with clean water, and 
discharging the dirty wash water into the sewer. This washing 
may be effected by flowing water over the sand bed, and at the 
same time raking it with rakes to a depth of several inches, or what 
is more common the removal of the surface layer of dirty sand to SYSTEM OF SLOW SAND FILTRATION 
137 
a depth of to in., and its subsequent washing in special 
apparatus designed for this purpose. When the sand is to be 
taken out of the filter the dirty surface layer is scraped off with 
thin flat shovels, and deposited in piles. The dirty sand is then 
either wheeled out of the bed in wheel-barrows, or moved in 
small, tram cars to a court where it is washed. In large modern 
plants (Fig. 36) the sand is usually shovelled into sand ejectors 
and transported by water under pressure through piping to 
washing machines outside the filter. In some of the plants re- 
WASHING AND STORAGE OF SAND 
Fig. 36.-Method of washing and storing sand, Washington filter plant. 
cently constructed, the dirty sand is washed by machines mov- 
ing slowly over the surface of the sand bed, or by movable wash- 
ers placed on the sand bed during the scraping process. 
Surface Washing of the Sand Bed.-Over 20 years ago Piefke1 
described a method for cleaning slow sand filters which consisted 
in stirring up the surface layer of sand while a thin sheet of 
water was allowed to flow rapidly over the sand to a drain. The 
experimental results were quite satisfactory when the filter was 
properly arranged for carrying out the method. Within the last 
few years this method has been tried again with considerable 
success in some of the small filter plants of the New York Water 
Department in Long Island in the Borough of Brooklyn. It has 
been termed the "Brooklyn method," and is carried out after 
draining down the filter until only 2 or 3 in. of water remain 
above the sand. Drains just at the sand level are then opened 
at one end of the bed, and the wash water allowed to flow over the 
surface of the bed from an inlet at the opposite end. Boards 
1 Vierteljahresschrift fUr offentliche Gesundheitspflege," 1891, p. 59. 138 
WA TER P URIFICA TION 
are used to form channels about 15 ft. wide, through which the 
water may be directed, and to increase its velocity. While the 
water is passing through these channels the sand is raked in 
order to loosen the dirt, so that it may be washed away by the 
water flowing over the surface of the bed. 
This method appears to be quite successful with filters clogged 
with certain kinds of light organic matter, such as might arise 
from a growth of microorganisms, for example, or from other 
vegetable matter; but for waters carrying any clay sediment, 
which penetrates the sand to some depth, it has not proven satis- 
factory. At the Pittsburgh slow sand filter plant, this method 
has been tried, but was abandoned on account of the imperfect 
washing which the sand received. At the Torresdale slow sand 
filter plant in Philadelphia, this method has also been used, 
but the deep penetration of the sediment produced shorter 
and shorter periods of service of the filters and has been 
discarded. 
Old Methods of Sand Washing.-There have been many 
methods devised for cleaning dirty sand which has been removed 
from filter beds, ranging from simple washing with a hose stream 
to mechanically driven drum washers. One of the earliest de- 
vices consisted of a box with a perforated false bottom, through 
which the wash water was forced. The dirty sand shoveled 
into the box was stirred in order to assist the rising water to carry 
off the dirt. Washing the dirty sand as it lay on an inclined 
platform of plank, brick or iron with a hose stream has also been 
used. From 2 to 4 cu. yd. of sand were washed at a time. The 
dirty wash water flowed over a weir at the foot of the incline and 
from thence to a drain. 
Mechanically driven drum washers of various designs have 
been very generally used in Germany. The cylinders were open 
at both ends; some of them were slightly conical in shape and 
were placed with their axes in a horizontal position. They were 
provided with screw blades on the inside which moved the sand 
from the lower to the higher end during the rotation of the drum. 
A stream of water was used to wash the sand as it worked its 
way upward through the cylinder. In some cases a narrow cyl- 
inder was used in which the sand was advanced by means of a 
screw in a direction opposite to a flowing stream of water. From 
2.5 to 4 cu. yd. of sand could be cleaned per hour, and the quan- 
tity of water used ranged from 1,500 to 2,500 gal. per cubic yard SYSTEM OF SLOW SAND FILTRATION 
139 
cleaned, according to the type of machine employed. From 2 
to 4 hp. were required to rotate these drums. 
Present Methods of Sand Transportation and Washing.- 
Although ejector (Figs. 37,37a and 38) sand washers have been used 
PORTABLE SAND EJECTOR 
SECTION ON C-D 
Female Hose Connection 
V alve 
Female Hose Connection 
Standard Screw Flange 
Fig. 37.-Portable sand ejector. 
SECTION ON AB 
PLAN 
To be turned 
Russia Iron, 
Standard Screw Flange^ 
Iron Strep 
to Hold 
Oak Bar- 
for many years, they have been the subject of a great deal of 
study within a comparatively recent period. They were first 
employed in England, and are now very generally used both for 
moving and washing sand. The hydraulics of ejectors and of the 
transportation of sand through pipe lines by water under pressure, 140 
WATER PURIFICATION 
has been carefully studied by Messrs. Hazen, Hardy, Longley, 
Knowles and Rice. 
For removing the sand from the filter, the latter is piped with 
high-pressure water mains (3-in. and 4-in. pipes), and with 
another set of pipes for carrying the sand. A portable ejector 
connected with the water and sand pipe lines is placed on the 
surface of the sand, and the piles of dirty sand which have been 
scraped up, are shoveled into it. The sand passes through the 
% tl Iron 
Ma Steel, 
Steel 
2M* Flange union 
2 Flange union 
Threaded to 
fit hose 
Threaded to 
fit hose 
KJ Steel 
PLAN AND SECTION ON AB 
Fig. 37a.-An easily made sand ejector. 
pipes with the water to ejector washers and is cleaned. It is 
then returned to the beds directly or is stored in courts or bins 
outside the filters. 
The usual form of portable ejectors for removing the sand 
scraped from the bed is that of a cast-iron bowl with a sheet-iron 
extension of the sides for holding the sand. The ejector nozzle 
may be made of hardened tool steel, the length of the wearing 
part being in., the outside diameter in., and the bore 
% in. The throat of the ejector is of chilled cast iron and about 
7 in. long. The inner portion is a Venturi tube with the smallest 
diameter 1% in. In a distance of 5 in. the diameter increases 
from 1% to 2% in. SYSTEM OF SLOW SAND FILTRATION 
141 
At the Pittsburgh slow sand filter plant 3-in. rubber hose is 
used to convey the sand to the iron pipe lines, and although the 
I 
SECTION ON C-D 
Auxiliary Jet J 
Fig. 38.-A sand washing hopper. 
SECTION ON E-F 
Holes for Bolts 
-I Beams 
SECTION ON A-B 
-Cust Iron 
brass couplings and expansion rings have been found to wear, 
the inner rubber tube of the hose is not apparently eroded. 
Wrought-iron pipe 3 and 4 in. in diameter has also been much 142 
WA TER P URIFICA TION 
used at this plant, and after several years' usage does not show 
appreciable signs of wear in the straight lengths. At the curves, 
however, wear on the piping has been noted, and by substitut- 
ing rubber hose at the bends for the curved pipe in several 
instances, less trouble from erosion has resulted. 
These sand ejectors will handle 8 or 9 cu. yd. of sand per hour 
with water pressures of 85 to 90 lb. per square inch. At the 
Pittsburgh filter plant, Mr. C. F. Drake states the cost during 
the first 7 months of 1910 for water for sand ejecting, washing, 
transporting and restoring to have been approximately 12 cts. 
per cubic yard of sand handled. 
Sand-washing Machines.-In order to avoid the removal of 
the dirty sand from the filter and the return of the cleaned sand 
to it, a number of devices have been invented for washing the 
sand practically in place. The Nichols separator or sand washer 
is a portable device which may be placed upon the surface of the 
sand in the filter. The bed is scraped in the ordinary way, and 
the dirty sand is shoveled into a portable ejector and conveyed 
by water under pressure to the separator. The latter consists 
of a closed steel cylinder 42 in. in diameter and 36 in. high, and 
having a cone-shaped bottom. To this cone-shaped bottom 
are attached water-pressure and sand-discharge pipe lines. The 
dirty sand enters near the top of the separator and is met by a 
rising current of water. As the latter ascends the sand descends, 
and is forced out of the bottom of the separator in a clean condi- 
tion. The dirty wash water is carried away at the top through 
hose connections to the drain pipes of the filter. The loss of 
sand from the separator consists of the finer material which is 
carried away by the rising wash water. It constitutes about 
3 per cent, of the sand washed in practice. It is kept at a 
minimum by properly proportioning the volume of sand and of 
wash water, and by means of a disk and series of baffles provided 
within the separator. The thickness of the sand bed is not 
reduced where these machines are used, as the sand is imme- 
diately returned to the bed and can be spread at once if desired. 
Transportation of the sand for long distances to stationary 
washers outside of the filters is avoided by using these machines. 
The separator weighs about 700 lb.; it can handle about 10 
cu. yd. of sand per hour, and requires about 1,200 gal. of water 
for washing each cubic yard of sand. The water pressure re- 
quired to operate the machine is about 65 lb. per square inch. SYSTEM OF SLOW SAND FILTRATION 
143 
The Nichols separator (Fig. 39) is used at the Philadelphia slow 
sand filter plants where it has given very satisfactory results. 
The depth of sand cleaned by these machines at this plant 
depends upon the penetration of the sediment, and may vary 
from 2 in. to 10 in. 
Fig. 39.-The Nichols sand-washing machine. 
Blaisdell Washing Machine.1-This apparatus (Fig. 40) con- 
sists of an inverted box about 4 ft. square and 2 ft. deep, which is 
sunk in the water on the filter to the surface of the sand, and is 
supported from above by an iron framework on which is a plat- 
form for operating the machine. The framework rests on wheels 
running on rails supported by the walls on either side of the filter. 
Within the box is a revolving rake, which consists of a hollow 
axle from which project hollow teeth, and through which water 
is forced under a pressure of 10 to 20 lb. per square inch when 
the axle is revolved by an electric motor. The box may be 
moved vertically or horizontally by means of a motor. 
In the operation of the machine, it is slowly moved along the 
rails by means of an electrically driven motor, causing the box 
with its revolving rake to slide over the surface of the sand bed. 
1 "Water Filtration and the Mechanical Washing of Filter Sand at Wil- 
mington, Del." Engineering-Contracting, Aug. 26, 1914. 144 
WATER PURIFICATION 
A suction pump draws away from the interior of the box slightly 
more water than is being discharged through the teeth of the 
revolving rake. The dirty wash water from the pump runs 
to waste through a gutter along the side of the filter. 
The machine may be used in filters designed for washing by 
this method and is probably more applicable to small than to 
large units. It has the advantage of cleaning the sand without 
Fig. 40.-Blaisdell's sand-washing machine. 
removing it and without even draining the filter, both of which 
operations materially decrease the net yield of filters cleaned by 
some of the other methods. This apparatus is used at the slow 
sand filter plant at Wilmington, Del., and in the preliminary 
filters at the Belmont plant in Philadelphia. 
Methods for the Partial Cleaning of Slow Sand Filters.-The 
presence of organic matter in a water in the form of microscopic 
organisms or of mineral matter in a fine state of division fre- 
quently clogs the sand of a filter very rapidly. These adverse 
conditions not infrequently occur with little warning to those 
in charge of a plant, and necessitate resorting to rapid methods 
for loosening up the sand in the beds in order to maintain a 
supply of filtered water. The efficiency of the filtering process 
may have to be sacrificed to some extent in such cases, and the 
methods employed are regarded as only temporary expedients. 
Raking.-In this method the filter is partially drained, and the 
sand is then lightly raked with garden rakes having teeth about SYSTEM OF SLOW SAND FILTRATION 
145 
in. long. Four men will rake an acre bed in about 8 hr. at 
a cost of about $8. The filter can be refilled immediately and 
filtration resumed. The prolongation of the period of service 
by raking depends upon the character and penetration of the 
clogging material. The second raking is less effective than the 
first, and the third is usually of little value. Much deeper 
layers of sand have to be removed to be washed when raking 
has been resorted to between cleanings. The bacterial efficiency 
of the filter is not as much affected as one would suppose. 
Spading.-Compacting of the surface of the bed may, under 
some conditions, proceed so far as to necessitate more effective 
loosening of the sand than could be obtained by raking. By 
going over the surface of the bed with a spade, pushing it down 
into the bed to a depth of 6 or 8 in., and at the same time working 
it backward and forward to break up the. surface sand layer, a 
slight benefit may be obtained. It is obvious that a sand bed 
requiring such measures to increase its output, would produce 
a poorer quality of water, and that the effect of the spading would 
not last long. Such has been the experience at the Torresdale 
filter plant at Philadelphia. 
Scraping and Piling.-Emergency conditions sometimes make 
the scraping, washing and resanding of a filter difficult, when the 
time of its withdrawal from service, which these processes in- 
volve, is considered. In such cases scraping off a layer of sand 
and piling it on the bed, until such time as it could be washed, 
has been tried with fairly good results. After scraping and pil- 
ing, the bed may be put back into service at once. Good results 
have been obtained by this method at the Torresdale filter plant 
in Philadelphia, but it of course is only a makeshift for delaying 
the complete process of cleaning until a more convenient season. 
Resanding Filters.-On account of the rather high cost of 
filter sand of good quality, it is almost invariably cleaned and 
returned to the beds for further use. Mr. George A. Johnson 
cites the only instance of which he has any knowledge where 
dirty filter sand is not cleaned, but is discarded. This is at 
Osaka, Japan, where low prices of labor (1 to 2 cts. per hour) 
enable clean dredged river sand to be pan-screened and prepared 
for the filter beds at a price of $0.65 per cubic yard. This is a 
price from one-fourth to one-half that for which sand may be 
usually obtained under favorable conditions. 
In those cases where the sand is removed from the filter for 146 
WA TEH P URIFICA TION 
cleaning, the beds are reduced in thickness by successive clean- 
ings. Never less than 12 in. of sand are allowed in slow sand 
filters, and usually the bed is quite a little thicker than this after 
the final scraping, and at the time the bed is refilled with sand to 
the original depth. When the sand is stored in courts or bins 
outside the filters, it is sent by means of ejectors and piping to 
the bed to be refilled. It is then spread by men with shovels. 
It is a common practice to send the dirty sand from a filter 
being scraped to the ejector washers, and from them to a filter 
to be refilled. No storage of sand is required in such cases, if 
the sequence of cleanings and refillings, and the arrangement of 
piping makes it possible. This is the practice at the Pittsburgh 
slow sand filter plant. At this plant the original elaborate sand- 
distributing machines have been abandoned, as it has been found 
much cheaper to spread the sand by hand than to operate the 
machines. 
At the Washington, D. C., filter plant, the average depth of 
sand replaced in refilling the filters during the period 1905-12 was 
14.5 in. The yearly averages during this period ranged between 
7.9 and 22 in. 
In those plants where the sand is cleaned in place, as for exam- 
ple with the Nichols separator, only respreading of the sand by 
hand is required. Where the Blaisdell sand-cleaning machine 
is used even respreading of the sand is unnecessary. The same 
is also true where the "Brooklyn method" of cleaning the sand 
is used, although this method of washing the sand is more prop- 
erly classed with those partial methods of cleaning described above. 
An excellent idea of the amount of sand which it is necessary 
to handle in operating slow sand filters may be obtained from 
data taken from the 1912 report of the operation of the Wash- 
ington, D. C., filter plant. 
Fiscal year 
Million gallons pumped to 
Number of filters 
Cubic yards of sand 
Filters 
Sand 
washers 
Scraped 
Raked 
Removed 
and 
washed 
Replaced 
in filters 
1907-08 
23,758.24 
76.98 
126.67 
86 
22,275 
21,393 
1908-09 
22,435.16 
119.16 
137.80 
36 
24,683 
23,487 
1909-10 
21,605.44 
91.78 
87.20 
51 
15,505 
16,876 
1910-11 
22,039.95 
79.92 
91.00 
71 
14,941 
13,218 
1911-12 
22,668.40 
84.86 
104.40 
83 
14,820 
16,095 SYSTEM OF SLOW SAND FILTRATION 
147 
COST OF CONSTRUCTION OF SLOW SAND FILTERS 
The cost of constructing slow sand filters varies greatly. It 
depends upon local conditions to such an extent that comparative 
figures may be misleading, unless given in considerable detail. 
It is not always possible to separate the costs of the various parts 
of a plant, owing to the manner in which the original cost ac- 
counts were kept. Changes in the cost of labor and material 
naturally influence the actual cost of plants which have been 
built at different times during the past 20 or 25 years. 
The cost per acre for slow sand filters will be found to range usu- 
ally from $45,000 to $75,000. Covered filters will generally cost 
about 50 per cent, more than open filters. An average cost for 
slow sand filters will be found to be about $60,000 per acre. In 
exceptional cases the above figures may be much less or much 
more than those stated above. The filter at Lawerence, Mass., 
built in 1893, cost about $27,000 per acre of filtering surface. 
Mr. G. A. Johnson1 states that an uncovered filter built in 1903 
is Osaka, Japan, cost $31,000 per acre. At the other extreme 
in costs may be cited some of the Philadelphia slow sand filter 
plants. The Torresdale covered filters cost about $145,000 per 
acre, and the Belmont covered filters approximately $187,000 
per acre, which latter, however, includes the cost of a clear-water 
reservoir of 16,500,000 gal. capacity. 
Detailed costs of the Albany, N. Y., Washington, D. C., and 
Pittsburgh, Pa., slow sand filter plants, as stated by their de- 
signers and as tabulated by Mr. G. A. Johnson in the above- 
mentioned paper, are as follows: 
Detailed Costs of Slow Sand Filtration Plants 
Albany, N. Y. 
Daily capacity 15,000,000 gal. 
Land $8,290 
Pumping station and intake, complete 49,745 
Filters, eight beds, each 0.7 acre in area; sedimentation basin, 
capacity 37,000,000 gal.; and filtered-water reservoir, 
complete 324,217 
Conduit and connections with Quackenbush Street Pumping 
Station 86,638 
Engineering and contingencies 31,000 
Total approximate cost of works $499,890 
1 G. A. Johnson: Water Supply Paper, U. S. Geological Survey, No. 
315/1913. 148 
WA TER P ERIE IC A TION 
Cost of filters per acre $45,600 
Cost of uncovered sedimentation basin per million gallons 
capacity 4,100 
Cost of filtered water reservoir per million gallons capacity. . . 15,000 
Cost of filters per million gallons gross daily capacity 15,200 
Cost of works per million gallons daily capacity 33,320 
Washington, D. C. 
Daily capacity 75,000,000 gal. 
Land $619,900 
Pumping station, including intake, Venturi meter, electric 
generating plant, stack, etc., complete 183,600 
Twenty-nine covered filters, each 1 acre in area, complete 2,197,000 
Filtered-water reservoir, capacity 14,200,000 gal., including 
gate house and regulating apparatus, complete 150,000 
Lower gate house and pipe line 24,300 
Engineering and clerical work 181,500 
Total cost of works $3,356,300 
Total cost excluding land $2,736,400 
Cost of filters per acre 75,700 
Cost of filtered water reservoir per million gallons capacity... . 10,600 
Cost of filters per million gallons gross daily capacity 25,250 
Cost of plant per million gallons daily capacity 44,750 
Pittsburgh, Pa. 
Daily capacity 120,000,000 gal. 
River crossing and connections $298,589 
Low-lift pumping station .. 357,513 
Low-lift pumping machinery, boilers, etc 284,169 
River-well and intake 207,673 
Brilliant pumping station (additional machinery) 221,472 
Pipe lines to Highland reservoir 555,250 
Filters (46 1-acre units, covered),1 and open sedimentation 
basins 3,586,245 
Filtered-water reservoir, 45,000,000 gal. capacity 460,060 
Total cost of works $5,970,971 
Cost of filters (including settling basins) per acre 78,000 
Cost of filtered-water reservoir per million gallons capacity.... 10,200 
Cost of filters (including settling basins) per million gallons 
daily capacity 26,000 
Cost of plant per million gallons daily capacity 42,600 
Note.-Above costs compiled from the award of the arbitrators to the 
contractors in July, 1910, and does not include engineering and a small 
portion of day work done by the City of Pittsburgh. 
1 Ten more acres of filters have been built since the first 46 filters were 
built, but have never been put in service. SYSTEM OF SLOW SAND FILTRATION 
149 
References 
1. Report Board of Water Commissioners of Wilmington, Del., for 
1911-12. 
2. Report Bureau of Water of Pittsburgh, Pa., for 1912. 
3. Report upon the Maintenance and Repair of the Washington Aqueduct, 
D. C., and Washington Aqueduct, D. C., Filtration Plant. Appendix 
DDD, 1912. 
4. G. A. Johnson: Water Supply Paper No. 315, U. S. Geological 
Survey, Dept. Interior. 
5. Trans. Amer. Soc. C. E., vol. 43, 1900; and vol. 57, 1906; pp. 294 
and 307, respectively. 
6. Engineering Record: 
{a) "Pittsburgh Filtration Plant," vol. 54, No. 24, 1906. 
(5) "Lower Roxborough Filters," Dec. 8, 1900 and Jan. 3, 1903. 
(c) "Upper Roxborough Filters," Apr. 13, 1901, Mar. 28, and Apr. 11, 
1903. 
(d) "Belmont Filters," Sept. 12, 1903. 
(e) "The Bernhart Sand Filters, Reading, Pa.," vol. 60, Nov. 13, 1909. 
(/) "Toronto Water-purification Works," vol. 60, Sept. 18, 1909. 
(#) "Purifying St. Lawrence River Water at Ogdensburg, N. Y.," vol. 
64, p. 642. 
(h) "Notes on the Wilmington Filters," vol. 63, p. 310, 1911. 
(i) "The Didelon Filter-bed Regulator," vol. 57, Apr. 11, 1908. 
(j) J. H. Gregory: "Slow Sand and Mechanical Filter Costs," vol. 
69, June 20, 1914. 
7. G. A. Johnson: "Present-day Filtration Practice." Jour. Am. Water- 
works Assn., 1914. 
8. Nicholas S. Hill, Jr.: "Modern Filter Practice." Jour. Am. 
Water-works Assn., June, 1913. CHAPTER XIII 
EFFICIENCY AND COST OF OPERATION OF SLOW SAND 
FILTERS 
EFFICIENCY OF SLOW SAND FILTERS 
The physical properties and sanitary quality of waters after 
passage through slow sand filters depend upon the character of 
the applied water, and upon the skill with which the plant is 
operated. When finely divided suspended matter in the raw 
water is applied to the filters for too long a period, an effluent of 
unsatisfactory appearance is sure to result. Usually such con- 
ditions also produce an effluent of high bacterial content, which 
may be unsafe, or at least undesirable to supply for consumption. 
Difficulties with microorganisms, which produce odors that are 
carried through the sand with the filtered water, are also 
sometimes encountered. 
In operating slow sand filters the limitations of their efficient 
action must be thoroughly appreciated. Only by the most 
careful observation of each filtering unit during its period of 
service, and by a comprehensive system of records showing the 
performance of the plant as a whole, can it be expected to supply 
a water of uniformly high quality. Even with the most con- 
scientious attention, plants which are incorrectly designed 
or inadequately provided with methods for supplying a water 
properly prepared for the filters may become difficult to operate, 
and may produce effluents of an unsatisfactory character. 
Periods of Service and Yield.-The length of time which slow 
sand filters may be operated efficiently varies greatly. The yield 
of the filters as regards quantity, as well as quality, depends upon 
the frequency of cleaning and upon its thoroughness. These 
points are well illustrated in the following table taken from a 
paper1 of Messrs. Francis D. West and Joseph S. V. Siddons on 
the operation of the Torresdale plant at Philadelphia. 
1 Jour. New England Water-works Assn., vol. 27, Sept., 1913, p. 358. 
150 OPERATION OF SLOW SAND FILTERS 
151 
Results of Cleaning Slow Sand Filters by Various Methods 
at Torresdale Filtration Plant, Philadelphia, Pa. 
No. of 
filters 
cleaned 
Yield per filter 
Yield per 
acre, 
millions 
of 
gallons 
Rate per 
acre 
Period of service, days, 
Ave. 
Max. 
Min. 
Average 
Average 
Ave. 1 Max. 
1 
Min. 
65 
40 
73 
21 
53 
First p 
1.6 
eriods of service 
43 1 77 
11 
609 
19 
67 
4 
C 
25 
leaning 
1.8 
jy Brookyln method 
14 I 49 | 6 
530 
35 
69 
17 
( 
47 
cleaning by scraping and piling 
2.4 I 19 I 39 10 (a) 
122 
40 
142 
3 
53 
4.0 
14 57 
2 (b) 
282 
39 
80 
19 
C 
52 
leaning by ejection from fill 
2.6 1 20 I 45 
er 
10 (a) 
235 
116 
379 
10 
155 
4.5 
34 1 105 
4 (b) 
1,012 
85 
182 
3 
113 
Clean 
5.0 
ing by raking 
24 1 50 
2 (1st)* 
408 
50 
79 
16 
67 
4.3 
16 | 24 
7 (2d)* 
62 
52 
39 
13 1 
.. (3d)* 
1,029 
141 
569 
49 
Cleaning by 
188 I 4.4 
Nichols' sand separator 
42 | 148 111 
1,486 
30 
80 
4 
40 
General averages. 
2.1 1 18 1 77 
6 (a) 
2,958 
96 
569 
3 
128 
4.4 
29 148 
2 (b) 
Notes.-(a) Before pre-filters were placed in service; (b) after pre-filters were placed 
in service. Asterisk refers to results obtained after "first raking," "second raking" and 
"third raking," respectively. Data summarized through 1912, except the maximum, 
minimum and rate of filtration data under "rakings," and a method of spading in 1912 are 
omitted from the averages. 
The authors conclude that: 
"A plant constructed as Torresdale, without any sedimentation 
basin, is utterly unable to cope for any prolonged period with water 
having a turbidity of over 100, that is, with the slow sand filters operat- 
ing at a 6,000,000-gal. rate. When such a condition is reached, the 
pre-filters fail to do their proportion of the work, and the final filters 
choke badly, allowing fine silt to pass through them. This choking 
necessitates cleaning for 24 hr. a day with 55 to 58 filters doing the work 
of 65, and depending on calcium hypochlorite to reduce the number 
of bacteria and destroy the pathogens. Fortunately the periods of 
turbid water occur but seldom and are of short duration." 
In spite of these occasional adverse conditions the average 
bacterial efficiency of this plant is over 99.5 per cent. 
Removal of Sediment and Bacteria.-The number of bacteria 
removed by well-operated slow' sand filters usually ranges between 
98 and 99 per cent, of the number originally present in the raw 
water. If preliminary treatment of the raw water, by some of WATER PURIFICATION 
152 
the various methods previously described, is employed, the 
effluent of the filters is likely to contain less than 100 bacteria 
per cubic centimeter for a greater part of the time. The turbidity 
of the applied water greatly influences the number of bacteria 
found in the effluent. This is well shown by the following tables 
taken from Messrs. West's and Siddons' paper on the operation 
of the Torresdale filters which was referred to in a preceding 
paragraph. 
Table Showing Range in Turbidity, 1912 
Turbidity 
Delaware 
Kiver 
Effluent 
of roughing 
filters 
Turbidity 
Filtered-water 
basin 
Parts per million 
Days 
Days 
Parts per 
million 
Days 
0-10 
154 
297 
0-1 
339 
11-25 
146 
41 
2-10 
18 
26-50 
38 
8 
11-25 
7 
51-100 
9 
6 
25-51 
2 
101-250 
9 
6 
Over 51 
0 
251-500 
3 
6 
501-750 
4 
2 
751-1,050 
Over 1,050 
3 
0 
0 
Bacteria 
Delaware 
river 
Effluent of 
roughing filters 
Bacteria 
Filtered, water 
basin 
Days 
Per c.c. 
Days 
Days 
Per c.c. 
0-100 
0 
0-5 
85 
101-200 
0 
6-10 
101 
201-300 
0 
1 
11-25 
107 
301-400 
0 
3 
26-50 
34 
401-500 
0 
6 
51-100 
15 
501-750 
1 
19 
101-200 
12 
751-1,000 
6 
55 
201-500 
4 
1,001-1,500 
16 
44 
501-1,000 
5 
1,501-2,000 
36 
42 
1,001-2,100 
2 
2,001-2,500 
27 
30 
Over 2,100 
0 
2,501-5,000 
92 
93 
5,001-10,000 
86 
36 
10,001-25,000 
67 
28 
25,001-50,000 
22 
6 
50,001-100,000 
10 
2 
Over 100,000 
3 
0 
Table Showing Range in Bacteria, 1912 OPERATION OF SLOW SAND FILTERS 
153 
The results obtained at the Washington filtration plant also 
show the difficulties encountered in the operation of slow sand 
filter plants when treating turbid waters. A series of large 
reservoirs, through which the Potomac River water passes before 
it reaches the filter plant, afford ample periods of sedimentation 
and storage, and make it possible to keep out of the system quite 
a large volume of the muddiest water which would otherwise have 
to be filtered. During the fiscal year July 1, 1911, to June 30, 
1912, the inlet gates from the river at Great Falls were closed 
21.14 per cent, of the time, and excluded 30.28 per cent, of the 
total suspended matter. Treatment with alum and subsequent 
settlement was also practised in order to reduce the turbidity 
of the water applied to the filters. The average turbidity in the 
filtered water for the month of January, 1912, was 1 part per 
million, although a maximum of 4 parts per million were present 
at times during the month. Before coagulation of the sedi- 
ment in the water with alum was attempted, the maximum tur- 
bidity of the filtered water sometimes reached 20 parts per 
million; and the yearly averages for the years 1906-10 were 
between 1 and 2 parts per million. 
The bacterial removals by this plant are exceptionally high, 
and are best shown by the following table of yearly averages: 
Average Number of Bacteria per Cubic Centimeter in Settling 
Reservoirs and in Filtered-water Reservior1 
Year 
Dalecarlia reservoir 
Georgetown 
reservoir 
outlet 
McMillan 
Park reservoir 
outlet 
Filtered- 
water 
reservoir 
Inlet 
Outlet 
1906-07 
4,850 
1,940 
1,680 
635 
31 
1907-08 
6,300 
2,700 
2,940 
1,250 
55 
1908-09 
3,130 
1,950 
950 
390 
21 
1909-10 
14,300 
13,850 
10,850 
6,820 
143 
1910-11 
4,820 
3,370 
2,080 
1,390 
38 
1911-12 
8,200 
6,000 
2,600 
1,100 
35 
The following table taken from the same report, indicating 
the percentage of positive tests for the Bacillus coli found in 
testing 1 c.c. of water from samples taken from the various parts 
of the system of reservoirs and from the effluent from the filter 
plant is also of interest: 
1 Report upon the Maintenance and Repair of the Washington Aqueduct, 
District of Columbia, and Filtration Plant, 1912. 154 
WATER PURIFICATION 
Percentage of Positive Tests for the Bacillus Coli Found in 1 C.C. 
of Water from Samples from Settling Reservoirs and Filter Plants 
Year 
Dalecarlia ressrvoir 
Georgetown 
reservoir 
outlet 
McMillan 
Park reservoir 
outlet (applied 
water) 
Filtered-water 
reservoir 
Inlet 
Outlet 
1905-06 
19.4 
23.2 
14.9 
8.3 
1.8 
1906-07 
43.6 
29.2 
29.8 
13.0 
2.1 
1907-08 
31.3 
12.3 
22.1 
9.4 
0.3 
1908-09 
20.3 
15.0 
8.5 
7.1 
0.0 
1909-10 
26.9 
24.0 
19.8 
10.4 
0.8 
1910-11 
16.8 
11.5 
3.6 
2.2 
0.0 
1911-12 
43.5 
34.3 
21.1 
10.5 
0.8 
At the Wilmington, Del., slow sand filter plant, the raw water 
does not carry so much sediment but is very high in bacteria. 
The following table condensed from the annual report for 1911-12 
of the Water Commissioners of Wilmington, shows the efficiency 
of this plant: 
Averages for 1911-12 
Raw water 
Pre- 
filtered 
water 
Applied 
water 
Combined 
effluents 
Turbidity, parts per million 
68 
36 
30 
2 
Bacteria per cubic centimeter 
49,533 
29,816 
21,795 
472 
Percentage efficiency 
39.81 
26.91 
99.05 
Percentage positive B. coli tests in 
1 c.c 
99.3 
94.40 
89.90 
21.90 
The wide variation in the results obtained at different purifi- 
cation plants indicates great differences in the amount of pollu- 
tion with which the plants have to contend, and also the effect on 
the water of various methods of treatment prior to its final pas- 
sage through the filters. 
The maximum, minimum and average reductions of both the 
turbidity and the bacteria by each of the slow sand filter plants 
of Philadelphia are well shown in the following table taken from 
the Report of the Department of Public Works, Bureau of Water, 
for the year 1912. The comparison is made between the water 
applied to the filters and their effluents. OPERATION OF SLOW SAND FILTERS 
155 
Source of supply 
Schuylkill River 
Delaware River 
Lower 
Rox- 
borough 
Upper 
Rox- 
borough, 
Belmont 
Queen 
Lane 
Torresdale 
per cent. 
per cent. 
per cent. 
per cent. 
per cent. 
Turbidity: 
Average reduction 
98.85 
99.15 
98.95 
91.25 
93.74 
Maximum reduction 
100.00 
100.00 
100.00 
97.50 
100.00 
Minimum reduction 
97.37 
90.00 
95.45 
88.33 
Bacteria: 
Average reduction 
99.49 
99.63 
98.96 
96.42 
99.03 
Maximum reduction 
Minimum reduction 
99.97 
96.91 
99.96 
97.50 
99.95 
95.05 
99.69 
75.00 
99.90 
96.90 
Note.-The report does not state whether disinfectants were used or not, 
but if they were used the bacterial reductions are probably the result of 
filtration and disinfection combined. 
COST OF OPERATION OF SLOW SAND FILTERS 
The principal item of cost in the operation of slow sand filters 
is that for cleaning and moving the sand. The handling of large 
quantities of sand obviously involves considerable labor, and 
even where modern methods are used for ejecting, cleaning and 
restoring the sand to the filter beds, the cost of these operations 
must of necessity form no inconsiderable part of the total cost. 
If preliminary filters are employed to prepare the water for the 
slow sand filters, the cost of operating them is properly charge- 
able to the process as a whole; and if chemical coagulants or dis- 
infecting compounds are used they must also be regarded as a 
charge to be included in the total cost. In open filters the cost 
of removing ice may be quite high, depending on the quantity 
which must be handled. The cost of pumping the water to the 
filters is sometimes included in the cost of operation of filter 
plants, but can be usually regarded as not chargeable to the 
process. 
In some of the older plants in the United States the cost of 
scraping, washing and restoring sand has been in the neighbor- 
hood of $1.50 per cubic yard. Mr. George W. Fuller1 some years 
ago tabulated some costs of handling sand, as follows: Lawrence, 
1 Trans. Am. Soc. C. E., vol. 46, 1901, p. 336. 156 
WATER PURIFICATION 
Mass., 81.70 per cubic yard; Mount Vernon, N. Y., 81.51 per 
cubic yard; and Albany, N. Y., 81.38 per cubic yard. Very much 
lower costs have been obtained in some of the larger and more 
recently built plants. This is well shown in the following tables 
taken from the annual report of 1911-12 on the operation of the 
Washington, D. C., filter plant. 
Fiscal years 
A. Per million gallons pumped to filters 
Scraping 
Ejecting 
Washing 
Smooth- 
ing 
Raking 
Resand- 
ing 
Total 
1905-06 
$0.06 
$0.29 
$0.02 
$0.06 
$0.04 
$0.47 
1906-07 
0.07 
0.20 
0.05 
0.02 
0.24 
0.58 
1907-08 
0.09 
0.14 
0.03 
0.01 
$0.02 
0.13 
0.42 
1908-09 
0.07 
0.15 
0.03 
0.01 
0.01 
0.14 
0.41 
1909-10 
0.05 
0.10 
0.01 
0.01 
0.01 
0.08 
0.27 
1910-11 
0.06 
0.08 
0.01 
0.01 
0.02 
0.03 
0.21 
1911-12 
0.06 
0.08 
0.00 
0.01 
0.02 
0.03 
0.20 
B. Per cubic yard of sand 
1905-06 
$0.07 
$0.35 
$0.04 
$0.07 
$0.14 
$0.67 
1906-07 
0.06 
0.19 
0.03 
0.02 
0.17 
0.47 
1907-08 
0.09 
0.15 
0.03 
0.01 
0.14 
0.42 
1908-09 
0.06 
0.14 
0.03 
0.01 
0.13 
0.37 
1909-10 
0.07 
0.14 
0.02 
0.02 
0.10 
0.35 
1910-11 
0.09 
0.12 
0.01 
0.01 
0.05 
0.28 
1911-12 
0.09 
0.12 
0.00 
0.01 
0.05 
0.27 
Average Cost of Labor for Sand Handling 
The cost of handling sand at the slow sand filter plant at Pitts- 
burgh, Pa., is somewhat higher than at Washington, D. C., as 
shown by the following table which has been condensed from a 
large table of unit costs given in the Report of the Bureau of 
Water for the year 1912. 
1912 
Per acre 
Per cubic 
yard 
Per million 
gallons of 
water filtered 
Cost of scraping sand 
$25.41 
$0,178 
$0,326 
Cost of ejecting sand 
0.245 
0.440 
Cost of washing and transporting 
sand 
0.120 
0.216 
Cost of restoring sand 
0.073 
0.128 
Cost of raking sand 
8.66 
0.078 
Total cost 
$0,616 
$1,188 
Note.-Cost of common labor is $2.10 per day of 8 hr. OPERATION OF SLOW SAND FILTERS 
157 
As 33,467,000,000 gal. of water were filtered at this plant dur- 
ing 1912 at a total cost of $114,238.16, the cost per million gallons 
filtered was $3,413. These figures include both maintenance and 
operation. 
At Wilmington, Del., the total cost for the year 1911-12 for 
operating the slow sand filter plant was $1,821 per million 
gallons, divided as follows: 
Preliminary filtration 80.375 
Slow sand filtration 0.916 
Laboratory charges 0.530 
Total 81.821 
These are gross operating costs and do not include interest or 
depreciation on the plant investment. 
The cost of operating five different slow sand filter plants in 
Philadelphia, Pa., during the year 1912, was as follows: 
Cost per million gallons 
Upper 
Rox- 
borough 
Lower 
Rox- 
borough 
Belmont 
Torres- 
dale 
Queen 
Lane 
Average 
Pre-filters 
Final filters. 
Pumping station 
Total cost 
$3.59 
4.40 
$2.10 
3.52 
$0.43 
3.45 
$0.24 
1.67 
$0.22 
2.31 
$0.30 
2.13 
0.18 
$7.99 
$5.62 
$3.88 
$1.91 
$2.53 
$2.61 
Some very valuable detailed costs of the operation of the slow 
sand filter plant at Washington, D. C., are given in the report 
on the operation of this plant for the fiscal year 1911-12. 
Fiscal 
years 
Office 
and 
labo- 
ratory 
Pump- 
ing 
sta- 
tion 
Filter 
operations 
Prelimi- 
nary 
treat- 
ment 
Care of 
build- 
ings and 
grounds 
Con- 
struc- 
tion 
(new 
work) 
Main 
office 
Experi- 
men- 
tal 
filters 
Total 
Sand 
hand- 
ling 
Inci- 
dentals 
and re- 
pairs 
A. Labor 
1909-10 
$0.84 
$0.70 
$0.27 
$0.23 
$0.36 
$0.02 
$0.14 
$2.56 
1910-11 
0.78 
0.65 
0.21 
0.07 
0.23 
0.29 
0.05 
2.28 
1911-12 
0.79 
0.62 
0.20 
0.07 
$0.05 
0.41 
0.09 
0.09 
2.32 
B. Material 
1909-10 
$0.13 
$0.69 
$0.02 
$0.05 
$0.17 
$0.16 
$0.02 
$1.24 
1910-11 
0.08 
0.69 
0.05 
0.08 
0.17 
0.36 
0.01 
1.44 
1911-12 
0.17 
0.62 
0.04 
0.03 
$0.48 
0.20 
0.15 
0.00 
1.69 
C. Totals 
1909-10 
$0.97 
$1.39 
$0.29 
$0.28 
$0.53 
$0.18 
$0.16 
$3.80 
1910-11 
0.86 
1.34 
0.26 
0.15 
0.40 
0.65 
0.06 
3.72 
1911-12 
0.96 
1.24 
0.24 
0.10 
$0.53 
0.61 
0.24 
0.09 
4.01 
Cost per Million Gallons Filtered 158 
WA TER P U RIF IC A TION 
If the cost of pumping is deducted from the above total costs, 
it will be seen that the remaining costs are $2.41, $2.38 and $2.77 
per million gallons, respectively, for the fiscal years tabulated. 
References 
1. F. D. West, and J. S. V. Siddons: "Cost and Efficiency Data on Tor- 
resdale Filter Plant, Philadelphia." Jour. New England Water-works 
Assn., vol. 27, September, 1913, p. 358. 
2. Report upon the Maintenance and Repair of the Washington, D. C., 
Aqueduct and Filtration Plant, 1912. 
3. Report of the Department of Public Works, Bureau of Water, Phila- 
delphia, Pa., 1912. 
4. George W. Fuller: Trans. Am. Soc. C. E., vol. 46, 1901, p. 336. 
5. Report of Water Commissioners of Wilmington, Del., 1911-12. 
6. "Filtration Results at Albany, N. Y." Eng. Record, June 22, 1912. 
7. "Bacterial Penetration in Slow Sand Filter." Eng. Record, June 22, 
1912. CHAPTER XIV 
RAPID SAND FILTRATION 
The rapid filtration of water through beds of sand, as dis- 
tinguished from slow filtration, is characterized not alone by the 
difference in the rates of flow of the water through the bed, but 
by several other features which naturally differentiate the two 
general methods of purification. A high rate of filtration necessi- 
tates the rapid formation on the sand grains of a colloidal coat- 
ing, so that the latter will entrap and prevent minute particles 
of suspended matter from passing through the sand bed. 
This is accomplished by the use of chemical coagulants. Since 
the rate of flow is usually from 100,000,000 to 125,000,000 gal. 
per acre daily, it is obvious that relatively small areas are re- 
quired for filtration as compared with those needed for slow sand 
filters. The sand also naturally becomes dirty very quickly, 
and rapid and effective methods for cleaning the sand beds are 
required. To meet these conditions a type of filter has been 
evolved which bears little resemblance to the slow sand filter. 
Preparatory treatment of water for rapid sand filters requires 
special basins for obtaining the effect of the chemicals applied 
to the water. These coagulation basins, as they are called, are 
now considered essential features of this process of purification. 
The latter phase of water purification has been discussed in a 
previous chapter, and will not be considered in that which follows 
except as it may bear upon the construction and operation of the 
filters. 
Development of Rapid Sand Filters.-The clarification of water 
containing suspended matter offers difficulties which the pioneers 
in this field did not fully appreciate, and which were overcome 
only by patient investigation and experiment. The skill and 
inventive genius displayed in the construction of the first rapid 
sand filters have not always been appreciated. The later scien- 
tific investigations utilized the earlier experiences of the builders 
of "mechanical filters," and placed this method of water 
purification upon a rational and scientific basis. 
Mechanical filters were first developed in the United States 
in connection xyith the paper industries, in which it was necessary 
159 160 
WATER PURIFICATION 
to remove the coarser particles of suspended matter in the water 
used in the manufacturing process in order not to injure the 
texture of the paper. These filters consisted of cylindrical tanks 
containing a layer of sand, and some form of strainer system at 
the bottom of the tank, through which the filtered water could 
escape. They were incapable of removing to any great degree 
the finely divided clay or the bacteria, and their adaptability 
to the purification of municipal water supplies was given little 
consideration. The employment of a chemical coagulant to pre- 
pare the water for filtration produced such good results, however, 
that many mechanical improvements were made in filters of this 
type. These improvements consisted chiefly of better strainer 
systems, which produced a more uniform draft on all parts of 
the sand bed when filtering, and which also enabled a more even 
distribution of wash water, when the filter was being washed by 
forcing water up through the sand; of ingenious methods for 
stirring the sand when washing the bed with water; of convenient 
arrangements of piping, valves and drainage gutters; and of 
provisions for treating the water with chemical coagulants in 
separate tanks, or compartments of the filter tank prior to 
filtration. 
The first rapid sand filter plants constructed for purifying 
municipal water supplies were either moderately successful, as 
measured by present ideas of efficient filtration, or dismal failures. 
The more successful plants handled waters which did not carry 
much suspended matter, except possibly for very short periods, 
while the unsuccessful plants attempted the treatment of waters 
which were constantly turbid and which were heavily charged 
with suspended matter for a considerable portion of the year. 
The method of operating these plants was entirely empirical, 
which led naturally to many failures, some of which were 
extremely costly to the promoters of this type of plant. In 
spite of the reverses met in the early development of mechanical 
filters, the successes achieved were sufficient to attract the atten- 
tion of investigators, who attacked the problems involved from a 
scientific standpoint. 
Experimental Work.-The experiments of Mr. Edmund B. 
Weston at Providence, R. I., were undertaken in 1893-94. 
They demonstrated the possibility of obtaining fairly satis- 
factory results in purifying the water of the Pawtuxet River by 
means of a small mechanical filter. This river is a small stream RAPID SAND FILTRATION 
161 
in New England, the water of which is practically free from tur- 
bidity, but is more or less colored from vegetable matter. From 
70 to 90 per cent, of the coloring matter was removed by the 
filter, and on an average about 98.5 per cent, of the bacteria. 
The experiments were not, however, entirely conclusive, and 
after a period of investigation covering about 10 months were 
discontinued. 
The river waters of New England carry as a rule but little 
sediment, and while the above experiments had demonstrated 
the ability of this type of filter to successfully purify such waters, 
its usefulness for clarifying and purifying very turbid waters, 
so that they might be made suitable for drinking purposes, had 
not been investigated. The turbid water of the Ohio River 
offered an excellent opportunity to test the efficiency of me- 
chanical filters, and the importance of establishing their real 
worth was recognized by Mr. Charles Hermany, chief engineer of 
the Louisville Water Co. of Louisville, Ky. By cooperating 
with the manufacturers of various patented filters, Mr. Hermany 
established an experimental station which was located at the 
river pumping station of the Louisville Water Co. on the bank 
of the Ohio River. A laboratory was also built and equipped 
for use in testing the water before and after filtration through the 
various filters which had been installed. 
The experimental work was placed in charge of Mr. George 
W. Fuller, who was assisted by a staff of engineers, chemists and 
bacteriologists. The filters were operated by representatives 
of the manufacturers. The experimental work extended over a 
period of about 2 years between the latter part of the year 1895 
and the middle of the year 1897. This extensive investigation 
is covered in a report by Mr. George W. Fuller.1 
The conclusions reached as a result of this work have been of 
much value in the development of this type of filter, and estab- 
lished the fundamental principles underlying the art. In general 
it was shown that turbid waters could be successfully clarified 
and purified by this process, but that preliminary subsidence 
with the assistance of coagulants was essential, and that longer 
periods for settlement than had previously been employed were 
necessary in order to obtain efficient results from the filters. 
The efficiency of the experimental filters in removing bacteria 
1 Report on the Investigations into the Purification of the Ohio River 
Water at Louisville, Ky., 1898. 162 
WATER PURIFICATION 
ranged from 97.5 to 98.5 per cent., when operating under proper 
conditions. 
In order to determine whether slow sand filters or any modifica- 
tion of this type of filter could be used to clarify and purify the 
Ohio River water, there were instituted a series of experiments 
early in 1897 at Cincinnati, Ohio. This work was undertaken 
by the City of Cincinnati through its Commissioners of Water- 
works, who were about to build a new water-works system, in- 
cluding a purification plant. The work was carried on under the 
direction of Mr. Gustave Bouscaren, chief engineer for the 
Commission, and under the immediate charge of Mr. George 
W. Fuller, assisted by a staff of experts. 
In this experimental work ample sedimentation periods were 
obtained in large settling tanks. The settled water was applied 
to slow sand filters constructed with sand beds of different 
depths. The results showed that satisfactory effluents could 
not be obtained by even prolonged periods of sedimentation, and 
subsequent filtration through a bed of sand even 5 ft. in depth. 
Modification of this process, whereby preliminary coagulation 
of the settled water preceded filtration, indicated that much 
better results could be obtained. In consequence a rapid sand 
filter for experimental use was installed in order to obtain a 
comparison between water treated by the two different processes. 
The main conclusion reached in this work was that both the modi- 
fied slow sand filter process and the rapid sand filter process 
were able to produce a water of satisfactory quality, but that 
the method with rapid sand filters was the more economical. 
The employment of rapid sand filters preceded by plain sedimenta- 
tion and followed by an adequate period of settlement after 
coagulation with chemicals was, therefore, recommended as 
being able to produce a filtered water which would be completely 
clarified, and which w off Id contain less than 100 bacteria per 
cubic centimeter on an average. 
In 1897 Mr. Allen Hazen conducted tests on some rapid sand 
filters at Lorain, Ohio. These filters treated the water of Lake 
Erie which carries little sediment, but which at this particular 
locality contained a large number of bacteria. The bacterial 
efficiency of the filters averaged a removal of 96.4 per cent.; the 
clarification of the water was hardly a question at issue in these 
experiments since the water was practically clear before filtration. 
Mr. Hazen carried out with experimental filters at Pittsburgh, RAPID SAND FILTRATION 
163 
Pa., in 1898, another series of experiments on Allegheny River 
water, which latter may be classed as a normally turbid water, 
although not as turbid as the water of the Ohio River. The 
results of this work did not differ materially from those obtained 
at Louisville, Ky. 
In 1899, Col. A. M. Miller, U. S. A., conducted a series of 
tests on the Potomac River water at Washington, D. C. This 
river water is at times very turbid, although large reservoirs in 
the supply system enable the water to be subjected to long periods 
of settlement. Rapid sand filters were recommended for puri- 
fying this supply, but were not adopted. The experiments 
showed that this type of filter would be able to produce satis- 
factory results, although it was also thought that a slow sand 
filter system assisted by preliminary coagulation would be able 
to produce the desired quality of water. It is of interest to note 
that after the Washington slow sand filter plant had been operated 
a number of years, the preliminary treatment with coagulants 
has been adopted, and has materially improved the quality of 
water delivered by the plant. 
An interesting series of experiments were conducted by Mr. 
R. S. Weston at New Orleans in 1901 on the Mississippi River 
water. Both slow sand and rapid sand filters were tested in 
connection with preliminary sedimentation. The excessive 
quantities of sediment carried by the river water required too 
long periods of settlement to economically prepare it for slow 
sand filters, although satisfactory effluents could be obtained 
from the latter type of filter if preliminary settlement was suffi- 
ciently prolonged and supplementary subsidence with coagulants 
was practised. The most economical process was that which 
included plain sedimentation, followed by settlement with 
chemical coagulants, and final filtration through rapid sand filters. 
The general conclusions reached in these early experiments 
in different parts of the United States may be briefly summarized 
as follows: 
First.-The physical, chemical and bacteriological character- 
istics of any given water should be well understood if it is to be 
successfully purified by this process. 
Second.-Those waters carrying much suspended sediment 
require preliminary settlement to remove the coarser particles, 
followed by a shorter period of sedimentation with chemicals 
in order to coagulate and precipitate the finer sediment. 164 
WATER PURIFICATION 
Third.-Rapid filtration through a sand bed is satisfactory 
only where the sediment in the water is properly coagulated, 
and where the quantity remaining in the water and which is 
applied to the filter bed is not too great. 
Fourth.-The rate of filtration should be carefully controlled, 
and rapid fluctuations in the rate made impossible. 
Fifth.-The amount of chemical coagulant applied should be 
properly adjusted to the character of the water to be purified, 
and its application should be uniform. 
Sixth.-Chemical and bacteriological tests should govern the 
operation of rapid sand filters, and intelligent supervision of their 
operation should be regarded as essential if a properly purified 
water is to be produced. 
The value of the investigations carried on between the years 
1893 and 1901 in the practical development of the rapid system 
of sand filtration can hardly be overestimated. They not only 
defined the general principles involved, but they pointed out the 
practical methods by which the process might be developed in 
conformity with those principles. Reports covering all the 
data obtained in these various investigations have been published, 
and reference to them should be made if a detailed study of this 
experimental work is desired. 
Early Types of Mechanical Filters.-There are two general 
classes into which rapid sand filters may be divided, viz., pres- 
sure filters and gravity filters (Figs. 41 and 42). Pressure filters 
were among the earliest filters constructed, but have never 
reached a state of development equal to that of gravity filters. 
Pressure Filters.-The filters of this class are closed cylinders 
of steel or iron, in which is placed a bed of sand, and through 
which the water is forced under pressure. Sometimes other filter- 
ing materials are used beside sand, such as charcoal or coke, but 
sand is more commonly employed. A compartment of the closed 
cylinder is often used as a small coagulation chamber from which 
the treated water flows to the filtering compartment. The latter 
is provided with a strainer system of some form through which 
the filtered water escapes after passing through the filter bed. 
The strainer system also acts as a distributor for the wash water 
during the cleaning of the sand bed. No method of agitation of 
the sand prior to or during the washing process was employed 
in the earliest pressure filters, but later on compressed air began 
to be used, and is still the more common method of assisting the 
wash water in cleansing the filter bed. RAPID SAND FILTRATION 
165 
Fig. 41.-Early type of gravity rapid sand filter. 
Fig. 42.-Gravity type of rapid sand filter, showing mechanical rakes. 166 
WA TER P URIFICA TION 
One of the first methods employed in applying the chemical 
coagulant to the water which was to be passed through these 
filters, consisted of a box filled with alum crystals (Fig. 43) or 
with pieces of sulphate of alumina through which was bypassed a 
small quantity of the water on its way to the coagulation com- 
partment or tank. The outflow from this chemicdl feed box was 
hand-regulated by means of'a valve, and sometimes metered in 
Roberts' Filter Mfg. Co. 
order to determine the rate of flow of the solution. There was 
no method of ascertaining the strength of this solution, and 
obviously its concentration was inversely proportional to the 
rate of flow through the feed box. In consequence the water to 
be treated was likely to receive its largest dosage of chemicals 
during the period when the volume of water flowing through the 
coagulation tank was the smallest and needed the least, and vice 
versa. Such crude methods of applying the chemicals have been 
improved upon in recent years for this type of filter, but are still 
frequently far from being satisfactory. 
Fig. 43.-Coagulant feed apparatus RAPID SAND FILTRATION 
167 
As pressure filters are closed receptacles (Figs. 44 and 45) 
and are usually operated under a pressure of 40 to 60 lb. per 
square inch, they may be located between the pumps, taking 
the water from the source of supply, and the filtered-water 
storage reservoir. Pumping directly through the filters to the 
supply mains themselves has also been used, and for small instal- 
lations is still commonly employed. Pressure filters are useful 
in connection with large buildings, where open tanks are unde- 
HANDHOLE 
Fig. 44.-Vertical rapid sand-pressure filter. 
sirable. They are, however, neither efficient nor reliable from a 
sanitary standpoint, and are not economical to operate. They 
have not been generally adopted for purifying public water 
supplies, although quite a number of municipal plants were in- 
stalled during the early development of this method of water 
purification. Their usefulness lies in the field of small plants 
for buildings, and where only a somewhat imperfect purification 
is permissible. 
Gravity Filters.-As the name of this class of filters signifies 
(Fig. 46) the water flows through the sand by gravity. The 
first gravity filters to be constructed consisted of open cylindri- 
cal tanks of wood or iron. Some of them contained a compart- 
ment for permitting the chemical coagulant to act upon the 
water before being discharged on top of the filter bed, while 
other makers provided separate tanks for this purpose. Regula- 
tion of the rate of flow through the filter was usually accom- 168 
WATER PURIFICATION 
Fig. 45.-Horizontal rapid sand-pressure filter. 
Roberts' Filter Mfg. Co. RAPID SAND FILTRATION 
169 
C, Sedimentation basin. D, Filter bed. A, Supply to filter. B, Wash to filter. E, Outlet for filtered water. H, Controller 
for regulating discharge of filtered water (butterfly valve not shown). J, Suction pipe for wash pump. K, Pump for washing 
filter bed. M, Float tank to regulate supply to filter. T, Agitating apparatus. U, Chemical tank. V, Alum feed pump, W, Alum 
feed pipe to filter. X, Propeller for operating alum feed pump. Y, Supply pipe for alum feed pump. 
Fig. 46.-General arrangement of gravity rapid sand filters, filter piping and filtered water reservoir. 
Reproduced by courtesy of New York Jewell Continental Filtration Co. WATER PURIFICATION 
170 
plished by means of hand operated valves. In some cases the 
water entering the filter was measured over a weir. Washing 
the filter bed was accomplished by forcing water upward through 
the bed, accompanied by a stirring of the sand by means of 
revolving rakes, the teeth of which projected downward into the 
bed. The movement of the rakes kept the sand bed loosened 
up, thereby permitting the wash water to come into contact with 
each particle of sand and preventing it from rising through the 
sand in separate channels. 
These stirrers or rakes were driven by a small engine from a 
belted pulley and appropriate gearing, and were so arranged that 
Fig. 47.-Arrangement of pipe manifold. 
they could be gradually lowered as they revolved, and thus stir 
the sand bed to within a few inches of the bottom. The rake 
arms revolved at a rate of six to nine revolutions per minute, and 
the wash water passed upward through the bed at a rate varying 
from 7 to 9 gal. per square foot per minute. 
The dirty wash water escaped through gutters or outlets from 
7 to 12 in. above the sand surface. A common position for the 
gutter was around the inner circumference of the tank. By a 
suitable arrangement of the raw water-supply pipes, sewer pipes 
and valves, the waste-water gutters or outlets also served for the 
inlets for the treated water to be filtered. 
The strainer systems and manifolds (Fig. 47) were varied in 
character, and presented, even as they do today, one of the most RAPID SAND FILTRATION 
171 
Fig. 48.-Types of strainers for rapid sand filters. 172 
WATER PURIFICATION 
difficult and puzzling problems in the design of this class of filters. 
Slotted tubes, perforated plates, and strainer cups or nozzles were 
all used. The slotted tubes or nozzles were usually set in a mani- 
fold of pipes consisting of a central collector pipe with numerous 
laterals, usually extending at right angles to the main collector. 
Perforated plates or false bottoms were also employed. In some 
cases a false bottom fitted with nozzles was used. Many of 
these strainer systems are still in use, and the most modern of 
strainer systems are but modifications of the older types. 
The method of applying the chemical coagulant to the water 
to be filtered in gravity filters was usually by means of small 
pumps. These pumps were sometimes automatically adjusted, 
where possible, to the stroke of the large pump supplying the 
water to the filter, but were more frequently hand-regulated. 
Solutions of definite strength were prepared in special tanks from 
which the chemical feed pumps drew their supply. 
The early types of mechanical filters showed considerable in- 
genuity on the part of their designers in overcoming some of the 
difficulties inherent in a process which, though apparently sim- 
ple, is in reality complicated. A knowledge of hydraulics and 
mechanics, to say nothing of the chemistry and bacteriology of 
the process, were necessary to solve most of the problems in- 
volved. In the rigid scientific examination and study of the 
process, undertaken between the years 1893 and 1901 by various 
investigators, were laid the foundations for the development of 
the modern rapid sand filter plant of today. 
References 
1. E. B. Weston: Report Rhode Island State Board of Health on Ex- 
periments in Mechanical Filters, 1894. 
2. Allen Hazen: Report Ohio State Board of Health on Mechanical 
Filtration Plant of the Public Water Supply of Lorain, Ohio, 1897. 
3. George W. Fuller: Report on the Purification of the Ohio River Water 
at Louisville, Ky., 1898. 
4. Report of the Filtration Commission of Pittsburgh, Pa., 1899. 
<5. Report on Purification of the Washington Water Supply. Senate 
Document No. 2,380, 56th Congress, Second Division. 
6. R. S. Weston: Report on Water Purification and Water-works System 
for New Orleans, La., 1903. 
7. Philip Burgess: "The Development of the Mechanical Filter Plant." 
Eng. News, vol. 59, p. 249, 1908. 
8. Allen Hazen: "The Filtration of Public Water Supplies." 
9. Turneaure and Russell: "Public Water Supplies." CHAPTER XV 
GENERAL ARRANGEMENT OF RAPID SAND FILTER 
PLANTS 
In the development of rapid sand filters the necessity for 
properly preparing the water for filtration by adequate periods 
Fig. 49.-General plan of Cincinnati low-service pumping station, settling 
reservoirs and filter plant. 
of settlement with coagulating chemicals soon made it evident 
that special basins of the proper size and shape for this purpose 
must form an integral part of the plant. Reservoirs or tanks 
for holding the filtered water were also found necessary, as a 
supply of the latter for washing the filters must be available. 
A modern filter plant (Fig. 49), therefore, is composed of at 
173 174 
WA TER P URIFICA TION 
least three distinct parts; i.e., coagulation basins for the treat- 
ment of the water with chemicals; a set of filter tanks with the 
necessary system of piping, valves and controlling apparatus; 
and a filtered-water reservoir. A building is usually con- 
structed over the whole or a part of the filter tanks. This build- 
ing also generally contains the solution tanks for the chemicals, 
the regulating, measuring, and controlling apparatus for the appli- 
cation of the chemical solutions and the operation of the filters, 
the wash water pumps, and the laboratory and offices. The 
particular method of arranging and grouping the various parts 
of the plant is dependent to a great extent on local conditions. 
Certain modifications of the above-mentioned parts of a rapid 
sand filter plant of the gravity type are necessary in the case of 
plants using pressure filters, but as the latter are not so generally 
employed for purifying municipal supplies, they need not be 
given further consideration at this point. 
General Arrangement of Typical Plants.-Those plants which 
have to purify very turbid waters are usually provided with 
preliminary settling reservoirs in addition to the other parts of 
the plant mentioned above. No chemicals are used in these 
reservoirs and the precipitation of the sediment is effected by 
simply reducing the velocity of flow of the water as low as pos- 
sible, thereby giving time for the suspended matter to drop out. 
These plain sedimentation reservoirs form an integral part of 
such plants as those, for example, at Louisville, Ky., and at 
Cincinnati, Ohio. 
Several typical plain sedimentation reservoirs and coagulation 
basins have been described in Chapters VI and VII, and cuts are 
given showing their location in relation to each other and to the 
filters. It will be noted that the topography and other condi- 
tions of a purely local character usually determine the relative 
positions of the various parts of the plant. 
Relative Position of the Coagulation Basins and Chemical 
House.-There are one or two special features in laying out the 
plan for a rapid sand filter plant to which it is desirable to pay 
considerable attention, and which, if incorporated in the plant, 
will make its operation less difficult. The inlet to the coagula- 
tion basins or to the mixing chamber, when the latter forms a part 
of these basins, is the usual point for the application of the 
coagulating chemicals. The preparation of the chemical solu- 
tions is generally carried on in a part of the filter plant building. RAPID SAND FILTER PLANTS 
175 
, Sta.14 + 28.8T S. 
M.H.No.7 
Fig. 49a.-General plan of Cleveland filtration plant. 
M H.No2 
SU.11+ 20.23 
Sta.ll+64.83 
M.H.No.3 
Sta.14 92.87 S> 
Sta.15 +14.87 S.^ 176 
WATER PURIFICATION 
PLAN OF ROOF 
PLAN 
AT ELE. 13.0 AT ELE. 18.75 
Fig. 496.-Sections of Cleveland filtration plant. 
SECTION N-N 
SECTION L-L 
PLAN AT ELE. 32.0 
SECTION K-K RAPID SAND FILTER PLANTS 
177 
The distance which these solutions must be conveyed should be 
as short as possible, in order to avoid the use of long pipe lines 
which may become choked up by deposits or by corrosion. These 
chemical feed lines should be readily accessible for their full 
length both for the purpose of cleaning or renewal. 
In a plant where a large amount of chemicals are used, such as 
is the case where the softening of hard waters are made a part of 
the process, it may be best to place the chemical house over or 
close to the inlet to the coagulation basins, and quite apart from 
the filters and the filter house. This is the plan followed in the 
purification plant at New Orleans. At the Columbus, Ohio, 
water-softening and purification plant the design which was 
worked out resulted in a very compact arrangement of the 
chemical house, mixing tanks and coagulation basins. In the 
small plants a compact arrangement is more easily effected than 
in the large plants. For example, in the large plant at Cincin- 
nati, Ohio, the chemical house had to be placed at a point con- 
venient for receiving shipments of chemicals by rail. This 
happened to be a point located about 700 ft. distant from the 
place where the water to be treated could most conveniently be 
brought in order for it to be controlled in its passage into the 
coagulation basins. 
Arrangements of Filters.-The filter tanks in the older plants 
were usually cylindrical in form, and were placed in two or more 
rows. This arrangement enabled a common water supply pipe 
to be laid between the rows of filters, as well as a common effluent 
pipe for the discharge of the filtered water. It also permitted 
the individual filters to be connected to a common drain or waste 
pipe to be used in draining the filters and when washing them. 
As the earlier plants were not provided with rate-of-flow con- 
trolling apparatus, the system of influent and effluent piping was 
comparatively simple. 
Round wooden or iron filter tanks are not economical of space, 
and are also limited in the diameters to which it is practical to 
build them. With the development of concrete construction, 
oblong or square tanks of this material have taken the place of 
the round wooden and iron tanks. The former can be built side 
by side with no waste space between them, and are much more 
durable. 
Concrete filter tanks are usually built side by side in rows, 
with a so-called pipe gallery between the rows In this space 178 
WA TER PU RIF IC A TION 
Fig. 50.-Mechanical filter plant at Baisley's Pond, Brooklyn, N. Y. RAPID SAND FILTER PLANTS 
179 
Extend through Bank 
Extend to Channel of River 
Fig. 51.-Little Falls, N. J., rapid sand filter plant. 
Extend to Channel of River - 
SCALE IN FEET 180 
WATER PURIFICATION 
are laid the influent and effluent piping or ducts. It furnishes 
space as well for the necessary valves, rate-of-flow controlling 
apparatus, gages, and so forth, required in operating the filters. 
An accompanying cut shows the old method of placing the round 
filter tanks (Fig. 50) and the requisite piping, while another cut 
illustrates the typical arrangement of a set of modern concrete 
filters (Figs. 51, 52 and 53). 
Fig. 52.-Baltimore rapid sand filter plant. 
Another desirable feature in the arrangement of this type of 
plant is that the controlling, measuring and indicating apparatus 
should be located conveniently close together in the filter plant 
building. The apparatus referred to consist of the gages in- 
dicating the water levels in the various tanks, basins and reser- 
voirs, the steam and compressed-air pressure gages, electrical -Coivcrete Cop 
.J>. dev. 2610 
Wash Water Buiiding - footings. Columns floorsand 
Tanks are nofin Contract. 
fitter building:- filtered Wafer Basins, filter 
Tanks c floors t "T not in Contract. 
Columns above [Jev.222.0cfoofs 
are to be built under this Contract. 
Brickwork of filter BuHding, including 
Clere stories is to be Plastered. 
Longitudinal .Section on Filter. & Wash Water. BujLpiNG$. 
Note: Nie end windows of filter Gallery and Transept 
Oere-sforia are to have Graniie sills. AH other 
Clerestory windorrs are to have Concrete sills, 
reinforced with 4- Wbars. 
Sills on rear of Wash Water Building will be 
concrete, reinforced with -4- Cfs'"Bars. 
Index > 
Concrete notin this Contract. 
Concrete or Cinder Concrete 
furnished under this Contract. 
Brickwork. 
' far Skuliqht Detail, 
see Sheet M n. 
Cross Section on Line 'A-A'. (Jjwkiws east). 
Fig. 53.-Baltimore rapid sand filter plant, cross-section of plant. 
(.Facing page 180.) RAPID SAND FILTER PLANTS 
181 
meters, water-meter registering devices, and valves for controlling 
the flow of water to the coagulation basins, filters and clear-water 
reservoir. 
By thus grouping this apparatus fewer attendants are necessary, 
besides enabling the operator to have a better control over the 
purification process. Rapid changes in the method of operating 
plants of this kind are not infrequently required, and apparatus 
for controlling and indicating the effect of these changes should be 
centrally located. The larger the plant the more difficult, of 
course, does this compact arrangement become, but this feature 
should never be lost sight of in designing the plant. 
Clear-water Reservoir.-The position of the clear-water reser- 
voir in relation to the other parts of the plant is not so important 
as is that of the coagulation basins and filters. It is frequently 
placed under the filters, but may also be placed some distance 
away from them. The reservoir is usually covered in order to 
prevent growths of microscopic plants in the filtered water. In 
some of the large plants, however, the reservoir is not covered, but 
in such abundant growth of algse and diatoms occur. The coagu- 
lation basins are sometimes covered in the smaller plants but are 
not usually so in the larger ones. 
If the clear-water reservoir is not conveniently located for ob- 
taining a supply of filtered water for use in washing the filters, 
provision is usually made for either drawing filtered water directly 
from the main effluent pipe from the filters, or from some pump 
well supplied from this main. In many modern plants separate 
wash-water storage tanks or reservoirs are provided for holding 
a supply of filtered water at a sufficient elevation to supply wash 
water under pressure during the washing of the filters, instead of 
by direct pumping during the washing process as is quite com- 
monly the case in the smaller plants. This tank is usually located 
at an elevation of 50 or 60 ft. above the filters and is covered. 
The filter-plant building generally contains the laboratories 
and offices of the plants, beside the rooms for the preparation of 
the chemical solutions, and the storage space for the chemicals. 
A heating plant must be provided for the filter-house building 
where the climate makes it necessary, and a power plant as well, 
unless steam or electric power is obtainable from some other 
source. 
From the foregoing description of the principal parts of a rapid 
sand filter plant it can be seen that considerable latitude may be 182 
WATER PURIFICATION 
allowed in their arrangement, but that certain important factors 
must be kept in mind in order that the various parts of the plant 
shall be properly coordinated. Local conditions will necessarily 
introduce factors which will require a modification of the ideal 
design, and which may even make an efficient and economical 
plant an impossibility. It, therefore, is the province of the 
designer to harmonize as far as possible the arrangement of the 
various parts of the plant in order that as good a plan as can be 
obtained may result under the conditions imposed. 
References 
1. "Standardization of Rapid Filter Design." Eng. Record, Mar. 8, 1913. 
2. F. B. Leopold: "Some Recent Advances in Rapid Filter Design." 
Engineering-Contracting, Mar. 12, 1913. 
3. Ralph Hilscher: "The Present Conditions of Small Water-purification 
Plants in Illinois, Fault in Design and Operation." Engineering- 
Contracting, Mar. 19, 1913. 
4. "Rapid Filtration for the Croton Water Supply of New York City." 
Eng. Record, June 29, 1912. 
5. "Rapid Filtration Plant at Columbus, Ind." Eng. Record, Mar. 
8, 1913. 
6. "The St. Louis Mechanical Filters." Eng. News, Oct. 23, 1913. 
7. "The Evanston, Ill., Filter Plant." Eng. Record, No. 22, 1913. CHAPTER XVI 
DETAILS OF RAPID SAND FILTER-PLANT CON- 
STRUCTION 
The important parts of a rapid sand filter plant have been out- 
lined in a general way in the preceding chapter. The present 
chapter will deal with the various portions of the plant in detail. 
Number and Size of Filter Tanks.-The number of filter units 
in any given plant should be sufficient to meet the maximum de- 
mand for water in conjunction with such storage capacity as can 
be provided, and to permit of easy operation of the plant as a 
whole. At the prevailing rates of filtration for filters of this type, 
viz., 1.6 to 2.0 gal. per square foot of filtering surface per minute, 
there would be needed 433 and 348 sq. ft. respectively, of sand 
surface for each million gallons of daily output. A commonly 
used area is 350 sq. ft. of filtering surface for each million gallons 
of daily capacity. The larger filter units are usually constructed 
for either a daily capacity of 500,000, 1,000,000, 2,000,000, 
3,000,000 or 4,000,000 gal. For municipal plants the gravity 
type of filter is not usually built of units under 500,000 gal. daily 
capacity, except for quite small plants, nor over 4,000,000 gal. 
even for the largest plants. Pressure filters are naturally limited 
in size, and are not usually built to exceed units of much over 
500,000 gal. daily capacity. 
The circular gravity filter tanks are constructed of either wood 
or steel, and of course vary in capacity according to their size. 
For a capacity of 150,000 gal. in 24 hr., the diameter of the tank 
is approximately 8 ft., while for a capacity of 500,000 gal. per day 
the diameter is about 15 ft. These tanks are from 7 to 9 ft. in 
height, although those which are combined with a settling cham- 
ber underneath may be as high as 16 ft. 
Pressure filter tanks are built of either cast iron or steel. Cast 
iron is used for the smaller sizes up to those which are about 2 ft. 
in diameter, while steel is used for the larger diameters. Pres- 
sure filters are built to be erected either vertically or horizontally. 
The vertical pressure filters range from 1 to 10 ft. in diameter, and 
from approximately 5 to 12 ft. in height. The horizontal pressure 
filters are usually constructed for the larger-sized units, and range 
from 10 to 25 ft. in length with a diameter of about 8 ft. 
183 184 
WATER PURIFICATION 
Concrete Filter Tanks.-Concrete filter tanks are constructed 
either square or oblong, and with open tops. They are frequently 
built over a filtered-water reservoir, in which case the tanks 
usually rest upon groined arches supported by piers whose foot- 
ings are in the floor of the reservoir. Even where the space 
under the filters is not used for the storage of water, but merely 
contains the pipe collecting system of the filter underdrains, 
groined-arch construction is sometimes employed. 
It is not uncommon to divide large concrete filter tanks into two 
parts by a central conduit. The latter serves to distribute in- 
flowing water to the sand bed through a number of lateral gutters, 
which usually extend at right angles to the central conduit, and 
also to act as a collector for dirty wash water during the process of 
cleaning the filter sand. In large tanks tie-beams extending 
across the top of the tank give it rigidity, and may also act as 
supports to walls and covers above in those cases where only a 
part of the filter tank is within the filter building. 
Waste-water Conduits and Lateral Gutters.-The central and 
lateral conduits within concrete filter tanks may be built of con- 
crete or of sheet steel. The former material is generally used in 
the large tanks, and the latter in the small units. The central 
conduit may or may not be covered, but the lateral waste-water 
ducts are always uncovered. The lateral gutters are generally 
spaced from 5 to 7 ft. apart, in order that the wash water being 
flushed off the surface of the sand shall not have to travel at the 
most a greater distance than 2.5 to 3.5 ft. in order to escape to the 
central conduit. The edges of the lateral gutters should be of 
uniform height and smooth, since they act as weirs to the escaping 
wash water, and should draw from all parts of the bed at approxi- 
mately the same rate. A departure from this type of gutter is 
found in the saw-tooth edged gutter, which consists of a series 
of V-shaped notches or weirs on either side of the trough, and 
which are intended to accelerate the horizontal flow of the dirty 
water to the gutter by providing a uniform series of restricted 
exits through which the dirty wash water escapes at a greater 
velocity than it could over the long weir furnished by the edge of 
the gutter of the usual type. 
Lateral waste-water gutters are made with a number of differ- 
ent cross-sections, each of which is supposed to have some 
particular merit because of the shape selected. A common 
cross-section is a V-shaped bottom with vertical sides. The end RAPID SAND FILTER-PLANT CONSTRUCTION 
185 
farthest from the central conduit is usually more shallow than 
the end next to it, as less water is carried by this portion of the 
gutter. Semicircular gutters are frequently used, especially when 
constructed of sheet steel. A cross-section of a concrete gutter, 
which has a shallow V-shaped bottom, and sides that incline from 
the vertical toward the center of the trough, is supposed to 
prevent loss of sand during the washing process. Another type 
provides wings, which project out from the upper edge of the 
gutter, and are designed to prevent sand being carried over the 
edge, especially where air is used in agitating the sand bed prior 
to or during the application of wash water. Air vents are placed 
at intervals along the edge of the gutter, which allow the escape 
of air used to agitate the sand and which would otherwise become 
pocketed underneath the wings. 
The carrying capacity of lateral gutters is frequently too small 
for the volume of wash water which must be conveyed away. 
Flooding of these gutters represents a loss of efficiency in washing 
the filter and should be avoided by providing them with ample 
carrying capacity. The spilling of the water along both sides of 
the gutter and for its full length, as in the usual type, produces so 
much loss of head in the water flowing in the trough that the usual 
computations for flow will indicate a size and pitch for the gutter 
which are much too small. 
In an appendix (B) will be found a formula for calculating the 
approximate discharge of water from wash troughs of rapid sand 
filters. This formula has been derived by Mr. C. N. Miller, 
Assoc. M. Am. Soc. C. E., from a consideration of the laws of 
hydraulics, and from certain experimental data obtained by 
the author, which permitted a constant for the formula to be 
determined. 
Influent Piping and Conduits.-It is undoubtedly the best 
practice if possible to so arrange the level of the filter tanks with 
reference to the coagulation basins, that the coagulated water can 
flow directly to the filter tanks with the least possible difference 
in head. Check valves, of any kind which are intended to regulate 
the discharge of the water applied to the filters, are frequently so 
irregular in their control of the flow that they permit a sudden 
rush of water onto the bed and thereby disturb the filtering sur- 
face. A large central wash-water conduit in the filter tank may 
act in a measure as a stilling chamber, but it is better not to rely 
on such an effect if it is possible to provide flow lines that will 186 
WATER PURIFICATION 
Fig. 55.-Plan and cross-section of filter piping and pipe gallery in St. Louis filter plant. PLAN ABOVE SAND 
DETAILS OF FILTER PIPING, 
WATER PURIFICATION WORKS, 
LITTLE FALLS, N.J. 
SCALE OF FEET 
.Bell Piece 
Bell Pieces from 16 Wash Line^ 
This half of line like other 
half shown at left. 
There are 4 lines like this.. 
Fig. 54 -Cross-section of filter piping and pipe gallery in Little Falls, N. J., rapid sand filter'plant. 
{Facing page 186.) RAPID SAND FILTER-PLANT CONSTRUCTION 
187 
permit the influent valve to the filter to be left wide open while 
the filter is in service (Figs. 54 and 55). 
Since the pressure upon the influent and effluent piping of 
filters of the gravity type is low, light cast-iron pipe may be used. 
In some modern plants concrete conduits have displaced the cast- 
iron influent and effluent pipe lines entirely, except for short con- 
nections where valves must be inserted. Where these concrete 
conduits can be made water-tight and where satisfactory connec- 
tions between them and the pipe that must be used can be made, 
there is probably some economy in this method of construction. 
However, the cracking of such conduits due to expansion and con- 
traction is not uncommon, and appears to be most likely to occur 
where wall nipples connect with valves or with short pipe. 
Where there are double- or triple-deck conduits, the upper one 
carrying coagulated water, the next lower conveying filtered 
water, and the lowest duct carrying away dirty wash water, as is 
the case in some plants, especial care is necessary to make the con- 
duits water-tight, since contamination of the filtered water by 
leakage from the upper or the lower conduit into the middle con- 
duit is not impossible under certain conditions. 
Effluent Piping and Conduits.-The system of pipes or channels 
which serve to collect the filtered water from the underdrains of 
the filter tank differ more or less as to their design. Cast-iron 
piping is commonly used with the large concrete filter tanks, 
while wrought-iron piping is more easily adapted to the small cir- 
cular wooden and steel tanks. In large concrete filters the under- 
drain system is sometimes divided into sections, each part being 
served by a large single or branched collector pipe, which is con- 
nected to larger pipes, and they in turn with the effluent main in 
the pipe gallery. Another method is to construct large con- 
crete collecting channels as a part of the underdrain system. The 
main channel collector is connected with the effluent pipe in the 
pipe gallery directly, and thus does away with a system of collect- 
ing pipes underneath the filter tank. 
Since the effluent pipes (Figs. 56 and 57) or conduits must serve, 
in conjunction with the underdrain system, both as a collector of 
the filtered water during filtration, and as a distributor of wash 
water during the washing process, they should be so designed as 
to keep the head lost in the flow of water at a minimum. Rela- 
tively large pipes or conduits (Fig. 58) are required to produce 
this effect, and much care is necessary to proportion them prop- 188 
WATER PURIFICATION 
Fig 56.-Pipe gallery and under drain system of Cincinnati filters. 
Fig. 57,-Wash water piping of Cincinnati filter plant. RAPID SAND FILTER-PLANT CONSTRUCTION 
189 
erly. Any lack of uniformity in the draft on all parts of the filter 
during filtration is undesirable, and any condition which prevents 
a uniform delivery of wash water to all parts of the bed must be 
avoided. Next to the strainer system, the collecting system of 
pipes and conduits forms the most important and the most 
difficult part in the design of rapid sand filters. 
In the main discharge line from the filter is now placed in all 
the large and well-designed plants a regulator in order to control 
the rate of filtration. This may be a separate device through 
which the water flows, or it may be an auxiliary mechanism 
which utilizes the effluent valve to regulate the flow. When the 
Fig. 57a.-Interior view of filter house in Cincinnati filtration plant. 
rate controller is an integral part of the effluent pipe line, it is 
placed just back of the effluent valve which cuts off the filter from 
the main filtered water pipe line or conduit. The design and 
operation of these regulating valves will be considered under a 
separate section. 
Waste-water Piping.-Since the discharge of dirty wash water 
in all filters is through some system of troughs or gutters, it is only 
necessary to connect such a system with a main waste pipe which 
will convey the dirty wash water to the sewer. Where a central 
wash-water conduit forms a part of the filter tank, the escape of 
the wash water is usually through the bottom or end of the con- 
duit to a pipe line leading directly to the sewer. In this line is 
placed the waste-water valve. It is also customary to make a 190 
WATER PURIFICATION 
Fig. 58.-Pipe gallery of Minneapolis filter plant. 
Fig. 58a.-Operating floor of Minneapolis filter plant. RAPID SAND FILTER-PLANT CONSTRUCTION 
191 
cross-connection between this pipe line and the filtered water 
pipe of the filter tank, and to provide it with a separate valve. 
This connection is for the purpose of draining the filter directly to 
the sewer, when necessary, without allowing the drainage water 
to pass into the filtered-water pipe line. 
Wash-water Piping.-Since the underdrain system for carrying 
away the filtered water also serves as the distribution system for 
the wash water, it is only necessary to provide an independent 
connection between the main effluent pipe from the filter and a 
separate pipe line for supplying the necessary wash water under 
pressure. The pipe connection is made back of the rate controller 
Fig. 59.-Isometric view of filter unit in Baltimore filter plant. 
when the latter forms a part of the pipe line. The connecting 
pipe must, of course, be provided with a valve. 
Air Piping System.-A method of agitating the sand bed during 
the washing process consists in forcing compressed air through the 
water underdrain system of the filter tank, and thence up through 
the strainer system, gravel and sand beds, or through a separate 
system of pipes usually laid on the top of the gravel layer and 
under the sand bed. This compressed-air piping is sometimes 
carried over the top and down into the filter tank from above. 
The main air supply pipe is brought through the pipe gallery 
with the other large pipes and conduits which supply or withdraw 
water from the filter tank. 192 
WATER PURIFICATION 
Pipe Gallery and Cellar.-From the foregoing description of 
the piping and conduits with which it is necessary to equip each 
of the several filter tanks of a rapid sand filter plant, it is evident 
that considerable space must be allowed between the rows of 
tanks in order to provide a place for them (Fig. 59). If piping 
passes out through the bottom of the tank, it should be accessible, 
and in consequence a cellar is sometimes provided which contains 
nothing but this underdrain piping. The space underneath the 
tanks is also frequently utilized as a filtered-water reservoir. 
This is especially common in the construction of the smaller 
filter plants. 
Fig. 60a.-Operating floor of Cleveland filter plant. 
The pipe gallery (Fig. 60) should be large enough to accom- 
modate all the main and connecting pipe lines, and allow enough 
space for making repairs and the inspection of piping and valves. 
Considerable ingenuity is oftentimes required in laying out a pipe 
gallery plan so that all the pipes, conduits and valves are included 
that are desired, and so that there are no interferences, and yet 
provides the extra space needed for inspection and repairs. 
Several typical plans and cross-sections of pipe galleries are 
shown in the accompanying cuts. For Continuation of hpingh 
Set Sheet No. !4,_ 
6 "Pressure Line 
See No. 26 
fittings to be Y 
placed before A 
concrete ispouret 
-6&32-C 192-0^ - 
Five Filters omitted; Piping 
is duplicate of that shown. 
Ct. Wall Sleeves to. 
pass spigot end 
ofWi WPipes 
Above plan shows piping in 
South Filter Gallery. 
Piping in North FilterGatterg 
is identical except arrangement 
of 30" Wash Water and 4'Pressure 
Line. See Sheet No. IS 
' Piping for Fitters in Fast Fnd ) 
same as other Fitters, except ( 
difFerent arrangement of 24 * 
Influent, 30" Filter Drain and 
12" Wash Wafer Drain fa tees 
See Sheet No. !3 
GENERAL PLAN OF SOUTH GALLERY. 
- Floor of Filter Tank and Cover over Clear Water 
Reservoir cut away to show piping beneath. 
Filter Tank 
for Detail of 14"Risers - see Dheet Ho 38- 
' Portion of 30*fitter Draifi 
omitted to short 12 'Wash. 
Water Drain. 
TYPICAL SECTION 
Fig. 60.-Piping of filters and in pipe gallery in Baltimore filter plant. 
{Facing page 192.) RAPID SAND FILTER-PLANT CONSTRUCTION 
193 
Valves.-Each filter tank is usually provided with five valves, 
viz., an influent valve for admitting the water to be filtered, an 
effluent valve for stopping the flow of filtered water from the tank, 
a wash-water valve for permitting the application of wash water 
to the underdrain system, a drain valve for allowing the escape of 
the dirty wash water, and a waste or "rewash" valve for wasting 
the filtered water after its passage through the tank, when it is 
desirable to do so. The first four valves are essential to the 
proper operation of the filter. The fifth valve or waste-water 
valve is not absolutely necessary except where it is desired to 
drain the water from a tank and from the underdrain system with- 
out permitting it to pass out through the effluent valve into the 
Fig. 606.-Operating floor of St. Louis filter plant. 
filtered-water main. It is commonly provided for in a system of 
piping and valves and may be found useful at times, although as 
a rule it is little used. 
In small plants the filter-tank valves are usually hand-operated. 
They are as conveniently placed as possible on the various pipe 
lines and are turned by means of hand wheels. Where the valves 
must be located in a pipe gallery at some distance below the top 
of the filter tank, extension stems with hand wheels attached are 
employed. 
In gravity filters it is important that the operation of the filter 
during both the processes of filtering and washing should be ob- 
served, and that the various valves controlling the movement of 194 
WATER PURIFICATION 
the water into and from the filter tank, should be easily and 
quickly accessible to the operator. The opening and closing of 
valves must usually be done rapidly. This is not possible with 
large valves when hand-operated. Because of these conditions 
in operating and because observation of the processes of filter- 
ing and washing must take place at the top of the tank, power 
operated valves, whose movements are controlled from a table or 
switchboard, have come into general use especially in the larger 
plants. Hydraulically and electrically operated valves are both 
used, but the former are much more extensively employed than 
the latter. 
Hydraulic Valves.-The hydraulically operated valve consists 
of the usual body of the valve proper, surmounted by a closed 
cylinder containing a movable piston (Fig. 61). This piston is 
Fig. 61.-Piston of hydraulic valve showing adjustable stop for use on wash- 
water valves. 
attached to a rod connected to the movable gates of the valve. 
Pipe outlets at the top and bottom of the cylinder permit water 
under pressure to be admitted to either side of the piston. The 
hydraulic pressure is usually controlled through a multiple-way 
cock, which when set in one position admits water to one side of 
the piston and releases it on the opposite side or vice versa, thus 
moving the gates of the valve to either an open or closed position. 
The small pressure pipes with their multiple-way cocks are usually 
brought to a table or board located at one end and at the top of 
the filter tank, where the handling of the valves and the observa- 
tion of the processes in the filter can be conveniently conducted. 
The position of the hydraulic cylinder of the valve should, if 
possible, be horizontal, or at least at some inclination from the 
vertical. In the vertical position any leakage which took place 
around the piston, and which reduced the pressure sufficiently to 
allow the piston and gates to drop by gravity, might close the RAPID SAND FILTER-PLANT CONSTRUCTION 
195 
valve, although it had been left open. The position of the valve 
gates is usually indicated by a hand moving over a dial. The 
hand is operated by cords passing through small pulleys and con- 
nected to the movable piston. The pressure piping and multi- 
ple-way cocks should all be of brass, as corrosion should be 
avoided, if possible. 
Mr. George W. Fuller states1 that 
"For filter service it is rather desirable to specify a double-gate 
parallel-seat valve, as wedged-seat valves for this type of service are 
more likely to stick. Valves should be bronze-fitted throughout and 
the operating cylinders should be bronze-lined. Operating pistons can 
to best advantage be fitted with double-cup leathers. In determining 
the size of the operating cylinder, good service may be obtained by 
making the cylinder of such size that under normal conditions the 
total effective moving force on the piston is equal to the total pressure 
on the valve disk, assuming this in no case to be less than 30 ft. This 
is roughly equivalent to the assumption that the coefficient of friction 
of the disk, including the friction of the stuffing boxes and piston 
packing, is approximately 100 per cent." 
Hydraulically operated valves usually give little trouble if 
properly designed and set, and if kept in constant use. Sticking 
of the piston in the cylinder sometimes occurs when the valve is 
left unused for some length of time. Ordinary care and atten- 
tion should be given the moving parts, stuffing boxes, etc., and 
any corrosion of bronze parts of the cylinder or rod carefully 
noted, and, if possible, remedied. Sticking and leaking of brass 
multiple-way cocks give more or less trouble, but well-designed 
cocks may be had which reduce such difficulties to a minimum. 
Electrically Operated Valves.2-Either direct (Fig. 62) or alter- 
nating-current driven motors may be used in operating valves of 
this type. Direct-current motors for operating valves are built 
with compound-wound field coils, or as plain series motors. If 
compound-wound just enough shunt field winding is employed to 
prevent the armature from reaching a dangerous speed, if oper- 
ated with no load. This type of motor gives a high starting 
torque compared with its normal rating, and is well suited for 
1 George W. Fuller: "Some Features of Detail in the Design of Rapid 
Sand Water Filtration Plants." Engineering-Contracting, Sept. 2, 1914. 
2 The author is indebted to Mr. J. S. Gettrust, Assoc. M. Am. Soc. E. E., 
for certain descriptive matter pertaining to electrically operated valves 
and registering apparatus. 196 
WA TER P U RIF IC A TION 
valve operation as the motor slows down and exerts a high torque 
when most needed in starting, and speeds up as the load decreases. 
Direct-current motors offer another advantage in the operation 
of valves in that they may be stopped very quickly by cutting off 
the motor from its supply of current and connecting a resistance 
across the terminals of the armature. The motor then becomes 
a generator, and the inertia of the moving parts is used in forcing 
current through the resistance where it is dissipated as heat. 
This is known as dynamic braking, and, when properly effected 
Fig. 62.-Electricially operated gate valves at Cincinnati filter plant. 
by automatic devices, enables the gates of the valve to be 
stopped at any desired point. It is thus possible when closing a 
valve to seat the gates without danger that the inertia of the 
moving parts will cause the gates to overtravel and become 
wedged in the seat. 
In the accompanying cut (Fig. 63) is shown a control panel for 
a direct-current motor used in operating large valves. The 
motor can be made to move the valve in either direction by throw- 
ing the four pole-double throw switch to either side, and then 
operating the movable contact arm. The two lamps are used to 
indicate whether the valve is closed or open. The panel is 
equipped with fuses, an overload circuit-breaker, no-voltage RAPID SAND FILTER-PLANT CONSTRUCTION 
197 
release, and shunt trip coil, which latter is operated by a limit 
switch geared to the motor. In Fig. 64 is shown a type of con- 
trol panel by which the valve gates may be stopped at any 
intermediate point between a wide-open and fully closed position 
by means of a dial contact switch 
placed on the panel just under the 
lamps. 
For automatically stopping the 
travel of the valve gates various forms 
of limit switches have been used. 
They consist essentially of a set of 
movable contacts which are operated 
by a screw shaft, which is geared to 
the valve. The contacts close cir- 
cuits at either end of their travel 
which operate the mechanism for dis- 
connecting the motor from the power 
circuit. The movement of these cir- 
cuit-closing contacts is proportional 
to the movement of the valve gates. 
Some electrically-driven valves are 
operated without limit switches. 
The increased load on the motor, as 
the valve reaches the limit of its 
travel, operates an overload coil, 
which opens the controller circuit. 
This type of apparatus requires a 
slow-speed motor and a special con- 
struction of the valve. 
Experience has shown that the 
tripping mechanism for opening the 
motor circuit, when the valve reaches 
the limit of its travel, is not always 
reliable. If the circuit is not opened the valve stem may be 
twisted and bent, or some part of the equipment such as the 
gear teeth, motor frame or valve bonnet, broken. By install- 
ing additional contacts on the limit switch which will short-cir- 
cuit the motor in case of the failure of the limit switch to act, this 
difficulty may be overcome. These contacts must be accurately 
set, and when so placed will promptly blow the fuses in the motor 
Fig. 63.-Panel from which 
direct-current valve motors 
may be operated. 198 
WATER PURIFICATION 
power line if the limit switch does not act, and thereby prevent 
more serious damage to the valve and motor. 
The polyphase squirrel-cage type of motor is the most simple 
form of motor for operating valves with an alternating current. 
It may be built to operate even when submerged in water. The 
characteristics, however, of this type of motor do not lend them- 
selves as well to the operation of valves as do those of the com- 
pound-wound direct-current motor. As their starting torque is 
Fig. 64.-Panel from which direct-current valve motors may be operated, 
and so arranged that valve may be stopped at intermediate points between 
wide open and shut. 
low as compared with a direct-current compound-wound motor, 
it is necessary to design them especially for a high starting torque, 
which necessarily involves the use of a larger motor than would 
be the case if a direct-current motor was used. 
References 
1. S. B. Russel: "Notes on Large Filters." Eng. Record, vol. 60, p. 
240, 1909. 
2. E. G. Manahan and J. W. Ellms: "The Cincinnati Water Purifica- 
tion Plant." Eng. Record, Apr. 6, 1907. RAPID SAND FILTER-PLANT CONSTRUCTION 
199 
3. "Construction of the Minneapolis Filters." Eng. Record, May 18, 
1912. 
4. James W. Armstrong: "Baltimore's New Mechanical Filters." 
Eng. Record, vol. 69, p. 520, May 9, 1914. 
5. "Rapid Sand Filtration at Trenton." Eng. Record, May 9, 1914, 
p. 541. 
6. "Water Purification at Cleveland." Eng. Record, June 6, 1914, p. 651. 
7. "Water Filtration and Liquid Chlorine Plant, Bound Brook, N. J." 
Eng. News, vol. 71, p. 1036, May 7, 1914 (Pressure Filters). 
8. F. H. Stephenson: "Filter Valves Electrically Operated at Cincinnati." 
Eng. Record, vol. 59, p. 82, 1909. CHAPTER XVII 
DETAILS OF RAPID SAND FILTER-PLANT CONSTRUC- 
TION (CONTINUED) 
STRAINER SYSTEMS 
The strainer system of the filter tank serves the double purpose 
of collecting the water which has passed through the sand and 
gravel beds during filtration, and of distributing the water applied 
to cleanse the filter bed during the washing process. As the 
velocity of flow of the water is so much greater in the latter proc- 
ess as compared with the former, it becomes the governing factor 
in the design. A uniformly distributed upward flow under the 
whole area of the filter bed is the essential feature of successful 
washing, and the efforts of designers to approximate such a con- 
dition are clearly shown by the history of the development of this 
part of rapid sand filters. 
The simplest form of strainer system is a perforated false 
bottom of metal on which the filtering medium rests. Fine wire- 
cloth screens were first used for this purpose, but their failure was 
usually due to the clogging of the screen with sand, which rested 
directly on it, and to uneven distribution of the wash water by the 
screen. As the size of the filter units increased these difficulties 
became more pronounced. When the false bottom was con- 
structed of brass wire cloth or of perforated sheet brass, the cost 
of the strainer system was considerable and became a factor of 
importance with the increasing size of the filter tanks. 
To overcome uneven distribution and to reduce the cost, small 
hemispherical or otherwise-shaped brass nozzles or caps, which 
were either perforated with small holes or provided with very 
narrow slots, were used. These nozzles were screwed into a false 
bottom made of either wood or iron, or into a manifold of iron 
pipes (Fig. 65) which formed a grid on the floor of the filter tank. 
The sand rested directly on these "strainer cups," as they were 
called. In spite of making the openings in the cups extremely 
small, more or less trouble from clogging of these openings was 
experienced. Better distribution of the wash water, however, 
200 RAPID SAND FILTER-PLANT CONSTRUCTION 
201 
resulted since the flow was throttled in its passage through the 
small orifices, and entered the sand bed under a diminished but 
fairly uniform pressure. Wherever a manifold of pipes was em- 
ployed, it was found necessary to make them of ample size so 
that too much head should not be lost in them by friction before 
the water reached the distributing nozzle farthest removed from 
the main supply pipe. By restricting the neck of the nozzle, so 
that the flow of water was throttled before it entered the small 
openings of the cup, even better action was obtained. 
Fig. 65.-Typical arrangement of strainer nozzles and manifold for circular 
filter tank. 
Gravel Layers.-The difficulty arising from the clogging of the 
strainer cup openings with minute particles of sand was met by 
introducing graded layers of gravel around and above the strainer 
system. The sand rested upon the gravel layer; and if the latter 
was properly graded and not disturbed in any manner, it formed 
an efficient barrier to the entrance of sand particles into the per- 
forations or slots of the strainers. It then became possible to 
make these perforations or slots larger, and to produce a freer 
flow of water than could be obtained through the extremely small 
openings previously employed. 
The principle of using gravel was not new, as the old Hyatt 202 
WATER PURIFICATION 
cone sand valves, invented nearly 30 years ago, had made use of 
this idea. In these valves or strainers the screen at the top of the 
cone contained perforations of such size that sand could readily 
pass through them, but was kept from entering the system of 
collecting pipes by layers of shot or gravel placed between the 
upper and lower perforated plates of the cone valve. This shot 
or gravel also aided in distributing the wash water, and prevented 
jet action occurring to a considerable extent. 
Jet action is in a measure inevitable in almost any strainer 
system, but the more pronounced it becomes and the further up 
into the filter bed the jets of water are projected, the more un- 
desirable is the effect produced. Jets of water shot directly up 
through the bed wash a limited area immediately above each 
individual strainer. This effect is quite marked in some of the 
older filter strainer systems, and shows itself by depressed areas 
over the surface of the sand through which the greater portion of 
the wash water passes, leaving intermediate spaces unwashed. 
Perforated Plate and Pipe Systems.-In order to more uni- 
formly distribute the wash water, and especially to avoid the bad 
effect of jets of water washing limited areas of the filter bed, per- 
forated pipes or plates (Fig. 66) extending in parallel rows either 
crosswise or lengthwise of the filter have been constructed. The 
perforated-pipe system (Figs. 67 and 68) was employed with 
marked success in the Harrisburg, Pa., filter plant in connection 
with a graded gravel bed above the pipes. At Cincinnati, Ohio, 
perforated plates covering concrete channels located at the bot- 
tom of trough-like depressions, which ran lengthwise of the filter, 
have proven entirely successful. 
Hopper-shaped Filter Bottoms with Inverted Pyramidal or 
Prismoidal Depressions.-The perforated metal or wire-cloth 
bottom, theoretically at least, provided for water being discharged 
underneath every square inch of the gravel and sand of the filter 
bed. The employment of perforated brass nozzles, however, 
spaced some distance apart, afforded dead areas under which the 
water could not be applied. Gravel or sand lying between the 
nozzles was, therefore, imperfectly washed. To remedy this 
defect the shape of the bottom of the filter tank has in some in- 
stances been changed, so that the water enters at the bottom of 
depressions which are filled with gravel. As noted above in the 
Cincinnati filter bottoms, consisting of long parallel troughs with 
slanting sides, is found an example of this type of construction. RAPID SAND FILTER-PLANT CONSTRUCTION 
203 
HALF SIDE BLOCKS 
Fig. 66.-Details of perforated strainer system designed for high velocity washing at the .Evanston, Ill., filter plant. 
HALF MIDDLE BLOCKS 
Taper holes 
□ bars 
SECTION B-B THRU MIDDLE BLOCKS 
z Taper holes to be 
l tilled with Cement 
SECTION A-A THRU SIDE BLOCKS 
- Top of Gravel 
12"Graded, Gravel 
PART PLAN OF FILTER 
Concrete Screen Blocks \ 204 
WATER PURIFICATION 
Inverted pyramidal-shaped depressions with the strainer 
nozzles located at the bottom of the depressions have also been 
used and with more or less success. The disturbance of gravel 
beds above such bottoms was insured against in the Cincinnati 
SKETCH SHOWING ARRANGEMENT OF 
SAND AND GRAVEL 
Fig. 67.-Showing arrangement of perforated pipes, gutters, and filtering material in Cleveland 
filter plant. 
CROSS SECTION THRU' FILTER ON LINE E-F 
LONGITUDINAL SECTION THRU' FILTER ON LINE C-D 
SECTIONAL PLA^ OF ONE FILTER ON LINE A-B 
filters by the use of a screen of brass wire cloth, placed between 
the gravel and the sand, and which was fastened to the tops of the 
longitudinal ridges formed by the sides of the parallel troughs. 
In connection with filter bottoms containing pyramidal-shaped RAPID SAND FILTER-PLANT CONSTRUCTION 
205 
VIEW AT FLANGE CONNECTION HALF SECTION THRU BODY 
ENLARGED SECTION 
SECTION B-B 
SECTION C-C 
Fig. 68.-Details of perforated strainer piping and manifold header of Cleveland filter plant. 
LATERAL SUPPORT-C. I. 
1152-Req'd as shown 
LATERAL SUPPORT-C. I. 
576-Req'd. as shown 
DETAIL OF WEDGE BOLT 
AT END OF LATERALS 
>6 Dia. x 1% 1g. sq. hd. bolt 
threaded as shown. 
^"sq. nut fig. threaded 
SECTIONAL PLAN 
SECTIONAL ELEVATION 
Cast Iron Lateral \ 
SECTION THROUGH MANIFOLD HEADER AND LATERALS 
DETAIL OF CENTRAL MANIFOLD-C.J. 
288-Req'd-(complete of this detail) 
VIEW OF FLANGE CONNECTION 
FULL SIZE SECTION OF 
LATERAL AT D-D 
INVERTED PLAN OF 
CONNECTION AT OUTLET 206 
WATER PURIFICATION 
depressions, a recent design of Mr. William Wheeler of Boston, 
Mass., is of considerable interest, and is described below in detail. 
DESCRIPTION OF TYPICAL STRAINER SYSTEMS 
Strainer Systems Used at Toledo and Youngstown, Ohio, 
Filter Plants.-The filter tanks of the Toledo plant are 22.5 
by 16 ft. by 8.5 ft. (deep), and have a nominal capacity of 
1,000,000 gal. per day. The underdrain system consists of a 
central cast-iron rectangular manifold, into which are leaded 
2-in. cast-iron laterals spaced 10 in. center to center. Slotted 
brass strainers of the so-called Norwood type are screwed into the 
laterals 6.2 in. from center to center. The most constricted part 
of the strainer is in the tee, which is bored out to % in. in diame- 
ter. The aggregate opening of the 837 strainers in each filter is, 
therefore, 92.4 sq. in., or 0.18 per cent, of the total area of the sand 
surface. The laterals in these filters are bedded in cement, and 
only the tee-shaped strainers project above the flat filter floor. 
This system of strainers is now being removed and replaced by 
a perforated-pipe system. 
In the Youngstown plant the tanks are practically of the same 
size as those at Toledo. The main effluent manifold is a stand- 
ard-weight 10-in. cast-iron pipe laid lengthwise of the tank at 
the center, and imbedded in the concrete floor. Transverse 1^- 
in. collector pipes of wrought iron placed 6 in. apart are screwed 
into this manifold, and are imbedded in the concrete of the floor. 
Into these transverse collector pipes are screwed brass strainers. 
The arrangement of the strainers is such that they are on 6-in. 
centers in each direction, making 1,176 strainers in each tank. 
In many circular steel tank filters of the gravity type a false 
bottom plate, supported by standards and riveted to the side of 
the tank, is used. The brass strainer nozzles are screwed into 
brass-bushed holes in the plate, and are spaced about 6 in. apart. 
Some recent designs have made use of the false metal bottom 
plate provided with nozzles as above described, in rectangular 
filter tanks of concrete. False bottoms of concrete have been 
suggested, but the difficulties of designing a bottom to withstand 
the stresses to which it may be subjected, have not led to its 
adoption. 
Cincinnati, Ohio, Columbus, Ohio, and Minneapolis, Minn., 
Filter-plant Strainer Systems.-In the filter tanks of the Cincin- RAPID SAND FILTER-PLANT CONSTRUCTION 
207 
nati plant the bottoms consist of concrete channels located at 
the bottom of trough-like depressions running lengthwise of the 
tank. Each tank has 28 channels 3 in. deep, 2^ in. wide at the 
top and 2J^ in. wide at the bottom, spaced on 12-in. centers be- 
tween the concrete wedge-shaped ridge blocks forming the sides 
of the trough. The channels are connected at 12J4-ft. intervals 
by 3^-in. riser pipes to a system of large cast-iron pipes under 
the bottom of the filter tank. The perforated plates which cover 
the channels are held down by nuts on the threaded end of ^-in. 
hook bolts, slipped over cross-bolts spanning the channel. 
Strainer plates lap about in. These joints were made with 
red lead and the edges of the plates were cemented to the shoul- 
ders on which they rest with a Portland-cement mortar of 1 part 
of sand and 1 part of cement. The rise in the curvature of the 
plate is % in. Sixty-two holes %2 in- in diameter were drilled 
in each lineal foot of plate and serve a sand area of 1 sq. ft., mak- 
ing the strainer openings equivalent to 0.3 per cent, of the filtering 
surface. 
In the Minneapolis plant the strainer system includes a series 
of waterways made of parallel ridges and grooves cast in concrete 
on the filter floor. The ridges instead of running lengthwise of 
the filter extend crosswise or perpendicular to the central wash- 
water gutter. They are spaced on 1-ft. centers and their tops 
are 13 in. above the filter floor. The channels at the bottom of 
these trough-shaped depressions are covered with perforated brass 
plates, and the area of the openings in them is 0.15 per cent, of 
the sand area. The filtered water collected in the channels flows 
inward from the sides toward the center of each half of the filter 
unit, and is delivered through eight 14-in. outlets in the filter 
floor to a manifold of 18-in. cast-iron effluent piping. 
The strainer system of the Columbus, Ohio, filter plant differs 
from those at Cincinnati and Minneapolis in providing circular 
strainer plates about 2^ in. in diameter, perforated with Xe-in. 
holes, and bedded in slabs of concrete placed over channels of 
like material built on the floor of the filter. These channels are 
5 in. deep and 2^4 in- wide at the bottom. They are spaced 8% 
in. from center to center, and the perforated brass circular plates 
in the slabs, which cover the channels, are also spaced on 8^-in. 
centers. The lateral channels connect with the eight main col- 
lectors formed in the concrete bottom of the tank, which in turn 208 
WATER PURIFICATION 
connect through iron castings set in the filter floor with piping 
below the tank.1 
Harrisburg, Pa., Filter-plant Strainer System.-In this plant 
is found a very simple strainer system (Fig. 69) which was based 
on the idea that if a gravel underdrain system could be prepared 
that would hold the sand up under the actual conditions of 
operation, strainer cups of any sort were unnecessary, and under- 
drainage piping could be considerably simplified. The. system 
Section A-A 
Sectional Plan showing Assembly 
Front Elevation 
Fig. 69.-Details of special castings and pipe in underdrain system of 
Harrisburg, Pa., filter plant. 
as constructed consisted of a series of parallel lines of 1^-in. 
galvanized-iron pipe laid 6 in. apart and running crosswise of the 
filter, the pipes being drilled along their under surface with a row 
of holes %2 in- in diameter and 3 in. apart from center to center. 
The pipes were capped at their outer ends, and at the center 
entered a tee of special form, which connected with the side of a 
cast-iron manifold laid in the concrete floor of the filter. The 
manifold was connected with the main effluent pipe of the 
filter. 
1 Eng. Record, vol. 53, Feb. 24, 1906. RAPID SAND FILTER-PLANT CONSTRUC 
TION 
209 
"extending berrear-fh 
2"F/oor P/ank 
3x4" 
TieBeam 
Piaf 
Washer 
^Top & Bottom Re Rods 
Intermediate Tie Rods 
Plan 
3/q Tie Rod 
Top of Water 
T'R- vTRpofSand 
■^3 Tie Rod 
Sides and Bottom 
of ^'Surfaced Plank 
3"Concrete Spheres 
Tie Pod 
% Brass Pipes, 4 "/ong. 
3*4 Sill to support Floor 
Ver+icod Section A-B 
Fig. 70.-The Wheeler bottom for rapid sand filters. WA TER P URIFICA TION 
210 
Belfast, Me., and Akron, Ohio, Filter-plant Strainer Systems. 
-There has been recently developed a form of strainer bottom 
which departs materially from the general principles involved in 
the systems previously described. This system was invented by 
Mr. William Wheeler of Boston, Mass., and as described (Fig. 
70) by Mr. R. S. Weston1 consists of inverted truncated pyramids 
placed 1 ft. apart on centers. They are not deep enough to meet 
at their sides and form sharp edges. Each pyramid is provided 
with a 24-in. outlet. Over each outlet is placed one sphere 3 in. 
in diameter, made of neat Portland cement, and above this a layer 
of four similar spheres. Above the four spheres are placed nine 
blue glazed earthenware marbles, eight of which are about 1^4 
in. in diameter, and one (at the center) about 1%6 in- in diameter. 
Above the marbles are placed 6 in. of graded gravel all of which 
has passed through a 1-in. ring. In other words, the underdrain 
system consists of a system of spheres arranged like an inverted 
pile of cannon balls, with its center ball directly over the outlet 
pipe. 
The sphere directly over the outlet produces the so-called "ball 
nozzle" effect, and the tendency of the water to adhere to the 
sides and surfaces of the balls is that which produces so uniform 
a distribution of the incoming wash water in the washing process. 
This distribution begins near the point of entrance, and it is the 
uniform arrangement of the balls, as well as the adhesion of the 
water to their surfaces and to the sides of the pyramid, that per- 
fects the distribution before the sand layer is reached. This sys- 
tem will be discussed again in the description of methods of 
washing. 
References 
1. "Filter Plant at Milford, N. J., Hackensack Water Co." Eng. Record, 
vol. 50, p. 572. 
2. "Youngstown Filter Plant, Youngstown, Ohio." Eng. Record, vol. 52, 
p. 410. 
3. "Toledo Filter Plant, Toledo, Ohio." Eng. Record, vol. 62, p. 602. 
4. "Filter Plant at Bangor, Me." Eng. Record, vol. 63, p. 65. 
5. "Filter Plant at Minneapolis, Minn." Eng. Record, vol. 64, p. 586. 
6. "Filter Plant at Niagara Falls, N. Y." Eng. Record, vol. 65, p. 602. 
7. "Filter Plant at Albany, Ohio." Eng. Record, vol. 66, p. 193. 
8. "Experiences with Filter Underdrains." Eng. Record, vol. 69, p. 529. 
9. R. S. Weston: "The Wheeler Filter Bottom." Eng. News, vol. 72, 
p. 22. 
1 Eng. News, July 2, 1914, p. 22. CHAPTER XVIII 
DETAILS OF RAPID SAND FILTER-PLANT CON- 
STRUCTION (CONTINUED) 
DEVICES FOR USE IN AGITATING THE FILTER BED DURING THE 
WASHING PROCESS 
In the early development of rapid sand filters the proper 
cleansing of the sand bed was soon recognized as of the utmost 
importance. The forcing of water through the filtering medium 
in a reverse direction from that during filtration cleared the sand 
bed of a great deal of its accumulated sediment, but did not al- 
ways do it as thoroughly as was desired. The upward rate of 
flow of the wash water ranged from 8 to 12 in. per minute, and 
caused some slight expansion in the volume of the sand bed. In 
order to be certain that wash water had access to all parts of the 
bed, and that no sand escaped the cleansing action of the water, 
stirring devices were placed in the bed and were rotated during 
the application of the wash water. 
Mechanical Rakes.-As the earlier rapid sand filters consisted 
of cylindrical tanks, rotating rakes for agitating the sand bed in 
the washing process were found to be quite satisfactory. Since 
the filter bed consisted entirely of sand, it was possible to stir the 
bed to practically its full depth. The agitating devices usually 
consisted of a set of rake arms, each arm being provided with 
several teeth. The rake arms were hung from a vertical shaft, 
which was provided at its upper end with a horizontal gear, en- 
gaging either a worm or bevel gear. 
A small steam engine usually provided the power which it trans- 
mitted to a driving pulley. By means of belts and pulleys or 
gears, the movement of the vertical-shaft gears mentioned above 
was effected. Devices were also employed which enabled the 
rakes to be lowered into the bed as they rotated, and to be raised 
again just before the wash water ceased to flow up through the 
bed. 
The teeth of the rake arms were from 3 to 4 ft. in length, and 
they revolved at a rate of eight or nine times a minute. The 
211 212 
WATER PURIFICATION 
teeth descended into the bed to within 2 or 3 in. of the bottom. 
The teeth were square or wedge-shaped in section, and were 
spaced about 6 in. apart. Sometimes shorter teeth were used, on 
the end of which were attached chains. These chains were long 
enough to drag on the sand and were used for surface agitation 
12-l'bolts-circle I7'd 
Cast Iron^ , 
Fig. 71.-Details of air and water systems designed for the Jerome Park filter plant. 
New York. 
Section FF 
Air Vent Check Valve 
. Section C-C 
Cast Iron - r-0 
Section B-B > 
Strainer and air systems 
Manifold Outlet 
Section D-D 
Section A-A 
Section E-E 
of the sand bed when it was not desired to agitate the sand by- 
driving the rake teeth into the bed. 
Systems of Air Agitation.-With the advent of concrete tanks 
for filter purposes, which could be built so much more economically 
in a rectangular form, and which could be made so much larger 
than the circular wooden or steel tanks, the rotating mechanical 
agitating devices were soon found to be unfitted for use in them, RAPID SAND FILTER-PLANT CONSTRUCTION 
213 
Raw Water Dram 
20"Filtered 
Water 
Copper Ridcje Roll 
Fig. 72.-Cross-section of Columbus, Ohio, filters showing air piping and water piping. 
Pump and 
Motor 
Section A-B. 
Wafer Line El, 60,0 
Concrete Coping 214 
WATER PURIFICATION 
as they were cumbersome and awkward to operate. The use of 
air under low pressure was suggested as offering a means of agitat- 
ing the filter bed, and its employment soon became quite general. 
Two general methods are in vogue for distributing air under 
pressure to filters for agitating purposes. In one method the air 
is supplied through a separate system of pipes and exits to the 
bottom of the sand bed (Fig. 71), and in the other the main 
supply pipes deliver the air to the underdrain system of water 
pipes of the filter, from which it escapes through the strainers into 
the filter bed. Because of the disturbing effect of the ascending 
air on the finer gravel layers of the filter bed, the air-delivery pip- 
ing system is sometimes placed between the gravel and sand 
layers. Obviously, this can not be done if the water strainer 
system also serves as an air-delivery system. 
The main air-supply pipe coming from the blower to the filters 
is usually of cast iron. Branch distributors pass over the tops of 
the filter tanks and then downward, connecting with the inde- 
pendent distribution piping (Fig. 72), or with the water-strainer 
system, when the latter serves the double purpose of distributing 
both air and water as described above. Cast-iron piping is 
usually employed for air distribution up to the laterals both in 
the separate and the combined systems. In the separate system 
the laterals are usually seamless brass tubing perforated with 
openings of ^6 in. diameter, and spaced about 6 in. apart. 
Air is usually forced into the bed at the rate of 3 to 5 cu. ft. per 
minute of free air per square foot of filter surface. Usually the 
application of air precedes the application of the water, but in 
some cases the two are applied together. 
Depth of Sand and Gravel.-In the earlier rapid sand filters 
when sand alone was used in the filter bed, its depth ranged from 
30 to 50 in. Usually more or less sand was lost in washing, de- 
pending upon the violence of the agitation with the mechanical 
rakes and the upward velocity of the wash water. The shorter 
the distance between the normal level of the sand and the edge of 
the waste-water troughs, the greater the amount of sand lost 
during the washing process. Consequently, variations in design 
and methods of operation resulted in considerable differences in 
the actual depth of the sand bed. Renewals of the sand bed were 
more or less frequent, and a bed which had originally a depth of 
4 ft. might decrease to 2^ ft. before more new sand was placed 
in the tank. RAPID SAND FILTER-PLANT CONSTRUCTION 
215 
Present practice usually provides a depth of sand ranging from 
30 to 36 in. Depths less than 30 in. may be used successfully if 
the sand is fine enough. Deeper layers are usually provided 
where the sand is relatively coarse. 
The size of the sand particles is of importance if good purifica- 
tion is to be obtained. Sand with an effective size of only 0.26 
mm., and a uniformity coefficient of 1.68 is being used in a 30-in. 
layer in a plant recently constructed. Sands having an effective 
size of 0.5 to 0.6 mm. are frequently used, and as they offer much 
less frictional resistance to the flow of water, are probably efficient 
with certain classes of waters. Coarse sand was quite commonly 
employed in the earlier filters. Later, the use of a finer sand 
came into vogue. Present practice appears to be again swinging 
to the use of the coarser sands. With the early filters the re- 
quirements as regards clarification and bacterial removal were not 
as rigid as they became later when they began to be used for the 
purification of municipal water supplies. The use of sterilization 
agents of late years has again relieved the filters of such exacting 
demands in the way of bacterial purification, and provided the 
sand is fine enough to properly clarify the water, it is usually suit- 
able to effect the degree of bacterial removal demanded in con- 
junction with disinfecting agents. 
The proportion of very fine material in a sand should not be 
too high. For example, it is sometimes specified in buying sand 
that not more than 1 per cent, shall be finer than 0.25 mm. 
Where it is possible to wash the sand bed at a high rate after it 
is placed, the amount of fine material need not be given very 
much consideration, as it is very easy to wash it out by repeated 
washings, assisted by a light scraping of the sand surface between 
washings. The hydraulic grading of the bed by this method of 
washing makes this separation of the fine material easy and 
economical. 
Gravel layers vary in depth, but usually range from 6 to 10 in. 
Much deeper gravel layers are used where certain methods of 
filter washing are in operation, and as high as 18-in. layers have 
been employed in some instances. For high-velocity methods of 
washing probably 14-in. layers are more commonly used, since 
this appears to be about the depth required to suppress the jet 
action of the rising wash water. Gravel layers are usually 
graded, and the sizes of the stones vary from 3 in. in diameter in 
the bottom layers to Xe in- in diameter in the top layers. 216 
WAT FAI PURIFICATION 
The following table gives the depths and sizes of sand and gravel 
used in four modern rapid sand filter plants: 
Depths and Sizes of Sand and Gravel Used in Various Rapid Sand 
Filter Plants 
Filter plants at 
Cincinnati, 
Ohio 
Harris- 
burg, Pa. 
Columbus, 
Ohio 
Columbus, 
Ind. 
Depth of sand 
30 in. 
30 in. 
30 in. 
30 in. 
Effective size 
0.34 mm. 
0.39 mm. 
0.41 mm. 
0.35 mm. 
Uniformity coefficient : . . . 
1.60 
1.40 
1.36 
1.60 
With or without wire screen between 
gravel and sand 
With 
Without 
Without 
Without 
Depth of gravel 
9K in. 
7 in. 
10 in. 
18 in. 
Sizes of particles, inches 
Depths of the various layers, inches 
IK to 3 
- 
- 
10 
1 to IK 
- 
- 
- 
4 
K to 1 
1 
- 
2 
- 
K to % 
- 
4 
- 
- 
7 
2 
3 
to 
K 
- 
3 
X 2 to K 
3 
1 
K2 tO K 
1 
- 
Ke to % 
- 
- 
3 
- 
Screens between the Gravel and Sand Layers.-The disturb- 
ance of the gravel layers of a filter is obviously undesirable, if 
dependence is placed upon them to maintain an efficient barrier 
between the sand bed and the strainer system. As a support 
for the sand bed, or as a distributor of wash water, the gravel bed 
must at all times remain properly graded. The partial "inver- 
sion of the bed," that is the formation of pockets of sand or of 
gravel reaching from the surface of the filter bed to the strainer 
system, may produce poor purification results. This condition 
of the filter bed has been frequently found. Filter beds thus dis- 
turbed could not be properly washed, since differences in frictional 
resistance to the flow of water between the disturbed and undis- 
turbed portions of the sand and gravel layers made uniform 
cleansing of them impossible. For the same reason uniform rates 
of filtration could not be maintained over the whole area of the 
filter bed. RAPID SAND FILTER-PLANT CONSTRUCTION 
217 
The cause for the initial disturbance of the gravel bed was 
usually found in the strainer system, which produced such a pro- 
nounced jet action at certain points in the bed as to gradually 
displace the gravel and permit the sand to drop back into the 
space thus produced. The injection of compressed air into a 
filter bed for the purpose of agitating the sand layer also assisted 
in displacing the gravel, especially when the air entered through 
the water-strainer system below the gravel bed. Injection of 
air and water together caused even greater displacement effects 
than air alone. 
To remedy these difficulties, the use of a brass wire-cloth screen 
placed and rigidly held between the gravel and sand layers was 
tried. The results obtained were entirely satisfactory, since jet 
action of the strainers was minimized, and since the gravel bed 
was maintained in the original condition in which it was laid. 
The use of the screen also led to another interesting development 
in the method of washing a filter, now known as the "high- 
velocity method," and which will be described later under 
methods of operation. 
The screen was first used in the Cincinnati filter plant, and 
resulted from extended hydraulic experiments undertaken prior 
to the design of this plant. In constructing the filter beds of this 
plant the gravel was first carefully placed in the longitudinal 
troughs, filling the latter to the top. The brass wire cloth was 
then laid on the surface of the gravel layer and fastened to the 
tops of the ridges with nuts and washers on brass bolts, which 
latter had been previously set in the tops of the ridges when they 
were molded. The bolts were in. in diameter and were spaced 
18 in. center to center. The wire cloth had been woven into 
rolls 50 ft. long and 26 in. wide. It was made with 100 meshes 
per square inch, and the wires were of size No. 20, "Old English 
gage." The brass contained 25 per cent, of zinc and 75 per cent, 
of copper. 
The 26-in. widths of wire cloth extended across two troughs 
or depressions. Where two widths lapped, the edges were further 
fastened by driving double-pointed copper tacks through the two 
layers. The points of the tacks bent over after passing through 
the cloth by striking the concrete, and thus clipped the edges 
together. The methods of fastening down the screen proved to 
be inadequate after a year or so of operation. At other large 
plants where the screen is used, better methods of fastening have WATER PURIFICATION 
218 
been employed and no trouble from tearing away of the screen 
from the bolts and fastenings at the seams has been experienced. 
The life of the brass wire cloth has proven to be very short with 
certain waters. At Cincinnati and Louisville, Ky., the brass wire 
cloth in the filters has corroded badly and has finally become 
worthless. At Grand Rapids, Mich., and at New Orleans, La., 
on the other hand, the screen has not corroded and is still in ex- 
cellent condition. In both of the two last-mentioned filter plants 
the method of treating the water produces a water having prac- 
tically neither free nor half-bound carbonic acid, while at the 
two plants where the screen has corroded badly, the method of 
treating the water is such that in one case no free CO2 is present, 
but half-bound carbonic acid is always present (Cincinnati), and 
in the other both free CO2 and half-bound carbonic acid are 
always present (Louisville). It would seem as though the cor- 
rosive action of the water was due to the small amount of half- 
bound carbonic acid present. 
It has been found possible to avoid the use of the brass wire 
cloth by deepening the gravel layers, and by making them of 
somewhat larger material at the bottom. This phase of the de- 
sign will be further discussed under methods of operation. 
Velocity of Flow of Water through Sand Strainers and Under- 
drains.-In the design of the Jerome Park filters for the purifica- 
tion of the Croton water supply of New York City, several types 
of strainer systems were worked out. Mr. George W. Fuller1 has 
briefly described these designs and given a table showing the work- 
ing velocities for the flow of the water during filtration and filter 
washing for the different types of design. 
"Type A consisted of cast-iron manifolds, cast-iron laterals with 
brass strainer cups for washing at the rate of 9 gal. per minute per square 
foot of sand surface, and a separate air system for air wash through 
cast-iron manifolds, and brass laterals, at the rate of 4 cu. ft. per minute 
per square foot of sand surface. 
"Type B provided for the application of both air and water through 
the same system, using the Williamson strainer. This strainer is a 
cup screwed into the top of the pipe with an elongated neck projecting 
downward to the bottom of the pipe. 
"Type C consisted of concrete channels and perforated brass plates 
for applying water only at the rate of 15 gal. per minute per square foot of 
sand surface. The sand and gravel were separated by a brass wire screen. 
George W. Fuller: "The Croton Water Supply: Its Quality and 
Purification." Jour. Am. Water-works Assn., March, 1914. RAPID SAND FILTER-PLANT CONSTRUCTION 
219 
"In the following table are given the working velocities through the 
various parts of the three strainer systems for filtration and washing." 
Jerome Park Filters. Strainer and Air Systems-Working Velocities, 
Feet per Second 
Filtration 
W ashing 
Air (free) 
Type A: 
Through sand 
0.00443 
0.02 
0.0668 
Strainer holes or air holes 
1.50 
6.75 
260.0 
Strainer neck 
2.20 
10.00 
Laterals 
0.71 
3.18 
86.0 
Manifold 
1.11 
5.00 
79.0 
10-in. supply , 
1.46 
6.60 
18-in. collector 
1.80 
8.15 
24-in. main 
2.03 
9.17 
16-in. air main 
68.8 
Type B: 
Through sand. 
0 00443 
0 02 
0 0668 
Strainer holes 
1.55 
7.10 
23.40 
Strainer neck 
2.15 
9.70 
360.00 
Laterals 
0.58 
2.60 
8.68 
Manifold 
1.11 
5.00 
8.32 
10-in. supply 
1.46 
6.60 
18-in. collector 
1.80 
8.15 
24-in. main 
2.03 
9.17 
16-in. air main 
68.80 
Type C: 
Through sand 
0.00443 
0.0333 
Strainer holes 
1.41 
10.80 
Laterals 
0.26 
1.98 
Manifold channel 
0.40 
3.08 
14-in. supply 
0.73 
5.60 
24-in. collector 
0.99 
7.61 
24-in. main 
2.00 
15.20 
1. George W. Fuller: "Report on Water Purification at Louisville, 
Ky." 
2. George W. Fuller: " Report on Water Purification at Cincinnati, Ohio." 
3. Eng. Record, vol. 63, p. 65. 
4. Eng. Record, vol. 64, p. 379. 
5. Eng. Record, vol. 69, p. 529. 
6. George W. Fuller: "Some Features of Detail in the Design of Rapid 
Sand Water Filtration Plants." Engineering-Contracting, Sept. 2, 1914. 
7. George W. Fuller: "The Croton Water Supply." Jour. Am. 
Water-works Assn., March, 1914. 
References CHAPTER XIX 
REGULATING, MEASURING AND INDICATING DEVICES 
FOR RAPID SAND FILTER PLANTS 
RATE-OF-FLOW CONTROLLING APPARATUS 
Apparatus for controlling the rate of flow of water through 
rapid sand filters are now usually found in all modern and well- 
equipped plants. Sudden variations in the rate of filtration, es- 
pecially where the change is from a lower to a higher rate, are 
likely to produce poor purification results. If, instead of a fixed 
rate of discharge for the whole period of service of the filter, a 
changeable rate of flow which meets variations in the consump- 
tion of filtered water, is desired, it is well that the change of rate 
should be gradual so as not to subject the filter bed to suddenly 
increased or decreased pressures. It can be easily understood 
that a suddenly increased pressure might cause the bed to break 
through. It is also conceivable that rapidly decreasing the rate 
of flow might be injurious, since air which may be held in the bed 
at the higher rate might be liberated at the lower. This air 
would escape by breaking through the surface film and produce 
in the beds points of lessened frictional resistance, through which 
poorly purified water could escape to the underdrains. 
The rates of flow generally permitted through rapid sand filterr 
vary considerably. They may range from 75,000,000 to 150,000,- 
000 gal. per acre per day, although the usual rates are from 
100,000,000 to 125,000,000 gal. per acre per day. As the initial 
minimum frictional resistance to the flow of water through a 
clean filter at the usual rates of filtration will vary from 1 to 3 
ft., and the maximum resistance of the dirty filter from 10 to 12 
ft., it is evident that controlling apparatus must be constructed 
so as to operate within these limits, either to produce a uniform 
or variable rate of flow as may be desired. 
The loss of head through the passages of the rate controller 
at its normal capacity should be low, that is from 6 to 9 in. 
Surging of the water through the controller indicates a lack of 
sensitiveness to changing rates of flow and is undesirable. The 
220 REGULATING RAPID SAND FILTER PLANTS 
221 
best designed controllers will maintain the rate of flow within 
2 or 3 per cent, of the mean rate for which the controller is set. 
Means for adjusting the rate by changing the size of the orifice 
through which the water is controlled are usually provided in all 
well-equipped machines of the orifice type. In the Venturi type 
of controller this adjustment of rate is taken care of in various 
ways as will be shown in the description of typical machines. 
Rate controllers may be calibrated by noting and marking the 
sizes of the openings in the measuring or controlling orifices when 
the volume of water flowing through them is that desired. If 
facilities for measuring the volume of water discharged through 
these orifices for various settings of the controller are at hand, 
this calibration may be done before the controller is set in place. 
If this is not possible, the filter tank itself may be utilized by 
noting the lowering of the water in the tank with the influent 
valve closed, and calculating from the size of the filter tank the 
volume of water flowing through the controller for any given 
length of time. 
A Simple Method of Rate Control.-The rate controller is 
placed in the main discharge pipe of the filter between the filter 
tank and the filtered water effluent pipe which collects the water 
from all the filters. The depth of water over the sand bed is 
usually maintained fairly constant. This latter condition is not 
complied with, however, in a very simple form of control recently 
designed.1 This consists of a disk inserted in the discharge out- 
let of the filtered water pipe. This disk has an orifice of such a 
size that the rate of flow desired will be at the maximum rated 
capacity of the filter when the sand is clean and the head of water 
above the sand at some minimum level. If another orifice plate 
under a constant head of water is placed in the influent pipe of 
the filter, the delivery of water to the latter can be maintained 
at the same rate as the discharge. As the filter becomes clogged 
the height of the water above the sand will increase, thereby in- 
creasing the head on the outlet orifice and tending to maintain 
a practically constant discharge. By this method a positive 
pressure is maintained at all times on the outlet orifice. The 
gradually rising level of the water above the sand becomes a 
direct measure of the frictional resistance produced by the clog- 
ging of the sand. When the water above the sand has reached 
1 "Revamped Water Works and New Purification Plant at Fort Smith, 
Ark." Eng. Record, Sept. 5, 1914, p. 262. 222 
WA TER P U RIF IC A TION 
its maximum level, the filter must, of course, be stopped and 
washed, if the rate of flow is to be maintained. 
The above principle of control is not the most common one in 
use. The rate at which water will pass through a clean sand 
bed and through the usual system of underdrains, filter effluent 
piping and valves is far in excess of the safe rated capacity for 
rapid sand filters. It, therefore, is necessary to restrict the rate 
of discharge by introducing some frictional resistance at the out- 
let of the filter. This is easily done when the filter is placed in 
service by partially closing the effluent valve until the maximum 
rate desired is produced. By opening this valve a little at a 
time during the period of service of the filter, the increasing fric- 
tional resistance in the sand bed, due to clogging of the interstices 
of the sand layers, may be compensated for by reducing the fric- 
tional resistance through the outlet valve. When the latter is 
wide open the rate can no longer be maintained if the head lost 
in the sand bed reduces the effective pressure below that required 
to produce the desired rate of flow. It is at this point that wash- 
ing the sand must take place if the rated capacity of the filter 
is to be maintained. However, it is sometimes the case that 
operators prefer to obtain as much filtered water as possible from 
a filter between cleanings, and, therefore, permit the filter to 
operate at a decreasing rate of discharge until the volume of 
water obtained becomes too small to make it worth while to 
continue filtering. 
Automatic Rate Control.-The proper hand regulation of the 
effluent valve is obviously dependent upon the skill and attention 
of the attendant. It can be much better effected by automatic 
control. There are quite a number of automatic devices now in 
use which successfully control the throttling of the effluent valve 
itself, or more commonly of a specially balanced valve which is 
more sensitive to the action of the auxiliary apparatus required to 
move the valve. The auxiliary apparatus usually depend for their 
action either upon a float operating in an open chamber (Fig. 73) 
and moving directly various types of valves which control the 
flow of water from the filter; or upon changes in the difference in 
pressure upon two sides of a fixed orifice, which force is directly or 
indirectly employed to actuate valves controlling the flow; or 
upon changes in the difference in pressure in the full and con- 
stricted portions of a Venturi tube (Fig. 74), likewise utilized to 
directly or indirectly operate a valve controlling the outflow of REGULATING RAPID SAND FILTER PLANTS 
223 
Fig, 73.-Rate controller operated by float of disk type. 224 
W A TER P U RIF IC A TION 
water from the filter. In the float type of controller a movable 
weir is generally used. This type of controller can not be oper- 
ated submerged. The other types of controllers, as a rule, trans- 
mit the slight differences in pressure produced by the orifice or 
Venturi tube to auxiliary devices, which in turn make use of 
electrical or hydraulic power to actually move the controlling 
valve. This class of controllers may be operated submerged. 
The manner in which these devices operate can be more easily 
understood by a description of some of the typical rate controllers 
now in use. 
Fig. 74.-Venturi type of rate controller. 
Weston Rate Controller.-One of the earliest rapid sand filter 
controllers to be used was that invented by Mr. E. B. Weston1 
(Fig. 75). This controller, as may be seen from the cut, is of the 
open type, and consists of a chamber containing a float which is 
attached by levers to butterfly valves, and which admit the water 
from the inlet pipe to the controller. The float in the controller 
body is mounted on a hollow stem, and moves in guides located 
at the top and bottom of the controller casing. Beneath the float 
is a deflector designed to quiet the water and to diminish the 
effect of currents, thereby giving a smooth entrance to the dis- 
charge tube, and being aided in this respect by the flaring ring at 
the mouth of this tube. Mounted also on the float stem at a 
fixed distance below it, so as to be maintained at a constant depth, 
is a disk, which is turned with a thin edge and sharp corners, and 
1 Eng. Record, Nov. 25, 1899, p. 596. REGULATING RAPID SAND FILTER PLANTS 
225 
of such diameter as will give the annular orifice between the disk 
and the walls of the discharge tube a predetermined area propor- 
tional to the desired rate of discharge. As the float is mounted 
on the stem at a fixed distance from the disk, and as the float is 
free to move up and down with the water, a constant head of 
water is, thereby, maintained above the annular orifice formed by 
the disk and walls of the discharge tube. The rising float tends 
to close the butterfly valves and thus decrease the inflow to the 
controller from the filter and vice versa. 
Fig. 75.-Weston rate controller. 
Controllers at the Hackensack Filter Plant.-A somewhat later 
design of controller than that of E. B. Weston's was installed at 
the Hackensack filter plant (Fig. 76). This controller is also of 
the open type and is operated by a float which moves a balanced 
inlet valve. Each controller was set in a concrete-steel tank 6 
by 6 ft. by 5 ft. 10 in. (deep) with its outer end rounded, and with 
its flattened inner end against the inner wall. The top of the 
tank was 2 ft. above the high-water level in the filtered-water 
reservoir. The inlet to the controller from the filter was at the 226 
WATER PURIFICATION 
bottom of the former through a 20-in. pipe casting. To the upper 
end of the pipe casting was bolted the body casting of the bal- 
anced valve, and to the upper end of the latter was attached the 
body of the controller. This portion of the device consists of an 
outlet pipe and an adjustable cylindrical brass weir, connected 
above to a float and below to the balanced valve previously 
mentioned. 
The balanced valve body has two sets of openings, one above 
the other, separated by a metal barrel 3 in. high. In each set are 
3-Arm Cl Spider 
fit "Diam Weir Rod to Wheel Stand 
in Operating Gallery 
Guide to prevent 
Rotation 
Effluent Inlet . 
from Filter. 
Fig. 76.-Rate controller used at the Hackensack, N. J., filter plant 
six openings 2J^ in. high, separated only by narrow vertical bars 
of the metal. The valve cylinder has but one set of openings at 
its middle. These openings are 2% in. high. The cylinder is 
so attached to the valve rod that when it is in its central position 
all the annular openings just mentioned are closed. The extreme 
travel of the cylinder is 2^ in. vertically up or down, and is oper- 
ated as previously stated by the float in the chamber above. 
A brass cylinder with a flaring top 20^ in. in diameter is at- 
tached to the stem connecting the float and balanced valve, and 
forms the weir over which the water passes to the outlet of the REGULATING RAPID SAND FILTER PLANTS 
227 
controller. By means of a hand wheel on a stem extending from 
the weir cylinder to the operating floor, the distance between the 
float and weir edge may be adjusted, and thereby fix the head of 
water upon the weir, and consequently the rate of filtration. 
When adjusted for any given rate the rising and falling of the 
float automatically decreases or increases the amount of water 
admitted through the balanced valve, and hence maintains the 
rate of flow for which the weir is set. 
Vivian Rate Controller.-In connection with the construction 
of the Cincinnati filtration plant a rate controller was designed 
Fig. 77.-Sectional view of Vivian rate controller 
by Mr. Simon Vivian, which in practice has given very satisfac- 
tory results. This controller is of the submerged type, and con- 
sists of a balanced valve (Fig. 77) moved by hydraulic pressure. 
Its distinctive feature is its positive control of the balanced valve 
by an external source of hydraulic pressure. This external pres- 
sure is applied and released by the change in the difference in head 
of the water, whose flow is being controlled, upon the up- and 
downstream sides of an orifice, the size of which may be adjusted 
by means of a gate. This control is effected through the me- 
dium of an auxiliary valve (Fig. 78), hung from a flexible rubber- 228 
WA TER P U RIF IC A TION 
cloth diaphragm, whose movement upward or downward is 
brought about through the above-mentioned change in the 
difference in head upon the up- and downstream sides of the ori- 
fice. The diaphragm and pendant valve are located without the 
main body of the controller in a separate casting. 
By reference to the cut (Fig. 79) it will be seen that the 
main body of the controller containing the balanced valve is at- 
tached directly to the effluent pipe of the filter. The water 
entering the controller strikes the dead end of the inner casting 
and is diverted upward and downward through ports in which 
Valve Body- 
■A uxiHory D Valve 
D Valve 
Cover"1 
D-Valve Body. 
Fig. 78.-Sectional view of auxiliary valve of Vivian rate controller. 
are moving the disks of the plunger throttling valve. The latter 
consists of a shaft to which are attached two circular disks, so 
placed on the shaft as to practically close the ports when entering 
the latter. The water after passing the ports enters the outer 
body of the controller and flows out through the adjustable rate 
valve or gate. 
The plunger throttling valve is moved upward by an external 
source of hydraulic pressure exerted within the cylinder at the 
bottom of the controller body, and acts on the lower end of the 
plunger shaft to move it upward. When the pressure is released, REGULATING RAPID SAND FILTER PLANTS 
229 
the plunger falls of its own weight. The cylinder at the top is 
merely a guide box to insure perfect alignment of the shaft. 
The application and release of this hydraulic pressure is brought 
about by any change in the difference in pressure upon the up- 
and downstream sides of the rate gate, and is communicated by 
pipes to a small cast-iron box divided into two parts by a flexible 
rubber-cloth diaphragm. Pendant from the diaphragm is a 
cylindrical valve provided with D-shaped ports. This valve is 
free to move up and down through a bronze-lined casting attached 
to the bottom of the diaphragm box. This valve box is provided 
with three pipe openings. One opening is connected to the source 
of hydraulic pressure, one with the bottom of the pressure cylinder, 
in which moves the end of the plunger throttling valve, and one 
with a waste or drain. At certain positions of the diaphragm, the 
auxiliary-valve ports permit water to pass from the pressure pipe 
to the cylinder, and thus raises the plunger of the throttling valve 
and cuts off or reduces the flow through the controller; at the 
other extreme position the communication from the pressure line 
is cut off, and that with the cylinder is opened through the ports 
Fig. 79.-Vivian rate controller, outside view. 230 
WATER PURIFICATION 
of the auxiliary valve to the waste pipe, whereby the pressure in 
the cylinder is released and the plunger throttling valve drops. 
This movement of the auxiliary valve depends upon the move- 
ment of the flexible diaphragm, the lower side of which is open to 
the pressure of the water in the body of the controller above the 
rate gate or orifice, and the upper side of which is open to the 
pressure of the water on the downstream side of the rate gate. 
A balance is thus effected, which tends to maintain, for any given 
opening of the rate gate, a constant difference in pressure between 
its up- and downstream sides. The diaphragm virtually floats 
in the water and tends to assume a neutral position, which by 
neither letting in water under pressure to the under side of the 
plunger shaft, nor letting it out, holds the latter in a practically 
fixed position for any given head acting through the inlet of the 
controller. A change in this head causes a readjustment of the 
position of the plunger. 
For example, an increase in the head above the rate gate raises 
the diaphragm and in consequence the plunger of the throttling 
valve, thus checking the flow and decreasing the head. On the 
other hand, an increase in the head below the rate gate depresses 
the diaphragm and releases the pressure of the water in the cylin- 
der of the throttling valve, and causes it to drop, thereby de- 
creasing the loss of head and again tending to establish a constant 
difference in head for any given opening of the rate gate. This 
is, of course, the condition necessary for a constant rate of flow. 
The auxiliary valve of this controller has been effectively em- 
ployed in connection with a hydraulic valve in place of the bal- 
anced valve, and using an orifice plate instead of a rate gate. 
This of course decreases the first cost of the apparatus, since the 
hydraulic valve may be used both as the effluent valve of the 
filter and the throttling valve to control the flow. 
Simplex Rate Controller.-The type of controller which de- 
pends upon changes in the difference in pressure in the full and 
contracted sections of a Venturi tube is illustrated by the Simplex 
rate controller (Figs. 80 and 81). In this apparatus a Venturi 
tube forms part of the effluent pipe of the filter. On its down- 
stream side is placed a balanced valve. This valve, as shown in 
the cut, consists of an outer and inner casting; a balanced valve, 
which moves up and down in the inner casting; annular ports 
for the escape of the water from the outer to the inner casting; 
a weighted arm attached to the stem of the valve in order to REGULATING RAPID SAND FILTER PLANTS 
231 
counterbalance its weight; and a flexible diaphragm attached to 
the lower end of the stem of the balanced valve. The diaphragm 
forms a closed chamber at the bottom of the outer casting. This 
chamber is connected by a pipe with the throat of the Venturi 
tube. The upper side of the diaphragm is open to the pressure 
of the water in the outer casting. 
The weight upon the arm attached to the stem of the balanced 
valve is movable, and by setting it in the proper position for the 
rate of flow desired will maintain the correct port openings in the 
balanced valve. An increased velocity through the constricted 
portion of the Venturi tube decreases the pressure at this point. 
This diminished pressure is transmitted through the pipe to the 
Fig. 80.-Simplex filter rate controller. 
under side of the diaphragm. For any balanced position of the 
valve this diminished pressure underneath the diaphragm would 
cause it to lower slightly, and thus pull down the balanced valve 
and partially close off the annular ports through which the water 
is escaping. By throttling the outflow in this manner, the veloc- 
ity through the Venturi tube would be diminished, and the pres- 
sure at the throat increased. This increased pressure, being 
transmitted to the under side of the diaphragm, would have the 
effect of moving it back to its original position. In this manner 
the size of the orifice is adjusted automatically to maintaining 
the desired rate of flow from the filter. 
Earl Rate Controller.-This controller, invented by Mr. George 
G. Earl, depends for its action upon the difference in pressures 232 
WATER PURIFICATION 
transmitted by a piezometer and Pitot tube, which arc located 
in the constricted throat of a pipe forming part of the effluent 
main of the filter; or it can as well be used with any other pres- 
Fig. 81.-Arrangement of Simplex filter rate controller with Simplex rate of flow and loss of head gauges. 
sure or pressure difference which varies with the volume of flow 
to be governed. Where a Pitot tube and piezometer are used the 
former carries water at a level corresponding to the static and 
velocity heads, while the latter carries water at a level equal to REGULATING RAPID SAND FILTER PLANTS 
233 
the static head only. The difference in heads, therefore, repre- 
sents the velocity head in the restricted throat and varies with 
the quantity of water passing through it. 
The mechanism by which this principle is made to control the 
flow of water consists of three tubes in which hang cylindrical 
Rate of Flow 
Million Gallows 
per 24 Hours 
Equalizing 
Connection 
Mercury Balances 
Venturi Head 
Fig. 81a.-Simplex rate of flow gage, showing connections to controller. 
weights. These weights are suspended by rods from a bar sup- 
ported on a knife edge. One end of the bar projects between two 
electrical contacts. When the bar is in contact with either ter- 
minal an electrical circuit is formed, whereby an electrically op- 
erated pilot valve causes the hydraulic throttling valve in the 234 
WATER PURIFICATION 
effluent pipe to be either slightly opened or closed, depending 
upon the terminal in circuit. The cylindrical weights are in 
balance when the water about weights 1 and 2 (see diagram) 
stands at the same level, whether that level be high or low, and 
Stop Valve 
^Throttle Valve 
fro Clear Well 
ASSEMBLY OF PARTS 
FOR THROAT 
Accelerator 
Battery 
^'Plug 
«-• Pressure 
£ Pipes 
"r Drain 
MASTER CONTROLLER 
|x5 Zero Filtration. 
Fig. 82.-Earl rate controller. 
Zero " 
Filtration 
I Maximumy 
Filtratk>n|T 
(Electrically oper- 
h ated Four-Way 
Hydraulic Valve 
Battery 
Maximum 
' Filtration 
Operating 
y ,^°°r 
System to be in balance 
when water stands at heights 
X1 X2 Xs X4 X5 
To all filter controllers 
FILTER CONTROLLER 
Maximum 
Water Level 
No.2 
Float Tube 
/ 
Minimum 
Water Level 
ELECTRICAL CONNECTION AT END OF BEAM 
HALF SIZE 
Maximum height 
of water in I 
clear well 
Top of Filter 
the water about weight 3 stands at the level marked "zero rate 
of filtration." 
By reference to the drawing (Fig. 82) it may be seen that tubes 
No. 1 and No. 2 extend above the maximum water level in the 
filters, and that they are connected to the Pitot tube and the 
piezometer, respectively, by pipes. Under operating conditions REGULATING RAPID SAND FILTER PLANTS 
235 
the water assumes levels in these two tubes corresponding to the 
respective pressures in the Pitot tube and the piezometer. The 
Filter Controller 
Furnished with each Filter 
Adjustable Balance 
- Rate of Flow Gauge 
Master Controller 
Furnished when Specified 
Adjustable BalanceZ 
Gauge showing Rate 
of Filtration 
Valve to Decrease Rate of 
Filtration Independent \ 
of Clear Well Floats 
Valve to Increase Rate of \ 
Filtration Independent 
• of Clear Well Floats 
Valve to Increase Rate ] 
Independent of Master Controller 
Valve to Decrease Rate Independent 
/ of Master Controller 
Valve to Open or Shut 
Automatic Operation 
Automatic Pilot Valve - 
Valve to Open or Shut 
Automatic Operation 
Thia Floor must be on or above 
the Level of the Maximum Elevation 
of the Water on the Filter 
Pressure - 
Drain - 
This Dimension to be 
as Short as 
Possible 
Meter Integrating the Rate^ 
of Filtration. This may he 
Placed at any Distant Point 
and can be Provided with 
Recording Attachment, 
Recording the Rate of Filtration, 
Etc. if so Ordered. 
Pressure 
Drain 
Tight Closing 
Hydraulic Throttle 
Valve 
Line to Ciear 
Well 
Line from , 
Filter - 
Effluents / 
Equal Pressure 
Diaphragm 
Valve 
Speed Regulating 
Valves 
DIAGRAM SHOWING 
AUTOMATIC VARIATION 
OF FILTRATION 
Depth of Clear W ell 
| Floats Placed 
1 Clear W ell to 
I Regulate Rate 
of Filtration 
Automatically 
Fig. 82a.-Improved form of Earl rate, controller. 
weights in tubes 1 and 2 are of equal area, but both are weighted 
in excess of their maximum displacement. They are connected 236 
WATER PURIFICATION 
through rods with the bar, which is supported on a knife edge, 
located on a stand placed on the operating floor level of the filters. 
The third tube with a cylindrical weight having about one-half 
the area of the other weights stands also on the operating floor. 
The suspended weights are adjusted in weight so that the beam 
is in balance when the water level is the same about weights 
1 and 2, and at any assumed "zero rate level" about weight 3. 
With water in tube No. 3 at the "zero rate level," the controller 
would operate so as to entirely close the main valve. By holding 
any fixed level of water in tube No. 3 below the "zero rate level," 
a constant rate of flow may be maintained, for the reason that 
the set of weights will be in balance only when the water about 
weight 2 stands above the water level about weight 1 by a distance 
proportional to the aforesaid reduction in level about weight 3. 
Earl Master Controller.-The Earl mechanism may be adapted 
to producing a variable rate of flow bet ween given limits, by means 
of a "master controller." For this purpose a single tube con- 
nected to the filtered-water reservoir is placed at such a level as 
will permit the water to rise and fall within the limits desired; i.e., 
between a minimum level below which it should not drop and at 
which point the filtration rate should be increased, and the maxi- 
mum level in the reservoir above which the water must not go, 
and where filtration should cease. The weight hanging in this 
tube is connected by a rod to a bar supported on a knife edge, and 
is in equilibrium with another weight in a second tube standing 
at the operating floor level. By means of contacts above and 
below the projecting end of the balanced bar, circuits may be 
closed when the bar is out of balance, which open pilot valves ad- 
mitting water to or from the upper tube, and thus restoring the 
balance. This tube is also connected with all the No. 3 tubes 
of the separate controllers, and thus regulates each individual 
filter, causing an increase or decrease in the rate of filtration ac- 
cording to the fall or rise of the water in the upper tube of the 
master controller, and corresponding to like fluctuations in the 
watei' levels in the lower tube and the filtered-water reservoir. 
An improved form of Earl rate controller has recently been 
designed in which an equal pressure diaphragm is used in place 
of the long float tubes first employed. Through this diaphragm 
the pilot valves regulate the throttle valve of the controller. 
In other respects the principles upon which the control of the 
rate of flow is based are the same as in the original design. An REGULATING RAPID SAND FILTER PLANTS 
237 
integrating rate of flow meter has also been made a part of this 
rate controller installation. 
1. "Revamped Water Works and New Purification Plant at Fort Smith, 
Ark." Eng. Record, Sept. 5, 1914, p. 262. 
2. "Weston Rate Controller." Eng. Record, vol. 40, p. 596, Nov. 25, 
1899. 
3. "Water Filtration Works at Anderson, Ind." Eng. Record, vol. 51, 
No. 5. 
4. "Mechanical Filter at Hackensack, N. J." Eng. Record, vol. 50, 
Nov. 19, 1904. 
5. "Vivian Rate Controller." Eng. Record, vol. 59, p. 764. 
6. "Regulating Devices for Filters." Eng. Record, Oct. 19, 1912, p. 428. 
References CHAPTER XX 
REGULATING, MEASURING AND INDICATING DEVICES 
FOR RAPID SAND FILTER PLANTS (CONTINUED) 
MEASURING APPARATUS 
It is important to know the rate at which water is passing 
through a rapid sand filter plant for several reasons. Th6 opera- 
tor should know the volume of water which is to be treated with 
chemicals in order that they may be applied in correct propor- 
tions. Knowledge of the velocity of the water flowing through 
the settling basins, and through the filters is necessary if proper 
periods of subsidence and rates of filtration are to be maintained. 
Legitimate uses of water about the plant, and losses by leakage are 
more readily controlled if measuring apparatus can be easily and 
quickly consulted. Such apparatus also serves as a check on 
pumping machinery which may be supplying to or taking water 
from the filtration plant. 
Modern and well-equipped plants are, therefore, usually pro- 
vided with a meter measuring the inflow to the plant, with 
rate-of-flow indicating devices on the filters, with a meter on the 
wash-water pipe line, and possibly with a meter on the main 
filtered-water effluent pipe line. Probably no one plant is 
equipped with all of these devices, since local conditions may 
make one or more of them unnecessary and still enable the opera- 
tor to accurately control the operation of the plant. The most 
necessary meter is doubtless the one measuring the inflow to the 
plant. The measurement of the effluent is not so important, 
although it affords an excellent check upon the water used and lost 
by leakage in the plant. Measurement of the wash water is quite 
desirable, and may be determined in a number of ways, depend- 
ing upon local conditions. For example, if water for washing the 
filters is drawn from a tank, it may be metered directly, or the 
volume used ascertained after each wash by the lowering of the 
water in the tank. If pumps are used to pump directly into the 
wash-water tank, the volume which they pump may be determined 
from the known capacity of the pumps. The employment of 
238 REGULATING RAPID SAND FILTER PLANTS 
239 
rate-of-flow indicating apparatus on the filters is a refinement of 
control that can be dispensed with if reliable rate controllers have 
been installed. On the other hand, they may in a measure re- 
place the use of automatic rate controllers, since hand regulation 
of the filters then becomes comparatively easy, although depend- 
ent upon the faithfulness of the attendant. 
Venturi Meters.-The measurement of large volumes of water, 
such as must be handled in a filtration plant, offers more or less 
difficulty. Under exceptional circumstances weir measurements 
may be possible, but the more usual condition requires the meas- 
urement of the flow in pipe lines. For this purpose no better 
PUMPING STATION OR REGISTER HOUSE 
-REGISTER 
. SERVICE BOX 
PRESSURE PIPES 
SLEEVE 
MAIN 
'VENTURI METER TUBE - 
Fig. 83.-Venturi meter tube. 
meter is at present available than a Venturi tube (Fig. 83). The 
well-known properties shown by water flowing through a pipe 
having a contracted section or throat joined to a converging cone 
on its upstream end, and a diverging cone on its downstream side, 
enable a fairly accurate measurement to be made of very large 
volumes of water. 
As the water passes through the throat of the Venturi meter 
its velocity is increased, and its pressure decreased. A differen- 
tial gage, connected on one side with the water at the entrance to 
the meter, and on the other side with the water in the throat of 
the meter,^will indicate the loss of head due to the increased 240 
WATER PURIFICATION 
velocity in the throat. There are a number of ways for making 
use of the general principles involved in order to obtain the rate 
of flow and the total volume of water which may have passed 
through the meter in any given period. Recording devices are 
also employed for making permanent records. 
Where the registering apparatus must stand above a point to 
which the water in the pipe would rise, it is necessary to use floats 
which rise and fall in pipes connected to the upstream chamber of 
the meter and to the throat. Cords from the floats are wound 
around spindles attached to a differential mechanism located in 
the registering case. The floats are counterbalanced by weights. 
The same amount of rise or fall of water at the same time in both 
float tubes will have no effect on the movable hand registering 
the rate of flow on the dial face. With the floats at the same 
level the hand will point to zero on the gage. If the pressures 
transmitted cause the floats to stand at different levels, the hand 
will move to a point indicating the rate of flow. 
Where the pressure of the water can be made to actuate the 
registering mechanism (Figs. 84,84a and 85) without the interven- 
tion of floats, as above described, it is usually transmitted through 
mercury wells, which are subjected respectively to the pressures 
of the upstream chamber of the Venturi and to the throat. In 
each mercury well is a heavy metal float which rests upon the 
mercury. The mercury flows from one well to the other in direct 
proportion to the difference in the two pressures; consequently 
one float rises as the other descends, and this movement is trans- 
ferred through a rack and spur gearing to the indicator dial hand- 
shaft. A cam on this shaft controls the position of the pen on 
the recording chart, and also the amount of movement of the 
counter-dial figures, where a register which totals the quantity of 
water passing through the meter forms a part of the instrument. 
Where Venturi meters can be installed under the most favorable 
conditions, they are accurate to within 1 or 2 per cent. If, how- 
ever, local conditions are unfavorable, or if they do not receive 
the moderate amount of care that is required to keep them in good 
working order, they may vary as much as 5 or 6 per cent, from the 
true rate of flow. Where muddy waters are being measured, the 
inlet pipes and annular throat chamber should be blown out quite 
frequently in order to prevent the vent holes to the throat be- 
coming clogged. The tubes in which the floats rise and fall 
should be blown out occasionally as well as all connecting pressure REGULATING RAPID SAND FILTER PLANTS 
241 
Fig. 84. 
Fig. 84a. 
Fig. 84.-Venturi meter register with counter dial. 
Fig. 84a.-Venturi meter register with recording and indicating dials. 242 
WATER PURIFICATION 
pipes. The differential mechanism requires a slight amount of 
oiling from time to time; and when cords attached to the floats 
break a readjustment of the spindles and cam is usually necessary. 
The clockwork which carries the recording charts needs some 
attention in the way of oil and regulation, and the pens should 
be kept clean and full of ink. Air may accumulate in the water 
pressure piping where mercury wells are used, and should be re- 
leased when this occurs through the vent cocks provided for this 
purpose. 
Fig. 85.-Venturi meter registers for St. Louis filter plant chemical feed 
controllers. 
A Venturi meter manometer is a simple and comparatively 
inexpensive instrument for measuring the rate of flow in Venturi 
tubes. It is essentially a U-tube partly filled with mercury, one 
side of the tube being connected with the inlet and the other 
with the throat of the Venturi tube. The rate of flow may 
be read directly from a graduated scale placed between the two 
mercury columns. The zero of the scale is placed on a level with 
the lower mercury level, and the rate is read opposite the upper 
mercury level. These scales are, of course, graduated for the REGULATING RAPID SAND FILTER PLANTS 
243 
particular Venturi tube through which the rate of flow is being 
measured. The instrument is not suitable for measuring rapid 
fluctuations in the rate of flow. 
Simplex Water Meter.-In this apparatus especially adapted 
registering instruments make use of the principles of a Venturi 
tube or of a Pitot tube in order to measure the flow of water in 
a pipe. It may also be adapted to a weir, canal or conduit if 
suitable pipe connections can be made between these open chan- 
nels and the registering apparatus. 
The registering apparatus (Figs. 86 and 87) consists of a mer- 
cury float chamber, containing a float of such variable cross- 
Glass Dome 
>'l "Cement Plaster 
Chamber 
Dimensions 
4'6"x 5'6 "x 7'o" 
3 Way Cock 
Lead Pipe 
:* Direction of Flow 
Fig. 86.-Diagram, of a^Simplex meter register connected to a Venturi tube. 
section that its movement is in direct ratio to the flow of water 
through the Venturi tube, pipe, conduit or weir. The movement 
of the float actuates a revolving shaft to which is attached a hand 
which moves over the face of a fixed dial having uniform gradua- 
tions. Attached to the shaft and moving in proportion to the 
angular deflections, is a pen in contact with a rectangular chart 
wrapped on a revolving cylinder. A traction wheel passes over 
the face of a revolving disk, and is geared to a train of^wheels 
operating a series of small dials, similar to those of an ordinary 
gas or water meter. The "cylinder and disk are operated by, a 
clock. The traction wheel is so constructed that its movement, 244 
WA TER P U RIF IC A TION 
while in contact with the face of the disk, is free from rubbing 
friction. As the movement of the traction wheel and the pen 
are in direct ratio to the angular deflection of the shaft, they are 
also in direct ratio to the movement of the float and of the flow 
of water. 
The chart is uniformly graduated, the abscissas representing 
hours and the ordinates the rates of flow. The area comprised 
by the datum line and the line described by the pen on the chart, 
Fig. 87.-Simplex meter register and recorder, showing connections to 
Venturi tube. 
is in direct ratio to the total flow, and by determining this area 
and multiplying by the coefficient given on the face of the chart, 
the total flow may be calculated. 
Rate-of-flow Gages.-Gages indicating the rate of flow of 
water through a filter are useful where close regulation is desir- 
able. Apparatus of this type may be readily installed where a 
Venturi-tube type of rate controller is used, since the Venturi 
affords a means of indicating the rate of discharge when used in 
connection with appropriate registering apparatus. By the use REGULATING RAPID SAND FILTER PLANTS 
245 
Rising Grade to 
Filter 
x-Float Travel 
; always 
; Proportional 
:to Loss of Head 
Equalizing 
Connection 
f Mercury 
I Balances 
' Loss of Head 
Rising Grade 
to Main of 
Venturi Tube 
Fig. 87ti.--Simplex loss of head gage, showing connections to filter and 
effluent piping. 246 
WATER PURIFICATION 
of a Pitot tube it is also possible to determine the rate of flow 
of water through a pipe, where no Venturi tube forms a part of 
the effluent piping. 
Loss-of-head Gages.-A knowledge of the frictional resistance 
offered to the flow of water through a filter by the accumulating 
sediment being strained out of the water is valuable to the opera- 
tor of a filter. By noting the loss of head when the filter is 
started, and also when it is no longer able to maintain the rate of 
flow at which it is intended to be operated, the head lost becomes 
a measure of the useful period of service of the filter. Where 
filters are operated at variable rates, this information is of some- 
what less value, unless recording instruments are employed, 
whereby the relation between the losses of head during successive 
portions of the period of service may be compared. 
Fig. 88.-Loss of head gages. 
The loss of head is measured by a gage, which records the dif- 
ference in level of the water above the sand line in the filter, and 
that to which it will rise in an open pipe connected directly to 
the effluent pipe of the filter. This is commonly effected by the 
use of floats placed in vertical tubes connected directly to the 
portions of the filter mentioned above. Cords from the floats 
pass over spindles connected to a train of gears, which operate 
hands rotating in front of dials marked with the two water levels 
in feet. In some gages a separate indicator (Fig. 88) is used for 
each water level, while a third hand indicates the difference be- 
tween the two levels. The floats are counterweighted. Silk 
cords or a small flexible copper-wound cable is used to transmit 
the movement of the floats to the train of gears. The gage is REGULATING RAPID SAND FILTER PLANTS 
247 
usually placed on the operating floor level, and at some conven- 
ient point so that it may be easily observed. 
In one type of loss-of-head gage a hollow float is inverted in 
mercury contained in the bottom of a cylinder. The float can 
move vertically and transmits its motion through a rod passing 
through the top of the cylinder. The space in the cylinder out- 
side the float is connected by a pipe with the water standing above 
the sand level in the filter, and the inside of the float above the 
mercury is connected by a pipe with the effluent main. Differ- 
ences in pressure are transmitted 
through the mercury, and cause 
the*float to rise and fall. The 
movement of the float is trans- 
ferred through the rod mentioned 
above to suitable registering 
apparatus. 
A type of loss-of-head gage 
which uses no floats, but draws 
the water to a mercury manometer 
by means of a vacuum pump, is 
also used. One leg of the manom- 
eter is connected through pip- 
ing with the water on the surface 
of the filter, and the other leg 
with the water in the effluent 
pipe. The decreasing pressure of 
the water in the effluent main, 
from the time the filter is started 
to the close of its period of service, 
is measured directly on a gage by 
the rise of the mercury column 
connected with the effluent pipe. 
Recording types (Fig. 89) of loss-of-head gages are useful if a 
careful study of the operation of the filters is desirable. They 
are by no means as necessary a part of the equipment as are the 
non-recording types. If sufficient attention is not given to them, 
so that they are kept in good working order, they are, of course, 
useless. 
Water-level Gages.-It is usually necessary to know the water 
levels in reservoirs, basins and tanks for the proper control of the 
operation of purification works. Where possible, ordinary pres- 
Fig. 89.-Recording type of loss 
of head gage. 248 
WATER PURIFICATION 
sure gages are usually sufficient, if properly placed. Recording 
gages are frequently desirable, as they give a continuous record 
and enable the operator to keep a permanent record of fluctua- 
tions in the water levels. 
Gages should be grouped together as much as possible, since 
water levels in different parts of the system may be thereby more 
conveniently compared. Where the gage is placed at a point 
below the water level it records, ordinary spring or diaphragm 
gages may be used. These gages had best be graduated in feet 
above a known zero elevation rather than in pounds, since com- 
parisons of water levels in different parts of the system are 
thereby more easily effected. 
When the gage must be placed above the water level to be 
recorded, electrical gages may be employed. An excellent type 
of gage of this character consists of a float which rises and falls in 
a tube connected to the water level to be measured. A cord 
attached to the float passes over a small wheel connected with a 
transmitting device. The movement of this wheel operates the 
mechanism inside the transmitter, and causes a momentary 
closing of one set of contacts for a certain predetermined move- 
ment of the float upward, and a similar closing of another set of 
contacts when the float drops. The transmitter can be made to 
close the contact ten or any other convenient number of times 
for a change in elevation of 1 ft. by the float. 
The transmitter is connected to the receiver by three wires. 
The middle wire acts as a common return for the current to both 
sets of contacts. The two outside wires are connected to two 
separate sets of electromagnets in the receiver. When the arma- 
ture of one set is drawn toward its magnet by the closing of one 
set of contacts in the transmitter, it moves the pointer and record- 
ing pen in one direction. The other set of magnets operate the 
pointer and recording pen in the opposite direction when the 
circuit is closed through the second set of contacts in the 
transmitter. 
The batteries used to operate the electromagnets may be 
placed either at the receiver, or at the transmitter, or at any con- 
venient place on the circuit between the two. 
The above-described gage, which is known as the "George E. 
Winslow Electrical Indicating and Recording Apparatus," is 
made in different styles to meet varying conditions. In some 
cases a sensitive pressure diaphragm, which operates the trans- REGULATING RAPID SAND FILTER PLANTS 
249 
mitter by means of a lever, may take the place of the float and 
counterweight. 
A similar long distance, electrical, indicating and recording 
apparatus (Fig. 90) is made by the Bristol Co. although based 
Fig. 90.-Bristol recording water-level gage. 
upon quite different principles. The principle used in this sys- 
tem of transmission is that of an induction balance. The appa- 
ratus consists of a sensitive diaphragm bulb, which is submerged 
TO DISTANT 
RECORDER 
OR INDICATOR 
TO A.C. SUPPLY 
TO DISTANT 
RECORDER 
OR INDICATOR 
OPERATIVE 
MECHANISM 
OF TRANSMITTER 
Fig. 90<z.-Bristol recording water-level gage showing electrical connections 
in the water to some known level; a copper tube connection 
between the bulb and a transmitter; and a three-wire circuit 
connecting the transmitter with the receiver and indicating and 250 
WATER PURIFICATION 
recording gages. It is necessary to use an alternating current 
to operate the transmitter and receiver. 
The transmitting and receiving instruments are both equipped 
with two pairs of coils, arranged to swing in a horizontal plane 
over laminated iron cores. In the transmitter the pressure trans- 
mitted by the sensitive bulb through the copper tube, moves a 
helical pressure tube, which in turn causes the mechanically 
balanced solenoids to be deflected. One solenoid moves off from 
its core as the other moves on over its core, thus increasing the 
inductance in one solenoid and decreasing it in the other, and 
thereby proportionately increasing the flow of current in one 
coil and decreasing it in the other. 
The variations of current at the transmitter will produce like 
variations at the receiver, and cause the pair of solenoids of the 
TO DISTANT 
RECORDER 
OR INDICATOR 
TO DISTANT 
RECORDER 
OR INDICATOR 
Fig. 906.-Bristol recording water-level gage showing balanced adenoids 
TO AX. SUPPLY 
receiver to seek positions corresponding to those of the trans- 
mitter. The movement of the receiving coils is transmitted 
through an appropriate mechanism to a pointer moving over a 
properly calibrated scale, or to a pen which can trace a line on a 
chart for the purpose of producing a permanent record. 
Water-sampling and Inspecting Devices.-Means for obtain- 
ing samples of water from each individual filter are usually de- 
sirable in the operation of rapid sand filter plants, since the efflu- 
ent of one filter, if it is operating improperly, may contaminate 
the output of a whole group of filters. For example, where 
muddy waters are being filtered, the turbid effluent of one filter 
may cause a slight opalescence in the combined discharge from 
all the filters, in wfflich case it becomes necessary to detect which 
filter is producing the poor effluent in order that it may be taken REGULATING RAPID SAND FILTER PLANTS 
251 
out of service. By sampling the water from each filter the one 
producing the poor effluent may be easily discovered. 
Sampling devices for each filter commonly form, therefore, a 
part of each filter's equipment. They consist usually of a small 
pump with its suction pipe connected to the effluent pipe of the 
filter. The pump is usually made to discharge the water through 
a faucet into a sink located in the filter operating table. The 
pumps are either electrically or hydraulically operated, and are 
started and stopped from the operating table. 
An elaborate sampling apparatus, consisting of a multiple 
pump centrally located with respect to the filters, and having as 
many pump cylinders as there are filters, so that water may be 
drawn from any filter desired, has been used in some plants. 
Sight tubes of glass through which water flows continuously 
from both the influent and effluent mains afford a method of ob- 
serving the character of the water applied to the filters as well 
as that being discharged from them. They form a rapid and 
easy means for gaging the operation of the plant, if located in 
some portion of the filter-plant building where they may be 
frequently and readily observed. 
The multiplying of devices for sampling or observing the char- 
acter of the water passing through a plant can be easily overdone. 
On the other hand, wherever the character of the water being 
pumped is undergoing rapid variations, which demand frequent 
changes in the quantity of the chemicals required to properly 
purify it, as well as modifications in the general method of hand- 
ling it, the operator should be provided with all the necessary 
registering and sampling apparatus that are needed. False 
economy in this respect is as bad as extravagance in the opposite 
course. 
Where economy in the installation of a plant is necessary, an 
ordinary faucet tapped into the influent and effluent pipe line of 
each filter tank is usually sufficient to supply the samples which 
may be needed from each filter. Some inconvenience in getting 
to such sampling places, because of their location in the pipe 
gallery, is more than offset by the relatively infrequent intervals 
at which such sampling is generally necessary. In practice, 
samples of the water being applied to the filters, and of the com- 
bined effluents from the filters, are usually quite sufficient for 
routine control, and for this service sample pumps should be 
installed for the collection of water at these points. 252 
WATER PURIFICATION 
Operating Tables.-The operations of starting, stopping and 
washing a rapid sand filter are most easily carried out at some 
point where the movement of the water in the tank can be 
observed. As this point must be at the top of the tank, and as 
the valves which control the above operations are at a somewhat 
lower elevation, it necessitates apparatus adapted to these con- 
ditions. Where hand-operated valves are used the extension 
Sewea Waste Wash flaw Effluent 
Gange Heai.and Mechanism furnished under another Contract. 
Holes for bolts, cords, etc., to be left in table top as required. 
OPERATING TABLE 
Make 18 as shown 18 opposite hand 
Sample Pump 
Oak Panel 
Float tubes 
furnished 
under 
another 
contract 
Marble . 
Floor 
Fig. 91.-Plan and elevation of filter operating table 
FRONT ELEVATION 
SECTION 
of the stems to the operating floor level does not always lend 
itself to a convenient grouping of the wheel stands, because of 
the positions which the valves must occupy on the pipe lines 
below. However, in small tank units the stands are quite close 
together, although somewhat irregularly spaced, and produce a 
convenient and inexpensive design. Separate stands for loss- 
of-head and rate-of-flow gages may also be provided in this 
arrangement without the use of tables. REGULATING RAPID SAND FILTER. PLANTS 
253 
In the more modern plants, where hydraulically operated 
valves (Figs. 91, 92 and 93) have taken the place of hand-operated 
valves, it has become the custom to place the controlling 
apparatus on a small table. This table or cabinet is provided 
on its top with a series of levers and indicators, which by means 
of multiple-way cocks located within the cabinet, move and 
indicate the position of the hydraulic valves. The piping from 
the hydraulic valves is brought up to the table and connected 
to the controlling cocks. Cords or wires passing over pulleys 
connect the tail-rod of the hydraulic valve with the indicator 
Fig. 92.-Filter operating table used in Minneapolis, Minn., filter plant. 
on the table. The indicator can thus be set to show the exact 
position of the valve. 
The loss-of-head and rate-of-flow gages, when the latter gage 
is used, are also usually located on the top of the operating table, 
or on a board fastened to the back of the table. Sampling 
faucets to which water is forced by small pumps, and in some 
cases sight tubes by which the character of the influent and 
effluent water may be observed, sometimes form a part of the 
apparatus on an operating table. In such cases bowls or sinks 
must be provided for waste water. 254 
WATER PURIFICATION 
In the case of electrically operated valves the assembling of 
the switchboard panels, from which the valves are moved, side 
by side at some convenient point at the top of the filter, takes 
the place in a measure of an operating table. A stand or table 
is in such cases required for holding the gages and sampling 
apparatus. 
Operating tables are usually constructed with marble tops and 
with hardwood paneled sides and front. The metal work of the 
controlling and indicating apparatus is usually nickel-plated, 
and the piping and cocks are generally of brass. 
Fig. 93.-Filter operating table used in St. Louis, Mo., filter plant. [ 
The centralizing of the control of the operation of a series of 
filters or of all the filters in a plant at one point has been tried 
but with no very great success. The principal reason for the 
failure of this method is that the operator can not see the water 
in the tank which he is operating. 
It is evident that the use of all the operating-table equipment 
described above, makes the handling of the filters easier; never- 
theless, its installation without due regard to the real needs of the 
plant may lead to extravagance of design not commensurate with 
the results which could usually be expected. REGULATING RAPID SAND FILTER PLANTS 
255 
1. George W. Fuller: "Some Features of Detail in the Design of Rapid 
Sand Water Filtration Plants." Engineering-Contracting, Sept. 2,1914. 
2. F. B. Leopold: "The Use of Recording Gages in a Filter Plant." 
Eng. Record, vol. 58, p. 330. 
3. "Humphrey's Manometer-type Loss-of-head Gage. Engineering-Con- 
tracting, July 29, 1914. 
References CHAPTER XXI 
EQUIPMENT FOR THE HANDLING AND STORING OF 
CHEMICALS AND FOR THE PREPARATION OF 
SOLUTIONS 
The handling and storage of chemicals used in the opreation 
of water-purification plants call for careful consideration on the 
part of the designer. The uninterrupted operation of the plant 
is usually necessary, and, in consequence, a constant supply of 
chemicals must be provided. Plants are not infrequently so 
located that direct railroad connections are impracticable, in 
which case dependence must be placed on hauling the chemicals 
by wagon with horses or with motor-driven trucks from the 
railroad to the plant. This latter method is entirely practicable 
for keeping small plants supplied, but will hardly answer for 
the large plants, where the quantities of chemicals used daily 
are measured in tons rather than in hundredweights. At the 
large plants railroad connections are virtually imperative, unless 
the plant happens to be so situated as to be able to receive its 
supplies by water. 
Whether railroad facilities are available or not, there should 
be ample storage capacity at the plant to care for its operation 
for at least 4 to 6 weeks, if, for any reason, its regular supply 
of chemicals should be cut off. 
The manufacture at the purification plant of certain of the 
chemicals used in water-purification processes has been proposed. 
In several plants plans for making sulphate of alumina from raw 
material have been completed, and the necessary apparatus has 
been purchased and installed. Only under rather exceptional 
conditions could such a practice be possible, and then only in 
the large plants where competent and experienced chemists 
could be placed in charge of the manufacturing. 
Chemicals Used and Their Transportation.-The chemical 
compounds commonly used in water purification are sulphate 
of alumina, sulphate of iron, caustic and hydrated lime, and 
carbonate of soda or soda ash. For disinfection purposes, cal- 
cium hypochlorite or bleaching powder, and liquefied chlorine 
256 HANDLING AND STORING OF CHEMICALS 
257 
gas are required. Copper sulphate in small quantities is also 
sometimes used for destroying minute microscopic plant growths. 
Some of these chemicals, like caustic lime, are best handled in 
bulk, although it is not uncommonly shipped in barrels. It has 
been handled in sacks of cotton duck with a fair measure of 
success, if the lime is first crushed before being placed in the 
bags, and if the period of transportation is not too long. Hy- 
drated lime can be obtained in paper sacks. Lime in this form 
packs closely and does not carbonate readily. On the other 
hand, its cost is greater than the caustic lime. As it contains 
about 30 per cent, of water, the freight charges for this amount 
of water add materially to the first cost.1 
Sulphate of iron is handled in bulk and in sacks, and the same 
is true of soda ash. Sulphate of alumina may be handled in 
bulk or in sacks, but is more frequently transported in wooden 
barrels. Copper sulphate is shipped in barrels. 
Bleaching powder is usually obtained in sheet-iron drums 
containing 500 to 600 lb. of the chemical. This chemical is 
also packed in wooden casks. The latter weigh about 300 lb. 
Sheet-iron drums holding about 100 lb. may also be obtained, 
but the larger drums containing five or six times as much are 
more common. 
Liquefied chlorine gas is transported in steel drums each 
holding about 100 lb. of the gas, or in larger cylinders which hold 
as much as 2,000 lb., and which weigh with the steel drum over 
3,000 lb. 
Best Methods of Handling Chemicals.-The best method of 
handling the various chemicals employed in water purification 
depends largely upon the properties of the chemical, the quanti- 
ties to be used, and the storage facilities. Chemical compounds 
that are hygroscopic must, of course, be protected as much as 
possible from moisture. Handling large quantities of chemicals 
like sulphate of alumina, sulphate of iron, or lime in bulk is, ob- 
viously, the cheapest method to follow, although it is not always 
practicable to do so. The smaller the package, as for example 
where the material is barrelled or bagged, the higher the cost 
and the easier it is handled. Convenient sizes for packages range 
from 100 to 400 lb. in weight, but if above 400 lb. the package 
becomes awkward to handle. For example, the customary 
1 Lime carbonates only after it has been hydrated; hence when kept in 
air-tight bins it does not deteriorate. 258 
WATER PURIFICATION 
weight of bleaching-powder drums is from 500 to 600 lb., which 
makes them somewhat troublesome to handle. On the other 
hand, packages weighing as high as 400 lb. can be quite easily 
moved on a level by one man on two-wheeled barrel or bag trucks. 
The elevation of chemicals from the unloading platform to stor- 
age space, near the top of the solution tanks, is usually necessary 
in all plants. Where the chemicals are received in bags or bar- 
rels, the ordinary endless-chain bag or barrel hoist operated by a 
small electric motor furnishes the easiest and quickest method 
of handling large quantities of chemicals. In small plants, a 
good-sized platform elevator operated by either hydraulic power 
or by an electric motor probably answers the requirements better 
than an endless-chain bag or barrel hoist. 
Unloading, Elevating and Conveying Chemicals in Bulk.-In 
those plants where large quantities of coagulating or softening 
chemicals are used, the delivery is frequently made in bulk in box 
cars. Lime and sulphate of iron have been successfully handled 
in this manner. Where lime is bought in bulk it is usually neces- 
sary to crush it before discharging it into storage rooms or bins. 
In such cases the lime is shoveled by hand or by an automatic 
power shovel directly from the car into the crusher, from which 
the lime drops into an elevator. Bucket elevators, chain drag 
and screw conveyors, and belt conveyors have been used for 
hoisting and conveying the lime to the storage space. Belt 
conveyors have been tried for both lime and sulphate of iron, 
but have not proven satisfactory, since it was found that the dust 
which collects in the casings below the belt increased the power 
required for operation, and if not cleaned out at regular intervals 
eventually stopped the machinery. 
In general it may be said that the simpler methods of handling 
chemicals are more reliable, although they involve more manual 
labor. This latter element of cost does not become of much con- 
sequence until the quantity of chemicals to be handled daily 
reaches so large an amount as to require a considerable force of 
men; in which case mechanical methods are probably cheaper. 
In selecting machinery for handling chemicals for any plant, 
preference should be given to that which is composed of the few- 
est and simplest working parts. The breaking down of this 
class of machinery may cause serious inconvenience in operating 
a purification plant, and its upkeep may become burdensome 
when a too elaborate equipment has been installed. HANDLING AND STORING OF CHEMICALS 
259 
The dust, which is practically unavoidable in handling nearly 
all of the chemicals employed in water purification, should be 
minimized as much as possible. Lime when ground or crushed, 
or when hydrated, is particularly disagreeable on account of 
the fine dust which gets into the air wherever it is being handled. 
Adequate ventilating systems should always be provided where 
much material is being moved in order to lessen as much as 
possible the volume of dust escaping to the air, and thereby 
reduce to a minimum the discomfort of the workmen who must 
handle it. 
The Storage Plant for Chemicals of the St. Louis Water De- 
partment.-A very large storage plant for lime and sulphate of 
iron was constructed and placed in service in 1909 at the St. 
Louis water-purification plant. 
Building.-The building is divided into three sections (Fig. 
94), the storage bins comprising the two end sections, and the 
operating department the third or central section. The cylindri- 
cal bins for the lime are at one end of the building, and those for 
the iron sulphate at the opposite end. Four bins are provided 
for each chemical. The capacity of these bins is approximately 
10,800 cu. ft., each, giving a total storage capacity of about 1,250 
tons of lime, and 1,600 tons of sulphate of iron, equal to a supply 
for 37 days of the lime, and for 115 days of the sulphate of iron 
at the rate at which the chemicals were being used at that time. 
The central or operating section of the building has two floors 
above the basement, one at the ground level called the pump- 
room floor, and a second or mixing-room floor. A lantern at the 
top of the building crosses the mixing room and partly covers the 
bins on either side. In the basement are the crushers for both 
lime and sulphate of iron, most of the conveyors for storage and 
reclaim, air compressor, pump, sump-pit, and a system of piping. 
On the pump-room floor are the heater tanks, pumps, switch- 
board, wash and locker rooms. On the mixing floor are the 
tanks for slaking the lime, and for making the sulphate of iron 
solution, and the auxiliary measuring and weighing devices. 
Crushing and Conveying Systems.-There are two unloading 
points for lime, and one for sulphate of iron. At each of the 
unloading points for the lime is a No. 2 Sturtevant crusher with 
a capacity of about 15 tons per hour. Each crusher with its 
side conveyor is driven by a 30-hp. motor. The crusher for the 
sulphate of iron is a No. 1 Williams hammer crusher driven by a 260 
WA TER P U RIF IC A TION 
15-hp. motor. The lime is shoveled into chutes which convey 
it to the crushers. From the crushers the lime is carried to 
elevators which discharge on to a scraper conveyor, from which 
Cross Section on Center Line - Looking South 
Fig. 94.-Cross-section of center of coagulant house in St. Louis filter plant. 
Cross Section on Center Line - Looking • North 
the lime is distributed to the storage bins. The elevator and 
scraper conveyor in the lantern are driven by a 7^-hp. motor. 
In reclaiming the lime from the storage bins, it is taken from HANDLING AND STORING OF CHEMICALS 
261 
the bottom of the latter, and by means of conveyors carried to an 
elevator, which raises it again to the lantern, and discharges it 
directly into an auxiliary storage tank or daily supply hopper set 
on columns in the mixing room above the slaking tanks. This 
daily supply hopper holds from 25 to 30 tons of material. 
Difficulties Encountered in Operating Equipment.--The belt 
conveyors first employed in the St. Louis plant were found un- 
suited to handling either the lime or iron sulphate on account of 
the collection of dust in the casings below the belt. They were 
replaced by conveyors of the scraper type. 
Another difficulty which was encountered in the storage of the 
sulphate of iron in the bins was due to the iron sulphate crystals 
cementing together and thus causing the mass to cake. Bins 
from which material was irregularly drawn caked worse than in 
those which were being constantly used. Moisture left in the 
chemical on account of improper drying was thought at first to be 
the only cause of the trouble; but it seems probable that the 
cementation of the crystals is in part due to the escape of water 
of crystallization and to the pressure to which they are subjected 
in the storage bins on account of the great mass of material stored 
in them. 
These cakes of sulphate of iron, if they were not broken up 
in the conveyors and elevators carrying the chemical to the daily 
supply hoppers, clogged the orifices of the feeders and necessi- 
tated constant attention. The installation of grinders just fol- 
lowing the discharge of the chemical from the supply hoppers 
made the uniform feeding of the sulphate of iron through the 
orifices more certain, but did not remedy the trouble entirely 
even at this point. 
In order to reduce the caking in the bins as much as possible 
and to avoid the disagreeable necessity of sending men into the 
bins to break up the caked mass, an experimental chain breaker 
was installed in one of the bins. It consisted of a horizontal 
girder supported from the bin roof, and geared to make two 
revolutions per minute, when driven by the reclaim conveyor 
motor, and carrying a heavy chain whose length was adjusted 
so that it scraped over the surface of the cemented iron sulphate. 
The chain loosened the iron sulphate so that disagreeable labor 
of breaking it up by hand in a suffocating dust was to some extent 
avoided. However, lumps too large to pass the discharge chute 
of the bin leading to the conveyor required them to be dislodged 262 
WATER PURIFICATION 
and broken by poking the material through bar-holes in the 
conical-shaped bottom of the bin. The chain breaker proved 
of no value if the material had been stored too long in the bin. 
Mr. W. F. Monfort in discussing the probable reasons for the 
caking of sulphate of iron crystals states that it is well known 
that at 70°F. crystals of iron sulphate of the composition FeSO4- 
7H2O, subjected to a stream of air, lose 3 molecules of water, 
and pass over into pale green crystals of the composition FeSO4- 
4H2O. By heating to higher temperatures oxidation of the 
salt commences and finally all of the water of crystallization is 
driven off. 
The loss of the first 3 molecules of water takes place at summer 
temperatures to a limited extent during the period of shipment, 
so that the material has a greenish-gray appearance. Its com- 
position is only slightly affected, but it is sufficient to cause the 
analyses to indicate a little more than 100 per cent, of the salt, 
FeSO47H2(). The partially dehydrated crystals, produced in 
the exposed layers in transit, become distributed throughout the 
mass when the shipment is transferred to the closed bins. If 
the temperature in the bins rises above 70°F. water of crystal- 
lization escaping from the fully hydrated crystals passes upward 
through the mass as a vapor, and, wherever it may happen to 
condense on cooler particles of the salt, is reabsorbed by partially 
hydrated crystals to form the completely hydrated salt, and in 
so doing unites with other fully hydrated crystals to form masses 
of considerable size. 
The author is inclined to believe that Mr. Monfort's explana- 
tion is correct, but that in the drying out of the crystals in the 
course of manufacturing the lower hydrate may be formed in 
considerable amount at times, and that the rehydration of these 
crystals with water derived from moisture in the air is frequently 
the source of the required water for this recrystallization and the 
cause for the cementing together of the mass. Caking will occur 
in bags stored for a considerable length of time in an open room. 
Lime Handling Plant at Columbus, Ohio.-The method of 
handling lime for the water purification plant at Columbus, 
Ohio, is similar to that at St. Louis, except that the plant is 
somewhat smaller and less elaborate. 
Bin.-The equipment consists of a concrete bin (Fig. 95) ele- 
vated on columns over the railroad tracks located at the side of 
the head house of the plant. The bin is of concrete 25^ ft. HANDLING AND STORING OF CHEMICALS 
263 
Fig. 95.-Section of lime-storage house at the Columbus, O., filter plant. 264 
WATER PURIFICATION 
square with bottom slopes of approximately 38° with the hori- 
zontal, and heavily reinforced. The center column of the eight 
columns which support the structure, is cored, and is utilized 
as a chute to reclaim lime from the storage bin and deliver it by 
means of an elevator and conveyor to the tanks in the head 
house. The bin has a storage capacity of 220 tons of lime, or 
about 22 days' supply at the usual rate at which lime is used. 
Unloading Lime from Cars.-A power shovel is used for un- 
loading the lime from the cars into the hopper of the conveyor. 
It is operated by a separate 5-hp. 220-volt direct-current electric 
R.R. Siding 
Third Floor 
Head House 
R.R. Siding 
Fig. 96.-Plan of tanks and conveyor system used at Columbus, O., filter 
plant. 
motor. Two men are required to manipulate the shovel and 
with it a 40-ton car of lime has been unloaded in about 3 hr. 
Conveying Equipment.-This apparatus (Fig. 96) consists of 
a continuous V-bucket conveyor and elevator with a capacity 
of 50 tons per hour. It runs in a trough between the tracks and 
through chutes each side of the bin, being enclosed throughout 
its entire length. The lime is discharged from the elevator into 
a 12-in. screw conveyor, which latter carries it to the top of the 
steel supply tanks holding all together about 22 tons or 2 days' 
supply. 
The lime from these supply tanks is fed into automatic weigh- (Facing page 204.) 
SECTION S-S 
Fig. 97.-Coagulant house of the Cleveland, O., filter plant. 
SECTION R-R 
SECTIONAL PLAN 
SECTION T-T HANDLING AND STORING OF CHEMICALS 
265 
ing scales. The lime is in small lumps about 2 in. in diameter. 
From the scale it is dumped directly into the slaking tanks. 
Automatic Scales.-Three "Richardson Automatic Scales" 
are in use at this plant for charging the lime-slaking tanks with 
the correct amounts of lime. On knife edges at one end of the 
scale beam is suspended the weight hopper, and at the other 
end of the beam, also on knife edges, is suspended the weight box 
or pan, in which are placed the standard weights for any given 
charge. 
The lime is fed to the weighing hopper of the scale down a 
sloping chute, in which is placed a reciprocating plate feeder 
to keep the lime from blocking. This feeder operates only while 
the scale is filling, being motor-driven and starting and stopping 
automatically. The lime leaves the chute through a gate, which 
partially closes when the weight hopper has received four-fifths 
of its load, the last fifth being slowly dribbled into the hopper. 
When the exact amount of lime has been discharged into the 
hopper, the gate closes and is locked. The scale hopper holds the 
load until it is time to discharge it. 
The time of discharge of these three scales is governed by an 
electric-clock device, and so arranged that one, two or three 
scales may be unloaded as desired by throwing in the proper 
switches. For example, if 100 lb. of lime are to be weighed out 
each minute, a "minute disk" is placed on the clock face, and 
at the end of this interval a switch is automatically closed which 
is connected to the scale through a solenoid. The solenoid 
opens the door of the weighing hopper and permits the lime 
to escape. The door then closes and locks* and in so doing auto- 
matically opens the inlet gate again and starts another weighing 
operation. Each scale is equipped with a stop counter-device, 
which automatically stops the weighing-out operation of the 
scale when the desired number of charges have been weighed out 
and dropped into the slaking tanks. 
Nearly all the steam rising from the slakers is kept out of the 
scale by being drawn through a system of pipes with a suction 
fan. To prevent dust and steam clogging the system of levers 
of the scales, they are placed outside of the scale proper, which 
is enclosed in a heavy cast-iron dustproof casting. All scale 
pins are made of phosphor-bronze, and bearings in which the 
pins work are bronze bushed to prevent corrosion. The clock 
device is in a dustproof casing with glass doors, which are kept 266 
WATER PURIFICATION 
locked to prevent tampering with the equipment when once 
adjusted for any desired dumping interval. 
MIXING AND STORAGE TANKS, AND PIPING FOR CHEMICAL 
SOLUTIONS 
Tanks.-The tanks (Fig. 97) for preparing and storing chemical 
solutions are made of wood, iron and concrete. The sulphates 
of iron and aluminum may be dissolved and kept in concrete 
tanks, and will cause little deterioration in this material unless 
appreciable quantities of free acid are present. In the smaller 
plants wooden tanks are quite commonly used for these chemicals. 
Hypochlorite of lime solutions are best kept in concrete tanks, 
although wooden tanks lined with Portland-cement mortar will 
last a number of years if the lining is kept intact. Iron tanks 
lined with concrete are also used for this chemical, as well as 
lead-lined wooden and porcelain-lined iron tanks. Acid-resist- 
ing paints for both wood and iron tanks are also used. Soda 
ash should be dissolved and kept in iron tanks as it attacks 
concrete. 
Lime should be slaked in iron tanks, but may be stored, if 
cold, as a milk of lime, or as a lime-water solution in concrete 
tanks. The heat generated by the slaking of lime in concrete 
tanks is liable to produce cracks in them, and thus cause them 
to leak and become useless. 
The capacity of chemical solution tanks should be properly 
proportioned to the total output of the plant. The method of 
applying the chemical solutions will affect the size of the solu- 
tion tanks, as this will depend upon whether they are used as 
real supply tanks which hold a definite volume of a standard 
solution, or whether they serve as dissolving tanks through which 
flow a constant volume of solution of variable but controlled 
strength. 
The character of the coagulating chemical used also plays a 
part, since the relative solubilities of the various chemicals 
employed differ widely. The slight solubility of lime, for 
example, requires that large tanks be used if a real lime-water 
solution is to be applied to the raw water. On the other hand, 
if the chemical is used as a milk of lime, relatively small tanks 
may be employed. The slight solubility of bleaching powder in 
water makes it necessary to provide fairly good-sized tanks for HANDLING AND STORING OF CHEMICALS 
267 
the storage of its solutions, even though the quantity of the 
solution used is quite small. 
In those plants where solutions of standard strength are 
measured through orifices, there should be sufficient tank capacity 
to provide treatment of the water for 8 to 12 hr. without pre- 
paring new solutions. Never less than two tanks should be 
available for holding any one chemical solution, and in large 
plants three or four units are usually provided. 
Mixing Tanks.:-The rate of solution of a chemical compound 
in water is affected by the size of the solid particles, by the 
amount of agitation to which they are subjected, and usually 
to the temperature of the water used. Chemical reactions with 
the water may also take place, as for example in the hydration 
of lime. All these factors have to be taken into consideration 
in providing the best method of treating the chemical to be 
dissolved. 
Stirring Apparatus.-The rapid solution of the solid chemical 
compound and the production of a uniform solution, as just 
noted, is facilitated by stirring, and this is usually done in small 
plants by hand with a paddle, but is preferably effected mechanic- 
ally. Rotating paddles furnish as simple a means as any for 
aiding solution, and are effective. Paddles may be driven by 
an electric motor or by a small waterwheel. If a waterwheel 
is used the water driving the wheel may be discharged into the 
solution tank, and thus aid in the preparation of the solution. 
Compressed air is frequently used to produce a mixing action 
in solution tanks by being blown in through perforated pipe at 
the bottom of the tank. 
Mixing Device Used in Cincinnati Filter Plant.-A mixing 
device used for aiding the solution of sulphate of iron in the 
Cincinnati filter plant consists of a small-bladed propeller driven 
by an electric motor and rotating in a pipe open at both ends and 
submerged in the center of the solution tank. Water is fed into 
the tank through a perforated pipe at the bottom, and the solu- 
tion that is formed overflows at the top of the tank through an 
opening into a screen compartment connected to the chemical- 
solution pipe line. The solid sulphate of iron is dumped through 
a grating into a branch of the pipe inside of which the propeller 
is rotating, and falls through the propeller blades and pipe to 
the bottom of the tank. As only 100-lb. charges are dumped 
into the tank at any one time, solution is rapidly effected, and 268 
WATER PURIFICATION 
takes place in part while dropping through the pipe and tank. 
The propeller blades act in a measure to lift the solution over the 
top of the submerged pipe, and in so doing draw liquid from the 
bottom of the tank and thus keep up a circulation between the 
top and bottom of the tank. 
Lime Slaking.-The preparation of milk of lime requires a 
somewhat different stirring apparatus than those used in dis- 
solving alum, copperas and soda ash. Where caustic lime is 
used it must be slaked, that is hydrated. This is carried out in 
2-g'fitted bolts 
Ball diameter,e aligning ball 
(seated washer thrust capacity 5000* 
Babbit meta! 
1^ Shaft:2!6rpm 
'Ball dia^ Thrust capacity 8000* 
HighCarbon 
\ Steel) _ 
"39370 Cast Iron 
. Cast Iron Wheel 
Section C-C 
Calk groove oi 
tight with lead/ 
wool before "' 
painting 
Hind ley Worm Gear 
Plan 
Section A-A 
Hit shaft- cold roll, steel 
Cast Iron- 
Section B~B 
Elevation 
Fig. 98.-Details of stirrer for lime slaking designed for Jerome Park filter 
plant in New York City. 
steel tanks fitted with revolving rakes, which keep the solid 
lime and water in contact until hydration is complete. A typical 
design for such a stirring apparatus is shown in the accompanying 
cut (Fig. 98). 
Ordinary concrete mixers have been used for preparing milk of 
lime. This method of lime slaking is used in the water-purifica- 
tion plant at Grand Rapids, Mich., and at Baltimore, Md. 
Solution, of Bleaching Powder.-For mixing solutions of 
bleaching powder (Figs. 99 and 100) there are a number of 
devices on the market, all of which aim to get as intimate a HANDLING AND STORING OF CHEMICALS 
269 
mixture of the powder and 
water as possible. As soon 
as the powder is wet it 
becomes pasty and lumpy. 
These pasty lumps must be 
rubbed up with water if a 
good solutionis to be made. 
An accompanying cut shows 
how this is done in one form 
of mixer. 
The disagreeable gases 
arising from solutions of 
calcium hypochlorite while 
being prepared led to an 
ingenious design to over- 
come this difficulty in con- 
nection with the plans for 
the Grand Rapids, Mich., 
filter plant. The sheet-iron 
drums of this chemical, 
weighing 700 or 800 lb. 
each were to be opened 
under water in a dissolving 
tank provided with a false 
bottom, consisting of cast- 
iron grate bars. The drums 
are lifted into position by 
means of a chain block. 
Sharp steel points are then 
driven into the head of the 
drum by means of a shaft 
extending through a stuff- 
ing box in the side of the 
tank. This shaft is forced 
home by means of a lever 
on the outside. By means 
of a ratchet and pawl the 
drum is turned while a large 
Fig. 99.-Details of a design 
for 24-inch motor-driven mixer 
used for dissolving bleaching 
powder. 270 
WATER PURIFICATION 
can opener cuts the can in two. This operation is effected after 
the tank has been flooded with water. A similar device has been 
used at the filter plant in Minneapolis, Minn. 
Piping, Valves and Pumps for Chemical Solutions.-The con- 
veying of chemical solutions from the supply tanks to the point 
where they are discharged into the water to be treated presents 
some difficulties which are not easily overcome. All pipe lines 
for this purpose should be as 
short as possible and as 
straight as it is practicable to 
make them. Where bends 
in pipe must be made crosses 
or plugged tees should be used 
so that the line shall be 
straight between bends, and 
so that cleaning points are 
provided. All chemical- 
solution pipe lines ought to 
be so placed as to be easy of 
access, both for cleaning or 
temporary removal As a 
rule, they should not be laid 
under ground, nor buried in 
concrete or masonry. 
As incrustations and de- 
posits are likely to form in 
pipes conveying certain of 
these solutions, the pipes 
should be of sufficient size so 
that the lessening diameter 
of the pipes, as incrustation 
proceeds, shall not render 
them inadequate for carrying 
the necessary volume of liquid with the pressures available. On 
the other hand, if pipes are too large the diminished velocity of 
the flowing liquid may only hasten the formation of deposits 
and the ultimate clogging of the pipes. The effect of corrosive 
liquids can be overcome only by the use of pipe material which 
is not attacked by the liquid to be handled. 
All chemical-solution lines should be provided with means for 
flushing them out with water under pressure. In some cases 
Cl Propeller with 
two blades "w- 
i Water Supply-' 
I Drain 
Fig. 100.-Details of agitator for 
bleaching-powder solution tank designed 
for Montreal, Quebec, filtration plant. HANDLING AND STORING OF CHEMICALS 
271 
steam and hot-water connections will also materially assist in 
removing certain kinds of deposits. Certain deposits, as for 
example those formed in pipes conveying milk of lime or lime- 
water solutions, can not be removed by flushing and must be 
scraped out with scrapers. Deposits of lime that have been 
accumulating for some time in pipes, are very difficult to remove, 
and not infrequently involve the removal of the pipe entirely and 
the substitution of new pipe in its place. 
For conveying sulphate of alumina solutions (Fig. 101), lead, 
special bronze or hard-rubber piping are the best to use. Lead 
and brass or bronze piping are probably slightly attacked by this 
solution, but are usually found satisfactory where good grades 
of the chemical are used. Hard rubber is not attacked at all, 
i'Calv.lron- 
Receiving Tank-, 
Coagulant Feed to Influent Chamber? 
To agitate water in 
influent chamber 
Sedimentation Basin 
Solution 
Tank 1 
Solution 
Tank 2 
Fig. 101.-Airlift for alum solutions. 
but is liable to crack, beside being expensive. Iron pipe may be 
used for lime and soda-ash solutions, as it is not attacked by 
these chemicals. Sulphate of iron solutions, if not too acid, 
may be safely run through either cast or wrought iron, or 
even steel pipes. Slightly basic sulphate of iron tends to deposit 
iron hydroxide in pipe lines, and its corrosive action is very 
slight. This is not the case, however, if appreciable amounts of 
free acid are present. 
Calcium hypochlorite solutions may be conveyed through iron 
pipes, but both corrosion and incrustation occur, and the life of 
the iron pipes is short. Tile, lead and hard-rubber piping may 
be used for conveying this solution, as none is attacked. Lead- 
lined iron pipe, brass, bronze and copper pipe have been used, 
as well as ordinary rubber hose. 272 
WATER PURIFICATION 
Dry chlorine gas may be conveyed through iron or brass piping 
without corroding them. If water either as a liquid or as a vapor 
comes into contact with the gas while in the pipe, corrosion will 
result. Lead is probably attacked less than most metals, and 
is used to deliver the gas into the water to be treated. Hard- 
rubber or tile pipe is also used for discharging the gas under 
water, and also small pipes made of silver. Ordinary rubber 
garden hose has been successfully used by the author, but in 
time the rubber is attacked and becomes brittle and breaks. 
However, the low first cost of the hose makes its renewal easy 
and inexpensive. 
In general, the fittings and valves for chemical-solution pipe 
lines are of iron, brass, special bronzes or hard rubber. Lead, 
copper, stoneware and glass valves have been used, but more 
especially for corrosive solutions like those of bleaching powder. 
Gate and globe valves are commonly employed, and sometimes 
brass cocks. The latter, however, are not as a rule desirable on 
account of their liability to stick. 
Pumps.-Small pumps of both the reciprocating and centrifu- 
gal types are used for forcing chemical solutions through pipe 
lines to some higher elevation or into the water to be treated. 
They should not be used if it is possible to avoid it, since they 
are, of course, subject to the corroding action of the liquids they 
handle with all the attendant difficulties. Pumps of hard rubber 
or of Monel metal are practically proof against corrosion, but 
are expensive. Pumps with cast-iron pump ends lined with 
lead are not infrequently used for the more corrosive solutions. 
Eductors.-Mr. George W. Fuller recommends the use of 
water eductors for handling chemical solutions. He states: 
"They have the double advantage of being of ideal simplicity of 
form, and of providing additional diluting water to the solution in its 
passage through the pipes, with a consequent reduction in the objec- 
tionable action of the chemicals on the piping. The proportions of 
the eductors can be varied practically at will for handling solutions 
against different heads and with different working water pressures. 
The relations between these variables can be expressed by the formula: 
= J? -1 
Q \h 
where q equals the quantity of solution to be pumped, h the head to 
which the solution is to be pumped, Q the quantity of pressure water HANDLING AND STORING OF CHEMICALS 
273 
used, and H the available head of the pressure water: thus if the pres- 
sure water is available at 200 ft. head, and the solution is to be pumped 
against a head of 40 ft., 1 volume of pressure water will handle 1.22 
volumes of solution."1 
These ejectors can probably be used satisfactorily with chemical 
solutions that do not tend to incrust. With milk of lime they 
have not proven entirely satisfactory. 
References 
1. George W. Fuller: "Some Features of Detail in the Design of 
Rapid Sand Water Filtration Plants." Engineering-Contracting, Sept. 
2, 1914, p. 229. 
2. C. P. Hoover: "The Manufacture of Aluminum Sulphate Syrup at 
Columbus, Ohio, Water Purification Plant." Eng. Record, May 8,1915. 
3. "The New Coagulating Works of the St. Louis Water Department." 
Eng. Record, vol. 61, Feb. 17, 1910. 
4. Annual Reports of Division of Water of Columbus, Ohio, 1913. 
5. Annual Reports of Commissioner of Water of St. Louis, Mo., for 1911, 
1912, and 1913. 
6. E. G. Manahan and J. W. Ellms: "The Cincinnati Water Purifica- 
tion Plant." Eng. Record, Apr. 6, 1907. 
7. "The Municipal Water Purification Plant at Grand Rapids, Mich." 
Eng. Record, Sept. 30, 1911, p. 379. 
8. "The Mechanical Filtration Plant at Minneapolis, Minn." Eng. 
Record, Nov. 18, 1911, p. 586. 
9. James W. Armstrong: "The Baltimore Filtration Plant." Eng. 
Record, May 8, 1915, p. 583. 
1 Engineering-Contracting, Sept. 2, 1914, pp. 225-229. CHAPTER XXII 
APPARATUS AND METHODS FOR APPLYING CHEMICALS 
AND THE PREPARATION OF SOLUTIONS 
Apparatus for Feeding Dry Chemicals.-The application of 
chemicals to a water should be obviously regulated with con- 
siderable care in order to prevent either under or overtreatment. 
The application of the solid chemicals in a dry state directly to 
the water to be treated is used to a limited extent. The method 
possesses undoubted advantages in eliminating standard solu- 
tions of definite strength and the large tanks necessary to store 
them; and also in doing away with all apparatus for measuring 
their rate of application. On the other hand, the physical 
characteristics of some of the chemicals used in water purifica- 
tion, and the imperfections of much of the apparatus that has 
been designed for feeding dry chemicals, have prevented the 
general adoption of the method. 
In spite of the mechanical difficulties which the method of 
dry feeding presents, an apparatus designed by W. Donaldson 
in 1910, in which friction drive of a worm gear actuated a feed 
screw and permitted a nice adjustment of the weight of the sub- 
stance through a wide range, was designed and is used in seven 
plants at the present time, eight feeders being used for hydrated 
lime and one for sulphate of iron. This particular device did 
not work well with soda ash. 
Dry-feeding apparatus (Fig. 102) usually works well with 
aluminum sulphate as it is not hygroscopic and does not cake 
and bridge over the orifice. Hydrated lime may be fed in this 
way with considerable success in spite of its tendency to bridge 
over an orifice. This latter difficulty occurs with any fine 
amorphous material, and of course causes intermittent feeding. 
It is practically impossible to feed bleaching powder successfully 
in dry-feed apparatus. 
Mr. Allen Hazen in conjunction with Mr. R. S. Weston has 
designed a dry-feed apparatus that has been successfully used 
for several years to feed crushed sulphate of alumina to a water, 
274 METHODS FOR APPLYING CHEMICALS 
275 
and it is claimed that experimentally the apparatus with some 
modifications has been also successfully employed in applying 
calcium hypochlorite. 
The Hazen apparatus consists of "a screw conveyor operated 
in the bottom of a large hopper containing the dry chemical in 
such a state of physical subdivision that it would be steadily fed." 
Motive power for driving the conveyor is furnished by an electric 
motor or by a Pelton waterwheel. If the latter is used, the water 
discharged from the wheel may be employed in the dissolving 
box into which the dry chemical is dropped. The speed of the 
screw is controlled by a system of multiple gearing, which permits 
Water Wheel 
Cut Spur Gear- 
1200 R. PM. 
Four sets af change gears 
for varying speed 
Helical Conveyor 
Worm 
Gearing 
Water 
Wheel 
Water Level 
Fig. 102.-Dry-feed chemical apparatus. 
of a wide range in speed between the motor and screw, and by 
changing the speed of the motor. 
In apparatus installed in plants at Springfield, Mass., and in 
Poughkeepsie, N. Y., a 4-in. screw driven by a variable-speed motor 
with a range in speed from 1 to 4 for the motor, and with gears 
with a range from 1 to 56, enables one screw to feed chemicals 
with a variation in speed from 1 to 224, a range considerably in 
excess of the variations likely to be required in any plant. A 
2-in. conveyor will discharge, according to the gears in use from 
1 to 3 tons of coagulant per day, or the equivalent of 1 grain 
per gallon to as much as 40,000,000 gal. of water; while on the 
other hand, it may be run slowly enough to treat as little as 276 
WATER PURIFICATION 
100,000 gal. of water per day. A revolution counter attached 
to the screw conveyor provides a method for keeping a record 
of the amount of chemical used. The screws which have been 
employed in this type of apparatus ranged from 2 to 4 in. in 
diameter, but the preference 
would seem to be for a screw 
of small diameter and operated 
at a relatively high speed. 
Tests of this apparatus have 
shown that the discharge of 
chemicals per revolution 
varies less than 3 per cent, 
from the lowest to the highest 
speed. 
To effect a rapid solution 
of the chemical it should be 
dropped into flowing water. 
By keeping the solid particles 
in motion until they are dis- 
solved, the settling out of the 
solid may be prevented. 
In Figure 102a is shown the 
cut of a dry chemical feeding 
machine1 which has been used 
successfully in feeding hy- 
drated lime at the Toledo, Ohio 
filtration plant. This machine 
consists of a feed hopper with 
an agitator, a revolving drum 
with a ratchet drive, an adj ust- 
able orifice, a mixing compart- 
ment and a motor for driving 
the drum. The motor used 
was driven by water under 
pressure, and the discharged water was used for washing the 
lime into the pipe line through which the lime was being applied 
to the raw water. By adjusting the stroke of the motor, and 
the aperture of the orifice, it was possible to obtain a fairly 
uniform feed. The rated capacity of the machine varied from 
90 to 14,000 pounds in 24 hr. 
Fig. 102a.-Dry-feed chemical appa- 
ratus (Pittsburgh Filter Mfg. Co.). 
1 Made by the Pittsburgh Filter Mfg. Co., Pittsburgh, Pa. METHODS FOR APPLYING CHEMICALS 
277 
Modification of the Method of Dry Feeding of Chemicals. 
St. Louis Method.-In 1904 a method of applying chemicals to 
the water supply of St. Louis, Mo., was tried, which proved to be 
quite successful. By this method the chemicals were weighed 
out in small quantities and dumped by hand at regular intervals 
into solution tanks. These tanks were small and the solution 
formed by the discharge into the tanks of a sufficient but un- 
measured volume of dissolving water, overflowed continuously 
into the water being treated. No accumulation of undissolved 
chemical was permitted, and the amount dissolved was, there- 
fore, equal to the amount applied, which latter was, of course, 
adjusted to the quantity and character of the water being treated. 
The chemicals, which were used, were sulphate of iron and 
caustic lime. The so-called "sugar sulphate of iron" was em- 
ployed on account of the ease with which it dissolves and because 
of its purity. Weighed quantities of the iron sulphate were 
dumped at regular intervals into a wooden dissolving box 4 ft. 
8 in. by 5 ft., by 3 ft. deep inside, and which was supplied with 
water through 20 pipe nipples, % in. in diameter, and closed with 
brass caps. Each cap had four %4-in. holes drilled in the side, 
and so placed that the issuing water made a fairly uniform sheet 
over the bottom of the tank. The solution escaped through 
overflow pipes near the top of the tank. The solution was fed 
through wrought-iron pipes directly into the uptake shaft 
through which the water was passing from the intake to the wet 
well of the low-service pumping station. With some few modi- 
fications the same apparatus and method are still followed in 
the new coagulant house previously described. 
The lime-slaking apparatus consisted of four tanks into which 
the weighed lump lime was dumped from wheel-barrows at 
regular intervals. A constant volume of water at a temperature 
of about 120°F. was fed into these tanks. The milk of lime 
formed had a temperature of 140° to 180°F., and overflowed into 
the uptake shaft, the same as did the sulphate of iron solution. 
On account of the lime incrustations on valves and pipes, this 
point of application of the milk of lime was abandoned, and cen- 
trifugal pumps were installed, which forced the liquid to a receiv- 
ing well beyond the pumps and between the latter and the set- 
tling basins. 
The stirring apparatus for each lime-slaking tank consisted of a 
vertical shaft with a crossarm about 1 ft. from the bottom of the 278 
WATER PURIFICATION 
tank, and provided with rakes. The rake had the form of a 
half-circle, swinging in a vertical plane about the crossarm and 
with an arrow head on the bottom end. A 5-lb. weight fastened 
at the back of the rake gave the necessary resistance to swinging 
and brought it back to position if temporarily raised by striking 
lumps of lime which would cause it to jam, while still holding it 
in place to move everything in front of it. 
This apparatus has been somewhat modified in the new coagu- 
lant plant of the St. Louis Water Department, although the gen- 
eral principles involved are the same. The slaking tanks are of 
steel instead of wood, the stirring apparatus is more strongly 
built, and is driven from below the tank from a single line shaft. 
An ingenious arrangement of the vertical shaft from the bevel-gear 
drive avoids a stuffing box in the bottom of the tank. The tops 
of the tanks are covered. They are designed to slake 80 lb. of 
lime per minute, but can readily slake 90 lb. per minute. 
In the old plant the volume of slaking water added was 30 gal. 
per minute per tank, irrespective of the amount of lime added. 
This made the maximum strength of the milk of lime 8^ lb. of 
water to 1 lb. of lime, or 7,000 grains of lime per gallon. The 
present method is to add about 3^ lb. of water to 1 lb. of lime. 
The resulting temperature of the milk of lime is about 200°F. As 
this liquid has to be cooled before it could be pumped, a steel 
heater tank was installed to utilize this heat to raise the tem- 
perature of the water used to slake the lime, and thus reduce the 
quantity of steam formerly used to heat the slaking water. The 
heater tank is 4 ft. 6 in. in diameter and 3 ft. 3 in. in depth, and 
contains 1^-in. copper pipe coils, and a stirring apparatus. The 
milk of lime from all the slaking tanks discharges into the top of 
this tank, and it in turn discharges the cooled milk of lime through 
an overflow pipe near the top of the tank into the pump box, 
where, after dilution with cold water, it is forced by pumps to 
the delivery well. 
The efficiency of this method of slaking lime was found by 
Mr. W. F. Monfort to be considerably affected by the initial 
temperature of the water. In the following table he shows a 
series of results obtained in experiments made with this appa- 
ratus.1 
By maintaining an initial temperature in the hot-water supply 
of 160°F., and a temperature of 200°F. or more in the slaking 
1 Eng. Record, vol. 56, July 27, 1907. METHODS FOR APPLYING CHEMICALS 
279 
Series 
Tank 
Initial temp., 
deg. F., 
milk lime 
Maximum temp., 
deg. F., 
milk lime 
Efficiency, 
per cent. 
A 
1 
110 
156.0 
70.30 
1 
120 
171.3 
78.13 
4 
122 
156.0 
72.85 
3 
172 
202.0 
83.65 
B 
4 
127 
165.0 
66.04 
3 
165 
200.0 
81.25 
C 
4 
117 
153.5 
79.35 
3 
172 
208.0 
94.60 
tanks, Mr. Monfort concludes that a 15 per cent, greater 
efficiency can be obtained in slaking the lime. 
At the Cincinnati filtration plant, where the same general 
methods of applying chemicals have been followed as at St. 
Louis, a comparatively weak milk of lime is used. The strength 
of this milk of lime will range from 1,000 to 2,500 grains per 
gallon, and will average about 1,500 grains per gallon. The 
water used to slake the lime is heated to a temperature of 140°F. 
The milk of lime is discharged directly into a 40,000-gal. tank 
through which settled water is constantly flowing at a rate of 
about 900 gal. per minute. The milk of lime enters with the 
water flowing in at the bottom of the tank, and becomes dissolved 
as it flows upward with the diluting water, and passes out over 
the weirs at the top of the tank. A slowly moving stirring device 
rotating at the bottom of the tank assists in diffusing the lime 
through the water and thus aiding in its solution. 
In this way there is produced a lime-water solution, which will 
have a strength of approximately 30 grains per gallon. The 
strength of this lime-water solution will, of course, vary, depend- 
ing upon the time interval at which 100 lb. of the ground lime 
is dumped into the slaking tank. It is evident that the quantity 
of lime being slaked for a single lime-water saturator tank must 
not exceed that amount of lime which the water passing through 
the saturator is able to dissolve, viz., about 60 grains of calcium 
oxide per gallon. 
The efficiency of this method of applying lime was shown by a 
series of tests made by the author to be about 92 per cent. The 
results of these tests as tabulated in Mr. George H. Benzenberg's 
report as chief engineer of the Commissioners of Water Works 
of Cincinnati, Ohio, are as follows: 280 
WATER PURIFICATION 
Pounds. 
Per cent. 
1. Soluble lime in lime water applied to raw water. . . 
54,728 
77.4 
2. Soluble lime used up by the raw water required 
for making lime-water solution 
2,834 
4.0 
3. Soluble lime in dry sludge taken from slaking 
tanks 
1,755 
2.5 
4. Insoluble material in sludge from slaking tanks. . . 
6,229 
8.8 
5. Insoluble material presumably carried into 
saturator tank with milk of lime 
5,154 
7.3 
Total 
70,700 
100.0 
Percentage of the total available soluble lime in lime-water solu- 
tion applied to raw water 92.2 
Average strength of lime-water solution equaled 28.8 grains of 
CaO per gallon. 
APPLICATION OF MEASURED VOLUMES OF CHEMICAL SOLU- 
TIONS OF STANDARD STRENGTH 
The more common method of applying chemicals to a water 
to be treated is by means of an orifice or weir, through or over 
which flows a standard-strength solution of the chemical to be 
applied. The volume of this solution must, of course, be regu- 
lated so as to be properly proportioned to the volume of flow of 
the water being treated. This is effected usually by varying the 
size of the orifice or the width of the weir, although in some 
apparatus it may be produced by a change in the head coupled 
with a change in the size of the orifice or weir. 
Orifice Feed Tank.-A typical orifice feed tank is shown in the 
accompanying cut (Fig. 103). 
It consists of a cast-iron porcelain-lined tank, into which the 
chemical solution is fed from a large storage tank. The solu- 
tion enters through a cock controlled by a glass float.' By 
means of the ball cock and float a uniform depth of solution is 
maintained in the orifice tank. An overflow pipe of hard rub- 
ber is shown in one corner of the tank. 
The orifice through which the solution escapes is located near 
the top of a vertical pipe of hard rubber which passes through a 
rubber stuffing box in the bottom of the tank. This pipe may 
be moved up and down by means of a hand wheel and threaded 
stem, and thereby submerge the orifice to any desired depth. 
The depth at which the orifice is placed is indicated directly on 
a dial. The threaded stem has 16 threads to the inch, and one 
revolution of the hand wheel corresponds to a movement of the 
pipe, and consequently of the orifice, of Xe in. Very accurate METHODS FOR APPLYING CHEMICALS 
281 
adjustments of the depth of the orifice is, therefore, possible. 
Placing the orifice in the vertical feed pipe, so that it is sub- 
merged and yet is above the bottom of the feed tank, diminishes 
the chances for its becoming clogged by floating or deposited 
material. 
Where it is desired to apply the chemical solution to suction 
pipe lines conveying water to the pumps, it is necessary to use a 
suction tank in connection with any gravity feed orifice or weir 
tanks, in order to prevent entrance of air to the pump suction 
Fig. 103.-An orifice feed tank. 
pipes. This is effected by discharging the chemical solution 
from the orifice feed tank into an intermediate tank directly 
connected with the suction pipe. By means of an auxiliary 
supply of water entering through a cock controlled by a float, 
this intermediate tank is kept filled with water, thus sealing its 
outlet to the atmosphere and preventing air entering the suction 
pipe of the pump. 
Many variations from the general type of orifice feed tank 
described above are in use. Rectangular and circular orifices 
with adjustable gates are commonly employed, and automatic 
adjustment of the openings to correspond with changes in the 
volume of water being treated is not unusual in the larger plants. 
Proportional Chemical-solution Feed Pumps.-The variation 
in the flow of water through a purification plant, in order to 
meet fluctuations in consumption, necessitates a proportional 
variation in the dose of chemicals to be applied. This change 282 
WATER PURIFICATION 
may be made automatically or by hand. In the smaller plants 
proportional chemical-solution feed pumps attached directly to 
reciprocating steam pumps, and actuated by the latter are 
sometimes used. 
The front and rear head of one water end of the steam pump 
is drilled and tapped for pipe connections, which lead to the top 
and bottom, respectively, of the water end of the chemical 
pump. Each stroke of the steam pump causes a corresponding 
stroke of the chemical feed pump. A change in the amount of 
chemical solution to be added is effected by lengthening or shorten- 
ing the stroke of the plunger in the chemical end of the feed 
pump. For any given setting of the pump the quantity of 
chemical applied is proportional to the water being pumped by 
the steam pump. 
The motor end of the feed pump may be of brass. The 
chemical end is usually of iron, which for corrosive chemical 
solutions is lined with lead. 
Control of Application of Chemical Solutions by Means of 
Venturi Tube/-Automatic regulation of chemical feed by means 
of the change in velocity of the flow of water to be treated as it 
passes through a constricted pipe is one of the more common 
methods employed, especially in the larger plants. The general 
principle involved in this method of control is clearly shown by 
the accompanying cut (Fig. 104). 
This apparatus as described and tested by Mr. Ralph Hilscher, 
the patentee, depends for its action directly on the decreased 
static head at the restricted section of a Venturi meter, which 
is measuring the water to be treated. It consists of a tank 
divided into two compartments by a wall extending nearly to 
the top of the tank. Resting on the top of this dividing wall is 
a pivoted horizontal arm, from which are suspended two floats 
of.equal dimensions, one in each compartment, and at equal 
distances from the pivot. The arm in compartment B extends 
beyond the float, and connects with a balanced valve on the 
end of the feed pipe, which supplies the chemical solution from 
a storage tank. The valve is made so that when the float B 
falls below the float in A, the downward movement of the arm 
will cause the valve to open and admit chemical solution to 
compartment B. 
Compartment B is directly connected to the reduced section 
of the Venturi tube. It contains a valve which may be adjusted. METHODS FOR APPLYING CHEMICALS 
283 
The full section of the Venturi tube above the constriction is 
connected directly to compartment A. When a flow of water 
occurs through the Venturi tube, unequal pressures at the full 
and reduced sections will tend to make the levels in A and B 
different, that in B being lower. This difference in head (/i) 
is proportional to the square of the rate of flow through the 
Venturi tube, and, if utilized to force the chemical solution 
Supply 
Adjustable Valve 
Venturi 
Tube 
Fig. 104.-Diagram of chemical feeder. 
through an orifice, will produce a flow through that orifice 
directly proportional to the flow in the pipe line. The adjustable 
valve acts as an orifice in this device, and the desired effective 
head on it is established by building up the level in B equal to 
that in A. 
In tests made of an experimental apparatus, in which the rates 
at which the water was treated varied by about 300 per cent., 
departures from the average ratio of the volume of water treated 
to the volume of chemical solution being applied ranged from 
7.7 per cent, below to 6.2 per cent, above the mean, but averaged 
only 1.7 per cent, on either side of the mean. 284 
WATER PURIFICATION 
Register 
■Indicator 
Weir Chamber 
Vent 
H.WL m Weir 
Chamber 
Storage Chamber 
Fig. 105.-Proportional flow weir for feeding chemical solutions. 
Overflow Box- 
Stirrer Arm 
.Cut-off Plate 
Chemical Regulator 
Chemical Pipe 
to Softening 
Tanks - 
I Return to 
Chemical Tanks 
From Chemical 
Pump 
Overflow Box 
From Chemical 
Pump. 
, 'Return to 
Chemical Tank 
Chemical Pipe 
to Softening 
Tanks 
Cut-off Plate 
Fig. 106.-Regulation of chemical feed by means of Sutro weir. METHODS FOR APPLYING CHEMICALS 
285 
Control of Application of Chemical Solutions by Means of 
Weirs.-An ingenious application of this method of control 
is shown by the use of a proportional-flow weir for measuring 
the water to be treated in combination with an automatic ad- 
justable weir for regulating the flow of the chemical solution. 
The Sutro weir1 (Fig. 105) is of the proportional-flow type, that 
is the head of water on the weir is in direct proportion to the 
Fig. 107.-Booth's chemical regulator. (Made by the L. M. Booth Co.) 
quantity flowing over it. By means of a float in a chamber con- 
nected to the weir box, the change in the volume of water to be 
treated is indicated by the rise and fall of the float. 
This movement of the float is communicated by a system of 
levers to a cutoff plate, which varies the width of the weir over 
which the chemical solution is flowing. The head on the 
chemical-solution weir remains constant, as the cutoff plate 
merely diverts more or less of a constant volume of flow of the 
chemical solution to the discharge pipe, the balance flowing 
1 Manufactured by the L. M. Booth Co. 286 
WA TEH P U RIF IC A TION 
through the return pipe to the storage tank. By thus varying 
the width of the chemical-solution weir, by means of this cutoff 
or diversion plate, while the head remains constant, it is evident 
that the volume varies in proportion to the lateral shifting of the 
plate, and consequently to the volume of water passing over the 
Sutro weir (Figs. 106 and 107). 
PREPARATION OF SOLUTIONS 
A simple and effective method for preparing solutions of definite 
strength is to place in a compartment within and at the top of 
the solution tank, the weighed amount of the chemical compound 
to be dissolved. From a perforated pipe just above the com- 
partment water may be sprayed over the chemical until it is 
dissolved, the solution overflowing into the large tank. By 
regulating the volume of the water used to dissolve the chemical, 
the large tank can be filled by the time solution is effected. 
Sulphate of alumina, sulphate of iron and soda ash are easily 
dissolved in this manner. Soda ash, if dumped into the bottom 
of a tank and solution attempted by covering it with water, 
tends to form a solid crystalline mass very difficult to dissolve. 
Sulphate of iron is quite easily dissolved, especially in the form 
of fine-grained crystals known as "sugar sulphate of iron." 
This salt has a tendency, however, to form hard masses of crystals 
at the bottom of a tank, when large quantities of the salt are 
being dissolved without adequate stirring. 
At the Columbus, Ohio, water-purification plant, Mr. C. P. 
Hoover, chemist in charge, has perfected a process whereby an 
aluminum sulphate syrup (Fig. 108) is manufactured directly 
from bauxite and sulphuric acid at the plant. The bauxite 
and sulphuric acid are boiled together in lead-lined tanks until 
a basic solution of aluminum sulphate is obtained. The solu- 
tion is then diluted with water and is measured as needed into 
alum solution tanks, in which it is further diluted with sufficient 
water to make a solution of the standard strength that is to be ap- 
plied to the water being treated. By this process it is claimed that 
five distinct steps in the manufacture of aluminum sulphate and 
its preparation for use in water purification are eliminated, 
namely, filtering, concentrating, crystallizing, grinding and re- 
dissolving. It is estimated that a very considerable saving 
in cost is obtained as compared with the cost of using the solid 
aluminum sulphate, and that even greater reductions in cost can METHODS FOR APPLYING CHEMICALS 
287 
9"Spira/ Conveyor 
Elevator 
Acid Meas- 
uring Tank 
Elevation A-A 
Line Shaft 
Drive for 
Crusher 
Sludge 
Tank 
Crusher 
Manhole to 
(Crusher 
Boiling 
Tanks 
''Elevator 
Pulverizer 
Measuring Tanks 
Fig. 108.-An alum, manufacturing plant at the Columbus, O., filter plant 
Sectional Plan B-B 288 
WA TER P U RIF IC A TION 
probably be obtained by utilizing a cheaper raw product known 
as halloysite as a substitute for the bauxite. 
The following is an estimate of the cost of producing 100 tons 
of 17 per cent. A12O;! alum in solution at the Columbus Plant: 
Estimated Cost of Producing 100 Tons of 17 Per Cent. A12O3 Alum 
in Solution 
468 tons 66°Bc. sulphuric acid @ $12.50 $5,850.00 
265 gross tons bauxite @ $9.90 2,623.50 
Lubricating oil 20.00 
Steam 100.00 
Electric current, 10,000 kw.-hr. @ 2 cts 200.00 
Repairs to plant 500.00 
Depreciation '. 600.00 
Interest on investment 600.00 
Total $10,493.50 
In preparing solutions of bleaching powder, the compound 
must be rubbed into a paste with water, and thoroughly agitated 
with the dissolving water, so that it is diffused throughout the 
latter in a finely divided form. The tendency of this material is 
to form pasty lumps, the interiors of which are not reached by 
the water used to dissolve them. The large amount of insoluble 
material always present, with the consequent formation of sludge 
in the bottom of the tank, adds to the difficulty in obtaining the 
full value of this chemical in aqueous solutions. Not more than 
8^ lb. of bleaching powder (1 per cent.) should be added to 
100 gal. of water in preparing solutions, and a better solution 
with less waste will be produced if only one-half this weight of 
powder is used to the above volume of water. 
The strength of solutions of coagulating chemicals that are 
commonly used varies more or less with local conditions, such as 
tank capacity, volume and character of water to be treated. A 
convenient strength for a sulphate of alumina solution is 2 per 
cent., while solutions of sulphate of iron may vary from 2 to 5 
per cent. Too strong solutions are undesirable on account of 
their greater corrosive and incrustation effects in tanks and pipe 
lines. The accumulation of sludge in chemical solution tanks 
should be avoided by frequent cleaning. The clogging of feed 
lines and drains with these deposits may be minimized, but not 
entirely avoided by frequent flushing with water under consider- 
able pressure. In some cases hot water and even steam may be METHODS FOR APPLYING CHEMICALS 
289 
found beneficial in clearing long lines of chemical feed pipes. In 
iron sulphate solution tanks and pipe lines, iron bacteria (Creno- 
thrix and Leptothrix) have been found growing in such profusion 
as to form slimy sludges which adhere to the sides of the tanks and 
clog the pipe lines. The author's attention was called to these 
growths by Mr. W. F. Monfort of St. Louis, and he was able to 
confirm his observations by finding abundant growths of Creno- 
thrix in the sludge of sulphate of iron solution tanks at the Cincin- 
nati filtration plant. 
References 
1. W. F. Monfort: "Dry Feed of Chemicals in Water Purification." 
Jour. Am. Water-works Assn., March, 1915. 
2. H. N. Odgen: "Dry-feed Apparatus." Eng. Record, May 8, 1915, p. 586. 
3. Allen Hazen: "Dry Feed for Chemicals Usedin Water Purification." 
Eng. Record, July 19, 1913. 
4. "Automatic Chemical Feeder" (Ralph Hilscher, patentee). Eng. Record, 
Sept. 6, 1913. 
5. "Tuning Up the Minneapolis Filter Plant." Eng. Record, June 21, 1913. 
6. E. E. Wall: "Method of Regulating and Applying Chemicals in the 
St. Louis Water Department." Trans. Am. Soc. C. E., vol. 54, Part D, 
p. 236. 
7. W. F. Monfort: "Notes on Water Purification at St. Louis." Eng. 
Record, July 27, 1907. 
8. H. P. Letton: "Operations of Small Rapid Sand Filters." Munici- 
pal Journal, Mar. 26, 1914. 
9. "The Sutro Weir." Eng. News, vol. 72, p. 462. 
10. "Inverted Weirs." Eng. News, vol. 73, p. 72. CHAPTER XXIII 
POWER PLANT, PUMPING MACHINERY, AIR COMPRESS- 
ORS, AIR TANKS, WASH-WATER TANK AND 
MISCELLANEOUS EQUIPMENT 
A properly equipped rapid sand filter plant requires more or 
less power in order to operate its pumps, air compressors, stirring 
devices, and other auxiliary apparatus. The location of a filter 
plant near a pumping station often makes it possible to obtain 
all the necessary power from the station. On the other hand, a 
filter plant may be so located that an independent power plant is 
necessary. The larger the filter plant the more desirable does an 
independent power plant become. 
Power Plant.-Where power is derived from a nearby pump- 
ing station which is operated by a steam power plant, it may be 
possible to pipe the steam under pressure directly to engines driv- 
ing the pumps, air compressors and other machines to be operated 
in the filter-plant buildings. It is generally easier, however, to 
make use of the steam at the pumping station to drive electrical 
generators, and to convey the current thus formed to electric 
motors, which can be used for operating all the machinery used 
in the plant. For short distances a direct current may be used 
without a too great cost for conductors; but for long distances 
an alternating current will be found the cheaper means for con- 
veying the power. 
Because of the ease with which power may be conveyed by an 
electric current, it lends itself admirably to nearly all of the needs 
of filter-plant operation. Wash-water pumps, air compressors, 
stirring devices, valves, refrigerating apparatus, and other de- 
vices for which motive power is needed, can all be handled with 
electric motors. Even low-service centrifugal pumps for supply- 
ing untreated water to the purification plant, and which in some 
cases may be regarded as almost an integral part of the filter 
plant, are usually motor-driven. Plunger pumps also, when pro- 
vided with proper gearing, may be motor-driven, and are fre- 
quently used. 
290 POWER PLANT EQUIPMENT 
291 
Where hydraulic power is available it furnishes a means for 
generating electric current at a low cost. Even comparatively 
low heads of water may be utilized, provided a sufficient volume 
of water is available. Even where the water power is not always 
sufficient, and must be supplemented at times by steam-driven 
units, the usual low cost of current obtained from generators 
driven by water power makes it well worth while to employ it 
wherever possible. 
Fig. 109.-Water wheels at the Cincinnati filter plant. 
Section A-B 
Section C-D 
Power Plant at Cincinnati Water-purification Plant.-A some- 
what unique utilization of water power (Fig. 109) is successfully 
employed at the Cincinnati filtration plant, where the water to 
be purified is drawn from settling and storage reservoirs to the 
head house of the filter plant. The available head of water is 
about 28 ft. The water from the settling reservoirs flows through 
two 60-in. pipes to the front of the head house. Six 36-in. pipes 
connect the two 60-in. mains with concrete channels in the base- 292 
WA TEH P U RIF IC A TION 
ment of the head house. On each 36-in. pipe line is a motor- 
operated gate valve, by which the flow into the head house is 
regulated. 
On each of three of these 36-in. pipes there has been installed 
a Trump horizontal waterwheel. Each wheel is rated at 83 hp. 
with a flow of 2,100 cu. ft. per minute, or when about 22,000,000 
gal. of water passes through each wheel in 24 hr. The water- 
wheels are connected by flexible couplings to Fort Wayne 50-kw., 
230-volt direct-current generators, and are operated at 360 revo- 
lutions per minute. The speed of each wheel is controlled by a 
Lombard type M oil-pressure governor. Speed regulation is also 
assisted by a cast-steel flywheel placed on the turbine wheel 
shaft, and which weighs 3,900 lb. 
A peculiar feature of the waterwheels is the external bypass 
on each wheel, which opens as the turbine gates close, due to any 
reduction of the load on the generators, and which, thereby, 
relieves the sudden pressure in the supply pipe, and at the same 
time maintains a constant rate of flow of water into the plant. 
The rate of flow does not vary more than 3 per cent., and when 
tripping the circuit-breakers at full load the variation in speed 
of the wheels is but 12 per cent. 
The original installation for power and lighting purposes for 
this plant consisted of three 150-kw. generators driven by De- 
Laval steam turbines. This power plant is located at the River 
Pumping Station of the Cincinnati water-works, which is about 
mile distant from the filter plant. They now form reserve 
units which may be used when extra power is needed, as for 
example, when cleaning settling reservoirs or coagulating basins. 
The waterwheels, however, are entirely sufficient to supply 
power for the routine operation of the plant, and for lighting 
and some minor power requirements at the River Pumping 
Station. 
Pumping Machinery.-The pumping station which supplies 
the untreated water to a filter plant can not, as a rule, be regarded 
as a part of the filter-plant installation, although usually located 
in close proximity to the latter. Unless storage reservoirs of 
large size intervene between the pumps and the filter plant, the 
operation of the two plants must be coordinated if proper purifi- 
cation of the water is to be effected. For this reason the rate 
of pumping must usually follow the rate of filtration within 
comparatively narrow limits. POWER PLANT EQUIPMENT 
293 
Types of Pumps.-Both displacement and centrifugal pumps 
are used for this service, and points of particular advantage may 
be found in each type of pump according to local conditions. 
Mr. R. L. Daugherty, in his book on "Centrifugal Pumps," 
discusses in an impartial manner the relative merits of the two 
types of pumps. The following statements have been freely 
drawn from his summaries of the advantages and disadvantages 
involved in the use of the two types of pumps. 
Mr. Daugherty states that in general it may be said that dis- 
Fig. 110.-Double-suction single-stage volute centrifugal pump. 
placement pumps have the advantage where the head pumped 
against is high, and the capacity small. They may be under 
some conditions much more economical of power. They are 
able to lift water from below when starting without being 
primed. Where either the head or the discharge varies within 
wide limits, and where they do not maintain definite relations 
with each other, the displacement pump is perhaps easier and 
more economical to operate. 
Centrifugal pumps may be broadly divided into two classes, 
viz., turbine pumps and volute pumps (Fig. 110) without dif- 
fusion vanes. The latter type of pump is more commonly em- 
ployed than the former in filter-plant installations. Centrifu- 294 
WATER PURIFICATION 
Fig. 110a.-Three-stage horizontal turbine pump. POWER PLANT EQUIPMENT 
295 
gal pumps are simpler in construction and generally much less 
difficult to operate than displacement pumps. A centrifugal 
pump can handle effectively water containing more or less sand. 
They are much more economical of space, lighter in weight, and 
less in cost than reciprocating pumps. They give a continuous 
and smooth discharge without the pulsations produced by a 
reciprocating pump. On the other hand, they are usually oper- 
ated at high speeds, requiring them to be motor-driven, or else 
direct-connected to a high-speed prime mover like the steam 
turbine. The motor-driven pump lends itself to a great variety 
of conditions, and in many places furnishes the most practicable 
and economical pumping unit, even where overall efficiencies 
are considered. 
With the exception of large pumping engines, which consist of 
slow-speed steam engines directly connected to displacement 
pumps, and which give a very high economy of steam consump- 
tion, the smaller reciprocating pumping engines are usually less 
efficient than the direct-connected centrifugal pump. The 
volume of water discharged by a centrifugal pump varies with 
the speed of rotation, and is in direct proportion to the square 
root of the head. The displacement pump also varies its dis- 
charge with the speed, but the speed has no relation to the head 
against which the pump is discharging. For a centrifugal pump 
the rated head and discharge for the pump will be those values 
for which the efficiency is a maximum. This value of the dis- 
charge is often designated as the normal discharge. These 
values differ for different speeds. 
Wash-water Pumps.-Unless it so happens that a pumping 
station, handling the filtered water from the filter plant, is located 
near the latter and is able to supply wash water under pressure, 
it is necessary to install pumps for this service. Since filtered 
water for filter-washing purposes is usually required in consider- 
able volume at varying intervals, and since this water when 
required must be delivered in a comparatively short space of 
time, it is advisable to provide a small reservoir or tank for stor- 
ing water for this purpose. The tank enables the pumps to be 
smaller and to be operated somewhat more economically for the 
reason that they may be run more nearly continuously. 
The motor-driven centrifugal pump is well adapted for wash- 
water purposes. Wash-water pumps usually obtain their filtered- 
water supply directly from the filtered-water reservoir, or from 296 
WATER PURIFICATION 
a pump well or supply pipe connected with this reservoir. If 
the pumps are required to lift the water, the suction pipes must 
be provided with foot valves so that priming of the pumps is 
not required each time they are started. It has been the author's 
experience that a clack or flap foot valve is more productive of 
shock in the pipe line when the pump stops and is more liable to 
leak, than a number of small circular rubber pump valves oper- 
ated by springs and set in a cage. The area of the foot-valve 
opening should be about twice that of the suction pipe. It is 
usually desirable to place a strainer at the end of the suction pipe 
in order to protect the foot valve. The openings of the foot 
valves should be at least 3 ft. under the water surface to prevent 
vortex action. 
The suction lift should not exceed 25 ft. at sea level, and it is 
always best to keep it as low as possible. The greater the suc- 
tion lift the greater the liability of air dissolved in the water 
being liberated. This air may collect at some point in the pipe 
line, and, as it is always richer in oxygen than the atmosphere, 
it more actively corrodes the iron with which it comes into con- 
tact. According to Mr. R. L. Daugherty, 90 per cent, of centrifu- 
gal-pump troubles will be found on the suction side of the pump. 
The remainder of the difficulties are largely due to end thrust. 
The capacity of wash-water pumps, when they are used to 
supply a tank, depends largely on the capacity of the latter and 
on the size of the filter plant, viz., on the maximum demand 
which the filters may make for wash water at any time. The 
pumping capacity should not be less than 10 per cent, of the 
filtered water output of the plant, and preferably 12 to 15 per 
cent. The pressures against which wash-water pumps operate 
vary with the local conditions. When utilized for filter-washing 
purposes only, the total head really needed probably never 
exceeds 50 ft., although other considerations may make it desir- 
able for them to be capable of pumping against a much higher 
head. 
Types of Centrifugal Wash-water and Flushing Pumps Com- 
monly Used.-The kind of centrifugal pump used for wash-water 
purposes is generally of the volute type as this lends itself better 
to a varying range of head and to low lifts. The horizontal split 
case affords easy access to the pump impeller, and the interior 
of the casing, and makes the removal of the shaft and impeller 
when necessary, less difficult than in the side-plate casing type. POWER PLANT EQUIPMENT 
297 
For single-stage pumps the double-suction inclosed impeller type 
offers the advantage of eliminating end thrust and considerable 
wear and friction on bearings, beside permitting a larger volume 
of water to be handled with a given diameter of impeller. 
The advance in late years in centrifugal-pump design enables 
builders to supply single-stage pumps for considerable higher 
heads than was formerly the case. From 100 to 200 ft. head per 
stage appear to be the usual limits within which design is being 
kept. Multistage pumps, therefore, are not usually necessary 
in filter-plant operation, where high pressures are not commonly 
required. The exception to this may be in the case of centrifugal 
pumps used for flushing out sediment in coagulation and settling 
basins, where pressures at the nozzles of hose lines should be from 
60 to 75 lb. per square inch, and where a pressure of 100 to 120 lb. 
per square inch may be needed at the pump. 
Centrifugal pump casings are usually of cast iron, and the im- 
pellers of hard bronze. The portion of the shaft in contact with 
the water is generally protected with bronze sleeves. Labyrinth 
rings used to prevent leakage of water from the discharge to the 
suction chambers are also made of bronze. Ring-oiled bearings 
are necessary for the high speeds at which centrifugal pumps are 
usually operated. 
Pressure Pumps.-In case hydraulic valves are employed, it 
is necessary to provide water under considerable pressure to 
operate them. Other requirements for water under high pres- 
sure may exist, such as for the operation of sample pumps, rate 
controllers, boiler feed lines, sand ejectors, hose lines and so 
forth. The quantity of water required is not usually large. 
For this service an ordinary steam-driven displacement pump 
may be used, or a motor-driven triplex, single-acting, outside- 
packed plunger pump. Pumps of the latter type are quite com- 
monly driven through gears, but may be belt- or chain-driven 
as well. The speed of triplex pumps is from 36 to 50 revolutions 
per minute. For filter-plant operations pressures in excess of 
100 lb. per square inch are rarely required. 
Motors for Centrifugal Pumps.-In Vol. 1 of their books on 
"American Sewerage Practice," Messrs. Metcalf and Eddy have 
given an excellent summary of the relative merits of different 
types of electric motors for pumping work. The following state- 
ments are in part quoted from this work. 
Electric motors for centrifugal pumps should not be too small, 298 
W A TER P U RIF IC A TION 
as they may become overloaded under certain conditions. Their 
capacity should be such as to meet the maximum conditions 
imposed by the pump. If thp head is variable, speed regulation 
of the motor is desirable. This is possible where direct current 
is available. For pumping into elevated tanks or reservoirs in 
most filter-plant installations, the variation in head is usually 
not great enough to require a variable-speed motor. Where the 
capacity of the pump is to be varied, and a standard induction 
motor, which runs at constant speed, is used, the discharge from 
the pump must be throttled. The shunt-wound direct-current 
motor is usually employed for driving centrifugal pumps, but a 
compound-wound motor is better adapted to those cases where 
the voltage or load fluctuates considerably. Automatically 
started motors should be compound-wound. The squirrel-cage 
type of alternating-current motor is also frequently used for 
driving centrifugal pumps, but it requires a high starting current; 
the slip-ring motor, which takes a very small excess current at 
starting, is better in this respect and is to be preferred. 
Air Compressors.-Where air is used for agitating the sand 
bed prior to or at the time of applying wash water during filter 
washing, a supply of compressed air is necessary. This is usually 
obtained from rotary blowers of the positive type. These blow- 
ers consist of two cam-shaped impellers, which, in rotating inside 
of a casing, meet in a line of rolling contact, and force the air out 
from between them. The impellers are operated through gears 
at a speed of about 200 revolutions per minute, and are generally 
motor-driven. 
It is usually required that the air be discharged at a pressure 
of 4 lb. per square inch. The capacity of these blowers should be 
approximately 4 cu. ft. of free air per minute for each square foot 
of filter surface that may be subjected to the washing process at 
any one time. 
Air Tanks.-A somewhat unique method of providing a stor- 
age for compressed air and wash water has been employed at 
several places in the United States and Canada. It consists in 
providing a telescopic tank built like a gas holder to contain both 
air and wash water. 
A tank of this character was built in connection with the 
rapid sand filtration plant at Trenton, N. J., and is described in 
the Engineering Record in the issue of May 9, 1914. POWER PLANT EQUIPMENT 
299 
"The tank is of steel and is 40 ft. in diameter. It provides water 
under a head of 24 ft., and air under a pressure of about 4 lb. per square 
inch. The floating air chamber is loaded with 300 tons of concrete 
to produce the required air pressure. The tank provides a uniform 
pressure of air, and approximately a uniform pressure of water without 
the necessity of installing a large wash-water pump and blower, and of 
maintaining a large amount of power ready for use at irregular intervals. 
The necessary capacities of wash-water pumps and blower are only 
about one-tenth of that which would be necessary in case of direct 
delivery of water and air to the filter beds. The tank will supply water 
Manhole for air riser' 
(}'2'Thickness of Concrete slob 
Telescopic 
Air-Holder 
-4O'OInside dia 
-4 2 'o*Inside dia 
O'Airlnlet' 
Pipe Colum ns To support head. 
Wash-Water 
/2 "Wash Wafer Pump D/schQrfy 
Fig. 111.-Telescopic tank for air and wash water. 
and air, without replenishing, for washing two filters in immediate 
succession; but is so arranged that the pump and blower automatically 
start to supply water and air to the tank as soon as washing commences. 
Water is supplied to the filters at the rate of 19 in. vertical rise per 
minute or 12 gal. per square foot of filter surface." 
The accompanying cut of the water and air tank of the tele- 
scopic type built for the Montreal Water and Power Co., illus- 
trates the details of this class of storage water and air tank (Fig. 
111). 
At Columbus, Ohio, the water-purification plant is provided 300 
WA TER P U RIF IC A TION 
with three storage tanks for air, which are kept under a pressure 
of 60 lb. per square inch, and from which the air is drawn through 
a pressure-reducing valve at a pressure of 2 to 4 lb. per square 
inch. The air compressor has a capacity of 100 cu. ft. of free 
air per minute, and is driven by a 15-hp. (220-volt) electric 
motor. The tanks are each 6 ft. in diameter and 30 ft. long, and 
are made from riveted-steel plate. They are of sufficient 
capacity to permit one filter to be washed about every 1.75 hr. 
for about 3 min. at a maximum rate of 3 cu. ft. per square foot of 
filter surface per minute.1 
Wash-water Tanks.-Since in the washing of rapid sand filters 
a large volume of water is used in a comparatively short space of 
time, and at irregular intervals, it is more economical to install 
a storage tank which may be filled between washes that it is to 
provide power and pump capacity capable of furnishing the vol- 
ume of water at the required rate during the washing process. 
Washing a 1,000,000-gal. filter unit of 360 sq. ft. at the rate of 
7.5, 10, 12, or 15 gal. per square foot per minute, equivalent to a 
vertical rise of the wash water of 12, 16, 19, or 24 in. per minute, 
requires that the total volume of water to be delivered shall be 
2,700, 3,600 ,4,320, or 5,400 gal. per minute, respectively. 
The storage-tank capacity should be governed by the size of 
the filter unit and by the frequency of washing. It is usually 
desirable to have the tank of sufficient size so that at least two 
filters may be washed without refilling the tank. The size of the 
pumps required to fill the tank depends upon the probable fre- 
quency of washing. It is apparent that they must be able to 
refill the tank between washes. The frequency of washing is 
governed by the character of the water being filtered. 
As an illustration of the required wash-water tank capacity 
for a filter plant having ten 1,000,000-gal. filter units, all of which 
are being operated, let it be assumed that filters must be washed 
for 5 min. at a rate of 5,400 gal. per minute, and that each filter 
is to be washed six times a day. Each washing would require 
27,000 gal. of water. This volume of water would be required 
at intervals of 24 min., or at the rate of 1,125 gal. per minute. 
The wash-water pumps, therefore, would have to have at least 
this capacity. If the tank is to hold enough water for two washes, 
then its capacity must be at least 54,000 gal. 
In designing the wash-water tank and pumps for a rapid sand 
1 John H. Gregory: Proc. Am. Soc. C. E., Vol. 36, No. I, January, 1910. POWER PLANT EQUIPAI ENT 
301 
filter plant, it should not be overlooked that the maximum de- 
mands of the plant and not the average must be provided for. 
The above illustration is perhaps an extreme, since the wash 
water would amount to 16.2 per cent, of the total output of the 
plant; nevertheless, as previously stated, at least 12 per cent, 
and possibly 15 per cent, of wash water should not at times be 
regarded as an impossibility. 
Wash-water tanks may be built of steel plates or of concrete. 
If there is no natural elevation upon which the tank may be 
located, it will have to be placed upon a tower or other suitable 
foundation, which will provide the head necessary for washing 
the filters. Wash-water tanks' should be covered over at the 
top in order to keep out the light and thus prevent microscopic 
plant growths in the water. They should be protected from 
frost in cold climates where the formation of ice might cause 
trouble. 
Steam Boiler Plant.-The heating of portions of the buildings 
of a rapid sand filter plant, the necessity for providing steam in 
the preparation of certain chemical solutions, and the need for 
a small amount of steam in the laboratories of the plant, make the 
installation of a steam boiler plant necessary even if not required 
for power purposes. The exception to this would be, of course, 
where steam could be conveniently and economically supplied 
for the above purposes by the boilers of a nearby power plant. 
The offices, laboratories, chemical house, and to a certain 
extent the filter house should all be heated during the colder 
months of the year. In tropical climates such a procedure is, 
of course, unnecessary. Where men are constantly at work in 
the buildings, the temperature should, if possible, be kept suffi- 
ciently high so that they are comfortable. The heating of the 
filter house, the pipe galleries and other parts of the plant in 
which men are not constantly at work, need only be kept at a 
moderate temperature to insure against freezing of water in the 
smaller pipe lines, valves and tanks. 
In preparing lime solutions the heating of the water used to 
slake the lime is practised in a number of plants. In such cases 
steam under a moderate pressure is required in order to raise 
the temperature of the slaking water to 150° or 160°F. Closed 
heaters may be employed for this purpose. 
For steam baths, autoclaves, drying ovens and for preparing 
distilled water, a supply of steam under moderate pressure is 302 
WATER PURIFICATION 
always of much convenience in the laboratories, although other 
methods of caring for these requirements may be provided. 
Since the demand for steam in a rapid sand filter plant, ex- 
clusive of that which might be needed for power purposes, is 
rather small even in the largest plants, the boiler plant can 
be of moderate size. It should be, however, of ample capacity, 
so that it may be able to meet the maximum demands which may 
be made upon it. The steam pressures required will probably 
be from 60 to 75 lb. per square inch. The quantity of steam 
needed will, of course, vary for each plant. 
Where a fire-tube boiler1 is used, it is customary to allow 10 
to 12 sq. ft. of heating surface per boiler horsepower, or, if 34.5 
lb. of water are evaporated per horsepower, then 1 sq. ft. of heat- 
ing surface will give an evaporation of about 3 lb. of water 
from and at 212°F. The heating surface in a fire-tube boiler is 
estimated as about two-thirds of the shell and tube sheets and 
the external surface of all of the tubes. 
The grate area needed must be proportioned to the kind of 
fuel used and to the draft. With ordinary chimney draft, the 
grate area should be sq. ft. per horsepower, or more. If forced 
draft is used this area may be diminished. Ordinarily the rate 
of combustion is 10 to 15 lb. of coal per hour per square foot of 
grate for anthracite coal, and 15 to 20 lb. for bituminous coal. 
With forced draft these amounts of coal can be much exceeded. 
As it is always desirable to reduce the smoke from a boiler 
plant to a minimum, the use of mechanical stokers may be ad- 
visable for this purpose, although strictly from the point of 
view of economy their installation in connection with boilers for 
a filter plant can hardly be considered necessary. The cost of 
upkeep for most mechanical stokers more than offsets the saving 
in fuel produced by more uniform firing. 
Since a steam-heating plant may be so arranged that a great 
deal of the condensed steam can be returned to the boiler feed 
pumps and again used in the boiler, little trouble from scale and 
incrustation derived from a hard water, which the filter plant 
may be obliged to purify, need be experienced. In those cases 
where little fresh water is used, the condensed steam may act 
corrosively on the iron with which it comes into contact, and 
should be guarded against by emptying the boiler at times of 
cleaning and refilling with fresh water. 
1 "American Civil Engineers' Pocket Book," p. 1300. POWER PLANT EQUIPMENT 
303 
Refrigerating Plant.-It is not uncommon that water-purifica- 
tion plants are so located that it becomes difficult to obtain ice, 
which is needed for cooling drinking water, and for ice chests and 
incubators in the laboratories. If sufficient power is available, 
it is possible to install a small ice-machine plant, which will not 
only supply ice, but can be also utilized to cool large refrigerators 
by means of pipes through which cold brine is kept circulating. 
Such a plant has been found a useful adjunct of the Cincinnati 
water-purification plant, although its refrigerating capacity is 
equal to only about 2 tons per day. The plant consists of a small 
vertical ammonia compressor driven by a 5-hp. variable-speed 
(180 to 250 r.p.m.) motor. The compressor, oil separator, con- 
denser, liquid ammonia receiver, brine cooling system and brine 
pump occupy but little space in a small room in the basement of 
the head house. The brine tank, in which are suspended the cans 
containing the water to be frozen, requires somewhat more floor 
space, and is located on the first floor of the head house. The 
cold brine is circulated by a small centrifugal pump through the 
tank containing the ice cans, and also through a system of piping 
in a large refrigerator located in the laboratories. These pipes, 
through which the cold brine is kept circulating, assist in keeping 
the refrigerator cold, and also in preventing the melting of the 
ice which is stored in the refrigerator after it has been taken from 
the cans in which it has been made. 
Two drinking fountains in the head house are also arranged so 
that the water passing to them is cooled by cold brine circulating 
through a pipe system in a tank of water in which the supply 
pipe to the fountains is coiled. All of the ice made is "raw 
water ice," that is it is filtered water frozen in 25-lb. cakes in the 
small ice cans hung in the brine tank previously described. 
Gas for Laboratories.-Laboratories can scarcely be called 
well-equipped unless gas is provided at various points along the 
work benches. Where natural or artificial gas is not available 
from a public supply, which is the condition usually encountered 
in filter-plant installations, gasoline gas supplied by a gas ma- 
chine for this purpose should be provided. A proper supply 
tank for the liquid gasoline, consisting of several shallow pans 
placed one above the other within a closed tank, is usually buried 
in the ground outside of the building. An air pump, which is 
operated either by a weight or by a water motor, forces air into 
the gasoline tank, where it becomes enriched with the vapors 304 
WA TER P U RIF IC A TION 
from the gasoline. Suitable arrangement of pipes from the air 
pump to the tank and from the tank into the building, furnishes 
the means for forcing the air into the gasoline tank and for con- 
veying the gas thus formed to the distribution piping in the labo- 
ratories. The tank should have a capacity of 100 to 500 gal. 
of gasoline, depending on the volume of gas likely to be needed. 
A highly volatile gasoline must be used in the tank if satisfactory 
results are desired. 
The author has found it convenient to provide a double pipe 
system in the laboratories, one conveying the gas, and the other 
air piped directly from the air pump. By joining these pipes 
together at each burner outlet, it becomes possible to regulate 
the supply of air for each burner, and thereby obtain a proper 
mixture of gas and air. When the tank is freshly filled with gaso- 
line the gas formed is likely to be too rich and to produce a 
smoky flame. By adding air at the burner this condition may be 
avoided and a better and more economical utilization of the gas 
may be obtained. 
References 
1. J. S. Gettrust: "Utilizing Water Power at Cincinnati Filter Plant." 
Eng. Record, May 4, 1912. 
2. R. L. Daugherty: "Centrifugal Pumps." 
3. Metcalf and Eddy: "American Sewerage Practice." Vol. 1. 
4. "The Rapid Mechanical Filtration Plant of the Montreal Water & 
Power Co." Eng. Record, Mar. 9, 1912. 
5. "The Rapid Sand Filtration Plant at Trenton, N. J." Eng. Record, 
May 9, 1914. 
6. John H. Gregory: "The Improved Water and Sewerage Works of 
Columbus, Ohio." Proc. Am. Soc. C. E., vol. 36, No. 1, January, 1910. 
7. "American Civil Engineers' Pocket Book." 
8. "Choice of Pumps for City Water-works." Engineering-Contracting, 
Feb. 24, 1915. CHAPTER XXIV 
THE COST OF CONSTRUCTING RAPID SAND FILTERS 
The cost of constructing a rapid sand filter plant is obviously 
largely dependent upon local conditions. Statements of cost can 
be only general in character, unless they are actual records of the 
cost of some plant that has been installed. It is not always easy 
to itemize the cost of a plant on account of the way in which 
contracts for the work may be grouped, and also because of in- 
adequate methods of cost keeping. The cost of coagulating 
basins was touched upon in a preceding chapter, as was also the 
cost of settling reservoirs. The cost of the filters and filter- 
plant equipment vary considerably and any attempt to general- 
ize is more or less unsatisfactory. 
Mr. G. A. Johnson in a paper entitled "The Purification of 
Public Water Supplies" (U. S. Geological Survey, Water Supply 
Paper No. 315) has listed the cost of a number of rapid sand 
filter plants, describing briefly in each case the salient construc- 
tion features of the plant, and the bearing which these structural 
features had upon the actual cost of the plant. Seven of these 
plants which he describes are listed below. 
Place 
Capacity of plant, 
million gallons 
daily 
Cost per million gallons of 
filtering capacity 
Little Falls, N. J 
32 
315,300 
New Milford, N. J 
24 
11^000 
Harrisburg, Pa 
16 
10^300 
Binghamton, N. Y 
8 
10,800 
Watertown, N. Y 
8 
11,250 
Lorain, Ohio 
6 
14,200 
Scranton, Pa 
6 
13'330 
Approximate Cost of Rapid Sand Filter Plants 
It will be noted that the approximate cost per million gallons 
of daily capacity is about $12,600. The two plants not listed 
are those at Cincinnati, Ohio, and Columbus, Ohio. At Cincin- 
nati two large settling reservoirs constitute a part of the purifica- 
305 306 
WATER PURIFICATION 
tion system, although they are utilized also for storage purposes. 
At Columbus the purification plant is more properly characterized 
as a water-softening plant, and in any consideration of cost 
figures this must be constantly borne in mind. 
In discussing a paper1 by Mr. John H. Gregory on the "Im- 
proved Water and Sewerage Works of Columbus, Ohio," the 
author compared the costs of the Cincinnati and Columbus filter 
plants. These remarks2 are quoted below for the purpose of 
showing the difficulty of drawing comparisons where the struc- 
tures constituting the plants are not used for exactly similar 
purposes. 
" It is obvious that no strict comparison of the costs of similar 
parts of these two plants is justifiable, since each was designed to 
meet local conditions. At Cincinnati the plant was required to 
clarify and purify a turbid water; at Columbus the same conditions 
were to be met, but, in addition, the water was to be softened. 
Apparently, the latter function was considered as governing the 
design of the works at Columbus." 
Table 11 in Mr. Gregory's original paper details the cost of 
the Columbus plant as follows: 
Columbus Water-purification Plant-Capacity 30 Million Gallons 
in 24 Hr. 
Total cost 
Cost per million gal- 
lons daily capacity 
Settling basins 
$168,770 
39,660 
3,470 
$5,630 
1,320 
Head house 
Air-wash equipment 
120 
Lime saturator house 
32,550 
1,080 
Mixing tanks 
44,230 
1,470 
430 
Storage house 
12,880 
15,280 
Office and laboratory 
510 
Filter gallery 
102,710 
3,420 
Filtered-water reservoirs 
98,300 
3,280 
Wash-water tank, pipe and shelter. . 
13,150 
440 
Supplies for preliminary operation. . . 
460 
20 
Expenses unclassified 
1,020 
30 
Total 
$532,480 
$17,750 
1 Proc. Am. Soc. C. E., January, 1910. 
2 Proc. Am. Soc. C. E., March, 1910. CONSTRUCTING RAPID SAND FILTERS 
307 
In the author's discussion of Mr. Gregory's figures the detailed 
costs of the Cincinnati filtration plant were given: 
Total and Unit Costs of Main Features of Work Done in Construc- 
tion of Water-purification Works at Cincinnati, Ohio 
Total cost 
Cost per million gal- 
lons of daily capacity 
Preparation of grounds 
$33,359.67 
$297.85 
Pipe lines between settling reservoirs 
and head house 
55,354.77 
494.24 
Head and chemical house 
141,989.85 
1,267.77 
Coagulation basins, gate houses and 
pipe lines 
304,913.05 
2,722.44 
Filters, filter house, piping, sand and 
gravel 
Piping, valves and gate house be- 
592,112.30 
5,286.71 
tween filters and clear-water res- 
ervoir 
29,701.91 
265.20 
Clear-water reservoir 
121,362.29 
1,083.59 
Total 
$1,278,793.94 
$11,417.80 
"At Columbus, the unit costs per million gallons of capacity in 24 
hr. appear to be considerably greater for the settling basins and mixing 
tanks combined, than for the coagulation basins at Cincinnati. The 
figures for the Columbus tanks and basins are $7,100 per million gallons 
of capacity, as compared with $2,722 at Cincinnati. In a general 
way, these parts of the two plants correspond; but it should not be for- 
gotten that at Columbus more elaborate baffling of tanks and basins, 
more divisions of the flow of the raw and treated waters, and more 
places for the primary and secondary applications of chemical solutions 
were needed and provided for, than were required at Cincinnati. The 
greater combined unit costs of the head house, lime-saturator house, 
storage house, wash-water tank, offices and laboratories at Columbus, 
than for the corresponding head house, chemical house, wash-water 
tank, offices and laboratories at Cincinnati, are similarly explained by 
the necessity for designing a plant for softening, as well as for clarifying 
and purifying the water. The combined unit costs for the items noted 
above for the Columbus plant amount to $3,784, as compared with 
$1,268 for the Cincinnati plant. 
The filters and piping in the Cincinnati plant cost more per million 
gallons of capacity than did those at Columbus. The figures for 
Columbus, which include the air-washing equipment, are $3,540, as 
compared with $5,287 for Cincinnati. However, the filtered-water 
reservoir at Columbus cost more than that at Cincinnati. The figures 308 
WATER PURIFICATION 
for the Columbus plant are S3,280 per million gallons, and for the 
Cincinnati plant, SI,084. At the latter plant, the clear-water reservoir 
is a separate uncovered reservoir, while at Columbus it is directly under 
the filter tanks, which latter form a protecting roof. Virtually, no 
great difference in costs exists, if the cost of the filters, piping, and clear- 
water reservoir of each plant be combined and then compared. 
The cost per million gallons of capacity for the whole purification 
plant at Columbus is stated to be SI7,750, which amount does not 
include engineering; the corresponding figures for the Cincinnati plant 
as shown above, is SI 1,418, and this also excludes the cost of engineer- 
ing. The difference of more than SO,000 per million gallons of capacity 
is doubtless due to the additional requirement demanded by the local 
conditions at Columbus, that is, for the softening of a very hard water, 
and one which is at times subject to rapid fluctuations in its physical 
characteristics." 
In a more recent paper, entitled "Present Day Filtration Prac- 
tice,''read at the annual convention of the American Water-works 
Association in 1914, Mr. G. A. Johnson discusses the relative 
costs of slow and rapid sand filters. He concludes that the aver- 
age cost per million gallons of daily capacity is $32,600 for slow 
sand filters and $12,100 for rapid sand filters. To these conclu- 
sions Mr. John H. Gregory took exceptions in his discussion of 
the paper.1 He cites the cost of several rapid sand filter plants 
to show that Mr. Johnson's average cost of $12,100 per million 
gallons of daily capacity is much too low. The Columbus, Ohio, 
plant considered as a rapid sand filter plant cost nearer $15,000 
than $13,000 per million gallons of daily capacity as given by 
Mr. Johnson. Mr. Gregory concludes that the Toledo, Ohio, 
filter plant,2 built for a capacity of 60,000,000 gal. per day, cost 
about $14,500 per million gallons. The 20,000,000-gal. plant at 
Grand Rapids, Mich., cost approximately $16,300 per million 
gallons of daily capacity; and the New Orleans, La., 40,000,000- 
gal. plant cost about $30,200 per million gallons of daily capacity. 
He estimated that the Jerome Park filter plant, which was to be 
constructed for purifying the Croton water supply of New York 
City, and which was to have had a capacity of 320,000,000 gal. 
per day, would have cost approximately $20,000 per million 
gallons of daily capacity. 
1 Reader is referred to original paper for discussion of slow sand filter 
costs. 
2 Eng. Record, Nov. 26, 1910. CONSTRUCTING RAPID SAND FILTERS 
309 
Referring to the omission by Mr. Johnson of the cost of the 
settling reservoirs (81,521,000) from the total cost of the Cincin- 
nati filter plant, and assuming that it might be permissible to 
charge at least half of the cost of these reservoirs to the purifica- 
tion plant, Mr. Gregory concludes, that the cost of the latter 
would be about $18,200 per million gallons of daily capacity. 
He sums up his conclusions by stating that the weighted average 
of the unit costs of the Columbus, Grand Rapids, New Orleans, 
Toledo and Cincinnati plants is $18,600 per million gallons of 
daily capacity, which is over 50 per cent, higher than the weighted 
average cost of $12,100 stated by Mr. Johnson. 
It should be noted that the first three of the five plants named 
above were designed not only as filter plants but as water-soft- 
ening plants as well, and that rather exceptional local conditions 
involving special engineering structures probably account for 
the greater costs at each of those plants where the unit costs are 
high. 
The original purification works of the City of St. Louis, Mo., 
were composed of a number of basins, which were utilized for 
plain sedimentation purposes. Later, these basins were remod- 
eled so that they might act as coagulation basins following the 
introduction of chemical coagulants. Additions to this plant have 
been made from time to time, the most recent being a filter plant 
with its appurtenances, and having a daily capacity of 160,- 
000,000 gal. The cost of the filters, buildings, equipment, and 
such modifications of the existing plant as were needed, was 
$1,357,289.15. 
According to the annual report of the St. Louis Water Depart- 
ment ending April 1, 1915, "the total cost of all parts of the puri- 
fication plant as now operated is estimated at $3,549,000 or 
$22,200 per million gallons rated capacity. This relatively high 
unit cost is explained by the fact that the capacity of the basins 
already in place is much greater than would have been necessary 
for the filter plant only, and the basin cost is, therefore, dispro- 
portionately high." 
A somewhat detailed table of costs for the 30,000,000-gal. 
filter plant at Trenton, N. J., is given below. The figures are 
based on the prices bid for the work (see Engineering Record, 
May 9, 1914, p. 543). 310 
WATER PURIFICATION 
Estimated Cost of Construction of Trenton Filter Plant 
Earth excavation, 19,000 cu. yd $16,530 
Rock excavation, 75 cu. yd 113 
Cement, 14,000 bbl 18,070 
Steel reinforcement, 650 tons 45,500 
Reinforced concrete, 8,000 cu. yd 53,600 
Concrete not reinforced, 1,500 cu. yd 10,050 
Cast-iron bell and spigot pipe, 450 tons 15,840 
Cast-iron bell and spigot specials, 42 tons 3,024 
Valve chamber 231 
Two 30-in. gate valves 850 
Screens and hoists 2,292 
All buildings 48,600 
Heating system 3,200 
Low-lift pumping equipment 45,225 
Filter equipment and piping 74,500 
Strainer system 33,000 
Filter gravel, 440 cu. yd 1,778 
Filter sand, 1,100 cu. yd 4,444 
Wash-water and air tank 9,503 
Engineering and miscellaneous 20,000 
Total estimated cost exclusive of land $406,350 
Cost per million gallons of daily capacity $13,545 
The estimated cost of the Baltimore filtration plant, as made 
by Mr. George W. Fuller in his preliminary report as consulting 
engineer, is as follows: 
Estimated Cost of Construction of Baltimore Filter Plant, Daily 
Capacity 128 Million Gallons 
Coagulating and settling basins and appurtenances, 
including land damages $370,000 
Filters and appurtenances, including land damages 900,000 
New covered filtered-water reservoirs 150,000 
$1,420,000 
Add 15 per cent, for engineering and contingencies. 213,000 
Total $1,633,000 
Cost per million gallons of daily capacity $12,760 
From the statements made and the figures given in the pre- 
ceding pages, it is apparent that no exact estimate of the average 
cost of rapid sand filter plants can be made, and that each case 
must be considered by itself. The special modifications of the CONSTRUCTING RAPID SAND FILTERS 
311 
design of the plant required to properly purify the kind of water 
that must be treated, the topography of the site on which the 
plant is to be built, the proportion of the cost of storage reservoirs 
and of pipe lines and tunnels conveying water to and from the 
purification plant proper and which should be charged to the 
cost of the latter, and the cost of the land, are all elements enter- 
ing into the computation, and concerning which engineers may 
not in some cases agree. The cost figures should, therefore, be 
considered as approximte only. 
References 
1. G. A. Johnson: "The Purification of Public Water Supplies." U. S. 
Geological Survey, Water Supply Paper No. 315, 1913. 
2. G. A. Johnson: "Present-day Filtration Practice." Proc. Am. Water- 
works Assn., 1914. 
3. Nicholas S. Hill, Jr.: "Modern Filter Practice." Proc. Am. Water- 
works Assn., 1913. 
4. Proc. Am. Soc. C. E., vol. 36, Nos. 1 and 3. 
5. J. W. Ellms: "Pure Water and Its Relation to Health." Jour. 
Assoc. Eng. Soc., vol. 48, No. 1. 
6. Engineering Record, vol. 69: "St. Louis Filter Plant." Mar. 21, 1914 
"Trenton, N. J., Filter Plant." May 9, 1914. "Baltimore, Md., 
Filter Plant." May 9, 1914. CHAPTER XXV 
RATES OF FILTRATION, LOSS OF HEAD AND 
WASHING OF RAPID SAND FILTERS 
There are three important features of the operation of rapid 
sand filters that require special consideration. These are the 
rate at which the filter is operated, the loss of head produced by 
the flow of water through the filter, and the method used in wash- 
ing the filter bed. 
RATES OF FILTRATION 
The chief distinction between slow sand and mechanical or 
rapid sand filters is the rate at which the water passes through the 
sand bed. The average rate of flow at which slow sand filters 
are now operated in the United States is considerably greater 
than formerly. This increase in the rate has been made possible 
in many instances by the employment of methods of preliminary 
treatment, consisting of sedimentation produced with the aid of 
coagulants, and by "roughing filters." In the United States a 
rate of 6,000,000 gal. per acre daily is not usually exceeded, and 
the average is about 4,000,000 gal. per acre per day. 
In the operation of rapid sand filters, however, the rate does 
not differ materially from that first established, namely, 125,000- 
000 gal. per acre daily. Higher rates than this are sometimes 
employed, but unless conditions are unusually favorable, are 
not often used. Mr. Nicholas S. Hill, Jr., in a paper on " Modern 
Filter Practice" read before a convention of the American Water- 
works Association in 1913, gives a table of statistics relating to 
quite a number of filter plants in the United States, and which, 
among other data, includes the rates of filtration employed. 
Mr. Hill, in commenting on this data, states in substance that 
with a few exceptions, such as at Elmira, N. Y., Hackensack, 
N. J., and Harrisburg, Pa., where the rates are 150,000,000, 180,- 
000,000, and 166,000,000 gal. per acre daily, respectively, it 
will be noted that the rate of filtration is almost always 125,000,- 
000 gal. per acre per day. The size of the filter units, and conse- 
quently their capacity, has been increasing during the past 15 
312 RATES OF FILTRATION 
313 
Location 
Date 
Plant capac- 
ity (mil. gal.) 
g 
E g 
6'3 
Z 3 
Unit size 
Unit capacity 
Rate per acre 
Filter 
material 
(inches) 
Sand 
Per cent, 
wash water 
Rate applied in 
rise per min. 
Air wash 
Dimensions 
Area (sq. 
ft.) 
Sand 
Gravel 
Eff. size 
(mm.) 
Un. coef. 
Elmira, N. Y 
1897 
7.0 
21 
13XDia. 
132 
0.33 
150 
30 
6.0 
0.56 
1.43 
2.8 
14.0 
no 
Little Falls, N. J. . 
1902 
32.0 
32 
24X15 
360 
1.00 
125 
30 
12.0 
0.44 
1.47 
3.5 
12.0 
yes 
Hackensack, N. J.. 
1904 
48.0 
16 
46X25 
1,000 
3.00 
180 
30 
12.0 
0.50 
1.30 
2.7 
8.5 
yes 
Ithaca, N. Y 
1904 
3.0 
6 
16X11 
176 
0.50 
131 
30 
? 
0.39 
1.50 
3.8 
? 
? 
Charleston, S. C.. . 
1904 
6.0 
12 
15XDia. 
176 
0.50 
100 
50 
12.0 
0.70 
1.30 
4.0 
13.0 
no 
Harrisburg, Pa 
1905 
12.0 
10 
27X16 
432 
1.00 
166 
30 
7.0 
0.38 
1.30 
2.3 
9.0 
yes 
Marietta, Ohio. . . . 
1906 
2.4 
8 
20X10 
200 
0.50 
81 
36 
? 
? 
? 
6.5 
11.0 
no 
Cincinnati, Ohio,. . 
1908 
112.0 
28 
28X50 
1,400 
4.00 
125 
30 
7.5 
0.32 
1.20 
3.0 
24.0 
no 
Columbus, Ohio. . . 
1908 
30.0 
10 
46X26 
1,089 
3.00 
125 
30 
10.0 
0.41 
1.36 
2.5 
12.0 
no 
Bristol, R. I 
1908 
3.0 
6 
15X12 
180 
0.50 
125 
30 
6.0 
0.45 
1.74 
7.0 
14.0 
yes 
Winfield, Kan 
1908 
2.0 
4 
18X10 
180 
0.50 
121 
48 
? 
? 
? 
4.0 
? 
no 
New Orleans, La... 
1909 
60.0 
10 
53X27 
1,431 
4.00 
101 
36 
8.0 
0.35 
1.65 
1.6 
24.0 
no 
Louisville, Ky 
1909 
36.0 
6 
72X30 
2,160 
6.00 
125 
24 
10.0 
0.36 
1.50 
2.0 
22.0 
no 
Sandusky, Ohio 
1909 
6.0 
6 
20X18 
360 
1.00 
100 
36 
10.0 
0.40 
1.30 
3.5 
21.0 
yes 
Toledo, Ohio 
1910 
34.0 
34 
22X16 
352 
1.00 
125 
30 
9.0 
0.40 
1.45 
2.1 
14.0 
yes 
Wilkinsburg, Pa 
1910 
12.5 
10 
22X20 
450 
1.25 
120 
36 
8.0 
0.40 
1.60 
2.3 
13.0 
yes 
Newport, R. I 
1910 
7.5 
6 
24X14.5 
348 
1.25 
156 
27 
18.0 
0.40 
1.70 
3.5 
16.5 
yes 
Kansas City, Kan.. 
1911 
12.0 
10 
32X20 
480 
1.25 
125 
30 
? 
? 
? 
? 
30.0 
no 
Cohoes, N. Y 
1911 
10.0 
10 
23X15 
24X18 
347 
1.00 
100 
24 
12.0 
9.0 
0.40 
0.59 
1 65 
4.0 
? 
? 
Bangor, Me 
1911 
8.0 
8 
432 
1.00 
100 
27 
1.25 
4.5 
12.0 
yes 
Fort Worth, Tex.... 
1911 
5.0 
4 
25X20 
540 
1.25 
125 
34 
11.0 
0.39 
1.42 
? 
14.0 
yes 
Ottumwa, Iowa 
1911 
4.0 
4 
24X14.6 
348 
1.00 
125 
27 
9.0 
0.50 
1.40 
3.5 
12.0 
yes 
Minneapolis, Minn. 
Grand Rapids, 
1912 
1912 
40.0 
12 
51X23 
1,173 
3.25 
125 
30 
8.0 
0.39 
1.65 
0.5 
24.0 
no 
Mich 
-13 
20.0 
10 
36X23 
738 
2.00 
118 
30 
8.0 
0.39 
1.60 
4.6 
24.0 
no 
Niagara Falls, N.Y. 
1912 
16.0 
16 
25X14.5 
362 
1.00 
125 
30 
9.0 
0.35 
1.60 
2.5 
10.0 
yes 
Evansville, Ind.. . . 
1912 
12.0 
12 
23X16 
368 
1.00 
125 
30 
9.0 
0.26 
1.83 
3.3 
14.0 
yes 
Fargo, N. D 
1912 
6.0 
6 
21X17 
357 
1.00 
125 
30 
10.0 
0.35 
1.40 
1.5 
11.0 
yes 
Rock Island, Ill... . 
1912 
6.0 
6 
21.5X17 
365 
1.00 
125 
30 
10.0 
0.32 
1.65 
1.2 
14.0 
yes 
Grand Forks, N. D. 
1912 
4.0 
8 
18X11 
198 
0.50 
110 
30 
6.0 
0.39 
2.10 
4.0 
? 
? 
Albany, Ore 
1912 
3.0 
6 
22.5X10.5 
184 
0.50 
125 
37 
9.0 
0.36 
1.30 
? 
12.0 
yes 
Minot, N. D 
1912 
1.5 
3 
14.6X12 
181 
0.50 
125 
34 
30.0 
? 
? 
3.0 
13.0 
yes 
Trenton, N. J 
19- 
30.0 
16 
30X24 
652 
1.87 
125 
30 
10.0 
0.39 
1.65 
? 
18.0 
yes 
Evanston, Ill 
19- 
12.0 
6 
36X23 
846 
2.00 
125 
30 
8.0 
0.65 
1.65 
? 
? 
yes 
New York, N. Y.... 
19- 
320.0 
80 
50X34 
1,700 
4.00 
125 
30 
8.0 
0.65 
1.65 
? 
? 
Mechanical Filters 
years. In 1900 the average daily capacity of a filter unit was 
about 500,000 gal., having a sand area of 175 sq. ft. At the 
present time units of 4,000,000 gal. (1,400 sq. ft. of sand) daily 
capacity are used in the largest plants, while the average capac- 
ity is probably between 1,000,000 (350 sq. ft.) and 2,000,000 
(700 sq. ft.) gal. per day. 
Uniformity of Rate of Filtration.-It is of the utmost impor- 
tance that the rate of filtration shall be uniform over the whole area 
of the filter bed. Any sudden increase in the rate of flow is likely 
to cause the bed to break through at some point, and thereby 
seriously impair the quality of the effluent. A sudden decrease 
in the rate of flow is of much less consequence, so far as it may 
affect the quality of the effluent, unless air, which may have 
come out of solution in the water or have entered the filter 
through small cracks, should be liberated from the bed in consid- 314 
WATER PURIFICATION 
erable quantities; in which case the disturbance of the surface 
of the filter bed is likely to result in an imperfectly purified 
effluent. Slow and steady changes in the rate of flow within 
certain limits are permissible, and are not infrequently made use 
of where insufficient storage capacity for the filtered water makes 
it necessary that the filters should follow closely the rate of con- 
sumption. How great a change in the rate of flow can be per- 
mitted, without affecting the quality of the effluent, depends 
Turbidity of Applied Water-P.P.M. 
Fig. 112.-Maximum, effective rate of filtration for different turbidities of 
applied water. 
Maximum Rate-Gals, per Sq. Ft. per Mln. 
upon the condition of the sand bed and the character of the 
water being filtered. A bed of fine sand upon which there had 
been deposited a good protective colloid coating, and to which 
was being applied a well-coagulated water that was not too 
heavily charged with sediment, could stand a much greater 
shock produced by a change in the rate, than if the reverse of 
these conditions existed. Local conditions must be carefully 
studied before attempting to modify rates of filtration to any 
marked extent, and especially is this necessary where fluctua- 
tions in the rate are to be used continuously in order to follow 
changes in the rate of consumption of the filtered water. RATES OF FILTRATION 
315 
An interesting illustration of the maximum effective rate of 
filtration with water of varying turbidity has been furnished the 
author by Mr. H. W. Streeter, who has plotted the data (Fig. 
112) obtained in operating a rapid sand filter plant at Clarks- 
burg, W. Va. It will be noted that rates of filtration varying 
from 1 to 3.4 gal. per square foot per minute, corresponding to 
approximately 62,700,000 and 213,000,000 gal. per acre daily, 
respectively, have been plotted against turbidities in the applied 
water ranging from about 8 parts to 62 parts per million. 
Mr. Streeter's comments are as follows: 
"The data for this curve were obtained from about 30 tests of indi- 
vidual filters, and extended over a period of about 1 year. In making 
a test the filter was started at a low rate, which was slightly increased 
at intervals of about 20 min., or approximately the time required to 
displace three times the volume of water passing through the sand. 
These successive increases in the rate were continued until a rate was 
reached which produced a turbid effluent. The next lower rate than 
this (usually 0.2 gal. per square foot per minute) previously tried and 
found effective under the above conditions was taken as the maximum 
effective rate for the given applied water turbidity existing at the time 
of making the test. 
"Tests were carried out at various losses of head, and the aim was 
to distribute them evenly, so as to have observations weighted for 
each loss of head, that is, to have about the same number of tests of a 
certain applied water turbidity for each loss of head. 
"The curve is well defined for turbidities of applied water below 50 
parts per million. Above this there were not so many observations 
possible, because the applied water turbidity was not allowed to exceed 
this limit more than a few times. Above 50 parts per million of tur- 
bidity there seems to be a reversal of the curve. This means, of course, 
that the filters could not handle the higher turbidities with the same 
degree of effectiveness as they did the lower turbidities, even with very 
low rates of filtration. With very low applied water turbidities, how- 
ever, disproportionately high rates of filtration were indicated as being 
possible. 
"The curve relates only to the removal of turbidity, and does not 
attempt to define what the permissible rates of filtration for satis- 
factory bacterial removals might be under the same conditions. How- 
ever, satisfactory results from this standpoint were always obtained, 
when the removal of turbidity was complete, by disinfection of the 
filtered water." 
Rate Controllers.-The maintenance of a uniform, or of a 
slowly changing, rate of filtration is made possible by automatic 316 
WATER PURIFICATION 
rate controllers. The mechanism of a number of these devices 
has been previously described, and only a few points of general 
interest concerning their operation and care need be mentioned. 
When first installed, rate controllers should be carefully cali- 
brated for the conditions under which they are to operate. This 
is easily accomplished by noting the distance the water falls in a 
filter tank in a given time when the supply to the tank is cut off 
and when the filtered water is discharging through the controller 
in the usual manner. By calculating accurately the volume of 
water filtered from the area of the tank and the observed fall of 
the water during the measured period of time, it becomes possible 
to estimate the rate of flow. Such measurements are facilitated 
by a gage placed in the water above the sand in the filter tank, 
and on which are a series of hooks placed one above the other at 
known distances. By using a stop watch the time required for 
the water level to fall from one hook to the next lower can be 
accurately gaged. 
From time to time the adjustment of rate controllers intended 
to operate at constant rates should be undertaken, in order to 
make sure that they are in proper working condition. Corrosion 
of metal surfaces in moving parts or incrustations or tubercula- 
tion of iron piping, or the wearing out of diaphragms or other 
short-lived parts, are the obvious reasons for periodical adjust- 
ments. Where variable rates of flow are desired, fully as much, if 
not more, care should be given to these machines in order to see 
that they are controlling properly. Checking the operation of 
any rate controller of the Venturi type is easily accomplished, 
if rate-of-flow gages form a part of the equipment. 
In those plants which are not provided with rate controllers, 
or if for reasons of cost they can not be installed in a new plant, 
the rate of filtration may be in a measure controlled by other 
methods, and of which advantage should always be taken. A 
maximum rate may be fixed by some form of orifice plate inserted 
in the effluent pipe or by skillful handling of the effluent valve 
itself; by throttling it when the filter is clean and gradually open- 
ing it as the filter becomes clogged, it is possible to maintain a 
fairly uniform discharge throughout the period of service of the 
filter. Inattention to the rates of filtration in plants not pro- 
vided with automatic controllers is undoubtedly the cause for 
the poor quality of the effluent which they so frequently produce. RATES OF FILTRATION 
317 
LOSS OF HEAD 
The pressure required for operating a filter varies with the 
type. Pressure filters are operated under considerably more head 
than those of the gravity type.1 As the period of service of a 
filter lengthens, or in other words as the sand bed becomes more 
and more clogged with the suspended matter strained out of the 
water, greater and greater pressures are needed if the rate of flow 
is to be kept uniform. 
A clean filter of the gravity type, when first put in operation, 
offers but little frictional resistance to the flow of the water, and 
rates of filtration far in excess of those which would be hygienic- 
ally safe could be maintained for short periods. To overcome 
this difficulty resistance to the flow of the water must be intro- 
duced. This may be accomplished by throttling the effluent 
valve of the filter, or better by the use of a rate controller, whose 
function is to automatically insert this frictional resistance to 
the flow of the water when the filter is clean, and to withdraw it 
as the friction head produced by the clogging of the sand bed 
increases. 
Total Head.-The difference between the level of the water on 
the surface of the filter and the level at which it discharges 
through a trapped outlet, or if the outlet pipe is submerged, at 
the level of the water in the tank or reservoir in which submerg- 
ence is effected, measures the total head under which the filter is 
operating. For the gravity type of rapid sand filters the total 
head available is usually from 12 to 15 ft. Variations in the 
available head are due to changes in the level of the water on the 
top of the filters and to fluctuations in the level of the water in 
the receiving basin or reservoir. 
Positive and Negative Head.-The total head of water avail- 
able for operating rapid sand filters of the gravity type is in- 
creased by prolonging the discharge pipe with its trapped outlet 
to a point below the level of the floor of the filter. This is simply 
a device for increasing the distance between the level of the water 
in the filter and that at which it discharges. This device has 
been the subject of patent claims, the validity of which have 
1 Mr. George W. Fuller, in the report on the investigations at Louisville, 
Ky., states that the Western pressure filter had a minimum available head 
of 115 ft. The filter was washed when the loss of head reached 50 ft. The 
sand in this pressure filter was 49.5 in. in depth, and had an effective size 
of 0.44 mm. h 318 
WATER PURIFICATION 
been the cause for much litigation. The water flowing through 
this so-called "downdraft tube" is claimed to exert "suction" 
on the filter bed by the production of a partial vacuum within 
the latter, and to produce a number of secondary effects of a 
beneficial character. When these conditions exist it has become 
customary to regard the total head acting in such filters as di- 
vided into two parts, namely, "positive head" and "negative 
head. " The significance of these claims will be somewhat better 
understood by a more complete discussion of the "loss of head" 
in a filter. 
Loss of Head.-The head lost in the passage of water through 
a filter is the sum of the heads required to produce the rate of 
Loss of Heat -Feet 
N.B. Filter contained 14 inches 
gravel and 30 inches sand. 
Rate of filtration approxi 
mately 112 million gallons 
per acre per 24 hours. 
Fig. 113.-Typical loss of head curve for a rapid sand filter. 
Hours of Service 
How or velocity head, and the head required to overcome all 
frictional resistance throughout the whole system, or friction 
head. The velocity head is small as compared with the friction 
head, which is by far the largest factor, and which is produced 
principally in the sand layer. 
The loss of head increases directly with the rate of flow, with 
the depth of the sand bed, and with the square of the effective 
size of the sand grains. The loss of head is greater at lower tem- 
peratures than at higher ones, being due to the increased vis- 
cosity of the water at the lower temperatures. The decrease in 
porosity of the filter bed is due to the cumulative effect of sus- 
pended matter strained out of the water, and under some circum- RATES OF FILTRATION 
319 
stances to the entrainment of air in the interstices of the sand 
bed. The effect of the sediment in increasing the loss of head 
is the factor of the greatest importance, and, naturally, the char- 
acter and quantity of sediment determine in a large measure the 
rate at which the head is lost. Trapping of liberated bubbles of 
air in any quantity in the bed causes a rapid loss of head. 
It is obvious that the zone of greatest frictional resistance in 
a filter bed during filtration is most likely to occur at or near the 
surface of the sand bed. As the period of operation of the filter 
lengthens, the resistance becomes greater. A typical loss-of- 
head curve taken from one of the Cincinnati, Ohio, filters is 
shown in the accompanying cut. 
This curve merely shows the rate at which the available head 
is used up. When the total head is used up the filter must be 
taken out of service and cleaned in order to restore it to its 
original condition. The maximum rate of flow of a filter operated 
at variable rates must gradually diminish as the filter becomes 
clogged, since the decrease in the available head necessarily limits 
the output of the filter even though the controller orifice is wide 
open. 
Partial Vacuum in Filter Bed.-By inserting glass tubes in the 
side of a rapid sand filter tank of the gravity type, so that the 
ends of the tubes project into the filter bed at different levels, 
it will be observed, that as filtration proceeds the water will 
lower in the tubes and finally disappear at the point where the 
tube enters the tank. If the tube is trapped, it will be noted 
that the water may drop below the level at which the tube enters 
the side of the tank, indicating a pressure less than that of the 
atmosphere at this point. The distance which the water drops 
in the trapped tube below the level of the point at which the 
tube enters the filter bed is a measure of the partial vacuum at 
this plane in the filter bed. 
Mr. George W. Fuller has clearly described the conditions in 
a rapid sand filter when a partial vacuum exists, in his testimony 
as an expert witness in the case of the New York Continental 
Jewell Filtration Co. vs. the City of Harrisburg, Pa. (1910). 
"A 'partial vacuum' occurs in a filter under those conditions where 
the pressure becomes less than atmospheric pressure. This is brought 
about in filters when the 'loss of head' through a section of a filter 
becomes greater under the given conditions of operation than is the 
depth of the water measured from the top of the water on the sand 320 
WATER PURIFICATION 
bed down to the plane or bottom of the section of the filter bed in ques- 
tion. A partial vacuum is most easily noted in the case of filters which 
have a series of glass tubes connected at various points on the side of the 
filter. If these observation tubes should not be trapped, but have a 
direct horizontal connection into the side of the filter, a partial vacuum 
would be found to exist when the water disappears in the glass tube 
connected to the filter at the point under consideration. So long as the 
water stands in the observation tubes above the level where connection 
is made with the filter, no partial vacuum exists. If the observation 
tubes connected with the filter at various points on the sides are trapped 
down below the elevation of the connection through the filter wall, 
then there is an opportunity for measuring the degree of partial vacuum 
which exists. Thus, if the water should stand in an observation tube 
1 ft. below the level where this tube is connected with the sand bed of 
the filter, then the degree of partial vacuum is stated to be equivalent 
to a column of water 1 ft. in height. » 
"For a partial vacuum to be secured in the practical operation of 
filters, it is necessary for atmospheric pressure to be excluded from 
within and below the filter, and this is done by preventing air entering 
the sides of the filter and by having the filter outlet either submerged 
in a body of water, or else trapped. The purpose of this, of course, is 
to prevent the entrance of air from below. If this were not done air 
would enter from below and displace in part the water within a filter 
below the section where a partial vacuum is developed. Looking at 
it in the light of the relation to the Toss of head' in a filter, it may be 
said that a partial vacuum occurs in a filter with a trapped outlet, when 
the total loss of head under the given conditions exceeds at any plane 
the depth of water measured from that plane up to the top of the water 
in the filter above the sand bed. Where the outlet is trapped and no 
air enters the sides of the filter, losses of head in excess of that just stated 
are utilized. This is accomplished by virtue of the partial vacuum 
from the utilization in part of the atmospheric pressure. The extent 
to which use is made of atmospheric pressure exerting itself upon the 
surface of the water above the sand bed in the filter for purposes of 
pushing the water through the filter is proportional to the degree of 
partial vacuum secured." 
Mr. Fuller points out in his testimony following that quoted 
above, that a 
"downdraft tube extending below the filter floor is not necessary for 
the production of a partial vacuum, and that the 'lower' portion of the 
filter itself serves as a downdraft tube as regards that portion lying 
beneath the plane where a partial vacuum first develops. In other 
words, if a partial vacuum should develop 1 in. beneath the surface of RATES OF FILTRATION 
321 
a sand layer in a filter bed, the lower portion of the filter tank enclosing 
the portion of the sand bed and other constructions below the plane 
cited would constitute a downdraft tube." 
In defining what is commonly understood as "negative head," 
Mr. Fuller states that it has become customary to divide the 
total head into two parts, namely "positive head" and "negative 
head." 
"Scientifically speaking, this is done accurately when the positive 
head is called that which exists above the uppermost portion of the filter 
wherein a partial vacuum is produced. A negative head is either ob- 
tained by difference, or it may be defined as a head causing the water 
to move from the top portion of the section containing a partial vacuum 
to the plane of the delivery pipe which controls the amount of head 
which can be made available as a total." 
The term "negative" applied to any portion of the forces 
tending to produce flow of water through a filter is confusing 
and unnecessary. The author is indebted to Mr. C. N. Miller 
for a clear and concise discussion of the hydraulics of the flow 
of water through a filter. In this discussion the particular case 
of the so-called "negative head" is treated from a mathematical 
standpoint. This matter will be found in an appendix (A). 
Washing Rapid Sand Filters.-The cleansing of the filter bed 
at the end of each period of service is an extremely important 
part of filter-plant operation. Unless this cleansing process is 
effective, a gradual deterioration in the condition of the bed 
must ensue, with the consequent production of a more or less 
poorly purified effluent. 
Probably no part of rapid sand filter-plant operation has re- 
ceived as much study as the process of washing. The mechanical 
construction of the strainer system, the method of distributing 
the wash water, and the manner of agitating the sand, all present 
problems of more or less difficulty. The construction of strainer 
and agitating systems has been described in a preceding chapter, 
and only the various methods of washing will be discussed here. 
Essentials for Efficient Washing.-Any method of washing, 
(Fig. 114) to be effective, must remove the sediment which has 
accumulated during the preceding period of filtration. This 
sediment may have penetrated into the bed for some distance, 
or it may be all practically within an inch or two of the surface, 
depending upon the amount and character of the sediment, the 
amount of the coagulant employed, the size of the sand particles 322 
WAT FAI PURIFICATION 
and the rate of filtration. The uniformity of the discharge of 
wash water is, of course, dependent upon the design of the strainer 
system. The loss of head through the openings through which 
the wash water enters the filter bed must be as uniform as pos- 
sible for all sections of the bed. In the modern type of filter 
the gravel layer, or its equivalent, acts as a second distributor 
of the upward-flowing wash water, and prevents jet action 
immediately above the individual strainer openings or wash- 
water inlets. 
Turbidity 
Fig. 114.-Relation between turbidity of wash water and length of time of 
washing. 
Time of Washing in Minutes 
Development of the Methods for Washing Rapid Sand 
Filters.-Since the sand in any properly designed filter bed should 
retain all of the applied suspended matter, it alone is really in 
need of cleansing. Consequently methods for agitating the sand 
bed prior to or during the upward flow of wash water have 
always formed a part of the washing operation. Three methods 
of agitation have been developed, one of which is now practically 
obsolete, while the other two are in active use today. 
The early type of rapid sand filter consisted merely of a RATES OF FILTRATION 
323 
cylindrical tank in which the sand bed rested directly upon the 
strainer system. In these tanks rotating rakes stirred the sand 
to nearly its full depth as the wash water flowed up through the 
bed. The bed increased in volume somewhat under this opera- 
tion, and the agitation produced by the rakes assisted the wash 
water in removing the particles attached to the sand grains. 
In order to prevent the clogging of the strainer openings by 
sand particles, it became necessary to place between the sand bed 
and the strainer system a layer of coarser material which was so 
graded in the size of its particles, that at the bottom they would 
be too large to pass through the strainer openings, and at the 
top too small to permit the sand to pass downward. Graded 
layers of gravel came into use for this purpose, and furnished a 
satisfactory protection to the strainer system, and a support for 
the sand bed. 
With a gravel bed which could not be disturbed and which sup- 
ported a sand bed which it was desirable to agitate in some man- 
ner in order to make the wash water more effective, it soon be- 
came evident that rakes mechanically rotated could not be safely 
employed. Moreover, the increasing size of filter units, and es- 
pecially the introduction of rectangular tanks in place of the cir- 
cular tanks, so increased the mechanical difficulties involved, that 
designers sought other means for agitating the bed. Except in 
old filter tanks, which may be still in use, or in quite small plants, 
where circular tanks are still sometimes installed, this method of 
agitation is no longer commonly employed. 
Velocity of Wash Water in Feet per Minute 
Gravity filters 
Agitation with rakes 
Average 
Maximum 
Minimum 
Jewell filter 
Warren filter 
Western filter 
0.62 1.16 0.19 
0.81 1.82 0.39 
No agitation except that produced by 
rising wash water 
1.01 1.26 0.74 
Note.-It is stated by Mr. Fuller that the washing of the Western filter, 
in which no rakes were used, was generally unsatisfactory. The effective 
size of the sand in these filters ranged from 0.43 to 0.51 mm., and the depths 
of the beds were from 27 to 31 in. 324 
WA TER P U RIF IC A TION 
The rate at which the wash water was applied in this method 
was relatively low, and, moreover, varied considerably. Mr. 
George W. Fuller in the report of the experiments with the Jewell, 
Warren and Western gravity filters at Louisville, Ky., gives some 
figures on wash-water velocities which have been recalculated to 
show velocities above the sand line and not in the bed itself. 
The results are shown in the preceding table. 
Compressed Air for Agitating the Sand Bed.-The use of com- 
pressed air for agitating the sand bed was advanced as a substi- 
tute for mechanical rakes, and soon came into general use. By 
this method of agitation the air is applied at a low pressure, and 
in a volume of 2 to 5 cu. ft. per minute per square foot of sand 
area. The air is usually applied to the tank after the water has 
been drawn down, so that only a few inches covers the surface 
of the filter bed. The air is generally applied for several minutes 
prior to the application of the wash water, and produces in its 
escape through the water a 11 boiling action," which gives an ob- 
server the impression that the sand is being violently agitated. 
This is not the case, however, and only a very slight movement 
of the sand is produced. 
The author was enabled some years ago to actually observe 
the effect of air and water applied to a sand and gravel filter bed 
in an experimental filter having glass sides. The following is 
quoted from the published account of these experiments, which 
appeared in the Report of Mr. George H. Benzenberg, Chief 
Engineer of the Commissioners of Water-works of Cincinnati, in 
Appendix C. 
"On top of the releveled layer of gravel a 30-in. layer of sand was 
placed, and the effect of air and water currents observed. Water 
admitted under a pressure of 5 lb. per square inch had no very marked 
effect on the gravel. The sand simply became less dense and floated 
up a little, as has been observed before. 
"Air was then forced in through the gravel and sand, the water being 
shut off. Under a gage pressure of about 3 lb. a volume of air was 
discharged into the filter equal to 6.86 cu. ft. The agitation of the 
sand was less marked than the writer expected, and other observers, 
when first seeing the effect of the air agitation, have noted this fact with 
considerable surprise. By watching the water above the sand it would 
be concluded that a very general agitation of the sand was taking place. 
"A peculiar effect of the air current was that beside an ascending 
current moving more or less sand with it, a descending current would 
often be noticed carrying sand in the opposite direction. This pecu- RATES OF FILTRATION 
325 
liarity is much more marked when air and water are being applied than 
with air alone. The gravel was more or less disturbed by the rising 
air currents. 
"Forcing in air and water together was next tried. It was imme- 
diately apparent that a far better agitation of the sand took place when 
the two were applied together. The less dense condition of the sand 
afforded a better play for the air currents, and they moved the sand 
about much more easily. Ascending and descending currents of sand 
could easily be traced. The gravel naturally suffered a great deal by 
the more violent agitation. At the end of the wash the sand and 
gravel were more or less mixed up, and presented a very irregular 
outline. It was quite apparent that the gravel layer ought not to be 
subjected to the violent action of the air current if it was to retain its 
uniform distribution over the bottom of the filter, unless it was held in 
place in some manner." 
The rate at which wash water is applied to a filter, in which 
compressed air is used for agitation, does not differ materially 
from those employed when rakes were used to agitate the bed. 
The velocity of the rising wash water is usually from 10 to 15 in. 
per minute, measured in the tank above the sand line. Lower 
velocities must be used if air and water are applied at the same 
time, unless ample distance between the normal sand line and 
the edges of the waste troughs is provided to prevent wastage of 
sand at the above velocities. The escape of the air from the 
rising wash water tends to carry the sand much higher than would 
the water alone. Under these conditions too high velocities may 
cause loss of sand over the waste troughs as stated above. 
The process of washing a filter may take from 10 to 20 min., 
depending upon the rate of application of the wash water, the 
amount of water required to clean the bed, the ease with which 
valves may be manipulated, and air compressors and pumps 
started and stopped. Hydraulically or electrically operated 
valves in the larger plants facilitate the operation greatly. The 
washing process should always be carefully watched in order to 
note whether the wash water is rising uniformly, whether the 
waste troughs are flooded, and whether the sediment is being 
properly carried away, i.e., does not remain suspended between 
the troughs instead of flowing toward and into them. 
High-velocity Method of Washing Rapid Sand Filters.-The 
slight agitating effect of compressed air upon a filter sand, as 
shown in the previously described experiments of the author, being 
regarded as unsatisfactory, further investigations were made in 326 
WATER PURIFICATION 
order to discover, if possible, a method of agitating and washing 
the sand which more nearly corresponded with the scrubbing ef- 
fects produced by the rotating mechanical rakes formerly used. 
As it was noticed that by increasing the velocity of the wash 
water in its passage through the sand bed, the latter was floated 
upward to higher levels, and that the sand grains moved more 
freely, rubbing against each other as they were forced upward 
by the rising currents of water, it seemed as though it offered 
another method for agitating the sand and had the added ad- 
vantage of cleansing the sand grains at the same time. 
As a result of the experimental work which necessarily followed 
these tentative conclusions, there was developed what is now 
known as the "high-velocity method of washing" rapid sand 
filters. The difficulties to be overcome in the new method related 
to determining the velocity at which wash water could be forced 
up through the sand bed without disturbing the gravel layer, 
and yet produce a proper agitation of the sand; to the distance 
to place the waste troughs above the normal sand level in order 
to prevent loss of sand while washing; to the way in which the 
wash water should be first applied to prevent a mixing of the 
dirty top layers with the cleaner sand below by local eruptive 
discharges of the wash water at points of less resistance; to the 
proper length of the washing period, and consequently to the 
volume of wash water which would be required; and lastly to 
the effectiveness of the whole process as shown by the absence 
of any evidence of the accumulation of sediment within the filter 
bed. 
The following quotation from the published record of this ex- 
perimental work will serve to show the results of these studies of 
the author: 
"The following conclusions appear to be justified by the result of 
this series of experiments: 
"1. That the application of wash water to a filtering bed composed 
of Ohio River sand, having an effective size of 0.33 mm., at the rate 
of 2.0 cu. ft. per square foot per minute, produces an effective and 
satisfactory agitation of the sand. 
"2. That the application of wash water at this rate thoroughly cleans 
the sand bed of any clay that may have been carried down into it. 
"3. That sand is carried over the edge of a gutter 31 in. above the 
sand line in too large quantities when the wash water is applied at a 
rate of over 2.5 cu. ft. per square foot per minute. RATES OF FILTRATION 
327 
"4. That at a rate of 2.5 cu. ft. per square foot per minute a small 
amount of sand is carried over a gutter 31 in. above the sand line, but 
the amount is probably negligible. 
"5. That over 96 per cent, of the sand carried over a gutter 31 in. 
above the sand line at a rate of 2.0 cu. ft. per square foot per minute is 
less than 0.45 mm. in diameter. 
"6. That a distance of 30 in. between the sand line and the edge of 
the gutter will probably prevent the loss of appreciable amounts of 
sand, when the wash water is applied at rates of less than 2.5 cu. ft. 
per square foot per minute." (Report of the Chief Engineer, Com- 
missioners of Water-works of Cincinnati, Ohio.) 
The successful use of this method of washing filters for over 8 
years at the Cincinnati filtration plant, and its gradual adoption 
by other plants establishes its reliability and usefulness. Its 
advantages lay in the absence of any need for the entire apparatus 
required where compressed air is used, and which necessarily con- 
sists of air compressors, storage tanks, delivery piping, and some- 
times a special distribution system of piping; in the greater sim- 
plicity of a single operation in place of two; in the fact that the 
agitating and washing are taking place at the same time and 
hence are more effective; and in the shorter period required to 
wash the filter, thereby lengthening the time a filter may be kept 
in effective service. The disadvantage of the method is chiefly 
in the somewhat larger sizes of pipes required in the wash-water 
distribution system, and which are necessary in order to convey 
the relatively large volume of wash water used in a period of 
usually less than 5 min. The method requires no greater percent- 
age of wash water than does the method using compressed air and 
a lower rate of application of the wash water. 
Mixing of the Sand and Gravel.-In the early experiments 
carried on by the author, the fear of mingling the sand and gravel 
by the application of the wash water at a high velocity, and the 
fact that such so-called "inversions" of filter beds were well 
known to have taken place in filters where low-velocity methods 
of washing were used, led to the use of a brass wire screen be- 
tween the gravel and sand layers. This screen was fastened in 
place, rigidly holding down the gravel and preventing its upward 
movement, while allowing free motion to the sand. The screen 
had openings of such size (100 meshes per square inch) that the 
gravel could not pass through them, but permitted the sand to 
do so. 328 
WATER PURIFICATION 
The screen fully demonstrated its effectiveness in the experi- 
mental work, and was consequently adopted as a part of the de- 
sign of the large filter tanks of the Cincinnati filtration plant. 
This screen was made of brass and contained over 70 per cent, of 
copper and about 30 per cent, of zinc. Within 2 or 3 years after 
the filters were put into service breaks in the screen became more 
or less frequent. These were due in part to imperfect methods of 
fastening, but principally to actual corrosion of the brass. Con- 
tinued corrosion of the brass further weakened the cloth so that 
breaks became quite frequent. 
The remedy sought, and confirmed by experiment, consisted in 
removing the screen and substituting a heavier and deeper layer of 
gravel. The original gravel layer of 7.5 to 8 in. was deepened to 
14 in., the lower layers of stones being composed of washed Ohio 
River gravel, the sizes of the stones of which varied from 1 to 2 
in. in diameter. The depths of the several layers were made as 
follows in the reconstruction of the filter beds: 
Depth of layer, 
inches 
Gravel passing screen 
with mesh of, inches 
Gravel retained on 
screen with mesh of, 
inches 
1st layer 
2 
1 
2d layer 
2 
1 
% 
3d layer 
3 
4th layer 
4 
5th layer 
3 
Reconstructed Filter Beds of Cincinnati Filter Plant1 
The depth of the sand layer is from 28 to 30 in. The sand 
originally had an effective size of 0.34 mm., and a uniformity 
coefficient of 1.5 to 1.6. The sand has become somewhat 
coarser by washing and now has an effective size of 0.38 mm., and 
a uniformity coefficient of 1.35. 
Loss of Head in Washing and Flotation of Sand.-From a series 
of experiments with an experimental filter, the losses of head 
obtained in washing filter beds of various depths, and the height 
of flotation of the sand, were observed, and are tabulated in the 
accompanying table. The gravel used was washed Ohio River 
gravel, consisting of rounded pebbles of varying sizes and placed 
in graded layers. The sand was also from the Ohio River and 
1 Eng. Record, vol. 71, p. 581. RATES OF FILTRATION 
329 
had effective sizes of 0.31 and 0.41 mm., and uniformity coeffi- 
cients of 1.45 and 1.41 respectively. 
Experimental Data on Loss of Head in Washing Filters by the 
High-velocity Method1 
Test 
num- 
ber 
Depth 
of 
gravel, 
inches 
Depth 
of 
sand, 
inches 
Up- 
ward 
veloc- 
ity of 
wash 
water, 
feet per 
minute 
Total 
head 
ap- 
plied 
under 
filter, 
feet 
Static 
head, 
feet 
Loss of head in feet of water 
Rise of 
sand, 
inches 
Through 
filter 
Through 
strainer 
plate 
Through 
gravel 
Through 
sand 
4t 
7H 
20 
1.0 
8.33 
6.40 
1.93 
0.58 
0.06 
1.29 
2.5 
4t 
7^ 
20 
1.5 
9.00 
6.42 
2.58 
1.20 
0.09 
1.29 
4.9 
4t 
7^ 
20 
2.0 
9.92 
6.43 
3.49 
2.05 
0.14 
1.30 
7.3 
5t 
7H 
25 
1.0 
8.70 
6.40 
2.30 
0.60 
0.06 
1.64 
2.8 
5t 
25 
1.5 
9.42 
6.42 
3.00 
1.22 
0.09 
1.69 
6.3 
5t 
7^ 
25 
2.0 
10.29 
6.43 
3.86 
2.03 
0.14 
1.69 
9.8 
6t 
7M 
30 
1.0 
9.00 
6.40 
2.60 
0.60 
0.06 
1.94 
2.6 
6t 
7^ 
30 
1.5 
9.81 
6.42 
3.39 
1.30 
0.09 
2.00 
6.6 
6t 
7^ 
30 
2.0 
10.63 
6.43 
4.20 
1.99 
0.14 
2.07 
10.7 
7 
7H 
20 
1.0 
8.40 
6.40 
2.00 
0.50 
0.06 
1.44 
2.7 
7 
7H 
20 
1.5 
9.07 
6.42 
2.65 
1.12 
0.09 
1.44 
5.3 
7 
7H 
20 
2.0 
9.93 
6.43 
3.50 
1.90 
0.14 
1.46 
7.9 
8 
7/ 
25 
1.0 
8.75 
6.40 
2.35 
0.50 
0.06 
1.79 
3.7 
8 
7/2 
25 
1.5 
9.42 
6.42 
3.00 
1.10 
0.09 
1.81 
6.8 
8 
7J4 
25 
2.0 
10.27 
6.43 
3.84 
1.90 
0.14 
1.80 
9.8 
9 
7H 
30 
1.0 
9.10 
6.40 
2.70 
0.50 
0.06 
2.14 
4.2 
9 
7^ 
30 
1.5 
9.76 
6.42 
3.34 
1.07 
0.09 
2.18 
7.8 
9 
7M 
30 
2.0 
10.57 
6.43 
4.14 
1.80 
0.14 
2.20 
11.8 
11 
14 
20 
1.0 
8.54 
6.40 
2.14 
0.60 
0.13 
1.41 
2.9 
11 
14 
20 
1.5 
9.38 
6.42 
2.96 
1.32 
0.20 
1.44 
5.7 
11 
14 
20 
2.0 
10.41 
6.43 
3.98 
2.22 
0.28 
1.48 
8.5 
12 
14 
25 
1.0 
8.80 
6.40 
2.40 
0.60 
0.13 
1.67 
3.7 
12 
14 
25 
1.5 
9.70 
6.42 
3.28 
1.27 
0.20 
1.81 
7.5 
12 
14 
25 
2.0 
10.75 
6.43 
4.32 
2.20 
0.28 
1.84 
11.4 
13 
14 
30 
1.0 
9.22 
6.40 
2.82 
0.57 
0.13 
2.12 
4.7 
13 
14 
30 
1.5 
10.03 
6.42 
3.61 
1.22 
0.20 
2.19 
9.2 
13 
14 
30 
2.0 
11.10 
6.43 
4.67 
2.20 
0.28 
2.19 
13.7 
15 
18 
20 
1.0 
8.70 
6.40 
2.30 
0.62 
0.15 
1.53 
3.0 
15 
18 
20 
1.5 
9.53 
6.42 
3.11 
1.32 
0.22 
1.57 
6.2 
15 
18 
20 
2.0 
10.58 
6.43 
4.15 
2.23 
0.32 
1.60 
9.4 
16 
18 
25 
1.0 
9.02 
6.40 
2.62 
0.64 
0.15 
1.83 
4.1 
16 
18 
25 
1.5 
9.87 
6.42 
3.45 
1.33 
0.22 
1.90 
7.7 
16 
18 
25 
2.0 
10.92 
6.43 
4.49 
2.20 
0.32 
1.97 
11.3 
17 
18 
30 
1.0 
9.40 
6.40 
3.00 
0.62 
0.15 
2.23 
4.9 
17 
18 
30 
1.5 
10.25 
6.42 
3.83 
1.30 
0.22 
2.31 
10.2 
17 
18 
30 
2.0 
11.22 
6.43 
4.79 
2.10 
0.32 
2.37 
15.6* 
t Brass wire screen in place. * Fine sand passed over wash-water troughs. 
The relatively low losses of head through the gravel layers and 
the comparatively uniform losses of head through the sand for any 
1 J. W. Ellms and J. S. Gettrust: "A Study of the Behavior of Rapid Sand Filters 
Subjected to the High-velocity Method of Washing." Proc. Am. Soc. C. E., January, 
1916. WATER PURIFICATION 
330 
given depth of the latter, even for variations in the velocity of the 
wash water from 1 to 2 ft. per minute, are of interest in this 
table, as well as the greater relative increase in the height to 
which the sand floated as the velocity of the wash water was in- 
creased. 
References 
1. Nicholas S. Hill, Jr.: "Modern Filter Practice." Amer. Water- 
works, Assn., 1913. 
2. George W. Fuller: "Report on Purification of Ohio River Water, 
Louisville, Ky." 
3. Report of the Chief Engineer, George H. Benzenberg, to the Com- 
missioners of Water-works, Cincinnati, Ohio. 
4. "Reconstruction of Cincinnati Filter Beds." Eng. Record, vol. 71. 
5. Jos. W. Ellms and John S. Gettrust: "A Study of the Behavior of 
Rapid Sand Filters Subjected to the High-velocity Method of Wash- 
ing." Proc. Amer. Soc. C. E., January, 1916. 
6. H. M. Pirnie: Discussion on "Washing Rapid Sand Filters, Negative 
Head, Etc." Proc. Am. Soc. C. E., April, 1916. 
7. Harold C. Stevens: "Pressure Filters." Jour. Am. Water-works 
Assn., June, 1916. CHAPTER XXVI 
THE PHYSICAL AND CHEMICAL CHANGES PRODUCED 
BY THE APPLICATION OF CHEMICAL COAGU- 
LANTS AND BY THE SUBSEQUENT 
FILTRATION OF THE TREATED 
WATER 
No intelligent operation of rapid sand filters is possible without 
a thorough knowledge of the chemistry of coagulation. The 
physical as well as the chemical characteristics of the water being 
treated must also be known, if the best purification results are to 
be obtained. Certain of the physical features of the coagulation 
of suspended impurities in water have been discussed in a pre- 
vious chapter. The present chapter will deal more fully with 
some phases of the subject relating to filtration, and to the 
chemical reactions involved in the application of coagulants. 
Coagulation.-The coming together in small flocculent masses 
of the finely divided suspended particles in a turbid water is in- 
duced by the addition of certain chemical salts, which react either 
with the salts naturally dissolved in the water or with alkaline 
bases that may be added with the coagulating chemical. The 
two coagulating chemical salts commonly used in treating waters 
are aluminum sulphate and ferrous sulphate. From the former, 
aluminum hydroxide is produced, and from the latter, ferric 
hydroxide. The gelatinous properties of these two hydroxides, 
and the ease with which they may be formed in the water, give to 
them their value for water-purification purposes. 
A portion of the coagulated impurities are precipitated in pre- 
liminary settling basins. The balance must be removed by the 
filters. It is evident that once the flocculent masses have been 
formed, they should not be broken up by agitation. In conse- 
quence, after a thorough mixing of the applied chemicals with 
the water has been effected, the subsequent flow of the water 
should be at a low velocity, and as free as possible from any form 
of agitation. Spilling a coagulated water into a filter tank is 
not, therefore, good practice, since it breaks up the flocculent 
331 332 
WA TER P UR IF IC A TION 
matter, and also tends to keep the surface of the sand bed in the 
immediate vicinity of the inlet washed free of its colloidal 
coating. 
Amount of Coagulated Sediment in Water Applied to Filters.- 
In the chapter on the efficiencies of settling and coagulating 
basins, a number of tables were given showing the residual tur- 
bidity of water passing from these basins. The range in the tur- 
bidity of the coagulated water, which was to be applied to rapid 
sand filters, varied from practically zero to 50 parts per million. 
The quantity of sediment which a filter will successfully handle 
obviously depends upon several factors; namely, the quantity and 
character of the colloids remaining in the water, the amount 
and kind of colloidal coating on the sand grains of the filter bed, 
the fineness and depth of the sand bed, and the rate of filtration. 
The larger the amount of coagulant applied, the thicker and more 
resistant will be the gelatinous coating on the sand surface. The 
lower the temperature which a water reaches, the less the reten- 
tive capacity of the colloidal hydroxide for sediment appears to 
be. The greater the rate of filtration, the farther into the sand 
bed will the minute particles be carried, and the sooner the filter 
bed will become overcharged. Finally, the character of the 
sediment itself plays an important part, since the finer the col- 
loidal particles are, the more difficult it becomes to restrain them 
from passing through the filter bed. 
Certain kinds of organic matter in the water materially assist 
the applied coagulant in preventing the passage of suspended 
particles. To illustrate, the author has found that with rates of 
filtration of 125,000,000 gal. per acre daily, usually not more than 
15 to 20 parts per million of suspended matter produced by the 
clay particles found in the Ohio River water can be successfully 
filtered through a sand bed 30 in. deep, the sand of which has an 
average effective size of 0.38 mm. On the other hand, it has been 
observed with these same filters, that a turbidity of 100 parts 
per million in the applied water, resulting from a rise in the 
Ohio River during mid-summer, was filtered without any diffi- 
culty whatsoever. 
The greater retentive capacity of the sand bed during the 
summer months is probably due to the organic colloids produced 
by the growth and death of microscopic plants which flourish in 
the river water during the warmer months of the year and which 
materially assist the chemically produced colloids in the filtra- PHYSICAL AND CHEMICAL CHANGES 
333 
tion process. During periods of muddy water, this is obviously 
an advantage as it materially reduces the amount of chemical 
coagulant which would otherwise be required. On the other 
hand, the presence of these natural organic colloids, when the 
water is quite free from suspended matter, is detrimental to the 
filtration process, since their gelatinous character unmodified by 
clay particles, causes them to clog the bed too quickly, and in- 
creases the amount of wash water required to clean the bed. 
Effect of Temperature on Coagulation.-In those climates 
where the temperature of the treated water during the cold season 
of the year reaches nearly to the freezing point, the coagulation 
of suspended matter during this period is imperfectly accom- 
plished. The diminished efficiency appears to arise from some 
modification of the gelatinous properties of the chemical colloids 
produced, and which may be the result of a retardation of the 
chemical reactions involved, and the formation of smaller flocs 
having a diminished adsorptive power. 
The practical effect of the low temperatures of the water upon 
the amounts of coagulant required is well illustrated by the curves 
in the accompanying cuts. These curves were prepared by Mr. 
H. W. Streeter, from observations made while operating a rapid 
sand filter plant, and are here published with his permission for 
the first time. Mr. Streeter's comments on these diagrams (Fig. 
115) are as follows: 
"These curves are based upon a series of diagrams prepared from 
data taken from daily records, and from many observations secured 
during about a year's operation of the plant. The diagrams shown were 
drawn for typical winter, spring and autumn, and summer conditions. 
Over each period a record of the temperature of the water was kept, 
and the mean was computed for each period. The variation in water 
temperatures for each period was not more than 5° from the mean. 
The following notation was employed in plotting: 
Tr = turbidity of raw water in parts per million. 
Tc = turbidity of coagulated water in parts per million. 
"Each line represents a certain value of Tr, and the amounts of 
aluminum sulphate necessary to give a coagulated water of a given 
turbidity (indicated by the value of Tc for each chart) are indicated for 
different temperatures of the water. The time of sedimentation, 
following the application of the coagulant, was about 5 hr. 
"The curves are most poorly defined for extremely low and high raw 334 
WATER PURIFICATION 
RELATION BETWEEN AMOUNTS OF 
COAGULANT, TEMPERATURE OF WATER, 
AND TURBIDITY OF COAGULATED 
AND RAW WATER 
Time of Sedimentation 5 hrs. 
Tr=Turbidity of raw water in p.p.m. 
Tc= " " coagulated water in p.p.m. 
Fig. 115.-Relation between 
amounts of coagulant, temperature 
of water, and turbidity of coagu- 
lated and raw water. 
Alum in Grains per Gallon 
Temperature of Water in Degrees Fahrenheit 
Alum in Grains per Gallon 
Alum in Grains per Gallon 
Temperature of Water in Degrees Fahrenheit 
Alum in Grains per Gallon 
Alum in Grains per Gallon 
Alum in Grains per Gallon 
Alum in Grains per Gallon PHYSICAL AND CHEMICAL CHANGES 
335 
water turbidities, and for very high applied water turbidities, since these 
were the conditions least frequently encountered. 
"It should also be stated that the particular values obtained were 
based upon local conditions from which there might be considerable 
variation in other plants, particularly as regards the size and arrange- 
ment of the basins, their depth, their method of operation, and the 
method of applying the coagulant. For circumstances differing from 
those under which the values were obtained, results might not agree 
closely. The curves bring out, however, some very interesting relations 
between the amounts of coagulant required to produce a given degree 
of clarification at different water temperatures. 
"The curves were found to be of great practical service in 'setting 
coagulant feeds' to meet sudden changes in the condition of the raw 
water. 
"The influence of the variations in the character of the suspended 
matter in cold and in warm seasons, particularly when the raw water 
turbidity was high, may be a factor affecting the general trend of the 
curves, aside from the effect of the temperature of the water." 
Chemical Reactions Involved in Coagulation.-When a water 
which is naturally alkaline from the presence in solution of the 
bicarbonates of calcium and magnesium, is treated with a solu- 
tion of aluminum sulphate, a reaction occurs which may be 
illustrated as follows: 
1. A12(SO)3 + 3H2Ca(CO3)2 3CaSO4 + A12(OH)6 + 6CO2 
Aluminum 
Sulphate 
Bicarbonate 
of Calcium 
Calcium 
Sulphate 
Aluminum 
Hydroxide 
Carbon 
Dioxide 
Soluble 
Soluble 
Soluble 
Insoluble 
in Water, 
coagulum 
desired 
Soluble 
The reaction is precisely analogous if magnesium bicarbonate 
reacts with the aluminum sulphate instead of the calcium bi- 
carbonate. In natural waters both of the carbonates of calcium 
and magnesium are held in solution principally by an excess of 
dissolved carbon dioxide. If insufficient amounts of naturally 
alkaline carbonates are present, then alkali must be added. This 
is usually provided by introducing a solution of sodium carbonate 
(soda ash) with the coagulating chemical. Caustic lime is some- 
times employed with aluminum sulphate, but has the disadvan- 
tage of making the water harder by 7.7 parts per million for 
each grain of aluminum sulphate neutralized. The reaction with 
sodium carbonate is as follows: 336 
WATER PURIFICATION 
2. A12(SO4)3 + 3Na2CO3 + 3H2O 3Na2SO4 + A12(OH)6 + 3CO2 
Aluminum 
Sulphate 
Sodium 
Carbonate 
Water 
Sodium 
Sulphate 
Aluminum 
Hydroxide 
Carbon 
Dioxide 
Soluble 
Soluble 
Molecules 
reacting 
Soluble 
Insoluble 
in water 
Soluble 
When ferrous sulphate or copperas is employed as the chemical 
coagulant, the reactions are somewhat different from those with 
aluminum sulphate. The iron being in the ferrous state, that 
is, in a condition in which it is capable of taking on more oxygen, 
abstracts the latter from the air dissolved in the water. Since 
it is ferric hydroxide, or the more highly oxidized iron compound 
that is the real coagulant, this oxidizing action must be favored 
and assisted. The best and cheapest method thus far found to 
assist this oxidation is by the addition of caustic lime. The 
reason for using this caustic lime may be better understood if the 
reactions involved are first given. Assuming that the sulphate 
of iron is added to the water before the caustic lime, then the 
following reaction occurs: 
3. FeSO4 + H2Ca(CO3)2 CaSO4 + H2Fe(CO3)2 
Sulphate 
of Iron 
Bicarbonate 
of Calcium 
Calcium 
Sulphate 
Bicarbonate 
of Iron 
Soluble 
Soluble 
Soluble 
Slightly soluble 
If caustic lime is now added to the water the following reaction 
takes place: 
Bicarbonate 
of Iron 
4. H2Fe(CO3)2 + 2Ca(OH)2 Fe(OH)2 + 2CaCO3 + 2H2O 
Calcium 
Hydroxide 
Ferrous 
Hydroxide 
Calcium 
Carbonate 
Water 
reacting 
Slightly 
soluble 
Slightly 
soluble 
Very slightly 
soluble 
Very slightly 
soluble 
5. 4Fe(OH)2 + O2 + 2H2O 2Fe2(OH)6 
F errous 
Hydroxide 
Oxygen 
Water 
Ferric Hydroxide 
Very slightly 
soluble 
From air 
dissolved in 
water 
Reacting 
molecules 
Insoluble in water; 
coagulum desired 
By reference to reaction No. 3 it will be noted that the iron 
and calcium have exchanged places, and that a bicarbonate of 
iron is formed. This latter compound like the bicarbonate of PHYSICAL AND CHEMICAL CHANGES 
337 
calcium is slightly soluble in water, principally because of the 
carbonic acid to which it is somewhat loosely bound. Its tend- 
ency to hold the iron in solution seems to produce a retarding 
effect on the oxidation of the iron to the ferric condition. In 
order, therefore, that the iron may become oxidized more quickly, 
the caustic lime is introduced to remove this loosely bound car- 
bonic acid, as showm by reaction No. 4. The ferrous hydroxide 
is easily oxidized to ferric hydroxide by the oxygen dissolved in 
the water. Reaction No. 5 illustrates this change. 
Even if no alkali is introduced, the instability of the bicarbon- 
ate of iron is shown by the partial hydrolysis of the compound, 
with the resultant formation of ferrous hydroxide. The reaction 
is markedly reversible. 
6. H2Fe(CO3)2 + 2H2O Fe(OH)2 + 2H2CO3 
Bicarbonate 
of Iron 
Reacting mole- 
cules of water 
Ferrous 
Hydroxide 
Carbonic Acid, or 
CO2 plus one mole- 
cule of H2O 
Slightly 
soluble 
Very slightly 
soluble 
Soluble 
The free carbon dioxide (CO2) is liberated by violent agitation 
combined with aeration; and it is possible to hasten the above 
reactions (Nos. 6 and 5) and thereby avoid the use of an alkali. 
This method has been proposed, but the difficulties of applying 
the principles in practice on a large scale have not as yet been 
overcome. 
Assuming that the lime is added before the ferrous sulphate, 
as is sometimes the practice, the free carbonic acid and the bi- 
carbonates of calcium and magnesium react with the caustic lime 
as follows: 
7. H2Ca(CO3)2 + H2CO3 + 2Ca(OH)2 3CaCO3 + 4H2O 
Bicarbonate 
of Calcium 
Free Carbonic 
Acid 
Calcium 
Hydroxide 
Calcium 
Carbonate 
Water 
reacting 
Slightly 
soluble 
Soluble 
Slightly 
soluble 
Very slightly 
soluble 
The completeness of reaction No. 7 depends upon the amount 
of caustic lime added. In water-softening plants this reaction is 
usually carried far enough to remove all of the free and half- 
bound carbonic acid, and may and often is carried even farther, 
if it is desired to precipitate as hydroxide the magnesium of the 338 
WATER PURIFICATION 
salts of this latter element. Under these conditions, there only 
remains to react with the ferrous sulphate that may be added, 
the small amount of calcium or magnesium carbonate which will 
remain in solution in a water even when carbonic acid is absent. 
If by chance an excess of caustic lime has been used, the reaction 
with the ferrous sulphate is, of course, direct. 
It is evident, therefore, that there may be in solution to react 
with the ferrous sulphate applied, bicarbonates only, or bicar- 
bonates and monocarbonates,1 or monocarbonates and calcium 
hydroxide. Where there exist only monocarbonates, or mono- 
carbonates and calcium hydroxide, the reactions may be expressed 
as follows: 
8. FeSCh + CaCOs FeCO3 + CaSO4 
Sulphate 
of Iron 
Calcium 
Carbonate 
Carbonate 
of Iron 
Calcium 
Sulphate 
Soluble 
Very slightly Very slightly 
soluble soluble 
Soluble 
9. FeSO< + Ca(OH)2 Fe(OH)2 + CaSO4 
Sulphate 
of Iron 
Calcium 
Hydroxide 
Ferrous 
Hydroxide 
Calcium 
Sulphate 
Soluble 
Slightly 
soluble 
Very slightly 
soluble 
Soluble 
10. FeCO3 + 2H2O Fe(OH)2 + H2CO3 
Carbonate 
of Iron 
Reacting molecules 
of Water 
Ferrous 
Hydroxide 
Carbonic 
Acid 
Very slightly 
soluble 
Very slightly 
soluble 
Soluble 
The oxidation of the ferrous hydroxide proceeds as shown by 
reaction No. 5. Theoretically each grain of ferrous sulphate 
requires 0.5 part per million of dissolved oxygen to effect oxida- 
tion. This amount is very small, and no water utilized for a 
water supply is likely to have less than 10 to 15 times this amount 
as a rule. 
Effects of Chemical Coagulants on Dissolved Constituents of 
Water.-From the foregoing reactions it is apparent that cer- 
tain changes are produced in the dissolved constituents of a 
1 Monocarbonates may be defined as those compounds which contain no 
excess of carbonic acid over and above that directly combined with the base 
to form the normal salt. PHYSICAL AND CHEMICAL CHANGES 
339 
water by the addition of the coagulating chemicals. The alumi- 
num, i.e., the metal is precipitated as the hydroxide, but the 
sulphuric acid with which it was combined remains dissolved 
in the water combined with the alkaline earth base. Analogous 
results are obtained with the ferrous sulphate. The iron is 
precipitated as the hydroxide, but the sulphuric acid remains 
dissolved. In both cases the acid unites with the calcium and 
magnesium in the water to form a sulphate. The carbonic acid 
liberated either remains dissolved in the water, or is precipitated 
as calcium carbonate whenever caustic lime is provided to unite 
with it. 
The practical effect upon the dissolved constituents of the water 
is, therefore, to replace carbonates of calcium and magnesium by 
sulphates of these elements, or in other words to convert the tem- 
porary hardness or alkalinity of the water into permanent hard- 
ness. The total hardness of the water is not modified, but the 
permanent hardness is increased in proportion as the temporary 
hardness or alkalinity is decreased. These general statements 
are strictly true only when the natural alkalinity of the water 
reacts with the applied chemical. If soda ash is used to assist 
the natural alkalinity of the water to decompose the applied 
sulphate of aluminum or sulphate of iron, or is present naturally, 
the sulphate of sodium that may be formed does not add to the 
hardness of the water. On the other hand, if calcium hydroxide 
is employed, the permanent hardness is increased. 
Where calcium hydroxide is employed merely to accelerate 
the decomposition of sulphate of iron, and is not needed to rein- 
force the alkalinity of the water, its effect upon the water depends 
upon the amount used and on the original quantities of the bicar- 
bonates of calcium and magnesium which were present. If the 
original alkalinity of the water is below 50 parts per million 
(expressed in terms of calcium carbonate), and the amount of 
calcium hydroxide added is just sufficient to neutralize the sul- 
phuric acid of the applied chemical coagulant, there will be no 
reduction in the alkalinity of the filtered water; if there is more 
caustic lime added than is needed by the chemical coagulant, but 
the amount is still only sufficient to bring the total alkalinity of 
the water up to about 50 parts per million, then the alkalinity of 
the filtered water will be increased. This means, of course, that 
the total hardness of the water will be increased to the extent 
that the alkalinity has been increased, since the additional per- 340 
WA TER P URIE IC A TION 
manent hardness is produced only by the destruction of an equiva- 
lent amount of alkalinity. 
On the other hand, if the alkalinity of the water due to bicar- 
bonates of calcium and magnesium exceeds about 50 parts per 
million and calcium hydroxide is used, the effect will be to reduce 
the total hardness of the water, since carbonates of calcium and 
magnesium in excess of about 50 parts per million will not remain 
dissolved when deprived of the carbonic acid which normally 
holds them in solution. Reference to reaction No. 7 will show a 
theoretically complete reaction between the bicarbonate of cal- 
cium and calcium hydroxide. It is obvious that where not 
enough caustic lime is used to effect a complete conversion of 
the bicarbonates of calcium and magnesium to the monocar- 
bonates of these bases, a mixture of bicarbonates and monocar- 
bonates will be present. This is actually the case where calcium 
hydroxide is used merely as an aid to the decomposition of the 
ferrous sulphate, and where the total alkalinity is below about 
50 parts per million. 
In the preceding paragraphs, 50 parts per million of residual 
alkalinity in a lime treated water is only to be regarded as ap- 
proximate, since a number of factors, such as temperature of the 
water and time of reaction affect the final amount of the calcium 
and magnesium carbonates that will remain dissolved. In 
practice, however, a hard water softened by means of calcium 
hydroxide will show after 6 to 10 hr. settlement about 50 parts 
per million of alkalinity. This figure is naturally higher than the 
more exact laboratory results given in succeeding paragraphs, and 
which are not obtained in practice. 
Theoretical Reduction of Alkalinity.-The application of 1 
grain per gallon of aluminum sulphate (A12(SO4)318H2O), to a 
water naturally alkaline with calcium and magnesium carbonates, 
will reduce the alkalinity, when expressed in terms of calcium 
carbonate, 7.7 parts per million. One grain of sulphate of iron 
(FeSO47H2O) per gallon will reduce the alkalinity 6.16 parts 
per million. These theoretical reductions are not always pro- 
duced in practice; first, because commercially chemicals are not 
produced of precisely the exact composition given in the formulas, 
either purposely or because of the difficulties connected with 
their manufacture on a commercial scale; secondly, because of 
the adsorption of the undecomposed chemical on the surfaces of 
colloidal particles in suspension; and thirdly, because of some PHYSICAL AND CHEMICAL CHANGES 
341 
direct combinations between the applied chemicals and the 
dissolved constituents of the water. 
For 1 grain of sulphate of iron only 0.2 grain of calcium oxide 
is required to effect the decomposition of the former, but con- 
sidering the impurities in the lime as usually obtained, and the 
losses in preparing solutions, it is usually necessary to use about 
twice as much (0.4) for each grain of the iron sulphate used. 
The theoretical reduction in alkalinity produced by sodium 
carbonate for each grain of aluminum sulphate applied is 8.1 
parts per million of NagCOs, and the amount of carbonic acid 
(CO2) liberated is 3.4 parts per million. It has been claimed 
that not all of the aluminum sulphate is changed to hydroxide 
(see reaction No. 2) by sodium carbonate, but that some basic 
sulphate of aluminum is produced, which is somewhat soluble 
and which might pass through a filter bed. Mr. George C. Whip- 
ple, in a paper1 on "Hot-water Troubles," in which he discusses 
the effect of alum-treated waters on iron piping, makes the follow- 
ing statement: 
"It is possible, also, that where the amount of alkalinity in the raw 
water is very low the alum reaction does not go to completion but a 
certain small amount of aluminum sulphate remains in the water. 
Conductivity experiments made by Meville C. Whipple have shown 
that whereas the conductivity of hard water is increased by a very 
small amount when alum is added, the conductivity of relatively soft 
water is considerably increased by the addition of the same amount of 
alum. Thus, in one series of experiments, in which 1 grain per gallon 
of alum was added to solutions of calcium carbonate of different 
strengths, it was found that the increase in conductivity in the solution 
containing 100 parts per million of calcium carbonate was only about 
2 per cent., whereas in the solution containing 50 parts per million it 
was 15 per cent.; in the solution containing 25 parts per millien, 20 per 
cent.; and in the solution containing 10 parts per million, 35 per cent." 
Solubilities of Calcium and Magnesium Carbonates and Mag- 
nesium Hydroxide.-Calcium carbonate and magnesium carbon- 
ate are only slightly soluble in pure water, but at ordinary tem- 
peratures and pressure will dissolve to the extent of about 1 gram 
per liter of the former, and 20 grams per liter of the latter, if the 
water contains sufficient carbon dioxide. In the absence of car- 
bon dioxide calcium carbonate is only soluble to the extent of 
about 13 parts per million. Magnesium carbonate is much more 
1 Proc. Am. Water-works Assn., 1911. 342 
WATER PURIFICATION 
soluble than calcium carbonate in the absence of carbon dioxide, 
but its exact solubility appears difficult to determine, since it 
appears impossible to prepare the pure carbonate of magnesia, 
it being invariably basic, that is, it contains both the hydroxide 
and carbonate of magnesium. The solubility of magnesium hy- 
droxide (MgO2H2), however, has been determined with consid- 
erable accuracy, and appears to be about 10 parts per million. 
It has been shown by Whipple and Mayer1 that lime water 
reacts more rapidly with magnesium bicarbonate than with cal- 
cium bicarbonate, and that this is due to the fact that the precipi- 
tate is colloidal in character in the former case, and crystalline 
in the latter, although colloidal at first for a short time. It re- 
quires some time for calcium carbonate to be completely precipi- 
tated in the crystalline form. This fact is of practical importance 
as it explains the so-called "after-deposits" of carbonate of lime 
on filter sand, in softened-water reservoirs, and in distribution 
pipe lines. This phenomenon is of more consequence in water- 
softening plants than in rapid sand filter plants, although this 
effect may always be noted to a greater or less degree, wherever 
caustic lime is employed. Practically the greater part of a pre- 
cipitate of calcium carbonate separates out within 6 hr. at ordi- 
nary temperatures, but the reaction appears not to be complete 
for a great many hours thereafter. Filtration of a water super- 
saturated with these compounds of lime and magnesia through 
sand reduces the quantity in solution, as shown by a reduction in 
the alkalinity of the filtered water. 
Two experiments given by Whipple and Mayer, in the paper 
cited, are sufficient to show the relative solubilities and reaction 
periods for both calcium carbonate and magnesium hydroxide, 
when bicarbonates of these elements are treated with lime-water 
solutions. 
In the first experiment just enough lime water was used to 
neutralize all the free CO2 and precipitate all the calcium car- 
bonate. 
The results obtained in precipitating magnesium hydroxide 
with lime water are complicated by the solubility of the calcium 
carbonate produced by the precipitation of the calcium of the 
1 George C. Whipple and Andrew Mayer, Jr.: "The Solubility of 
Calcium Carbonate and Magnesium Hydroxide and the Precipitation of 
These Salts with Lime Water." Jour. Infect. Diseases, Supplement No. 
2, February, 1906. PHYSICAL AND CHEMICAL CHANGES 
343 
Experiment No. 
Time of standing in 
hours 
Temperature, 
degrees C. 
Parts per million 
of CaCOs 
I 
1.0 
22 
92 
II 
3.5 
22 
35 
III 
6.0 
22 
26 
lime water. The results obtained in two parallel experiments 
made at 35°C., and for the same periods of time by precipitating 
a solution of magnesium bicarbonate and a solution of calcium 
bicarbonate with lime water, and taking the difference between 
the observed alkalinities of the solutions as the amount of dis- 
solved MgO2H2, are given in the following table: 
Time in hours 
Alkalinity in parts per million 
Calcium carbonate 
Magnesium hydroxide 
and calcium carbonate 
Difference (MgOzHj) 
0.25 
55.0 
85 
30.0 
0.50 
40.0 
66 
26.0 
1.00 
33.0 
60 
27.0 
2.00 
29.0 
55 
26.0 
3.00 
28.0 
53 
25.0 
5.00 
25.0 
50 
25.0 
10.00 
22.0 
48 
26.0 
24.00 
18.5 
39 
20.5 
48.00 
16.0 
36 
20.0 
Vegetable Coloring Matter.-The colloidal particles which pro- 
duce color in a water are derived from the vegetable matter with 
which the water has come into contact. The coloring matter is 
composed of compounds of humic, gallic and tannic acids, derived 
from the decomposition of vegetable matter in swamps and low- 
lands partially covered with water. 
The adsorption of this coloring matter by other colloids is the 
practical method of removing it from a water. For this purpose 
aluminum sulphate is used, since the aluminum hydroxide result- 
ing from its decomposition produces a colloid with adsorption 
properties that seem able to withdraw quite a proportion of the 
coloring matter from the water. Usually not all of the color 
can be thus removed, unless excessive amounts of aluminum 
sulphate are used. 344 
WATER PURIFICATION 
Iron salts from which ferric hydroxide would normally be pre- 
cipitated are not adapted for removing color, since they unite 
with the organic acids and tend to produce colloidal iron com- 
pounds which deepen instead of diminish the color. A color of 
20 parts per million or less in a water is not objected to by most 
people for drinking purposes. Efforts to reduce highly colored 
waters to less than a color of 10 parts per million is not generally 
advisable on the score of cost, since large amounts of coagulant 
are necessary to remove this residual coloring matter. 
Presence of coloring matter appears to prevent in some meas- 
ure the theoretical reduction in the alkalinity of water treated 
with aluminum sulphate, in the same manner as does the pres- 
ence of large quantities of colloidal clay. Mr. George C. Whipple, 
in commenting on this fact in his paper on " Hot-water Troubles," 
says: 
"It has been frequently noticed that the reduction in alkalinity is 
somewhat less than the theoretical. Fuller found this to be true with 
the turbid waters of the Ohio River, and explained it as being due to 
the absorption of a certain amount of the sulphate of alumina by 
the clay. The writer has found it to be the case with colored waters 
and especially with soft colored waters. It is possible that here also 
it may be explained by the absorption of the alum by the colloidal 
organic matter present in the colored water, but as it occurs also with 
waters that have been filtered and the suspended matter removed, it is 
perhaps more reasonable to believe that some alum unites chemically 
with the organic matter. The higher the color of the water, the larger 
the amount of aluminum sulphate that is combined in this way. It 
was found that when the amount of alum added was insufficient to 
Effect of Coloring Matter in Water on Reduction of Alkalinity 
Color 
Original alkalinity 
10 P.p.m. 
75 P.p.m. 
Reduction in alkalinity for each grain per gallon of aluminum 
sulphate used 
P.p.m. 
P.p.m. 
P.p.m. 
0 
6.6 
8.0 
50 
5.3 
6.6 
Note.-Experiments made by adding 0.75 grain per gallon of aluminum 
sulphate to distilled water to which calcium carbonate had been added to 
give the desired alkalinity, and extract of leaves to give the desired color. PHYSICAL AND CHEMICAL CHANGES 
345 
coagulate and precipitate the coloring matter, the reduction in alkalinity 
was low. 
"From the observations and experiments it seems reasonable to 
believe that with colored waters a double reaction takes place, and that 
a small part of the alum unites with the organic matter directly, while 
the larger part reacts with the calcium carbonate in the manner above 
mentioned. The nature of this aluminum organic compound is not 
known. Possibly it exists as a colloid." 
The results of one of Mr. Whipple's experiments are as follows: 
Oxygen Dissolved in Water.-All natural surface waters con- 
tain more or less air dissolved in them. The quantity varies 
with both the temperature and pressure. At 32°F. and atmos- 
pheric pressure, 63.8 parts per million of air are dissolved in 
water. At the same temperature and pressure 14.7 parts per 
million of oxygen are dissolved. At 86°F., the amount of air 
dissolved falls to 33 parts per million, and the oxygen to 7.6 
parts per million. Approximately 1 part per million of dissolved 
oxygen by weight in 1,000,000 parts of water corresponds with 
0.7 c.c. of oxygen per liter of water. 
The amount of oxygen required to saturate water at a pressure 
of one atmosphere and at various temperatures is given in the 
following table:1 
Quantities of Dissolved Oxygen in Parts per Million by Weight in 
Water Saturated with Air at the Temperature Given 
Temperature, 
degrees C. 
Oxygen 
Temperature, 
degrees C. 
Oxygen 
0 
14.70 
16 
9.94 
1 
14.28 
17 
9.75 
2 
13.88 
18 
9.56 
3 
13.50 
19 
9.37 
4 
13.14 
20 
9.19 
5 
12.80 
21 
9.01 
6 
12.47 
22 
8.84 
7 
12.16 
23 
8.67 
8 
11.86 
24 
8.51 
9 
11.58 
25 
8.35 
10 
11.31 
26 
8.19 
11 
11.05 
27 
8.03 
12 
10.80 
28 
7.88 
13 
10.57 
29 
7.74 
14 
10.35 
30 
7.60 
15 
10.14 
1 "Standard Methods of Water Analysis." Report American Public 
Health Association. 346 
WATER PURIFICATION 
A water may be supersaturated with oxygen at any given tem- 
perature, or it may contain less than the normal amount. Gill1 
has shown that a water may be heated 10° above its saturation 
point for 24 hr.-in some cases 48 hr.-without giving up entirely 
the excess of oxygen. Depletion of oxygen in a water is usually 
due to its removal by oxidizable substances in solution or in 
suspension. Supersaturation is not uncommon even in surface 
waters. Stored and impounded waters that are in contact with 
much organic matter, as for example, with the deposited sedi- 
ment from muddy waters, or with the unstripped soil of reservoir 
bottoms, are quite commonly lower in dissolved oxygen than 
they are capable of holding on account of the using up of the 
oxygen by the decomposing matter undergoing slow oxidation. 
The depletion of a water's oxygen on account of the use of 
ferrous sulphate as a coagulant is so slight (0.5 part per million 
for each grain of ferrous sulphate used per gallon) that it may be 
disregarded. Aeration to restore oxygen to filtered waters, which 
have been purified with the aid of this chemical is not, therefore, 
necessary and is not practised. Aeration of water to remove 
obnoxious gases or volatile matter producing offensive odors is 
frequently required in water-purification processes, and is usually 
undertaken after the water has passed through the filters. 
1. "Water Purification at Louisville." Report by George W. Fuller. 
2. "The Purification of the Ohio River Water at Cincinnati." Report 
by George W. Fuller. 
3. Paul Rohland: "The Colloidal and Crystalloidal State of Matter." 
4. George C. Whipple: "Hot-water Troubles." Proc. Am. Water-works 
Assn. 
5. George C. Whipple and Andrew Mayer, Jr.: "Solubility of Calcium 
Carbonate and Magnesium Hydroxide and the Precipitation of 
These Salts with Lime Water." Jour. Infect. Diseases, Supplement 
No. 2, February, 1906. 
6. "Standard Methods of Water Analysis." Report Committee American 
Public Health Association. 
7. Aug. H. Gill: "Upon the Difficulty with Which Water Parts with 
Its Dissolved Oxygen." Technology Quarterly, vol. 5, No. 3, October, 1892. 
8. Thos. M. Drown: "Amount of Dissolved Oxygen Contained in the 
Water of Ponds and Reservoirs at Different Depths in Winter under 
the Ice." Report Mass. State Board of Health, 1892. 
9. George F. Catlett: " Colloidal Theories Applied to Colored Waters." 
Eng. Record, June 3, 1916. 
'Aug. H. Gill: "Upon the Difficulty with Which Water Parts with 
Its Dissolved Oxygen." Technology Quarterly, vol. 5, No. 3, October, 1892. 
References CHAPTER XXVII 
EFFICIENCY AND COST OF OPERATION OF RAPID SAND 
FILTERS 
The highest efficiency in the operation of rapid sand filters 
can be obtained only when the filters are of correct design, and 
when they are handled intelligently. The more complete the 
mechanical equipment of the filters, the easier become their 
operation. But no degree of perfection thus far attained in this 
direction can be regarded as a substitute for skillful manipula- 
tion. The latter must be based upon a thorough knowledge of 
the processes employed, and an appreciation of the limitations 
of those processes and of the equipment being used. Excellent 
results maybe obtained with badly designed and poorly equipped 
plants by skilled operators; on the other hand, very low effi- 
ciencies may be obtained in well-designed and equipped plants 
by unintelligent and careless management. 
The proper preparation of the water for filtration is of the 
utmost importance, and this fact can not be too strongly empha- 
sized. Overloading of filters, by the application of water con- 
taining too large an amount of imperfectly coagulated sediment, 
is a frequent cause for the production of poorly purified effluents. 
The fact that the ability of filters to bear overloading may vary 
from time to time makes the most careful observation of operating 
conditions imperative at all times. 
Period of Service of Rapid Sand Filters.-The length of time 
a rapid sand filter may be kept in service is obviously influenced 
by several factors. These are, as previously noted under "Loss 
of Head" in a preceding chapter, the rate of filtration, size of 
sand particles and depth of bed, character of applied sediment, 
entrainment of air within the bed, and the temperature of the 
water. In the gravity type of filter, where the total available 
head is usually in the neighborhood of 12 ft., it is apparent that 
the head may be either slowly or rapidly used up, depending 
upon the factor or combination of factors tending to clog the 
sand bed. Periods of service, therefore, actually vary from 1 hr. 
to over 500 hr., depending upon the particular factors involved. 
347 348 
WA7W PURIFICATION 
If the rate of filtration is uniform, and is at or near the average 
for filters of this class, viz., 125,000,000 gal. per acre per day, the 
period of service will be much shorter, other conditions being 
equal, than if a uniformly low rate or a variable rate below the 
above-mentioned average rate is maintained. Beds of relatively 
fine sand (0.35 mm. effective size and under) will clog quicker 
than beds of coarse sand (0.35 to 0.60 mm. effective size). Deep 
beds or even quite shallow beds, which are washed by the "high- 
velocity method," whereby a preponderance of their finest par- 
ticles are brought to or near the surface, will produce shorter 
periods of service, other things being equal, than shallower beds 
or those composed of coarser particles. 
The chief factor influencing the length of the period of service 
is the character and amount of the suspended matter to be 
strained out of the applied water. As this material varies greatly 
both in quantity and kind in different localities, and even in the 
same locality at different seasons of the year, it causes the greatest 
variation in the service periods of the filters. For example, 
waters containing colloidal clay particles vary greatly at differ- 
ent seasons of the year in the amount of this kind of suspended 
matter. Under flood conditions the quantity in suspension is 
oftentimes very large, and sometimes the particles are very small 
and difficult to coagulate. Under such conditions the periods 
of service must necessarily be short if a clear effluent is to be 
maintained. 
Variable Rates of Filtration.-The relation between a variable 
rate of filtration and the period of service is well illustrated by the 
following table taken from the Report of the Sewerage and Water 
Board of New Orleans for 1914 on the operation of their two 
filtration plants. 
Length 
of run 
in hours 
Corresponding 
net rate of fil- 
tration, M. G. 
per A. per 24 
hr. 
Corresponding 
quantity of 
water filtered, 
gallons 
Corresponding 
per cent, of 
wash water 
used 
Corresponding 
length of run 
at rate of 125 
M. G. per A. 
per 24 hrs. 
Maximum.... 
512.3 
68.6 
46,480,000 
0.2 
278.9 
Minimum 
22.7 
69.3 
2,125,000 
1.9 
12.8 
Average 
129.6 
82.1 
14,170,000 
0.5 
85.0 
Carrollton Filter Plant EFFICIENCY AND COST OF OPERATION 
349 
Length 
of run 
in hours 
Corresponding 
net rate of fil- 
tration, M. G. 
per A. per 24 
hr. 
Corresponding 
quantity of 
water filtered, 
gallons 
Corresponding 
per cent, of 
wash water 
used 
Corresponding 
length of run 
at rate of 125 
M. G. per A. 
per 24 hrs. 
Maximum.. . . 
353.8 
83.4 
5,057,000 
0.2 
236.0 
Minimum 
27.3 
58.5 
280,000 
0.6 
13.5 
Average 
188.2 
88.9 
2,870,000 
0.4 
134.0 
Algiers Filter Plant 
The extremely long periods of service obtained in the opera- 
tion of the filters of these two plants is unusual, and can be ex- 
plained only by the character of the applied sediment, which is 
apparently much more easily strained out of the applied water 
than is generally the case. 
The character of the water handled at the Carrollton plant, 
and the purification results effected, so far as the removal of 
turbidity is concerned, are shown in the following table: 
Turbidities 
Year 
Mississippi River water 
Effluent grit reservoir 
1910 
1911 
1912 
1913 
1914 
1910 
1911 
1912 
1913 
1914 
Parts per million 
Maximum 
1,700 
1,400 
2,200 
1,900 
2,400 
1,450 
1,250 
1,950 
1,800 
2,000 
Minimum 
150 
190 
120 
60 
130 
160 
130 
50 
Average 
550 
500 
750 
675 
600 
450 
425 
775 
600 
575 
Effluent coagulation res. 
Filters 
Year 
1910 
1911 
1912 
1913 
1914 
1910 
1911 
1912 
1913 
1914 
Maximum 
525 
280 
1,700 
650 
425 
1 
0 
5 
5 
3 
Minimum 
2 
5 
5 
2 
0 
0 
0 
0 
0 
0 
Average 
32 
32 
40 
28 
32 
0 
0 
0 
0 
0 
The rapid sand filters of the Cincinnati filtration plant are op- 
erated at all times at a rate of 125,000,000 gal. per acre per day. 
They purify, like the New Orleans filters, a turbid water. Never- 
theless the periods of service are much shorter on an average as 
may be seen from the following table: 350 
WATER PURIFICATION 
Periods of Service and Percentage of Wash Water of Cincinnati 
Filters 
Year 
Yearly average 
Periods of service, hours 
Percentage of wash water 
1909 
22 50 
2 78 
1910 
15.56 
3 50 
1911 
19.14 
3.05 
1912 
20.32 
2.51 
1913 
17.26 
2.65 
1914 
19.50 
2.24 
The maximum and minimum monthly averages of the periods 
of service and of the wash water for the above years for the Cin- 
cinnati filters are as follows: 
Year 
Maximum monthly average 
Minimum monthly average 
Month 
occurred 
Period 
service, 
hours 
Per cent, 
wash 
water 
Month 
occurred 
Period 
service, 
hours 
Per cent, 
wash 
water 
1909 
June 
30.50 
2.42 
August 
14.97 
3.81 
1910 
June 
24.19 
2.49 
August 
7.13 
6.42 
1911 
October 
29.94 
1.96 
August 
13.36 
4 16 
1912 
August 
32.38 
1.34 
March 
10.44 
3.80 
1913 
November.... 
28.17 
1.55 
August 
9.43 
4.59 
1914 
September... . 
35.39 
1.26 
June 
10.12 
3.36 
Effect of Microscopic Plant and Animal Life on Periods of 
Service.-The presence in a water of microscopic plants and ani- 
mals in any considerable amount, and especially at certain per- 
iods of their growth, have a very marked effect upon the periods 
of service of rapid sand filters. The comparatively short life 
cycle of these microorganisms, whereby they multiply with great 
rapidity, live for a few days or weeks and then disappear, is usu- 
ally paralleled in the filter plant purifying the water by shortened 
periods of service of the filters. The clogging effect of these 
organisms is even greater than that of colloidal clay particles, 
although the former condition is less liable to be productive of 
poorly purified effluents than the latter. 
In this connection it may be noted that the shortest periods 
of service of the Cincinnati filters occurred, as a rule, during EFFICIENCY AND COST OF OPERATION 
351 
August. The minimum period of service in 1914 was in June. 
These low averages were caused by the clogging action of micro- 
scopic plants and of other colloidal organic matter. The mini- 
mum period of service in 1912, which occurred in March, was 
due to the high turbidity (360 parts per million) of the Ohio 
River water during that month. However, in August of this 
same year, the maximum period of service was reached, although 
the turbidity of the Ohio River water averaged 385 parts per 
million, or 25 parts per million higher than the average for March. 
The difference in the efficiency of the filters during these 2 
months can only be explained by the difference in the character 
of the applied sediment, and by the condition of the sand of the 
filters. The sediment carried by the flood waters during the 
first 3 or 4 months of the year contains little colloidal organic 
matter; the very fine clay particles are not readily coagulated, 
and have a tendency to pass through the filter bed and produce 
a turbid effluent. On the other hand, the turbid water passing 
down the stream in the summer months carries considerable or- 
ganic matter, which materially assists the sand bed in straining 
out the suspended particles. Moreover, the sand grains acquire 
during the warmer months of the year a heavy gelatinous coating 
which is particularly effective in restraining the passage of finely 
divided clay particles. These differences in the character of the 
sediment and in the condition of the sand bed seem to offer a 
reasonable explanation for the varying efficiencies of the filters. 
The effect of varying amounts of turbidity on the yield of 
filtered water for each foot loss of head is well illustrated in the 
accompanying curve (Fig. 116) kindly furnished the author by 
Mr. H. W. Streeter, which he plotted from data obtained by him 
in the operation of a rapid sand filter plant. The rapid increase 
in the yield after the turbidity of the applied water falls below 40 
parts per million is quite marked. 
Volume of Wash Water Required.-The proportion of the 
total volume of water filtered by a rapid sand filter plant, which is 
used for cleaning the filter beds, varies considerably. The longer 
the periods of service, the less proportionately will be the volume 
of wash water required. At the Carrollton plant in New Orleans, 
the percentage wash water required after one of the filter runs 
was but 0.2 per cent, of the total yield of filtered water for the 
entire period of service. The period of service of this filter was 
512.3 hr. 352 
WA TER P U RIF IC A TION 
The lowest monthly average percentage of wash water used by 
the entire Cincinnati filter plant was 1.26 in November, 1914; and 
the average for the 6 years recorded in the accompanying table 
is approximately 2.8 per cent, (see page 350). The maximum 
percentage of wash water as shown by the monthly averages for 
the Cincinnati plant for the above period was 6.42 in August, 
1910. 
The average percentage of wash water used during 1914 at the 
Turbidity of Applied Water P.P.M. 
Yield in Gals, per Sq. Ft. per Ft. Loss of Head 
Fig. 116.-Average yield per foot loss of head for different turbidities of 
applied water. 
Louisville, Ky., filter plant,1 which like the Cincinnati plant 
purifies Ohio River water, was 3.16; the maximum monthly 
average was 5.48 per cent. (November), and the minimum 1.77 
per cent. (April). 
As an example of the percentage of wash water used in a rapid 
sand filter plant that purifies a water somewhat different in 
character from that of the Ohio River, the following table taken 
from the Report of the Board of Commissioners for the Water 
1 At the Louisville filter plant during 1914 a number of new filters were 
put into service which required considerably more wash wafer than did the 
old filters. The new filters required 4.55 per cent, of wash water and the 
old filters 2.97 per cent.; the average being 3.16 per cent. EFFICIENCY AND COST OF OPERATION 
353 
and Lighting Department of Harrisburg, Pa., for 1912, is of 
interest. 
Year 
Turbidity, 
p.p.m. 
Sulphate of 
aluminum, 
grains per 
gallon 
Length of runs 
Wash water, 
per cent. 
Hours 
Minutes 
1906 
110 
0.95 
18 
20 
2.0 
1907 
76 
1.06 
12 
20 
2.6 
1908 
52 
1.09 
13 
29 
2.6 
1909 
42 
0.84 
14 
22 
2.4 
1910 
19 
0.61 
18 
40 
2.2 
1911 
32 
0.95 
23 
7 
2.3 
1912 
59 
0.77 
21 
54 
2.3 
Average.. 
54 
0.895 
20 
20 
2.34 
At the purification works at Columbus, Ohio, the softening of 
the water is a prominent feature of the process. The following 
table compiled from the 1913 Report of the Division of Water of 
this city shows the relation between turbidity, color, amount of 
applied chemicals and percentage of wash water. 
Turbidity, p.p.m. 
Color, 
p.p.m. 
Chemicals appliec 
per gallon 
■ „ Percent- 
, grains age wasjj 
water 
1913 
River ' Settled 
River 
Filtered 
T- „ 1 Soda 
Llme | ash 
Alum. 
water | water 
water 
water 
sulph. 
Jan 
198 16.0 
34 
7.0 
6.3 2.9 
2.9 2.7 
Feb... . 
32 3.0 
24 
3.0 
7.7 6.0 
1.8 2.1 
Mar.. . . 
280 15.0 
28 
5.0 
6.0 2.5 
2.4 2.3 
Apr.. . . 
178 12.0 
38 
6.0 
6.6 2.6 
2.0 2.4 
May.... 
20 0.0 
21 
2.0 
8.5 5.2 
1.1 1.7 
June... . 
21 0.0 
28 
5.0 
8.9 5.9 
0.9 1.2 
July.... 
18 0.0 
26 
6.0 
8.2 5.1 
0.7 1.2 
Aug. . . 
18 0.0 
29 
9.0 
7.4 5.7 
0.7 1.5 
Sept.... 
16 2.0 
19 
8.0 
7.4 7.5 
0.8 1.6 
Oct 
14 11.0 
16 
5.0 
8.3 8.6 
1.7 1.5 
Nov.... 
20 5.0 
15 
3.0 
8.1 8.8 
1.6 1.6 
Dec.... 
11 7.0 
21 
6.0 
8.2 11.7 
1.7 2.1 
Average yearly results 
1909 
86 
42 
8.7 
7.9 
3.8 
1.76 
4.1 
1910 
37 
6.3 
29 
7.0 
7.6 
5.3 
1.02 
2.3 
1911 
68 
3.2 
28 
7.0 
7.5 
4.3 
1.57 
1.3 
1912 
84 
7.5 
28 
6.4 
7.0 
3.4 
1.90 
2.1 
1913 
69 
6.0 
25 
5.0 
7.6 
6.0 
1.50 
1.8 354 
WA TER P U RIF IC A TION 
Relation of Turbidity of Applied Water to Quantity of Wash 
Water.-The relation between the turbidity of the water applied 
to filters and the percentage of wash water required to cleanse 
them is well brought out by the accompanying curve prepared by 
Mr. H. W. Streeter from data secured during the operation of a 
rapid sand filter plant. His comments on the diagram (Fig. 117) 
are as follows: 
Per cent of Water to Wash Filters 
Turbidity of Applied Water-P.P.M, 
Fig. 117.-Relation between turbidity of applied water and amounts of 
water necessary to wash filter. 
"The curve is based upon a year's daily operating records. The 
time required for the application of the wash water was very carefully 
recorded, and the rate of application was frequently checked by noting 
the rise of the wash water in the filter tank by means of a gage placed 
therein. The rate of application of the wash water was subject to 
very accurate control. 
"The turbidity of the applied water was taken as the mean prevail- 
ing throughout the filter run which terminated with the washing being EFFICIENCY AND COST OF OPERATION 
355 
considered. Usually the variation in the turbidity was slight for the 
comparatively short period involved. 
"The curve is not considered as reliable for turbidities of applied 
water above 50 parts per million as for those below this turbidity. 
Below a turbidity of 50 parts per million there were more observations 
available and these lay along a well-defined line." 
The following data obtained in the operation of the filtration 
plant at Little Falls, N. J., and tabulated by Mr. G. A. Johnson in 
his paper on "Present-day Water Filtration Practice"1 is a good 
illustration of the efficiency of a rapid sand filter plant handling 
for the most part a highly colored water rather than one with 
much turbidity. The bacteria given in the original table are 
omitted. 
1912 
Sulphate 
aluminum, 
grains per 
gallon 
Per cent, 
wash water 
Turbidity 
Parts per million, color 
Average 
Highest 
day 
Average 
Highest 
day 
R 
F 
R 
F 
R 
F 
R 
F 
Jan.... 
0.74 
1.7 
7 
0 
9 
0 
32 
5 
38 
8 
Feb.... 
0.65 
1.8 
13 
0 
30 
0 
35 
6 
53 
11 
Mar.. . 
0.98 
2.7 
26 
0 
118 
0 
52 
6 
92 
15 
Apr 
1.07 
2.8 
8 
0 
13 
0 
41 
6 
46 
9 
May... 
1.46 
3.3 
8 
0 
13 
0 
55 
7 
66 
11 
June... 
1.60 
3.3 
10 
0 
12 
0 
56 
8 
65 
11 
July.. . 
1.65 
3.4 
9 
0 
11 
0 
46 
9 
55 
11 
Aug. 
1.62 
3.4 
8 
0 
17 
0 
43 
9 
52 
10 
Sept.... 
1.47 
3.1 
7 
0 
10 
0 
41 
8 
64 
10 
Oct... . 
1.93 
3.1 
10 
0 
23 
0 
56 
7 
78 
10 
Nov... 
3.16 
3.0 
10 
0 
22 
0 
61 
14 
71 
22 
Dec.... 
1.86 
2.3 
8 
0 
12 
0 
40 
11 
53 
19 
Remarks: R = river water. F - filtered water. 
Mr. Johnson comments as follows on this data: 
"The records given in the last table show that while on one day in 
1912 the turbidity of the water rose to 118 parts per million, the yearly 
average turbidity of the raw water was only 10 parts per million. The 
filtered water was free of turbidity at all times. The color of the un- 
filtered water reached a maximum of 92 parts per million in March, 
while the average for the year was 46 parts. The filtered water color 
averaged 8 parts for the entire year, and here it should be noted that 
at this plant the aim is to reduce the color so that the filtered water shall 
lJour. Am. Water-works Assn., March, 1914. 356 
WA TEH P U RIF IC A TION 
not contain more than 10 or 15 parts. In November and December the 
color of the filtered water was slightly in excess of this upper limit for 
short periods." 
The average amount of sulphate of aluminum used at the Little 
Falls plant during 1912 was 1.51 grains per gallon, or 216 lb. per 
million gallons of water treated. The average percentage of wash 
water was 2.8. The filters are washed with water applied at the 
rate of 8 gal. per square foot per minute, and compressed air is 
used for the agitation of the sand bed. 
Hardness.-The effect of purification upon the dissolved salts 
causing hardness in a water is obviously dependent upon the kind 
and amount of coagulants and softening agents used. The addi- 
tion of sulphate of iron or of aluminum to a water containing 
sufficient alkalinity to decompose them necessarily increases its 
content of sulphate of calcium and of magnesium, and hence its 
permanent hardness. If softening agents, such as lime and soda 
ash are applied to a water containing considerable quantities of 
salts of calcium and magnesium, a reduction in the dissolved 
hardening constituents may be effected. How much change in 
the hardness of the purified water is produced by a number of 
purification plants is shown by the following table given by Air. 
George A. Johnson in the same paper quoted from above.1 
Hardness of River Waters at Various Places Before and After 
Coagulation and Filtration 
City 
Aver- 
age 
for 
year 
Kind 
filter 
Amount 
coag. used, 
grains per 
gallon 
Parts per million 
Increase of 
incrustants 
in filtered 
water due to 
the use of 
coagulants 
Total 
hardness 
Incrustants 
River 
Filt. 
River 
Filt. 
Springfield, Mass.... 
1912 
s 
(d) 0.24 
11 
2 
4 
2 
2 
Little Falls, N. J... 
1903 
R 
(d) 1.38 
31 
30 
7 
14 
7 
Louisville, Ky 
1912 
R 
(d) 1.73 
95 
91 
29 
38 
9 
Cincinnati, Ohio.. . 
1910 
R 
(a) 0.84 
(6) 1.79 
76 
89 
32 
41 
9 
New Orleans, La.. . 
1912 
R 
(g) 4.41 
(6) 0.33 
(a) 7.50 
111 
60 
21 
25 
4 
Columbus, Ohio... . 
1910 
R 
(c) 4.30 
270 
85 
111 
35 
Softened 
(d) 1.57 
Note.-R designates rapid sand filters and S slow sand filters, (a) 
Lime; (5) Sniphate of iron; (c) Soda ash; (d) Sulphate of aluminum. 
1 G. A. Johnson: "Present-day Water Filtration Practice." Jour. Am. 
Water-works Assn.,^March, 1914. EFFICIENCY AND COST OF OPERATION 
357 
At the first four purification plants listed in this table, the 
use of chemicals is simply and only for the purpose of coagula- 
tion in order to prepare the water for filtration. At the last two 
plants, however, chemicals are used to effect a softening action 
in conjunction with those applied for coagulation purposes. 
Bacterial Purification.-The efficiency of rapid sand filter 
plants in the removal of bacteria ranges from 90 per cent, to 
over 99 per cent, in well-operated plants. With few bacteria 
in the applied water, the percentage removal will not be as high 
as when the number is large. With large numbers of bacteria 
in the applied water, however, the number of bacteria actually 
passing through the filter bed may be and generally is much 
higher than where the number of bacteria in the applied water 
is low. 
The following tables taken from the reports of the water de- 
partment of Harrisburg, Pa., and Cincinnati, Ohio, indicate the 
bacterial content of the water at these two purification plants in 
various stages of the process, and the percentages removed. 
Year 
Bacteria per cubic 
centimeter 
Percentage Removals by 
River 
water 
Settled 
water 
Filtered 
water 
Coagula- 
tion 
basin 
Filters 
Total 
1906 
12,372 
3,228 
94 
73.91 
97.06 
99.24 
1907 
10,710 
4,504 
44 
57.96 
99.03 
99.59 
1908 
4,949 
1,662 
19 
66.43 
98.86 
99.62 
1909 
5,762 
1,083 
17 
80.87 
98.28 
99.70 
1910 
7,843 
62 
5 
99.02 
91.94 
99.94 
1911 
10,357 
58 
7 
99.04 
87.93 
99.94 
1912 
5,115 
35 
2 
99.32 
94.29 
99.97 
Average 
8,239 
1,519 
27 
82.36 
95.34 
99.71 
Harrisburg Filtration Plant 
Note.-Calcium hypochlorite has been used since late in 1909, and is 
applied to the raw water as it enters the settling basin. This use of a dis- 
infectant somewhat obscures the real efficiency of the plant so far as the 
effect of the coagulating chemicals and the filters are concerned; but results 
prior to 1909 are not thus affected. 358 
WATER PURIFICATION 
Average number of bacteria per cubic centimeter 
1908 
1909 
1910 
1911 
1912 
1913 
1914 
River water 
7,000 
9,300 
8,900 
13,790 
11,130 
16,100 
16,500 
Settled water 
3,400 
2,500 
3,200 
3,140 
3,945 
2,965 
3,570 
Applied water 
700 
475 
750 
776 
898 
465 
670 
Filtered water 
100 
75 
75 
39 
26 
56 
36 
Average percentage removal 
98.5 
99.2 
99.2 
99.7 
99.8 
99.7 
99.8 
Cincinnati Filter Plant 
Note.-Disinfectant used in filtered water during the first 3 to 6 months 
of each year beginning in 1911, but was not used prior to that year. 
Some very complete data obtained in operating the New 
Orleans (Carrollton) filter plant are given below. 
River 
Effluent grit reservoirs 
1910 
1911 
1912 
1913 
1914 
1910 
1911 
1912 
1913 
1914 
Maximum 
Minimum 
Average 
41,000 
275 
4,200 
8,500 
95 
2,400 
32,000 
400 
3,300 
20,000 
100 
1,900 
9,000 
400 
1,800 
50,000 
140 
3,500 
8,500 
40 
2,000 
19,000 
475 
3,400 
30,000 
230 
1,700 
8,500 
190 
1,700 
Effluent coag. reservoirs 
Filters 
1910 
1911 
1912 
1913 
1914 
1910 
1911 
1912 
1913 
1914 
Maximum 
Minimum 
Average 
11,000 
12 
1,500 
7,000 
5 
400 
12,000 
23 
1,000 
11,000 
10 
350 
4,000 
13 
190 
1,300 
1 
(a) 80 
(b) 26 
3,000 
2 
la) 90 
(c) 38 
3,200 
2 
(a) 140 
(c) 40 
550 
1 
21 
325 
1 
19 
Bacteria Per Cubic Centimeter 
(a) For entire period. 
(b) Excluding March and April. 
(c) Excluding period of abnormal counts due to algae growths, and use of hypochlorite 
of lime. 
Efficiency in the Removal of the Bacillus Coli.-The identi- 
fication of fecal bacteria in a water has become a part of the rou- 
tine practice in the laboratories of water-purification plants; and 
the results obtained by tests of the raw and filtered waters are 
regarded as a measure of the efficiency of the purification process. 
In the absence of direct methods for enumerating organisms of 
this class, more or less indirect methods must be used. The EFFICIENCY AND COST OF OPERATION 
359 
smaller the sample of water which gives a positive test indicating 
the presence of these organisms, the greater is its degree of 
pollution. On the other hand, the larger the sample in which 
positive tests are obtained, the less likelihood is there that it con- 
tains pathogenic organisms which may have been associated 
with the fecal bacteria. The tests for B. coli, therefore, may be 
used, if properly interpreted, to indicate the purity of the water 
tested, and hence may be made a measure of the efficiency of the 
purification processes. 
Some data obtained in the operation of the filter plant at 
Louisville, Ky., during 1914, is of interest in this connection. 
Out of 2,190 samples of Ohio River water tested 1,762, or 80.5 
per cent., gave positive evidence of the presence of the B. coli. 
Out of 2,195 samples of filtered water only 26, or 1.2 per cent., 
gave positive tests for this organism. These results were pre- 
sumably obtained by testing 1-c.c. samples. Disinfectants were 
used 130.5 days during the year, and hence the figures given 
include the effect of both filtration and disinfection. 
At the Columbus, Ohio, filtration and water-softening plant 
during the year 1913, 204 1-c.c. samples of the raw water gave a 
positive test for B. coli; and 2, 14 and 19 samples of the filtered 
water of 1-c.c., 10-c.c. and 50-c.c., respectively, also gave posi- 
tive tests. The filtered-water results, the same as in the Louis- 
ville filter plant data, include the effect of the use of disinfectants. 
The following table taken from the 1912 Report of the Board 
Table Showing the Percentage oe Positive Tests for B. Coli-com- 
MUNIS IN THE UNFILTERED, FILTERED AND TAP WATERS AT 
Harrisburg, Pa., Filter, Plant 
Year 
Unfiltered 
Filtered 
Tap 
1906 
71.87 
2.73 
4.95 
1907 
64.02 
1.02 
2.64 
1908 
65.72 
1.07 
1.62 
1909 
63.15 
1.00 
1.60 
1910 
55.44 
0.17 
0.82 
1911 
77.30 
0.61 
1.46 
1912 
46.90 
0.81 
2.81 
Average for 7 
years 
63.49 
1.06 
2.27 
Note.-These results are all on 1-c.c. samples. Disinfection of the co- 
agulated water has been practised since late in 1909. 360 
WATER PURIFICATION 
The following table gives the total number of tests, the number and percentage of positive tests on various 
quantities, and the B. coli index of the river and of the filtered water: 
Month, 1914 
River water 
Filtered water 
Bacillus coli 
B. coli 
index 
No. 
per 
c.c. 
Bacillus coli 
B. coli 
index 
No. 
per 
c.c. 
No. of 
test 
days 
0.01-c.c. tests 
0.1-c.c. tests 
1.0-c.c. tests 
No. of 
test 
days 
1.0-c.c. tests 
20.0-c.c. tests 
Total 
No. 
No. 
+ 
Per 
cent. 
+ 
Total 
No. 
No. 
+ 
Per 
cent. 
+ 
Total 
No. 
No. 
+ 
Per 
cent. 
+ 
Total 
No. 
No. 
+ 
Per 
cent. 
+ 
Total 
No. 
No. 
+ 
Per 
cent. 
+ 
January 
31 
31 
3 
9.68 
31 
19 
61.29 
31 
31 
100.00 
16.21 
31 
31 
0 
0.00 
31 
1 
3.23 
0.00281 
February 
28 
28 
3 
10.71 
28 
15 
53.57 
28 
28 
100.00 
15.46 
28 
28 
0 
0.00 
28 
1 
3.57 
0.00476 
March 
31 
31 
1 
3.23 
31 
15 
48.39 
31 
31 
100.00 
8.26 
31 
31 
0 
0.00 
31 
0 
0.00 
0.00032 
April 
30 
30 
3 
10.00 
30 
16 
53.33 
30 
30 
100.00 
14.80 
30 
30 
0 
0.00 
30 
0 
0.00 
0.00078 
May 
31 
31 
5 
16.13 
31 
15 
48.39 
31 
29 
93.55 
19.81 
31 
31 
0 
0.00 
31 
7 
22.58 
0.01516 
June 
30 
30 
' 1 
3.33 
30 
4 
13.33 
30 
24 
80.00 
5.00 
30 
30 
0 
0.00 
30 
1 
3.33 
0.00589 
July 
31 
31 
1 
3.23 
31 
11 
35.48 
31 
20 
65.42 
6.75 
31 
31 
0 
0.00 
31 
1 
3.23 
0.00484 
August 
31 
31 
9 
29.03 
31 
16 
51.61 
31 
29 
93.55 
31.71 
31 
31 
1 
3.23 
31 
6 
19.35 
0.04488 
September 
30 
30 
3 
10.00 
30 
12 
40.00 
30 
26 
86.67 
13.47 
30 
30 
0 
0.00 
30 
6 
20.00 
0.01567 
October 
31 
31 
6 
19.35 
31 
18 
58.06 
31 
29 
93.55 
23.58 
31 
31 
2 
6.45 
31 
16 
51.61 
0.08945 
November 
30 
30 
0 
0.00 
30 
1 
3.33 
30 
20 
66.67 
0.97 
30 
30 
0 
0.00 
30 
6 
20.00 
0.01611 
December 
31 
31 
4 
12.90 
31 
28 
90.32 
31 
31 
100.00 
20.74 
31 
31 
1 
3.23 
31 
26 
87.10 
0.07531 
Total 
365 
365 
39 
365 
170 
365 
328 
365 
365 
365 
89.88 
4 
71 
19.50 
0.0230 
Average 
10.63 
46.43 
100.00 
14.73 
100.00 
100.00 
1.10 
1.08 
100.00 
19.45 
Per cent. time.. 
100.00 
100.00 
10.69 
100.00 
46.58 
89.87 
Note.-Positive presumptive tests of 1-c.c. samples were confirmed in the usual way: all the results of other tests are presumptive, and were obtained 
with lactose bile. 
From Jan. 1 to May 27, 438 samples of filtered water showed 92.69 per cent, of tests to be positive in 100 c.c. before disinfection, and 23.97 per cent, 
positive in 100 c.c. after disinfection. EFFICIENCY AND COST OF OPERATION 
361 
of Commissioners of Harrisburg, Pa., shows the percentage of 
positive B. coli tests obtained in the raw and filtered waters and 
in the tap water for a period of 7 years. 
The data obtained in the laboratories of the Cincinnati, Ohio, 
filtration plant as to the presence or absence of B. coli, are given 
in some detail in the Report of the Water Department of Cin- 
cinnati for the year 1914. 
The following table shows the numbers of bacteria per cubic 
centimeter and the percentage of positive B. coli tests on vari- 
ous quantities of filtered water before and after the application 
of 0.1 part per million of chlorine: 
Month, 1914 
Bacteria per 
c.c. 
Per cent, positive B 
coli 
1.0 c.c. 
20.0 c.c. 
100.0 c.c. 
Before 
After 
Before 
After 
Before 
After 
Before 
After 
January 
300 
26 
3.23 
0.00 
58.06 
3.23 
94.62 
16.13 
February.. . . 
370 
87 
0.00 
0.00 
57.14 
3.57 
94.05 
33.33 
March 
305 
39 
0.00 
0.00 
12.90 
0.00 
77.53 
3.23 
April 
80 
13 
16.67 
0.00 
83.33 
0.00 
97.78 
7.78 
* May 
75 
22 
14.81 
0.00 
70.37 
18.52 
100.00 
55.55 
Average 
226 
37 
6.94 
0.00 
56.36 
5.06 
92.80 
23.20 
* Chlorine applied from Jan. 1 to May 27, 1914. 
Percentage removal of bacteria by disinfection, 83.63. 
The general significance to be attached to these figures for B. 
coli, when used as a definite measure of purity of a water will be 
discussed in a succeeding chapter. 
COST OF OPERATION OF RAPID SAND FILTER PLANTS 
The cost of operating a rapid sand filter plant is obviously 
dependent upon the completeness of its equipment, upon the 
character of the water being treated, and on the skill with which 
the plant is handled. A poorly equipped plant will require more 
men to operate it,than one well provided with labor-saving de- 
vices. A water which requires a large amount of chemicals to 
properly purify it is necessarily expensive to treat. Unintelligent 
management of both men and material is productive of high costs. 
The cost for labor and chemicals constitute the largest part of the 
total cost of operation and maintenance. 
Mr. G. A. Johnson1 has tabulated the operation and mainte- 
1 "Present-day Water Filtration Practice," Jour. Am. Waler-works Assn., 
March, 1914. 362 
WATER PURIFICATION 
nance costs for a number of both slow sand and rapid sand filtra- 
tion plants. 
Year 
City 
Kind of 
filter 
Average volume 
of water 
filtered daily, 
gallons 
Cost of operation 
and maintenance 
per million gal- 
lons of water 
filtered 
1911 
Albany, N. Y. 
Slow sand 
20,000,000 
$2.50 
1912 
Pittsburgh, Pa. 
Slow sand 
100,000,000 
3.41 
1911 
Philadelphia, Pa. 
Slow sand (a) 
9,000,000 
5.62 
1911 
Philadelphia, Pa. 
Slow sand (6) 
13,000,000 
3.59 
1911 
Philadelphia, Pa. 
Slow sand (c) 
38,000,000 
3.88 
1911 
Philadelphia, Pa. 
Slow sand (d) 
202,000,000 
1.91 
1912 
Washington, D. C. 
Slow sand 
62,000,000 
4.01 
1912 
Cincinnati, Ohio 
Rapid sand 
50,000,000 
4.12 
1911 
Harrisburg, Pa. 
Rapid sand 
9,000,000 
3.93 
1912 
Little Falls, N. J. 
Rapid sand 
30,000,000 
3.20 
1912 
Louisville, Ky. 
Rapid sand 
25,000,000 
3.48 
1912 
New Orleans, La. 
Rapid sand 
16,000,000 
6.32 
Weighted average, 
Slow sand 
$2.86 
Weighted average, 
Rapid sand 
4.04 
(a) Lower Roxborough; (6) Upper Roxborough; (c) Belmont; (d) Torresdale. 
As an illustration of the effect upon costs of operation, where a 
difficult water is to be purified, the detailed figures given in the 
1913 Report of the Division of Water of Columbus, Ohio, are of 
interest. 
Total cost 
Cost per million 
gallons of water 
purified 
Per cent, of 
total cost 
Labor 
Repairs to lands and 
buildings 
Repairs to machinery 
Repairs to other equipment 
Office and laboratory 
supplies 
Heavy chemicals 
General supplies 
Oil, waste and machine 
supplies 
Miscellaneous 
$23,760.13 
260.07 
32.00 
1,016.33 
1,571.07 
69,999.06 
2,707.74 
169.71 
1,411.44 
$3.78 
11.17 
23.5 
69.4 
Total 
$100,927.55 
$16.08 
Note.-Cost per million gallons based on 6,276,675,000 gal. of water 
delivered for consumption. EFFICIENCY AND COST OF OPERATION 
363 
The cost of labor and chemicals constitute 92.9 per cent, of the 
total cost, leaving but 7.1 per cent, for all other purposes. 
The Carrollton plant costs at the New Orleans filter plant for 
the year 1914 are also given in much detail, and are as follows: 
CARROLLTON 
Operation 
1. Labor, attendance and supervision 
Total cost 
$23,319.68 
Cost per 
million 
gallons 
$2.86 
2. Labor-unloading, crushing and storing 
chemicals 
1,510.25 
0.19 
3. Lime 
14,745.93 
1.81 
4. Iron 
2,532.96 
0.31 
5. Supplies, tools, petty cash, car fare, 
telephone, ice, stamps, bond premiums, 
etc 
1,678.82 
0.21 
6. Machinists' labor furnished by pumping 
station 
114.95 
0.01 
7. Labor and material furnished by pump- 
ing station for power, heating and 
lighting 
1,300.00 
0.16 
Total cost of operation 
$45,202.59 
$5.55 
Credit for sale of filling and sand 
465.08 
0.06 
J Net cost of operation 
$44,737.51 
$5.49 
Betterments and Additions: 
1. Labor (includes $582.97 machinists' labor) 
$ 3,179.27 
2. Material 
1,267.06 
$ 4,446.33 
Care and Maintenance: 
Park and Grounds 
1. Labor-care of grounds, etc $ 3,879.90 
2. Labor-watchman 894.40 
3. Labor and material furnished by pump- 
ing station for lighting 550.00 
4. Material and supplies 116.25 
$ 5,440.55 
Betterments and Additions: 
1. Labor $ 205.15 
2. Material 119.00 
$ 324.15 364 
WA TER P U RIF IC A TION 
"All figures for cost are exclusive of interest and depreciation charges. 
"All figures above for cost of purification per million gallons are 
based on actual quantity of water treated during year, as shown by 
corrected Venturi meter readings, viz., 8,147,000,000 gal., and are 
exclusive of all charges for high- and low-lift pumping. 
"Cost of wash water, at $12.75 per million gallons used, based on 
amount of water pumped by high-lift pumps, including all pumping 
costs, was 7 cts. per million gallons of water filtered. 
"Cost of all water for cleaning reservoirs was 3.6 cts. per million 
gallons of water filtered, excluding treated water wasted in draining 
reservoirs for cleaning. 
"This makes the net cost of filtered water pumped $12.86 per million 
gallons. 
"Total gross cost of delivering filtered water to the distribution 
system, at the plant, exclusive of interest and depreciation charges, 
based on quantity actually delivered, viz., 8,010,000,000 gal., was $13.06 
per million gallons, to cover all costs of pumping and purification, or 
$14.33 per million gallons, if the betterments and care and maintenance 
of the park and grounds, etc., are included." 
The purification plants at New Orleans and at Columbus, Ohio, 
not only clarify but soften the water. The cost of softening is 
high compared with the cost for clarification. The two plants 
on the Ohio River at Louisville, Ky., and Cincinnati, Ohio, are 
better examples of plants which clarify and filter the water with- 
out attempting to soften it. 
Total cost 
Cost per million 
gallons 
Superintendence and laboratory pay roll.... 
$ 4,980.00 
$0.50 
Filter operators' pay roll 
8,250.00 
13,324.31 
2,679.23 
2,095.96 
426.77 
0.83 
Coagulant 
1.35 
Wash water 
0.27 
Heat, light and power 
0.21 
Supplies 
0.04 
Repairs 
353.17 
0.04 
Incidentals 
788.00 
0.08 
Total 
$32,897.44 
$3.32 
Cost of Filter Operation in 1914 at Louisville, Ky. 
Note.-Volume of filtered water on which above calculations of cost 
per million gallons are based, was 9,892,887,622 gal. EFFICIENCY AND COST OF OPERATION 
365 
A condensed table of the cost per million gallons for operating 
and maintaining the Cincinnati filtration plant for 7 years, based 
upon the volume of water delivered for consumption (not the 
total water filtered), is as follows: 
Year 
1908 
1909 
1910 
1911 
1912 
1913 
1914 
Average for 
7 years 
Operation 
$4.19 
4.17 
3.91 
3.77 
3.46 
3.44 
2.99 
$3.70 
Maintenance.. 
0.05 
0.09 
0.28 
0.35 
0.38 
0.48 
0.39 
0.29 
Total 
$4.24 
4.26 
4.19 
4.12 
3.84 
3.92 
3.38 
$3.99 
Of these total costs per million gallons there were expended 
the following amounts for coagulating chemicals. 
Year 
1908 
1909 
1910 
1911 
1912 
1913 
1914 
Average for 
7 years 
Cost per M. G. 
Percentage of 
Total cost... 
$1.72 
40.6 
1.89 
44.4 
1.93 
46.0 
1.86 
45.1 
1.78 
46.4 
1.67 
42.6 
1.21 
35.8 
$1.72 
43.1 
The cost of operating the Harrisburg, Pa., filter plant during 
the year 1912, as given in the yearly report of the Water Depart- 
ment, is quoted below. 
" The operating expenses were $17,865.50, coagulant costing $4,390.95; 
coal, $1,171.09; supplies, $1,377.57; material and repairs, $1,687.10; oil 
and waste, $163.07; chemist and laboratory, $1,212.38; salaries, 
$7,854.59; and boiler repairs, $8.75. This makes the cost per million 
gallons $5.53, coagulant costing, $1.35; coal, 36 cts.; supplies, 42 cts.; 
repairs, 52 cts.; oil and waste, 5 cts.; chemist and laboratory, 37 cts.; 
and salaries, $2.46. 
The fixed charges for interest, sinking fund and State tax on the 
filtration loan was $22,423.33, a cost per million gallons of $6.94. 
Fixed Charges.-Beside the cost of operating and maintaining 
a filter plant, there should properly be charged against it the 
interest on the amount of capital invested and a reasonable charge 
for depreciation. The ordinary operation charges will range 
from $4 to $6 per million gallons, and all other charges, including 
interest, sinking fund and depreciation will add from 75 to 125 
per cent, to the above costs. The whole cost of filtration will, 
therefore, probably range from $7 to $14 per million gallons. 366 
WATER PURIFICATION 
1. Reports of the Sewerage and Water Board of New Orleans, La. 
2. Reports of the Water Department of Cincinnati, Ohio. 
3. Reports of the Division of Water, Columbus, Ohio. 
4. Reports of the Board of Water and Light Commissioners, Harrisburg, 
Pa. 
5. Reports of the Louisville Water Co. 
6. G. A. Johnson: "Present-day Water Filtration Practice." Jour. 
Am. Water-works Assn., March, 1914. 
7. H. P. Letton: "Operation of Rapid Sand Filters." Municipal 
Journal, 1914. 
8. George W. Fuller: "The Proper Operation of Filters." Eng. 
Record, vol. 58, p. 498. 
References CHAPTER XXVIII 
DISINFECTION OF WATER SUPPLIES 
The removal from a water of disease-producing organisms, as 
well as those that are harmless, by the methods described in the 
preceding chapters, is accomplished largely by settling and by 
mechanically straining or filtering them out of the water. The 
pathogenic organisms which by chance pass through a filter may 
or may not be injurious from a sanitary point of view, since their 
ability to infect depends in a great measure upon their number 
and virulency. Those water supplies which at present appear 
to be only slightly polluted, and which are still used unfiltered, 
may be at times dangerous to consumers through the presence 
temporarily of pathogenic forms. The direct application of 
germicidal agents to a drinking water, therefore, has always 
appealed to sanitarians as a prophylactic measure. 
Historical.-Disinfection and even sterilization may, of course, 
be effected by boiling the water, but is virtually an impractical 
method on account of its cost where any considerable quantity 
of water must be treated. For this reason other methods of dis- 
infection have been sought. 
The application of chemical compounds to water and sewage 
for disinfecting purposes has received considerable study at 
different times during the past 50 or 60 years. In these in- 
vestigations the halogens and their derivatives, compounds of 
copper, permanganates, caustic lime, and other agents capable 
of exerting an oxidizing action, have each and all been used 
from time to time on both water and sewage. It is only within 
the last decade, however, that the practice of disinfecting water 
supplies with chemicals has become quite general. The agent 
now commonly employed is chlorine, either in the gaseous form, 
or as calcium hypochlorite or bleaching powder. 
For the disinfection and deodorization of sewage, English in- 
vestigators employed calcium hypochlorite or bleaching powder 
as early as 1854. It was also recommended by the American 
Public Health Association in 1885 as the most efficient and cheap- 
est disinfectant available. Certain German scientists also en- 
367 368 
WA TER P U RIF IC A TION 
dorsed its use. As early as 1892, the efficacy of the hypochlorites 
as water sterilizing agents was quite well known as the subject 
had received considerable laboratory study. 
Bleaching powder was first applied to a public water supply 
in Maidstone, England, in 1897, following a typhoid fever epi- 
demic. It was used to disinfect the distribution pipe lines of the 
city. Later in 1904 and 1905, it was again employed at Lincoln, 
England, for a like purpose. In 1896, Janies Hargreaves of 
Liverpool, England, in a paper read before the Liverpool Poly- 
technic Society, discussed the disinfection of sewage by means 
of chlorine manufactured by the electrolysis of salt. In 1906-07, 
Earle B. Phelps in this country experimented with chlorine and 
its compounds for the disinfection of sewage and water. In 
1910, Major C. R. Darnall, U.S.A., made a series of experiments, 
using chlorine gas for the disinfection of drinking water. The 
development of apparatus for the practical application of chlorine 
gas has all taken place within the last 5 years. 
The investigations of Johnson and Leal on the continuous ap- 
plication of small quantities of bleaching powder to the water 
supply of Jersey City in 1908, and its successful employment in 
the same year on the excessively polluted water of Bubbly Creek 
at the purification plant in the stock yards of Chicago, established 
its value in this country, and marks the beginning of its general 
adoption for the disinfection of public water supplies. 
From a comparatively recent series of questions sent out by 
Mr. Francis F. Longley,1 it appears that in the United States over 
2,000,000,000 gal. of water per day are treated with either bleach- 
ing powder or chlorine gas. Four out of every five plants were 
using bleaching powder for disinfecting the water, while the re- 
mainder used liquid chlorine. In about 75 per cent, of the plants 
listed, the water supply was drawn from rivers, 20 per cent, from 
lakes, and the balance from the ground. 
CHLORINE AND ITS COMPOUNDS 
The cheapest and most easily obtained compound of chlorine 
is calcium hypochlorite or bleaching powder. As a byproduct 
of the soda-ash industry in Europe and of the electrolytic 
caustic-soda manufacturing process in this country, it may be 
prepared at a low cost. The use of bleaching powder in the 
1 Rep. Com. on Water Supplies. Am. Jour. Public Health, Dec. 4, 1914. DISINFECTION OF WATER SUPPLIES 
369 
paper and textile industries creates a continuous demand for this 
chemical. 
Preparation and Composition of Bleaching Powder.-Bleach- 
ing powder is prepared by passing chlorine gas over slaked lime. 
The latter absorbs and combines with the gas, forming a mixed 
salt in which calcium is united to two different acids. The com- 
mercial article is far from being pure as the following analyses 
will show: 
Per cent. 
Calcium oxychloride,1 CaOCl2 64.93 
Calcium chloride, CCal2 1.28 
Calcium chlorate, Ca(C10s)2 0.34 
Calcium hydroxide, Ca(0H)2 19.64 
Calcium carbonate, CaCOs 1.51 
Calcium sulphate, CaSO^ 0.25 
Oxides of Na, K, Mg, Al, Fe and Si 2.52 
Moisture 9.95 
100.42 
Per cent. 
1 2 
Available chlorine2 37.00 38.30 
Chlorine as chlorides 0.35 0.59 
Chlorine as chlorates 0.25 0.08 
Lime 44.49 43.34 
Iron oxide 0.05 0.04 
Magnesia 0.40 0.31 
Alumina 0.43 0.41 
Carbon dioxide 0.18 0.31 
Silica, etc 0.40 0.30 
Water and loss 16.45 16.33 
100.00 100.00 
It is evident that commercial bleaching powder is a mixture of 
chemical compounds, but that its chief active constituent is a 
mixed salt having the composition CaOCl2. The activity of the 
compound is expressed in terms of the percentage of "available 
chlorine" that it contains, and which ranges from 30 to 37 per 
cent. The powder absorbs moisture gradually. The dry powder 
weighs about 49 lb. per cubic foot. It is not readily soluble in 
water (by weight 1 part in 20 of water), and solutions should 
not be made stronger than 0.5 to 1.0 per cent., if difficulties with 
Charles G. Hyde: "The Sterilization of Water Supplies by the Use 
of Hypochlorites." Pamphlet, 1911. 
2Lunge's "Sulphuric Acid and Alkali." Vol. 3, p. 642. 370 
WATER PURIFICATION 
sludge are to be avoided. Exposure to the air causes the com- 
pound to lose strength. Its offensive odor, corrosive properties 
and liability to deterioration on exposure, have been the chief 
objections to its use. 
Sodium Hypochlorite.-By the electrolysis of a sodium chloride 
solution, sodium hypochlorite (NaOCl) is formed when the prod- 
ucts of the electrolysis are not separated but are allowed to 
react upon each other. The hypochlorite solution thus produced 
is rather unstable and can not be transported without serious 
loss of strength. The cells in which electrolysis may be carried 
on are simple, and where the electric current can be obtained at a 
sufficiently low figure, the cost of the solution is not high. The 
control of the preparation of the solution, however, is not as 
easy as might at first appear. Inability to store the solution, on 
account of its unstable character, and its higher cost as compared 
to bleaching powder, have prevented its coming into practical 
use in spite of the considerable amount of experimental work 
done to develop it for water-sterilization purposes. 
Liquid Chlorine.-The chlorine gas produced by the electrolysis 
of sodium chloride in the manufacture of caustic soda is used in 
large quantities to produce bleaching powder, as heretofore 
stated. Of late years, however, an outlet for more or less of this 
gas in the liquefied form has been found. The purified gas is 
dried and liquefied by pressure, and is put on the market in 
steel drums or cylinders holding from 100 to 140 lb. each. The 
chlorine is very pure, containing no impurities except traces of 
oxygen, carbon dioxide and nitrogen. It is free from moisture. 
The pressure in the drums varies with the temperature, and 
ranges from 54 lb. per square inch at 32°F., to 216 lb. atl22°F. 
Chlorine gas is dangerous to inhale because of its corrosive action 
on the mucous membrane of the throat and lungs. Great.care 
should be exercised in dealing with it, and especially when it is 
under pressure. Leaky fittings or apparatus should be repaired 
immediately. Storage of cylinders containing the gas under 
pressure should be kept in as cool a place as possible. Supply 
tanks and apparatus for measuring and conveying the gas should 
be placed, if possible, in rooms maintained at a temperature of 
not much less than 50°F., since the conversion of the liquid 
chlorine to the gaseous form in the supply tanks requires heat, 
which is, of course, drawn from the metal container in which the 
liquid stands. The more rapid the evaporation of the liquid DISINFECTION OF WATER SUPPLIES 
371 
chlorine, the greater the lowering of its temperature, and con- 
sequently the lower the initial pressure of the gas being formed. 
This is of practical importance in adjusting pressure reducing 
valves in order to maintain a uniform flow of the gas. 
CHEMISTRY OF THE ACTION OF CHLORINE AND ITS COMPOUNDS 
IN WATER 
Bleaching powder may be regarded as a calcium-chloro- 
hypochlorite, or better as a mixed salt in which the two acids, 
hydrochloric and hypochlorous, have the common base calcium 
to neutralize them. This may be graphically shown as follows: 
/C1 
Ca 
N)C1. 
When this compound is placed in solution in water, it under- 
goes a molecular rearrangement, and appears to become a true 
hypochlorite, plus calcium chloride (2CaOCl2 = Ca(OCl)2 + 
CaCl2). It has been found that aqueous solutions of bleaching 
powder show the presence of Cl, CIO and Ca ions. 
When a bleaching powder solution is applied to a natural water 
containing the bicarbonates of calcium and magnesium, the 
following reaction takes place: 
Ca(OCl)2 + CaCO3 + H2CO3 2CaCO3 + 2H0C1. 
The hypochlorous acid being a weak acid has no appreciable 
effect on the calcium carbonate formed. This acid, however, 
will decompose into hydrochloric acid and oxygen with libera- 
tion of a considerable amount of energy in the form of heat. 
This occurs whether the acid is present alone in the water, or 
with other substances. If some oxidizable substance is present, 
the decomposition of the hypochlorous acid is rapid, and the 
energy of the reaction accounts sufficiently for the destructive 
effect produced on living cells like those of the microorganisms. 
If chlorine gas is applied directly to a water, the following 
reaction occurs: 
Cl2 + H2O HC1 + H0C1. 
This reaction being strongly reversible permits of the formation 
of but small quantities of H0C1 at any time, and requires the 
removal of one or both of the products of the reaction in order 
to bring it to completion. This occurs in two ways in a natural 372 
WA TEH P U RIF IC A TION 
water which is slightly alkaline from the presence of calcium and 
magnesium carbonates, and contains oxidizable organic matter as 
well. The HC1 reacts with the carbonates to form calcium chlo- 
ride and carbonic acid. The H0C1 is decomposed into HC1 and 
oxygen, which latter acts upon any oxidizable matter that may 
be present. 
(a) CaCO3 + 2HC1-»CaCl2 + H2CO3 
(6) 2H0C1->2HC1 + O2. 
Equation (b) represents the liberation of oxygen from hypo- 
chlorous acid, which latter is formed when either a hypochlorite 
or chlorine is dissolved in water. The energy liberated by the 
decomposition of the hypochlorous acid, as previously stated, 
explains the powerful oxidizing action of the evolved oxygen, and 
the destructive effect upon the microorganisms. Chlorine or the 
hypochlorites are, therefore, merely agents for the production of 
oxygen under conditions which render it extremely active. 
The term "available chlorine" in hypochlorites refers to the 
oxygen-releasing value of the chlorine in these compounds. 
From an examination of the reaction between chlorine gas and 
water, it will be noted that 2 atoms of chlorine are required to 
decompose 1 molecule of water and set free its atom of oxygen; or 
in other words, that 16 parts by weight of oxygen have been 
liberated by 70.9 parts of chlorine. This is in the ratio of 1.00 
part by weight of oxygen to 4.43 parts by weight of chlorine, 
It will also be seen that pure hypochlorous acid or a pure hypo- 
chlorite, when decomposed, liberate but 1 atom of oxygen for each 
atom of chlorine, or double the quantity liberated when pure 
chlorine reacts with water. Solutions of the hypochlorites are, 
however, always mixtures, containing 1 molecule of the chloride 
for each molecule of the hypochlorite; viz., CaCl2, Ca(C10)2 or 
NaCl, NaClO. The active or oxygen-releasing chlorine in the 
hypochlorites, therefore, bears the same relation to the total 
chlorine that it does in solutions of the pure element in water. 
The "potential oxygen," as it is sometimes called, which is 
produced by the decomposition of bleaching powder, forms from 
7 to 8 per cent, of the weight of the latter. On the other hand, 
where pure chlorine gas is used, it produces theoretically 22.5 per 
cent, of its weight of "potential oxygen." Various losses in 
actual practice lower the proportion of "potential oxygen" pro- 
duced in using either the hypochlorites or pure chlorine gas. DISINFECTION OF WATER SUPPLIES 
373 
Application of Chlorine and Its Compounds.-The general 
methods employed for the preparation and application of bleach- 
ing-powder solutions are described in more or less detail in Chap- 
ters XXI and XXII. As will be noted in reading these chapters, 
aside from providing for the corrosive properties of the solution, 
and for its offensive odor, the tanks and orifice apparatus may be 
similar to or the same as those used for solutions of other chem- 
icals that are commonly prepared and applied to water in purifi- 
cation processes. 
Sodium Hypochlorite Apparatus.-The probability that sodium 
hypochlorite will ever come into use as a disinfecting agent for 
Fig. 118.-Cell for electrolyzing a common salt solution. 
public water supplies is slight, in spite of the considerable experi- 
mental work which has been carried on to determine the best 
methods for controlling its production, storage and application. 
It has one marked advantage over the use of bleaching powder in 
not having to dispose of an offensive sludge. Its preparation is 
relatively simple, but the control of the strength of the solution is 
not particularly easy, nor can any storage of the liquid be made, 
except for relatively short periods, without a considerable de- 
crease in strength. 
The apparatus for preparing sodium hypochlorite consists of a 
tank for holding a solution of common salt, cells for the elec- 
trolysis of the salt solution, and the necessary apparatus for 374 
WATER PURIFICATION 
applying the electric current. An ordinary orifice box to control the 
quantity of the solution to be applied to the water is also necessary. 
The electrolytic cell may be made of soapstone or procelain. 
Carbon (graphite) electrodes are used. One cell now on the 
market consists of a small soapstone tank filled with a series of 
graphite electrodes which are separated by glass baffles. The 
salt solution enters a separate compartment at one end of the cell, 
and then passes up and down between the faces of the carbons, 
and out at the opposite end of the cell. The electric current 
enters the large positive electrode or anode at one end and passes 
out at the corresponding negative electrode or cathode at the 
opposite end. The opposite faces of the intermediate carbons 
act as positive and negative electrodes, and thus really form a 
number of cells in series. The following reaction occurs as the 
current passes through the salt solution: 
NaCI + H2O^C1 + NaOH + H. 
The chlorine is liberated at the anode and the hydrogen at the 
cathode. Around the cathode is also produced a solution of 
caustic soda. The latter is the result of the action of the sodium 
on the water molecule, viz., Na + H2O -> NaOH + H. The 
chlorine also reacts with a molecule of water to form hypochlor- 
ous acid (see page 373), which in turn reacts with the caustic soda 
to form sodium hypochlorite. 
NaOH + HOCl^NaOCl + H2O. 
The continuous flow of the brine solution through the cell 
makes possible the interaction of the products of the electrolysis. 
Other secondary reactions occur, such as the formation of sodium 
chlorate, and the reduction to the chloride of a portion of the 
hypochlorite and chlorate first formed. In order to minimize the 
effect of these secondary reactions, it is necessary to keep the 
brine solution below a temperature of 100°F. 
Theoretically 1 amp. of current flowing for 1 hr. will produce 
1.322 grams of chlorine from 2.182 grams of sodium chloride. 
On account of the low efficiency of cells of the type described 
above, however, it has been found in practice that it required 
from 4 to 5 amp. of current and about 8 lb. of salt to produce 1 lb. 
of chlorine. Experiments made by Mr. G. A. Johnson indicated 
that with a consumption of electrical energy equal to 3.2 
kw.-hr., 1 lb. of chlorine could be produced from 8.3 lb. of salt at DISINFECTION OF WATER SUPPLIES 
375 
a cost of 6.46 cts. per pound, where the current cost 1.5 cts. per 
kilowatt-hour, and the salt $4 per ton. 
From experiments made in the laboratories of the Ohio State 
Board of Health with cells for electrolyzing a sodium chloride 
solution, the following deductions were made and published in 
its Quarterly Bulletin, vol. 1, No. 4, 1909. 
"1. The cost of production of chlorine electrolytically depends on 
the cost of the salt and the cost of the electricity. 
"2. Since the cost of the current is based on kilowatts or the product 
of voltage times amperage, and since the voltage of the cell remains 
practically constant, it follows that to obtain a low cost of operation 
the amperage must be kept down. 
"3. Amperage can be lowered by increasing the density of the cur- 
rent, by lowering the temperature of the cell, and by an increased rate 
of flow. A limit is placed on the first and third factors by the cost of 
the salt, and again on the third factor by the capacity of the cell. In 
the practical working of the cell it would be important to have a supply 
of cold water at little or no cost for cooling purposes. 
"4. The highest efficiency at ordinary temperatures was obtained 
with a concentration of brine of about 40,000 parts per million with a 
rate of flow of about 1,300 c.c. per minute. 
"5. At an artificially lowered temperature, the best results were 
obtained at 46°F. with a concentration of 50,000 parts per million and 
a rate of flow of 1,680 c.c. per minute. 
"6. Hot weather conditions show the cell at its lowest efficiency. 
"7. No attempt was made to determine the point of the highest 
efficiency of the cell or to determine the amount of chlorates and side 
reactions formed since it became evident that the cell could not be run 
under practical conditions at a figure that could compete with chloride 
of lime." 
Liquid Chlorine Apparatus.-There have been several devices 
placed upon the market for applying chlorine gas obtained from 
liquid chlorine. The apparatus now in successful operation has 
been the outcome of considerable experimental study, and re- 
quired the overcoming of difficulties arising from the pressure 
variations of the gas at different temperatures, and from its ex- 
tremely corrosive effect on metals wherever it came into contact 
with the latter in the presence of moisture. The slight solubility 
of the gas in water has also required the use of special diffusion 
devices, or of absorption apparatus. 
The principle involved in the construction of most of the appara- 
tus now being used for the application of chlorine gas is that of 376 
WATER PURIFICATION 
maintaining by means of suitable pressure reducing valves a con- 
stant drop in pressure across some form of orifice. An apparatus 
of this type now on the market1 consists essentially of a reducing 
pressure valve in the pipe leading from the supply tank, a second 
regulating valve following the first valve, an orifice plate and an 
absorption tank. The first valve re- 
duces the initial pressure of the gas 
coming from the supply tank to 
about 15 lb. per square inch. The 
gas then passes through the second 
adjustable reducing pressure valve by 
which any desired pressure may be 
maintained over the orifice plate in the 
pipe line. Between the second regu- 
lating valve and the orifice plate is a 
branch pipe to which a chlorine pres- 
sure gage is attached. This gage is 
calibrated in pounds per hour, and 
thus serves to indicate the rate of dis- 
charge of the gas for any given setting 
of the second regulating valve. 
After passing the orifice plate, the 
gas is conducted to the bottom of an 
absorption tower. Water admitted 
at the top of the tower flows down 
over broken material, which must be 
of such a nature as to be unaffected 
by the chlorine and water, and which 
offers a large surface for the absorp- 
tion of the ascending gas by the de- 
scending water. The chlorine solu- 
tion thus produced is conducted to 
the water to be treatde. 
Another apparatus,2 essentially 
along the same lines as that just 
described, consists of a gage to show the pressure in the sup- 
ply tank, a pressure compensating valve for maintaining a con- 
stant drop in pressure across a control valve, a control valve, a 
chlorine gas flow meter which has been calibrated empirically, 
Fig. 119.-Manual control 
chlorinator, direct-feed type. 
(Wallace & Tiernan Co.) 
1 Made by the Electro-Bleaching Gas Co., New York, N. Y. 
2 Manufactured by the Wallace & Tiernan Co., Inc., New York, N. Y. DISINFECTIONjOF WATER SUPPLIES 
377 
a visible glass orifice through which the chlorine is measured, 
a check valve, a back pressure gage which will indicate any stop- 
page of the flow of gas, and a chlorine absorption chamber. In 
some types of apparatus the absorption chamber is not used as 
the solution of the gas is not attempted in the device itself, but 
is fed directly through a "carborundum diffusor" placed in the 
water to be treated. The "diffusor" becomes saturated with 
water due to the capillary effect of the minute passages within 
Fig. 120.-Automatic control chlorinator, direct-feed type. (Wallace & 
Tiernan Co.) 
the carborundum disk. The pressure of the chlorine entering 
the center of the "diffusor" enables the gas to force its way 
through these small passages and to become saturated with mois- 
ture. On emerging from the "diffusor" in minute bubbles, it is 
readily dissolved by the water surrounding the carborundum 
disk. 
For feeding very small quantities of chlorine gas. a small 
volumetric pulsating glass meter has been designed, which utilizes 
the pressure of the entering gas to displace a definite volume of 378 
WATER PURIFICATION 
water in a siphon connected directly with the absorption chamber 
of the apparatus. 
All of the types of apparatus described above are provided 
with flexible tube connections leading to the supply tanks, since 
Fig. 121.-Liquid chlorine apparatus in head house of St. Louis filter plant 
{Electro-Bleaching Gas Co.) 
the latter are usually placed upon platform scales, whereby-the 
loss of weight of the gas from the cylinders may furnish a check 
upon the quantity of gas passing through the apparatus. Other 
attachments to these devices enable them to operate automat- 
ically in unison with and proportional to variations in the rate of 
flow of the water being treated, through connections to floats 
or Venturi meters. DISINFECTION OF WATER SUPPLIES 
379 
In another chlorine apparatus which was invented for the dis- 
infection of water and sewage, a different principle from that em- 
ployed in the previously described devices, was utilized. In this 
device the action depends upon the loss of weight of the gas from 
a supply tank, instead of upon the maintenance of a constant 
drop in pressure across an orifice. 
Fig. 122.-The Leavitt-Jackson chlorine apparatus 
The liquid chlorine to be applied is stored in steel flasks or 
cylinders containing about 100 lb. This flask or tank is sus- 
pended on a hook from a sensitive scale beam, mounted on hard- 
ened steel knife edges. The counterbalancing weight on the 
beam can be fed forward either at a constant rate, or in propor- 
tion to the rate of flow of the water being treated. The beam 
communicates through a system of levers with a controlling 
valve connected by rubber tubing to the suspended cylinder. 380 
WATER PURIFICATION 
Any change in equilibrium of the system, due to the weight being 
set forward along the beam, causes the valve to open or close 
automatically to the proper size of orifice. This allows the 
exact weight of gas to escape continuously from the tank, and 
keeps the scale beam in balance. This apparatus is known as 
the Leavitt-Jackson chlorinator. 
For conveying dry chlorine gas iron or steel pipe may be used. 
Wherever chlorine and water in any form are in contact with 
iron, brass, copper, tin or any metal attacked by hydrochloric 
acid, corrosion will take place. Silver tubing is used for convey- 
ing the gas into water, as the silver chloride first formed on the 
tube acts as a protective coating to the metal underneath it. 
Glass, hard rubber and stoneware are also commonly employed 
for delivering the gas or its aqueous solution to the water to be 
treated. The author has used ordinary garden hose for convey- 
ing the gas directly into the water with considerable success. 
OZONE 
Ozone is a gas containing three atoms of oxygen. It is much 
more soluble in water than is oxygen, since at 12°C. 100 volumes 
of water will dissolve 50 volumes of ozone at atmospheric pres- 
sure. Ozone is relatively stable only when mixed with much 
oxygen. When dry, cold, oxygen gas is subjected to the action 
of electric waves, only about 7.5 per cent, of the gas is converted 
into ozone. Ozone is a more active oxidizing agent than oxygen, 
and like the hypochlorites its activity is due to the fact that it 
contains much more energy than oxygen. 
Formation of Ozone.-A silent electric discharge passed 
through air or oxygen gas will produce ozone. A spark or arc 
discharge of electricity passed through air, however, produces 
oxides of nitrogen, but little or no ozone. The necessity for pro- 
ducing a silent discharge of the electric current in the formation 
of ozone was noted by the earliest investigators, but the precise 
reason for it is not well understood even today. Commercial 
ozonizers are operated by high-tension alternating currents, and 
the chief feature of their design is the care taken to avoid any 
spark or arc discharge. 
As early as 1875 Werner von Siemens invented an ozonizer 
consisting of two concentric glass tubes, one slightly larger than 
the other, coated with tin foil on the external and internal faces, DISINFECTION OF WATER SUPPLIES 
381 
respectively. The two coatings were connected with two poles 
of a high-potential electrical machine, and a current of dry air 
was passed along the annular space between the two glass tubes. 
Upon the discharge of the current between the poles during the 
passage of the air, ozone was produced, the amount formed being 
dependent on the rapidity with which the air was withdrawn, 
whereby heating was avoided, and upon the absence of sparking. 
Modern Ozonizers.-Practically all of the modern apparatus 
for the commercial production of ozone is based upon the general 
principles developed in Siemens' original experiments. The 
Siemens and Halske (Fig. 123) ozonizers are of the tube type, in 
which air is drawn along an annular space in which high-tension 
SECTION OF 
STERILIZER 
Fig. 123.-Siemens-Halske ozone apparatus. 
electrical discharges occur. One glass tube has, however, been 
dispensed with, and an internal metal cylinder, separated by a 
slightly larger glass tube from the external cylinder of tin foil, 
forms the ozonizing unit in each apparatus. The outer cylinder 
of the apparatus is surrounded with water in order to keep down 
the temperature, and all the external parts are kept at the earth 
potential to reduce risks from shocks from the high-voltage 
currents employed. The Otto (Fig. 125) ozonizer of the latest 
type consists of a series of glass plates, coated on alternate sides 
with tin foil, and separated by narrow strips of insulating mate- 
rial, which allow air spaces between adjacent sheets of the glass. 
The Rosenberg ozonizer differs from the two preceding forms 382 
WATER PURIFICATION 
of apparatus in that glass is replaced by micanite, and the continu- 
ous sheets of tin foil, which form the electrodes in other appara- 
tus, are replaced by copper or by an aluminum-alloy gauze of 40 
Fig. 124.- 
-Vosmaer system for ozonizing water, 
meshes to the inch. Current at 4,500 volts is said to pass be- 
tween the electrodes in this apparatus without sparking. De- 
Fries' ozonizer is similar to Rosenberg's, and consists of a hori- 
Main Supply 
| Aqueduct 
Main Water Supply 
Fig. 125.-Otto system for ozonizing water. 
zontal brass trough fitted with a plate-glass cover from which are 
suspendedjhalf-disks of brass with serrated edges. The brass 
trough is water-cooled. A silent discharge results between the DISINFECTION OF WATER SUPPLIES 
383 
edges of the disks and the surface of the trough, when a current 
of 2,000 volts is passed through the apparatus. The air drawn 
through the apparatus is ozonized as it passes between the elec- 
trodes while the electric current is discharging. 
Ozonizing apparatus for the disinfection of water has been 
used in America in a few places, the best known plants being those 
installed at Ann Arbor, Mich., Lindsay, Canada, and at the 
Note- Anneal Aluminum Tubes in Hot 
Solder or Babbit before Rolling 
Cover Plate 
Electrode 
Support Plate 
9^-20 B.& S.Aluminum Tubef 
\1H-20 B.& S.Aluminum Tube 
SECTION OF TUBE 
2-14 B.& S.Aluminum Tube 
\ 
Fig. 126.-Details of ozone generator. 
Herring Run supply of the Baltimore County water system 
(Fig. 126). The company owning the latter water supply system 
has used the Bridge type of ozone generator consisting of cylin- 
drical aluminum electrodes and micanite dielectrics. 
Yield of Ozonizers.-In order to obtain the best yield (Fig. 
127) from an ozonizer, the entering air must be dry. This may 
be effected by refrigeration, and apparatus for this purpose 384 
WA TEH P U RIF IC A TION 
may, therefore, be considered as an essential part of the ozone 
plant. 
The Siemens-Halske ozonizers of the type shown in the accom- 
panying cut requires for its operation 1 hp. per hour, and produces 
13.5 to 27 grams of ozone per hour according to the amount of air 
Ozone per hour and apparatus 
30 Concentration, Grams 
ozone per cbm. air 
Ozone per Hp. hour 
Cbm, air per hour 
Fig. 127.-Curves of yield of an ozone apparatus. 
passed through it, and the degree to which it has been previously 
dried. A Gerard ozonizer (10-tube unit) used at Great Falls, 
S. C.1 is capable of producing about 20 grams of ozone per hour. 
By actual test these units required about 400 watts each, when 
the air passed through the ozonizer at the rate of 2 cu. ft. per 
minute. 
1 Eng. Record, vol. 64, July 1, 1911. DISINFECTION OF WATER SUPPLIES 
385 
Application of Ozone.-The ozonized air produced by the 
generators is usually applied to the water to be treated by means 
of scrubbers. These are towers from 12 to 15 ft. in height, in 
which the water entering the top of the tower flows down through 
a bed of coarse gravel. The ozonized air is blown in at the bot- 
tom of the tower, and as it ascends comes into contact with thin 
films of the water trickling down through the gravel bed. The 
ozone which is absorbed exerts its oxidizing action upon the or- 
ganic matter that is dissolved and suspended in the water, in- 
cluding, of course, the bacteria. The longer the contact which 
the water has with the ozone, the greater the purification effected. 
Any excess of ozonized air may be returned to the ozonizers again 
if desired. The purified water is drawn off at the bottom of the 
towers. 
In some cases the ozonized air is merely pumped into the bot- 
tom of a column of water, and allowed to rise through it. This 
method does not produce as complete a contact between the water 
and the ozone as do the scrubbing towers, and, beside, requires 
more power to overcome the pressure of the water column. 
At the Herring Run plant of the Baltimore County Water and 
Electric Co., the water flows to tne ozone plant directly from the 
reservoirs, and is treated with ozone by means of aspiration. 
The falling water sucks the ozonized air directly from the genera- 
tors, and then passes through a mixing chamber to the open 
suction well of the pump. Any excess of ozonized air escapes in 
the open suction well. 
The largest ozone plant that has probably ever been installed 
is at Petrograd, Russia. This plant contains 128 Siemens and 
Halske ozonizers of the tube type, and five sterilizing towers. 
A refrigeration plant dries the air before it is sent to the ozonizers. 
Injectors operating at a water pressure of 14 ft. suck ozonized air 
from the ozonizers, and drive it mixed with water to the ster- 
ilization towers. Absorption of the ozone by the water takes 
place in the injectors at the top of the sterilization towers, and in 
the passage of the water downward through the towers, where it 
comes into contact with ozonized air rising from the bottom of the 
towers. The ozonized water flows over cascades to remove the 
excess of air before entering a storage tank. 
The ozonizers are supplied with a current having a voltage of 
7,000. The volume of water treated is about 13,000,000 gal. 
per day. 386 
WATER PURIFICATION 
Quantity of Ozone Used.-The amount of ozone required to 
properly disinfect different waters must of necessity vary con- 
siderably. The extent of the original pollution of the water, and 
the degree of purification effected prior to treatment with ozone, 
will naturally affect the amounts required. The best results, so 
far as the destruction of the bacteria are concerned, will be ob- 
tained with those waters which contain the least amount of dis- 
solved and suspended oxidizable matter. Turbid waters should, 
therefore, be settled and filtered before attempting disinfection. 
Those waters containing vegetable coloring matter in colloidal 
solution will require more ozone than those free from it. 
At certain German plants1 the average consumption of ozone 
for fairly clear waters was found to be 1.3 grams per cubic meter, 
Slerilizafton Tourer J» 
Transformers 
and 
Ozonizers. 
f^ntilatinaToorrv 
uit/i 
Zitrified, ttaterJSasin. 
^a^in (Zontainintr 
tfa ter to be Treated. 
Fig. 128.-The Paderborn, Germany, water ozonizing plant. 
equivalent to about 5,000 grams per million gallons. Probably 
from 3,000 to 8,000 grams of ozone per million gallons will be 
sufficient for disinfecting most waters, if effective methods of 
application are used. 
Extent of Use of Ozone.-In America ozone is not used for dis- 
infecting water to any great extent. In Europe, however, there 
are several water-works plants using this agent. The largest 
plant is at Petrograd, Russia, as previously noted. Plants treat- 
ing over 1,000,000 gal. of water per day may be found also at 
Wiesbaden, Germany; Florence, Italy, and at Chartres, France. 
At Nice and St. Maur, in France, plants handling 10,000,000 and 
6,000,000 gal. of water per day, respectively, are sterilizing with 
this agent. A number of smaller plants are also to be found in 
different parts of Europe. 
1 Paderborn and Wiesbaden, Germany. DISINFECTION OF WATER SUPPLIES 
387 
ULTRA-VIOLET LIGHT 
The favorable effect of light upon most forms of life is common 
knowledge. On the other hand, the destructive action of sun- 
light upon many of the lower forms of plant life is also well known. 
The harmful effects of artificial light rays produced by the elec- 
tric arc under certain conditions have received considerable atten- 
tion by scientists during the last 
decade. The results of these 
studies indicate that the shorter 
wave lengths of light exert a marked 
destructive action upon bacterial 
life. 
Historical.-By exposing bacte- 
rial cultures to the light rays from 
different portions of the spectrum, 
or by using filters and screens which 
will only permit rays of certain 
wave lengths to pass, it is possible 
to note directly the germicidal 
effects produced by light rays of 
different wave lengths. For ex- 
ample, it has been found that 
colonies of bacteria growing upon 
an agar plate were not affected by 
the red rays of the spectrum, but 
that growth was prevented when 
exposed to the violet rays. 
The experimental work of Henri, 
Helbronner and von Reckling- 
hausen, at the Sorbonne University 
in Paris, demonstrated quite con- 
clusively that the shorter the wave 
length of light the greater its abiotic 
power. The shorter wave lengths 
of light are emitted at the violet end of the spectrum. The in- 
visible rays of light beyond the violet rays have the shortest wave 
lengths and are the most effective bactericidal agents. Rays hav- 
ing a wave length of 0.0003 mm. and less were found to destroy 
bacteria in a comparatively short space of time. Many of the 
pathogenic forms can be destroyed by an exposure of 10 to 40 sec. 
SECTION A-A 
Fig. 129.--Ultra-violet ray ster- 
ilizer for water. 388 
WA TER P URIFICA TION 
Even Paramecia and yeast cells were destroyed by exposures of 
3 to 5 min. Exposures of even Xo sec. are frequently sufficient 
to kill some bacteria. 
SECTION ON C.L. 
SECTION ON C.L. 
Fig. 130.-Sectional view of ultra-violet ray water sterilized. 
Production of Ultra-violet Light.-The mercury vapor quartz 
lamp has been found to emit light rich in ultra-violet rays, and 
Fig. 131.-Ultra-violet sterilizer for use with water under pressure. 
the conditions under which the largest number of these rays were 
produced have been carefully investigated by von Recklinghausen 
and his associates. This lamp, which consists of a fused rock- DISINFECTION OF WATER SUPPLIES 
389 
crystal (quartz) tube containing mercury, is the only practical 
source of ultra-violet light that up to the present time has been 
adapted to the conditions required for the sterilization of water. 
In lamps of this type metallic mercury forms the cathode and 
the arc passes through rarified mercury vapor. The light pro- 
duced is not due to the temperature of the arc, but to the lumines- 
cence of the vapor. The lamps may be operated on the usual 
distribution voltages for direct currents. An alternating cur- 
rent must be first rectified in the usual way before being sent to 
the lamps. 
The Pistol Lamp of Max von Recklinghausen.-A recent im- 
provement in lamps for sterilization purposes has been invented 
by Dr. Max von Recklinghausen. This is known as the " pistol 
lamp" because of its shape. It operates on a 500-volt current 
and requires 3 amp. Its luminous tube is U-shaped, the two 
branches of the U being very close together. The luminous por- 
tion of the lamp is enclosed in a 2-in. quartz tube, which forms 
the lamp chamber, and which prevents the water coming into 
contact with the lamp proper. By making the quartz tube U- 
shaped instead of straight, it has been possible to utilize all of the 
light emitted instead of only 60 per cent, of it as in former lamps. 
The efficiency of this lamp as compared with those first pro- 
duced has been increased not only on account of its shape, but 
also because of its greater production of ultra-violet light. About 
10 times as much ultra-violet light is produced in the pistol lamp 
as in the old 220-volt lamp, although the voltage is only doubled. 
Compared with the 3.5 amp. lamps, the ultra-violet rays are 50 
times as powerful, although the wattage is only four times as 
great. 
Conditions Necessary for Efficient Lamp Operation.-The 
hotter the mercury quartz lamp becomes, the more ultra-violet 
light it will produce. By properly cooling the electrodes and by 
having large electrode containers, it has been found possible to 
control the temperature of the lamp, if the tubes are of the 
correct dimensions and a proper amount of resistance is inserted. 
The heat must not be so great as to cause the quartz tube to 
disintegrate and become opaque. Under normal conditions the 
luminous tube of the lamp has a temperature of about 800°C. A 
temperature of 700°C. has been successfully maintained for many 
thousand hours. As soon as a lamp is worn out, it frequently 
extinguishes itself, and must be started again. This should serve 390 
WATER PURIFICATION 
as a warning to those in charge that a renewal of the lamp is 
needed. 
Since close and direct contact of the bacteria with the ultra- 
violet rays is productive of the best results, efforts have been 
made in designing the apparatus to so arrange it that the water 
shall flow slowly through the illuminated zone. In some appara- 
tus, where the lamp is suspended immediately above the water, 
the latter is made to pass twice through the light zone by means 
SECTION ELEVATION ON C.L. 
ELEVATION 
Fig. 132.-Multiple lamp flame for ultra-violet ray sterilization of water. 
of baffles. A better arrangement, especially where large volumes 
of water are to be treated, is to provide a flume through which the 
water flows. The lamps are placed in the sides of the flume, and 
of course must be made accessible for purposes of adjustment and 
renewal. Baffles also are used in order that a mixing action 
may be produced, thereby lessening the chances for any of the 
bacteria escaping a sufficient exposure to the light rays. 
Since actual exposure to the light rays is absolutely necessary, 
if the microorganisms are to be killed, the presence of any con- DISINFECTION OF WATER SUPPLIES 
391 
siderable quantity of suspended matter or of colloidal vegetable 
coloring matter in the water, which will screen or diminish the 
intensity of the light rays, can not be permitted. The presence 
of as much as 20 parts per million of suspended matter in a water, 
or of 40 parts of color, have not prevented successful sterilization 
according to Dr. von Recklinghausen. However, it is preferable 
to have the water as clear and colorless as possible, before at- 
tempting to sterilize it by this method. 
Fig. 133.-Micro-organisms 
that may produce disagreeable 
odors and tastes in drinking 
water. (From left to right, 
Anabsena, Asterionella and 
Clathrocystis.) 
Algicides.-The pollution of public water supplies by growths 
of algae, diatoms and other microorganisms is not uncommon, and, 
at those periods when the growths are particularly abundant, 
may require the use of remedial measures. These growths may 
become extremely troublesome in the operation of filters by 
clogging the filtering medium. 
By decomposing on the surface of the filter bed these organ- 
isms may also produce obnoxious odors and tastes in the filtered 
water. The algae, and especially the blue-green algae (Cyano- 
phyceae), are more frequently the cause of bad odors and tastes, 392 
WATER PURIFICATION 
although the diatoms and the infusoria are both sometimes found 
to be the source of these troubles. Certain of the tastes and 
odors are characteristic of the growth of these organisms, while 
others, which are usually extremely disagreeable, are the results 
of the decomposition of the organic matter after the death of the 
organisms. 
The following table taken from a paper by D. D. Jackson and 
the author " On Odors and Tastes of Surface Waters with Special 
Reference to Anabaena" {Technology Quarterly, vol. 10, Decem- 
ber, 1897), describes some of the characteristic odors of growth 
and decay. 
Microorganisms 
Natural odor 
Odor of decay 
Diatomace^: 
Asterionella 
Tabellaria 
Meridion 
Cyanophyce^ (Blue-green 
Algae): 
Aromatic to fishy. 
Aromatic. 
Aromatic. 
Anabaena 
Mouldy, grassy. 
Pig pen. 
Rivularia 
Mouldy, grassy. 
Pig pen. 
Clathrocystis 
Sweet, grassy. 
Pig pen. 
Coelosphserium 
Sweet, grassy. 
Pig pen. 
Aphanizomenon 
Chlorophyce je : 
Faintly grassy. 
Pig pen. 
Vol vox 
Fishy. 
Eudorina 
Faintly fishy. 
: h 
Ai 
Fandorina 
Faintly fishy. 
i n 
Infusoria : 
Uroglena 
Synura . .. 
Dinobryon.. 
Bursaria 
Peridinium 
Cryptomonas 
Mallomonas 
Fishy and oily. 
Ripe cucumbers. 
Fishy, like rockweed. 
Irish moss or salt marsh. 
Fishy, like clam shells. 
Candied violets. 
Faintly fishy. 
The research work of George T. Moore and Karl F. Kellerman, 
of the Bureau of Plant Industry of the U. S. Department of 
Agriculture, called attention some years ago to the value of cop- 
per sulphate in preventing and in destroying growths of micro- 
organisms in water. Its value for this purpose has been amply 
demonstrated in practice, and is now a recognized method for DISINFECTION OF WATER SUPPLIES 
393 
rectifying difficulties of this character. The following tables, 
compiled by Dr. Karl F. Kellerman,1 showing the frequency of 
occurrence of various genera of microorganisms in public water 
supplies, and the amounts of copper sulphate necessary to kill 
these organisms, are of interest in this connection. 
Table Showing the Occurrence of Genera of Microorganism most 
Frequently Reported as Causing Trouble in Reservoirs and Ponds 
Anabaena 
Beggiatoa 
Asterionella 
Chara 
Cladophora 
Clathrocystis 
Conferva 
Crenothrix 
Fragillaria 
Navicula 
Oscillatoria 
Spirogyra 
Arkansas 
4- 
4- 
4- 
California 
+ 
+ 
4- 
4- 
4- 
4- 
4- 
4- 
4- 
Colorado 
4- 
4- 
4- 
-4- 
Connecticut 
+ 
4- 
4- 
4- 
4- 
4- 
Dist. of Columbia 
- 
- 
- 
+ 
- 
- 
+ 
Georgia 
4- 
Idaho 
4- 
4- 
4- 
4- 
Illinois 
-• 
- 
-- 
- 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
Indiana 
4- 
4- 
4- 
4- 
4- 
Kansas 
4- 
4- 
4- 
4- 
Kentucky 
4- 
+ 
4- 
4- 
4- 
4- 
4- 
4- 
Louisiana 
4- 
4- 
Maine 
4- 
4- 
Maryland 
•4- 
4- 
4- 
4- 
Massachusetts 
4- 
4- 
4- 
4- 
4- 
4- 
+ 
4- 
Michigan 
4- 
4- 
4- 
4- 
Minnesota 
+ 
4- 
4- 
4- 
4- 
Missouri 
+ 
- 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
4- 
Montana 
4- 
4- 
Nebraska 
4- 
4- 
4- 
New Hampshire 
+ 
4- 
4- 
New Jersey 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
- 
+ 
+ 
New Mexico 
4- 
4- 
New York 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
Ohio 
- 
4- 
4- 
4- 
4- 
4- 
4- 
4- 
Oklahoma 
4- 
4- 
Pennsylvania 
+ 
+ 
+ 
+ 
- 
- 
+ 
- 
+ 
+ 
+ 
+ 
Rhode Island 
- 
4- 
South Carolina 
- 
- 
- 
- 
- 
+ 
- 
- 
- 
- 
- 
South Dakota 
- 
- 
- 
- 
- 
- 
+ 
- 
- 
+ 
+ 
+ 
Tennessee 
4- 
Texas 
- 
- 
- 
+ 
- 
- 
+ 
- 
- 
+ 
+ 
+ 
Vermont 
- 
- 
+ 
+ 
+ 
Virginia 
- 
+ 
- 
- 
- 
- 
+ 
+ 
+ 
+ 
- 
+ 
Washington 
4- 
4- 
West Virginia 
+ 
Wisconsin 
- 
- 
- 
- 
+ 
- 
+ 
4- 
+ 
+ 
+ 
+ 
1 Karl F. Kellerman: "The Rational Use of Disinfectants and 
Algicides in Municipal Water Supplies." Paper read before the Eighth 
International Congress of Applied Chemistry, vol. 26, p. 241. 394 
WATER PURIFICATION 
Table Showing Quantity of Copper Sulphate Required to Kill 
Various Forms of Odor-producing Organisms 
Anabaena 
0.09 
Kirchneriella 
5.00 to 10 
Asterionella 
0.10 
Leptomitus 
0.40 
Beggiatoa 
5.00 
Microspora 
0.40 
Chara 
0.20 to 5 
Navicula 
0.07 
Cladophora 
1.00 
Oscillatoria 
0.10 to 0.40 
Cladothrix 
0.20 
Peridinium 
2.00 
Clathrocystis 
0.10 
Scenedesmus 
5.00 to 10 
Coelosphaerium .... 
0.30 
Spirogyra 
0.05 to 0.30 
Conferva 
0.40 to 2 
Ulothrix 
0.20 
Crenothrix 
0.30 
Uroglena 
0.05 
Euglena 
1.00 
Volvox 
0.25 
Fragillaria 
0.25 
Zygnema 
0.70 
Hydrodictyon 
0.10 
Copper sulphate expressed as parts per million 
The use of too large quantities of copper sulphate is dangerous 
to fish life. The following table taken from Dr. Kellerman's 
paper shows the safe limit for treating water with copper sulphate 
when certain fish are present. 
Table Showing Safe Limit for Treating Water with Copper Sulphate 
When Certain Fish Are Present 
Copper sulphate expressed as parts per million 
Black bass 2.10 
Carp 0.30 
Catfish 0.40 
Goldfish 0.50 
Perch 0.75 
Pickerel 0.4 
Suckers 0.30 
Sunfish 1.20 
Trout 0.14 
A not infrequent result following the application of copper 
sulphate to a water supply infested with microorganisms is the 
appearance of large numbers of bacteria. These are harmless 
forms of water bacteria that multiply rapidly, probably on ac- 
count of the sudden increase in food supply caused by the 
decomposition of the organic matter resulting from the killing 
of the microorganisms. 
While copper sulphate has been proposed as a bactericidal 
agent, its efficiency in this connection is so much less than other 
disinfectants, such as chlorine and its compounds, for example, 
that it has never come into practical use. CHAPTER XXIX 
DISINFECTION OF WATER SUPPLIES (CONTINUED) 
THE CAUSE OF THE DESTRUCTION OF MICROORGANISMS BY 
DISINFECTANTS 
The exact cause for the destruction of microorganisms by any 
of the methods of disinfection described, namely by the hypo- 
chlorites, ozone or ultra-violet light, is none too well understood. 
Hypochlorous acid or its salts are, as has been pointed out, really 
agents for the production of oxygen in a very active state. 
Ozone, also, by its decomposition, produces oxygen which is 
liberated in an extremely active condition. By the passage of 
ultra-violet light rays through oxygen gas ozone is known to be 
formed. It is not difficult to conceive, therefore, that ultra- 
violet light rays may convert some of the dissolved oxygen always 
present in drinking water into ozone, which in breaking down sets 
free active oxygen in exactly the same manner as does the ozone 
prepared in ozonizers and applied as previously described. 
Should this theory of the action of ultra-violet light rays prove 
true, then all of the methods for disinfection described can be 
explained as oxidation methods, and the destructive effect of 
even the minute quantities of oxygen produced could be attri- 
buted to the energy set free by the breaking down of the molecule 
of hypochlorous acid or of ozone. This free energy would seem to 
afford a sufficient reason for the bactericidal effects produced. 
Another theory which has been advanced is that the ultra- 
violet rays produce hydrogen peroxide, which like ozone acts as 
an oxidizing agent, and which is the cause of the death of the 
microorganisms. Dr. von Recklinghausen contends, however, 
that sterilization is not produced with ultra-violet light by the 
formation of hydrogen peroxide in the water, as the quantity 
that could be formed would be too small to kill the bacteria. He 
also maintains that these bactericidal effects can be brought 
about in a water free from oxygen, and that they are not due to 
any chemical produced in the water, but are the result of the 
light itself. 
395 396 
WA TER P V RIF IC A TION 
Selective Action of Disinfectants and Its Effects.-It is evident 
that any of the disinfecting agents described may be more de- 
structive of some microorganisms than of others. Those bacteria, 
for example, that may be in the spore state, will not be as sus- 
ceptible to disinfecting agents as those that are not in this con- 
dition. Bacteria that do not produce spores, in which class are 
found most of the pathogenic forms dangerous to human beings, 
should be, therefore, more easily killed. 
A peculiar result sometimes produced by the use of the hypo- 
chlorites is that of the reappearance in large numbers of bacteria 
in a water that had apparently been sterilized, or if not rendered 
entirely free from bacteria had had the number of organisms 
materially reduced. Experiments by the author, which were 
made under controlled conditions, reproduced this phenomenon, 
which is not of infrequent occurrence in water-works operation. 
It would appear as though either all or nearly all of the original 
organisms had been so adversely affected by the toxic influence 
of the disinfectant, that they could not grow even under favor- 
able conditions on culture media, thus causing the conclusion to 
be drawn that the water was sterile or practically so; or that the 
less virile organisms had been actually killed, and only a few of the 
more hardy forms had revived and become the nuclei for a large 
secondary growth under more favorable environmental conditions. 
It is not improbable that disinfectants may exert to some extent 
merely a temporary inhibitory action so far as reproduction upon 
culture media is concerned; but it is also true that many organ- 
isms are actually destroyed, and usually those of a pathogenic 
character. 
Secondary growths are not regarded as of any particular sani- 
tary significance. Enormous bacterial growths of this character 
not infrequently follow the application of copper sulphate, where 
the latter is used for destroying alga? and diatoms in a water. 
The death and decay of this microscopic plant life merely affords 
the optimum conditions for bacterial activity. In other words, 
the equilibrium of the bacterial flora is disturbed by the action of 
the disinfectant, and species, which under normal conditions 
would only reproduce slowly, find an opportunity to multiply 
rapidly without hindrance from other plant and animal forms. 
Tastes and Odors Produced in Water Treated with Chlorine 
and Its Compounds.-The taste of chlorine or its compounds in 
the small quantities generally employed in the disinfection of DISINFECTION OF WATER SUPPLIES 
397 
water has been described as "medicinal," or like the taste of car- 
bolic acid. The taste is usually detected before the odor. Only 
those persons who are particularly sensitive to tastes and odors 
will be able to detect the smaller amounts. The average ob- 
server will not taste as little as 0.4 part of chlorine per million, 
although 0.5 part would be more frequently noted, and 0.6 part 
per million would be an amount which could be generally 
detected. 
The odor of chlorine in a bleaching-powder solution, according 
to Lederer and Bachmann,1 could be identified when it contained 
1.8 parts per million of "available chlorine," while it was noted 
in a solution of chlorine gas in water when there was present only 
0.9 part per million. 
While it may be true that these tastes and odors are attribu- 
table to the calcium hypochlorite or the chlorine gas added, 
there is some evidence to indicate that they may be due to organic 
compounds produced by the reaction between the chlorine and 
the soluble organic matter naturally present in most drinking 
waters. The author has made extremely disagreeable odors and 
tastes in an artificial leaf infusion by treating it with varying 
amounts of a bleaching-powder solution. 
It is of interest to note that there is a very wide variation in 
the amounts of bleaching powder that may be used in different 
waters without producing disagreeable tastes and odors. Mr. 
F. F. Longley2 has compiled some statistics on this point which 
he obtained from answers to questions sent to over 100 different 
water-works plants in the United States. He found that in one 
plant as much as 37 lb. per million gallons had not produced any 
taste or odor, and that in many places 20 to 30 lb. of bleaching 
powder per million gallons have not been noticeable. On an 
average about 14 lb. per million gallons did not appear to give 
any trouble. 
For the removal of tastes and odors due to overdosing with 
bleaching powder, Lederer and Bachmann3 recommend the use of 
sodium thiosulphate in quantities equal to one-half of the chlo- 
rinated lime applied. This chemical is cheap and adds about 40 
per cent, to the cost of the bleaching powder. 
1 Proc. Illinois Water Supply Association, March, 1912. 
2 Report of Committee on Water Supplies of Sanitary Eng. Sect, of 
A. P. II. A., December, 1914. 
3 Proc. Illinois Water Supply Association, March, 1913. 398 
WATER PURIFICATION 
Note.-Assuming bleaching powder had 33 per cent, of available chlorine, 
each 2.5 lb. per million gallons of water is equivalent to 0.1 part of chlorine 
per million; hence 25 lb. per million gallons equals 1 part per million of 
chlorine. 
Point for Application of Disinfectants and Amounts Used.- 
The most effective point in a water-supply system for the appli- 
cation of disinfecting agents naturally varies with local condi- 
tions, with the character of the water to be treated, and with the 
particular method of disinfection employed. Those waters 
which are not subjected to any form of purification prior to dis- 
infection will usually require somewhat greater amounts of the 
disinfecting agent or a longer period of contact with it than those 
that have been purified. For example, waters high in soluble 
coloring matter will use up more chlorine, more ozone, or require 
a longer exposure to ultra-violet light rays, than if they are color- 
less. The same is true of waters containing suspended matter in 
any considerable amount. 
After a water has been disinfected it should be delivered as 
quickly as possible into the distribution system for consumption. 
Long storage periods are undesirable, but where they can not be 
avoided, it is advisable to store in covered reservoirs, unless the 
expense involved is prohibitive. There is, of course, more dan- 
ger from atmospheric and surface drainage contamination in 
open than in covered reservoirs, and once a water has been pre- 
pared for consumption, it should not be subjected to conditions 
that may again render it unsafe. On the other hand, sufficient 
time must elapse for the full effect of the disinfectant to be pro- 
duced. Fortunately this period is quite short for practically all 
of the disinfecting agents described. Practical disinfection with 
chlorine and its compounds may be effected in 15 to 30 min., and 
more than 1 hr. would probably never be required. Provision 
for about 1 hr. of contact between the ozonized air and the water 
to be treated was provided for in one of the German ozone plants, 
and this is doubtless sufficient in most cases. Where ultra-violet 
light is used only a few minutes exposure in the light zone is 
claimed to be necessary to produce absolute sterility. 
Mr. Francis F. Longley in the report previously referred to 
states that 40 per cent, of the water supplies about which they 
obtained information concerning the use of the hypochlorites for 
disinfection had no storage following disinfection; 18 per cent, 
of them had less than 1 hr.; 9 per cent, of them had from 1 to 3 DISINFECTION OF WATER SUPPLIES 
399 
hr.; 5 per cent, from 3 to 12 hr.; 11 per cent, from 12 to 24 hr.; 
and 17 per cent, more than 24 hr. 
When a water supply is filtered, the most economical point to 
apply the disinfecting agent is in the filtered-water reservoir. In 
many rapid sand filter plants, no common point for the mixed 
effluents of all the filters is reached until the water leaves the 
filtered-water reservoir. In such cases as this there is no alter- 
native but to apply the disinfectant at this point, and depend 
upon the contact period in the pipe line to produce the effect 
desired. In the case ultra-violet light rays are used, there must 
also be some common point past which all the filtered water 
flows, and where exposure to the light is possible. 
Mr. Longley found that 37 per cent, of the cities giving replies 
to his questions used a disinfectant without other treatment, and 
that the balance used it as an adjunct to some other form of 
purification, usually in connection with filtration. In 57 per 
cent, of the plants it was used as a final treatment in connection 
with filtration; in 26 per cent, it was used after coagulation or 
sedimentation and before filtration, and in 17 per cent, it was 
applied before coagulation and filtration. In about 90 per cent, 
of the plants the use of the disinfectant was continuous, and in 
the remainder intermittent. 
The use of chemical disinfectants prior to coagulation and filtra- 
tion requires the application of larger amounts than would be 
the case if they were applied to the clear filtered water. There 
would appear to be little justification for this procedure, unless 
odors and tastes were lessened, or prevented in the filtered water. 
The danger of destroying the colloidal coating upon the sand 
grains of the filters by using too large amounts of hypochlorites 
in the applied water is a possibility that the author knows from 
experience can happen. 
Bacterial Reduction by Disinfection.-The efficiency of disin- 
fection processes depends, as in other methods of purification, on 
the care with which the process is carried on, and with the char- 
acter of the water being treated. In general, a reduction in the 
number of bacteria to less than 10 per cubic centimeter can be 
usually expected. In some plants no effort is made to reduce the 
bacteria to the lowest possible number, since this procedure 
might require too great an amount of the disinfecting agent, 
which would unduly increase the cost without rendering the 
water much safer from a sanitary standpoint. Where chlorine 400 
WA TEH P URIE IC A TION 
or hypochlorites are employed, the difficulty from tastes and 
odors from excessive amounts of these agents is always a matter 
to be given consideration. The reduction in the number of B. 
coli, as indicated by the usual tests for this organism, should be 
given due weight in the interpretation of the effects produced by 
disinfection. 
Relative Effects of Disinfectants on Vaiious Species of Bac- 
teria.-A large amount of experimental work has been carried on 
during the past few years to determine the ability of various 
bacteria to withstand the action of disinfectants. The resistance 
of spore-bearing bacteria has been especially studied. Mr. N. 
S. Hill, Jr., states that bleaching powder applied as a disinfectant 
to water does not destroy the spore-bearing bacteria, such as B. 
aerogenes capsulatus, B. butyricus, and B. cadaveris sporogenes. 
Lederer and Bachmann found that 25 parts per million of avail- 
able chlorine effected only a 3 per cent, reduction in the number of 
B. subtilis in 15 min., and that 400 parts per million of chlorine 
acting for 15 min. would produce a 95 per cent, reduction. B. 
mesentericus rosei and B. mesentericus fuscus required a con- 
tact of 15 min. with 5 parts per million of chlorine to be reduced 
in number 80 per cent, and 26.4 per cent., respectively. These 
Percentage Reduction1 on Disinfection of the Group of Intestinal 
Bacteria with Calcium Hypochlorite by a Contact of 15 Min. 
Bac- 
teria 
per c.c. 
160,000 
9,500 
3,000 
8,000 
180,000 
180,000 
500 
Parts 
per 
million 
of chlo- 
rine 
B. cloacae 
B. fecal, 
alkalig. 
B. 
paratyph. 
B. prot. 
mirabilis 
B. 
enteritidis 
B. 
lactis 
aerog. 
B. chol- 
erse suis 
ap- 
plied 
0.1 
99.98 
27.3 
0.2 
99.69 
99.99 
•99.97 
45.5 
99.83 
99.17 
95.8 
0.3 
99.75 
100.00 
100.00 
63.7 
99.98 
99.98 
100.0 
0.5 
100.00 
72.7 
100.00 
100.00 
0.7 
63.7 
1.0 
63.7 
3.0 
90.9 
5.0 
90.9 
1 Proc. Illinois Water Supply Association, March, 1912. DISINFECTION OF WATER SUPPLIES 
401 
same investigators found that intestinal bacteria were much 
more susceptible to disinfectants as the preceding table will show. 
Similar experimental data are available for the effects of ozone. 
The following table taken from the Report of the Imperial Board 
of Health of Germany, Book XVIII, vol. 3, 1902, gives the results 
of tests made by Drs. Ohlmuller and Prall. 
Bacteria per Cubic Centimeter 
Before 
ozonization 
After ozonization 
Species 
Ordinary forms 
Pathogenic forms 
38,330 
8 
0 
B. cholera? 
16,590 
9 
0 
B.typhi 
N. B.-Ordinary Spree and aqueduct water infected with cholera germs, 
and boiled water infected with typhoid fever germs. 
Dr. Max von Recklinghausen1 has recently given some data 
showing the length of time required to kill various species of 
bacteria by exposure to rays of ultra-violet light. 
Staphyloccus aureus 
B.ChoIerae 
B.Tpyhi 
B.Dysenteriae (Shigo) 
B. " " (Dopter)__ 
B.Coli „ 
B. Anthracis (sporogenes).. 
B. Pneumoniae (Fried) 
Sarcinae alba 
B. Aerogenes capsulatus- 
B.Tetani- 
B. Megatherium 
B.Phleole 
B.Subtilis 
B. Sarcinae luter. 
Paramecia 
Y east 
Seconds 
It was found in other experiments undertaken by Dr. von 
Recklinghausen, that water infected with B. coli could be ren- 
dered sterile by the light rays from a 220-volt 3-amp. mercury 
vapor quartz lamp acting for 1 sec. at a distance of 4 in.; in 4 sec. 
at a distance of 8 in.; in 15 sec. at a distance of 16 in.; and in 30 
sec. at a distance of 24 in. 
1 Jour. New England Water-works Assn., June, 1915. 402 
WATER PURIFICATION 
The variability in resistivity to ultra-violet light appears to 
be considerably less than in the case of chemicals like chlorine 
and ozone. 
Practical Results Obtained by Disinfection.-The actual re- 
sults obtained with disinfecting agents in practice are shown by 
the following data taken from different sources. 
During the first 4 months of 1913, there were used at the Cin- 
cinnati filtration plant over 10 tons of bleaching powder. This 
amount was applied to over 500,000,000 gal. of water, and was 
equivalent to practically 0.14 part per million of available chlo- 
rine. The disinfectant was applied in the form of a solution to 
the filtered water. The results produced upon the bacterial 
content of the water are shown in the following table. 
1913 
Ohio River 
water 
Settled 
water 
Filtered 
water 
Disinfected 
water 
January 
61,500 
16,500 
470 
70 
February 
16,140 
3,550 
150 
41 
March 
25,120 
4,060 
110 
14 
April 
9^575 
3,260 
80 
26 
Average Number of Bacteria per Cubic Centimeter 
The average removal of bacteria by settlement and filtration 
was 99.3 per cent., and by disinfection 81 per cent. 
At the Cincinnati plant in 1915 liquid chlorine was used during 
the whole year. The percentage removal of the bacteria as a 
whole and of the B. coli for each step in the process of purification 
is given in the following table. 
Percentage Removal of All Bacteria and of B. Coli for Each Step 
in Process 
Process 
All bacteria 
B. coli 
Based on number 
in raw water 
Based on number 
applied at each 
step in process 
Based on 
number in 
raw water 
Based on num- 
ber applied at 
each step in 
process 
By plain sedi- 
mentation 
81.17 
81.17 
85.81 
85.81 
By coagulation.. 
) 18-12 
™ ' QC ) 96.28 
13.67 
96.32 
By filtration 
2.78 J 
79.86 / 
By disinfection. . 
0.50 
71.03 
0.35 
66.80 DISINFECTION OF WATER SUPPLIES 
403 
The quantity of chlorine applied at the above plant averaged 
0.12 part per million. 
As an illustration of the effect of bleaching power on an effluent 
from slow sand filters, the result of treating the water at the 
Torresdale filter plant in Philadelphia is of interest. Fron 0.16 
to 0.33 part per million of available chlorine was used during the 
2 years tabulated. 
Filtered water basin. Tap at Lard- 
ner's Point, per cent, positive tests 
for B. coli 
1911 
1 c.c. 
10 c.c. 
1 c.c. 
10 c.c. 
Untreated water, May-Nov., inclusive... . 
14.5 
58.0 
10.0 
48.6 
Treated water, Jan., Apr. and Dec., 1912.. 
4.0 
13.3 
2.4 
12.9 
Treated entire 12 months, 1913 
4.1 
22.0 
2.0 
16.6 
Treated for 7 months, Jan.-July, inclusive.. 
0.9 
3.3 
0.6 
6.2 
Over 0.16 part of chlorine used continuously. 
An example of the effect of bleaching powder on the water of 
Lake Erie, used without purification other than disinfection, as 
the water supply of Cleveland, Ohio, is given by Mr. D. D. 
Jackson in his Report on the Sanitary Condition of the Cleve- 
land Water Supply (1912). The amount of available chlorine 
applied varied from 0.50 to 0.86 part per million. 
1912 
Bacteria per cubic centimeter 
Untreated water 
Treated water 
March 
18 
7 
April 
31 
19 
1912 
B. coli per cent, of times found in dilution 
indicated 
Untreated water 
Treated water 
Cubic centimeter tested 
0.1 
1.0 
10.0 
0.1 
1.0 
10.0 
February 
0 
6 
53 
0 
0 
0 
March 
0 
13 
52 
0 
0 
16 
April 
0 
17 
43 
0 
0 
7 404 
WA TER PURI EIC A TION 
The effect of ozone on bacteria, and incidentally on the reduc- 
tion of color in a water, is well illustrated by tables taken from a 
paper by Mr. S. T. Powell1 describing the treatment of an unfil- 
tered colored water obtained from the Herring Run supply of 
the Baltimore County water system. 
According to Mr. Powell the following table illustrates the 
bacterial efficiency of ozone at low concentrations when a good 
contact between the gas and the water is effected by proper mix- 
ing. The water treated was unfiltered. 
Bacteria per c.c. 
Percentage 
removed 
Temperature 
air, deg. F. 
Humidity 
Ozone concentra- 
tion, grams per 
cubic meter of air 
Raw water 
Ozonized 
water 
2,720 
25 
99.1 
30 
90 
1.05 
1,600 
16 
99.0 
24 
80 
1.77 
1,400 
8 
99.5 
34 
76 
1.26 
1,580 
56 
96.4 
54 
70 
1.27 
740 
16 
97.8 
78 
64 
0.58 
4002 
40 
90.0 
80 
76 
0.84 
The effect of the ozone on vegetable coloring matter in the 
water with low concentrations of the gas and under varying 
weather conditions is shown by the following table. 
Color in parts per million 
Percentage 
removed 
Temperature 
air, deg. F. 
Humidity 
Ozone concentra- 
tion, grams per 
cubic meter of 
air 
Raw water 
Ozonized 
water 
35 
15 
57.2 
48 
88 
1.69 
30 
20 
33.3 
40 
58 
2.54 
28 
18 
35.7 
68 
68 
0.76 
28 
20 
28.5 
80 
68 
0.90 
28 
20 
28.5 
80 
76 
0.84 
28 
20 
28.5 
78 
80 
0.63 
50 
40 
20.0 
82 
80 
0.63 
The following bacterial results were obtained some years ago 
in testing ultra-violet ray apparatus designed to be used to ster- 
1 "The Use of Ozone as a Sterilizing Agent for Water Purification." 
S. T. Powell: Jour. New England Water-works Assn., March, 1915. 
2 High turbidity in raw water. DISINFECTION OF WATER SUPPLIES 
405 
ilize the water supply of Marseilles, France. The water came 
from the River Durance and was filtered before sterilization. 
The volume of water treated was about 158,500 gal. in each test.1 
B. coli per 
Bacteria 
B. coli2 per 
Bacteria2 
100 cubic 
per cubic 
100 cubic 
per cubic 
Water treated in 24 
hr., gallons 
hours per 
centimeters 
centimeter 
centimeters 
centimeter 
Before sterilizing 
After sterilizing 
158,000 
98.4 
160 
0.80 
158,000 
98.4 
100-200 
240 
0 
2.00 
158,000 
98.4 
20 
0 
1.00 
158,000 
98.4 
500 
37 
0 
0.00 
147,940 
106.0 
80 
20 
0 
0.07 
158,400 
98.4 
50 
48 
0 
0.00 
158,400 
98.4 
50 
23 
0 
2.00 
158,400 
98.4 
200 
29 
4.20 
158,400 
98.4 
500-1,000 
51 
0 
0.17 
Dr. Max von Recklinghausen gives the following table of bac- 
terial results, obtained by sterilizing with ultra-violet light, in the 
Journal of the New England Water-works Association (June, 1915). 
Before sterili- 
zation, bac- 
teria per c.c. 
After sterili- 
zation, bac- 
teria per c.c. 
Operator 
Sewage bacteria... 
Sewage bacteria.. . 
Sewage bacteria... 
B. coli communis.. 
Sewage bacteria.. . 
B. typhosus 
Sewage bacteria.. . 
8,000 
2,840 
80,000 
11,000 
3,740 
35,000 
273,000 
25,000 
20,000 
0.0 
0.6 
0.0 
12.0 
0.0 
0.0 
0.0 
0.0 
0.0 
j Westinghouse Laboratory. 
Glaser, Austrian Army Med- 
ical Service. 
Burgess, London. 
Thresh and Beale, London. 
Jurist, New York. 
Westinghouse Laboratory. 
Philadelphia Clinical Labora- 
tory. 
Bengal Sanitary Committee. 
These tests were made with somewhat different types of 
apparatus. The first five tests were made with the B-2 type of 
sterilizer, the sixth and seventh tests with the B-5 type, and the 
eighth test with the C-3 type. 
1 Eng. Record, vol. 62, Dec. 10, 1910. 
2 Mean of results from two to five samples. 
3 Mean of results from two to seven samples. 406 
WATER PURIFICATION 
Hygienic Effects of Sterilization.-The marked effect of dis- 
infection upon the sanitary quality of drinking waters, as meas- 
ured by the stopping of epidemics of water-borne diseases like 
typhoid fever, or by the permanent reduction of endemic typhoid, 
has been most striking. The following table compiled by Mr. 
C. A. Jennings1 brings out this effect quite clearly. 
City 
Average typhoid fever death rates per 100,000 persons 
Began using cal- 
cium hypo- 
chlorite 
Before using 
After using 
Period 
Rate 
Period 
Rate 
Baltimore 
June, 
1911 
1900-10 
35 2 
1912-13 
22 8 
Cleveland 
Sept., 
1911 
1900-10 
35 5 
1912-13 
10 0 
Des Moines 
Dec., 
1910 
1905-10 
22.7 
1911-13 
13 4 
Erie 
March 
1911 
1900-10 
38 7 
1912-13 
13 5 
Evanston 
Dec., 
1911 
1907-10 
26.0 
1912-13 
14.5 
Jersey City 
Sept., 
1908 
1900-07 
18.7 
1909-13 
9.3 
Kansas City 
Jan., 
1911 
1900-10 
42.5 
1911-13 
20.0 
Omaha 
May, 
1910 
1900-09 
22.5 
1911-13 
11.8 
Of the eight cities tabulated, the maximum reduction in the 
typhoid death rate, 72 per cent., was in Cleveland, and the mini- 
mum, 35 per cent., in Baltimore. The average reduction was 
51 per cent. The above cities contain a combined population 
of over 2,000,000 people. 
Cost of Apparatus for Disinfection.-The cost of the equip- 
ment for disinfection varies widely, and depends on the particular 
process employed, and whether a simple or elaborate design is 
adopted. Local conditions also naturally affect the cost, as for 
example, the separate housing of the equipment in some instances 
and not in others, the carrying of liquid disinfectants longer or 
shorter distances, or the provision for independent power plants, 
as in the case of ozone and ultra-violet ray plants, instead of 
adapting or using existing sources of power. 
The cost of the equipment for applying bleaching-powder solu- 
tions includes usually the cost of the solution tanks, orifice tanks 
or apparatus, water-supply pipes lines, solution pipe lines and 
sewer lines for the disposal of sludge. Pumps may be required 
in some cases to handle the hypochlorite solution. A building 
for the apparatus may or may not be necessary. Frequently 
1 Proc. Illinois Water Supply Association, March, 1914. DISINFECTION OF WATER SUPPLIES 
407 
space in or slight modifications to buildings already on the ground 
obviate the necessity for separate structures. 
The cost of bleaching-powder disinfecting plants, based upon a 
capacity for treating 1,000,000 gal. of water per day may vary 
from $25 to $500. The larger the capacity of the plant the smaller 
the unit cost, and vice versa. For plants in which 5,000,000 
gal. of water or less are to be treated each day, a total outlay of 
$300 to $500 should be ample if no building to contain the appa- 
ratus is included. For plants having a capacity of 5,000,000 to 
20,000,000 gal. per day, the cost will not probably exceed $100 
per million gallons of daily capacity, and for plants treating above 
20,000,000 gal. daily, the cost may drop as low as $50 or $60 
per million gallons of daily capacity. 
Apparatus now on the market for applying liquid chlorine as a 
gas does not vary materially in size or design for any given type, 
whether it is intended to be used to treat a few hundred thousand 
gallons, or many millions of gallons of water daily. Single 
units cost approximately from $350 to $700 each. A recent 
installation for a large city which was intended to treat 60,000,- 
000 gal. of water per day and which consisted of three units, 
cost $2,000. Ten chlorinators placed in the purification plants 
in Philadelphia several years ago cost $9,750. 
The cost of ozonizing plants properly includes the ozonizers 
with the power plant to operate them, the refrigerating plant to 
cool and dry the air, and the sterilization towers. The cost for 
such a plant will naturally vary greatly, depending upon the 
local conditions. Mr. S. T. Powell in estimating the operation 
costs for the Herring Run plant of the Baltimore County Water 
& Electric Co. takes as a basis $5,000 per million gallons as the 
original cost for 10,000,000-gal. units or for larger installations, 
and which would include roughing filters and a refrigerating 
plant. 
The cost of ultra-violet ray apparatus is about $750 per thou- 
sand gallons per hour capacity for the larger machines, and rising 
to $900 for the smallest. Large installations will, of course, cost 
much less in proportoin to capacity. The estimated cost for 
the Niagara Falls, N. Y., plant, which would have treated 16,- 
000,000 gal. daily, was $19,800 for the canals and lamps, and 
$2,200 for transformers and building to house the same. This 
plant has not been installed at Niagara Falls, although the proj- 
ect was given careful consideration and study. 408 
WATER PURIFICATION 
Cost of Operation of Disinfection Plants.-The cost of operat- 
ing disinfection plants varies widely, as would be expected from 
the differences in the initial costs of the plants, and of the various 
agencies which they employ. The character of the water being 
treated also affects the cost, since the amounts of chemicals used 
or of power expended to produce disinfection is obviously not 
equal for all waters. 
For disinfecting with bleaching-powder solutions the cost per 
million gallons of water treated will vary in the United States 
from 10 to 50 cts., the average cost being about 25 cts. These 
variations are due in part to the range in the cost of the bleaching 
powder in different parts of the United States. Between the 
Atlantic coast and the Mississippi River the normal price for 
bleaching powder is about $1.78 per hundred pounds, while from 
the Mississippi River to the Pacific coast, the cost ranges from 
$2.50 to $3 per hundred weight. The smaller plants buy in 
relatively small packages, and hence must pay a higher price than 
do the large plants buying in large drums and in considerable 
quantities. The depreciation in the value of bleaching-powder 
plants is high, and, of course, adds to the operation and main- 
tenance costs. 
Where liquid chlorine is used, the costs do not differ much from 
those incurred by the use of bleaching powder. The depreciation 
in value of apparatus is not, perhaps, quite as great as where 
bleaching powder is employed, but, nevertheless, must be taken 
into account in any true estimate of the cost of using this agent. 
With a normal price of about 8 cts. per pound for liquid chlorine, 
the cost per million gallons for disinfection at the Cincinnati 
filtration plant varies from 8 to 12 cts. for the chemicals alone. 
The cost of disinfection with ozone is much higher than where 
the hypochlorites or liquid chlorine are used. To produce a con- 
centration of 2.5 grams of ozone per cubic meter of air at the 
Petrograd plant in Russia, the ozonizers consumed 57-kw.-hr. per 
million gallons of water treated, and the compression of the 
ozonized air in order to force it into the towers required 76 kw.-hr. 
more of electrical energy. In addition to these costs are those 
resulting from drying and cooling the air by refrigeration. It 
was estimated that the total cost of filtration and disinfection 
with ozone at the Petrograd plant was $15 to $17 per million 
gallons, of which about one-half was chargeable to the ozone 
process. DISINFECTION OF WATER SUPPLIES 
409 
Mr. S. T. Powell of the Baltimore County Water & Electric 
Go., estimates that the cost of ozone sterilization in this country 
will be between $2.50 and $4 per million gallons. His estimate 
"is based upon a maximum installation cost of $5,000 per million 
gallons on 10,000,000-gal. units, or larger installations, which includes 
roughing filters and refrigeration plant, interest on the investment at 
5 to 5J$ per cent., depreciation at 5 per cent., current at a maximum 
of 3 cts. per kilowatt-hour, as well as supervision and management." 
The consumption of electrical energy in operating ultra-violet 
ray apparatus is estimated by Dr. Max von Recklinghausen not 
to exceed 100 kw.-hr. per million gallons. The maintenance cost 
of the lamps must be added to the cost of the current. Lamp 
renewals may cost from 50 to 60 cts. per million gallons. Even 
with a low cost for the electrical current, it would not appear as 
though disinfection by this process could cost less than 75 cts. 
per million gallons. With current at 1 ct. per kilowatt-hour and 
a lamp renewal cost of 50 cts., the cost per million gallons would 
be $1.50. 
References 
1. George A. Johnson: "The Purification of Public Water Supplies." 
U. S. Geological Survey, Water Supply Paper, No. 315. 
2. Chas. G. Hyde: "Sterilization of Water Supplies by the Use of 
Hypochlorites." Proc. League California Municipalities, October, 1911. 
3. "Report of Committee on Water Supplies." Am. Jour. Public Health, 
December, 1914. 
4. "J. W. Ellms: "Disinfection of Public Water Supplies." The American 
City, December, 1913. 
5. A. Lederer and F. Bachmann: "Some Interesting Observations on 
the Disinfection of Lake Waters with Calcium Hypochlorite." Proc. 
Illinois Water Supply Association, March, 1912. 
6. Proc. Illinois Water Supply Association, March, 1913 and 1914. 
7. F. D. West: "Liquid Chlorine Disinfection at Philadelphia." Eng. 
Record, May 16 and 23, 1914. 
8. Eng. Record, vol. 62, Dec. 19, 1910; also vol. 69, p. 674. 
9. S. T. Powell: "The Use of Ozone as a Sterilization Agent for Water 
Purification." Jour. New England Water-works Assn., March, 1915. 
10. Max von Recklinghausen: "Ultra-violet Rays for Water Purifica- 
tion." Jour. New England Water-works Assn., June, 1915. 
11. "Violet-ray Sterilization for Niagara Falls Filters." Eng. Record, 
June 20, 1914. 
12. Max von Recklinghausen: "Sterilization of Water with Ultra- 
violet Rays." Eng. Record, Aug. 2, 1913. 
13. C. A. Jennings: "What Hypochlorite Sterilization of Water Supplies 
Has Accomplished in Several Cities." Proc. Eighth International Con- 
gress of Applied Chemistry. CHAPTER XXX 
THE REMOVAL OF DISSOLVED MINERAL MATTER FROM 
WATER 
WATER SOFTENING 
The origin and general character of the mineral matter dissolved 
in natural waters were discussed in Chapter II. 
The amount of mineral matter dissolved in natural waters 
varies greatly. The reasons for desiring to reduce the quantity of 
these mineral compounds in solution are to render the water more 
suitable for domestic use, or to make it better fitted for industrial 
purposes. If the water is being used as a public water supply, 
both of these reasons may be quite properly urged as sufficient 
grounds for installing some form of purification. 
It is obvious that the nature of the dissolved impurities deter- 
mines the character of the purification method to be employed. 
It is also apparent that means may be used for treating a water 
for industrial purposes, that are not permissible in connection 
with a public water supply. The hardness of a water, that is 
caused by salts of calcium and magnesium, does not in itself 
make it unfit for drinking purposes. In fact, a small amount of 
hardness in a drinking water is probably beneficial to those who 
consume it, and does not render it unfit for most industrial uses. 
Only when the amount of the dissolved salts of calcium and mag- 
nesium become excessive is it necessary to resort to methods of 
purification. On the other hand, even comparatively small 
amounts of iron, manganese, or of free acid are objectionable 
for domestic purposes, and for most industrial uses. Waters 
containing large amounts of saline matter are unfitted for both 
drinking and technical purposes. For example, great quantities 
of the soluble salts of sodium, more especially the chloride, sul- 
phate and carbonate of this element render a water practically 
worthless, so far as being able to make it fit for most purposes by 
any methods of purification now known, other than by distilla- 
tion. This latter means of removing the dissolved salts is ob- 
viously of limited value where considerable quantities of water 
are to be purified. 
410 REMOVAL OF DISSOLVED MINERAL MATTER 
411 
According to Deussen and Dole1 waters that have a hardness 
that does not exceed 200 parts per million, or that do not contain 
enough mineral matter to have a disagreeable taste, are accept- 
able for drinking and cooking. A hardness greater than 1,500 
parts per million renders a water undesirable for cooking, while 
a water containing approximately 250 parts per million of chloride 
has a slightly salty taste. Somewhat less of the carbonate and 
more of the sulphate may be tolerated. However, a drinking 
water containing more than 300 parts per million of carbonate, 
1,500 parts of chloride, or 2,000 parts of sulphate is unhealthful 
to most persons. On these bases a sodium chloride water show- 
ing more than about 3,000 parts per million of mineral matter, or 
a sulphate water showing more than 3,500 parts of mineral matter 
is reasonably classed as unfit for domestic use. The drinking 
of such water is usually characterized by intestinal disturbances. 
The expense to a community using a hard water for soap alone 
is considerable, and may be sufficient in itself to warrant the cost 
of softening. Lessened troubles with plumbing for hot-water 
systems, the reduction of scale in boilers, and the greater value of 
a soft water for many manufacturing processes, are all excellent 
reasons for resorting to methods of purification for reducing ex- 
cessive quantities of lime and magnesia, or eliminating com- 
pounds of iron and manganese. 
Removal of Salts of Calcium and Magnesium.-The process of 
removing the bicarbonates, sulphates and chlorides of calcium 
and magnesium is commonly termed 11 water softening," that is 
the reduction of the " hardness " of a water produced by the pres- 
ence of these salts in solution. Temporary hardness,2 caused by 
the bicarbonates of calcium and magnesium, and in some cases 
iron, is that portion of the "total hardness" of a water that may 
be removed by boiling. The remainder of the total hardness is 
commonly known as the "permanent hardness" of the water, 
and is almost always entirely due to the sulphates and chlorides 
of calcium and magnesium. In some cases nitrates of these 
elements may contribute more or less to the permanent hardness 
1 Alexander Dettssen and R. B. Dole: "Ground Waters in LaSalle 
and McMullen Counties, Texas." Water Supply Paper No. 375, U. S. 
Geological Survey. 
2 For a scientific classification of waters with reference to their dissolved 
mineral constituents, the reader is referred to Chapter II. The terms used 
above are those commonly employed in describing this class of waters, and 
are not strictly accurate. 412 
WATER PURIFICATION 
of the water. Liquid waste products from industrial processes 
that contain free mineral acids or sulphates of iron, aluminum or 
manganese are not infrequently discharged into streams, and 
may create a "hardness" in the water that renders it unfit for 
almost any use, unless the "acidity" produced is neutralized. 
It is evident, therefore, that water-softening processes must be 
adapted to the particular character of the compounds producing 
"hardness," and that different reagents are required to obtain 
the results desired. 
Chemistry of Water Softening.-The solubility of the carbon- 
ates of calcium and magnesium in water is greatly increased by 
the presence of carbonic acid in solution. Under these conditions 
the carbonates are in reality bicarbonates of these elements. 
This free and loosely bound carbonic acid is easily removed by 
boiling, but may also be removed by caustic lime. The use of 
lime for removing the free and half-bound carbonic acid, thereby 
converting the bicarbonates into normal or monocarbonates, is 
the basis of one of the oldest and most efficient water-softening 
processes known. The chemical reactions involved are as 
follows: 
1. H2CO3 + Ca(OH)2^CaCO3 + 2H2O. 
2. CaCO3, H2CO3 + Ca(OH)2^2CaCO3 + 2H2O. 
3. MgCO3, H2CO3 + 2Ca(OH)2^2CaCO3 + Mg(OH)2 
+ 2H2O. 
There are two features of these reactions which should be noted, 
namely, that the caustic lime added is itself precipitated as the 
carbonate of calcium, the same as is the calcium carbonate de- 
prived of its half-bound carbonic acid; and that the magnesium 
is precipitated as the hydrate. The molecular proportions as 
shown by the above reactions must exist if the greatest reduction 
possible is desired. On account of the slight solubility of calcium 
carbonate in water, when no free or half-bound carbonic acid is 
present, the complete precipitation of the lime is impossible. 
Magnesium carbonate in the absence of free and half-bound 
carbonic acid is even more soluble in water than calcium carbon- 
ate, and unless the caustic lime applied is in sufficient amount to 
produce magnesium hydrate, a considerable quantity of the car- 
bonate still remains dissolved. The solubility of the normal 
carbonates of lime and magnesia, and the slow precipitation of 
these salts as the reaction approaches theoretical completion, 
accounts for the residual alkalinity which waters softened with REMOVAL OF DISSOLVED MINERAL MATTER 
413 
caustic lime always exhibit. The so-called "after-precipitation 
effects" in waters softened with lime and soda ash, are due to the 
physical properties of the precipitates produced by the softening 
reactions. It is simply a case of the slow deposition of salts that 
have been in colloidal solution. 
The removal of permanent hardness may be accomplished'by 
another reagent, namely, sodium carbonate or soda ash. The re- 
actions which take place are as follows: 
1. CaSO4 + Na2CO3^CaCO3 + Na2SO4. 
2. (a) MgCl2 + Na2CO3r±MgCO3 + 2NaCl. 
(&) MgCO3 + Ca(OH)2<=±Mg(OH)2 + CaCO3. 
The reactions involving sulphates, chlorides and nitrates of 
calcium and magnesium, and sodium carbonate, all consist of an 
exchange of acids, whereby the insoluble carbonates of these bases 
are precipitated, and a corresponding amount of the soluble sul- 
phates, chlorides and nitrates of sodium remain in solution. The 
soluble salts are not reduced by the treatment, but the less 
objectionable salts of sodium are substituted for those of calcium 
and magnesium. 
Free acid or the sulphates of iron, aluminum and manganese 
may be neutralized with sodium carbonate. The hydroxides of 
the bases just mentioned are formed, and on account of their in- 
solubility may be easily removed by filtration. In the case of iron 
and-manganese, however, oxidation of the hydroxides first formed 
are usually necessary in order to effect complete precipitation. 
Reactions between Natural and Artificial Zeolites and Hard 
Waters.-By the action of hard waters on certain complex sili- 
cates of aluminum and sodium, an exchange of the calcium or 
magnesium for the sodium of the zeolite may be effected. The 
soil studies of Dr. R. Gans led him to divide zeolites containing 
aluminum into two general classes, which he describes as: (1) 
"double silicates of aluminum," and (2) "aluminate-silicates." 
The first class of zeolites exhibits little absorptive power for any 
exchange of bases as noted above. The second class of zeolites, 
however, appears to possess this property in quite a marked 
degree. Gans was able to produce artificial zeolites which showed 
even greater absorptive power than the natural zeolites. The 
artificial zeolite of Gans has the following probable composition: 
2SiO2.Al2O3.Na2O + 6H2O. 
Under the trade name of "Permutit" this artificial zeolite has 414 
WATER PURIFICATION 
been placed upon the market as a water-softening agent. The 
nature of the reaction between this compound and a water con- 
taining compounds of calcium and magnesium may be expressed 
as follows: 
1. 2SiO2.Al2O3.Na2O + CaCO3, H2COM2SiO2.Al2O3.CaO + 
2NaHCO3. 
2. 2SiO2.Al2O3.Na2O + MgSOM2SiO2.Al2O3.MgO + Na2SO4. 
The zeolite may be regarded as a solid reagent, exchanging one 
of its constituents for a base in the solution on which it is acting. 
Since the zeolite remains insoluble, it may be regenerated by a 
reversal of the reaction. This is accomplished by the action of 
sodium chloride, a strong solution of which is permitted to react 
on the solid zeolite for a period equal to that to which it was sub- 
jected by the hard water. The sodium of the salt substitutes 
itself for the previously absorbed calcium or magnesium, and the 
resulting chloride of the alkaline earth base goes into solution, 
and may be washed out with clean water. The reversible 
character of the reactions is simply the result of molecular concen- 
tration, or as it is sometimes called the law of mass action. 
The reaction between the sodium chloride and the zeolite in 
which the sodium has been replaced by the alkaline earth base 
calcium or magnesium may be shown as follows: 
2SiO2.Al2O3.CaO + 2NaCM2SiO2.Al2O3.Na2O + CaCl2. 
Artificial zeolite as prepared by Gans was made by fusing to- 
gether feldspar, kaolin, silica and sodium carbonate in definite 
proportions. After extracting the melt with water, the porous 
material which remained was found to contain 46 per cent, of 
silica, 22 per cent, of alumina, 13.6 per cent, of sodium oxide and 
18.4 per cent, of water. 
Other Chemical Reagents Used in Water Softening.-The 
reagents described above are safe and suitable for softening a 
water for domestic purposes. They are also commonly employed 
for reducing the hardness of waters that are to be used for tech- 
nical purposes. There are, however, quite a number of effective 
chemical agents that may be used for softening waters for techni- 
cal purposes that could not be safely employed in treating waters 
that were to be used for drinking. For example, caustic soda, 
the hydrate, carbonate or aluminate of barium, the silicate or 
oxalate of sodium and tri-sodium phosphate are all chemicals 
that have been used as water-softening agents. Aqueous vege- REMOVAL OF DISSOLVED MINERAL MATTER 
415 
table extracts of oak, chestnut, logwood and quebracho are also 
employed on account of the tannin which they contain. 
For assisting the softening of water in conjunction with heat, 
principally for the purpose of modifying the physical properties of 
the precipitates, numerous mechanical agents are used, such as 
graphite, talc, petroleum oils, and gelatinous or starchy substances 
like Irish moss and dextrine. These materials are usually 
employed in connection with some chemical agent such as soda ash 
or caustic soda. Many of the above-mentioned substances form 
the bases of the numerous "boiler compounds" so commonly 
employed in softening water within a steam boiler. Although 
the function of a boiler is to convert water into steam, it unfor- 
tunately too often has to play the role of a "softening tank" as 
well, when only hard waters are available for steam raising pur- 
poses. Unless a water contains only a comparatively small 
amount of the salts of calcium and magnesium, their precipitation 
by any process should be carried out in suitable apparatus out- 
side of the boiler instead of within it. The boiler may then be 
kept free from scale or deposits, however they may be formed, and 
can be operated at its highest efficiency. 
The object in using practically any of the reagents named is 
to produce a deposit of such a character that it will not form a 
scale on the iron within the boiler. The carbonates of calcium 
and magnesium, and the hydrate of the latter element are usually 
precipitated within the boiler in a finely divided condition, and 
do not adhere to the iron very closely. On the contrary, the 
sulphate of calcium produces an extremely adherent and porce- 
lain-like scale, which forms a non-conducting heat layer on the 
iron, and thereby prevents the transfer of heat from the iron to 
the water. Overheating of the iron not infrequently results, 
therefore, thus ruining the tubes or sheets of the boiler, and 
causing leaks and possible blowouts of a much more danger- 
ous character. If preliminary treatment of the water is effected 
in special apparatus, such as softening tanks and boiler feed- 
water heaters, these troubles with the boiler may be practically 
eliminated. 
The corrosive character of waters containing considerable 
amounts of magnesium chloride, or of magnesium salts in con- 
junction with sodium chloride is due to the formation of hydro- 
chloric acid at the high temperatures that may be produced within 
a steam boiler. Some of the magnesium chloride is hydrolyzed, 416 
WA TER P URIF IC A TION 
decomposing into magnesium oxide and hydrochloric acid. The 
latter passes into the steam, while the oxide remains in the water. 
For this reason sea water can not be used in ships' boilers. 
The excessive amounts of dissolved salts found in some natural 
waters, as for example in certain mineral water springs, or in the 
"alkali waters" of the Western plains, can not usually be re- 
moved by any method of purification now known, other than by dis- 
tillation as previously noted. The use of such waters in boilers 
may produce frothing and foaming, and thus become the cause 
for the evils arising from the passage of water mixed with steam 
into the steam pipes. Treatment of such waters with chemical 
agents can not usually effect any reduction in the total amount 
of dissolved solids, although it may be able to change their char- 
acter to some extent. For example, the use of barium hydrate 
in a water containing large amounts of sulphates will precipitate 
the latter as barium sulphate, but leaves the equivalent amounts 
of the hydrates of the bases deprived of their sulphate ion. 
Mr. R. B. Dole1 states that: 
"In locomotive service, it is in general economical to treat water 
containing 250 to 850 parts per million of incrustants, and to treat 
those containing less than 250 parts if the scale formed contains 
much sulphate. As the incrusting solids may commonly be reduced 
to 80 or 90 parts per million, the economy of treating boiler waters 
deserves consideration in a region where many supplies contain 300 
to 500 parts per million of incrusting matter. 
"The amount of mineral matter that makes a water unfit for boiler 
use depends on the combined effect in boilers of the softening reagents 
used with such waters and of the constituents not removed by softening. 
Sodium salts added to remove incrustants or to prevent corrosion 
increase the foaming tendency, and this increase may be great enough 
to render a water useless for steaming. It is not of much benefit to 
soften a water containing more than 850 parts per million of non- 
incrusting material and much incrusting sulphate. Trouble from 
priming in locomotive boilers begins at a concentration of about 1,700 
parts per million of foaming constituents, and the limit of safety for 
stationary boilers is reached at a concentration of about 7,000 parts. 
Though waters containing as high as 1,700 parts per million of foaming 
1 U. S. Geological Survey, Water Supply Paper No. 398. 
Proc. Am. Ry. Eng. and Maintenance of Way Assoc., vol. 8, p. 601, 
1907. 
Proc. Am. Ry. Eng. and Maintenance of Way Assoc., vol. 6, p. 610, 
1905. REMOVAL OF DISSOLVED MINERAL MATTER 
417 
constituents have been used, it is usually more economical to incur 
considerable expense in replacing such supplies by better ones." 
Practical Methods Employed in Water Softening.-The soft- 
ening of large volumes of water with a chemical reagent necessi- 
tates the proper and thorough mixing of the reagent with the 
water, the requisite period of time for the reaction to complete 
itself, and provision for the adequate disposal of the waste prod- 
ucts of the reaction. The control of the temperature conditions 
under which the reaction should proceed is highly desirable, but 
is not usually practicable. 
Application of Chemical Reagents.-The general methods and 
apparatus employed in applying chemicals have been discussed in 
Chapter XXII. Only those features relating more especially 
to water softening with lime and soda ash, and to the technical 
features of water softeners of this type need be described at this 
point. 
The lime is applied to the raw water as milk of lime or as a 
lime-water solution. The latter method is undoubtedly the one 
to be preferred, but is not so generally used on account of the 
large storage space required for the lime-water solution. Milk of 
lime suspensions must be constantly stirred if a fairly uniform 
strength of the mixture is to be maintained. Where soda ash is 
also used, it is not infrequently dissolved with the lime in the 
same tank. The mixture, therefore, actually consists of a sus- 
pension of hydrated lime in a caustic soda solution. As a by- 
product of the reaction calcium carbonate will, of course, be also 
found in suspension. 
The methods of measuring the quantity of applied chemical 
solution are varied. Submerged orifices, proportional feed pumps, 
Sutro weirs, and even dry feeding of the chemicals are all in 
actual use. This class of devices was described in a preceding 
chapter. Another automatic arrangement not infrequently used 
in water-softening apparatus, more especially in those machines 
which soften water for industrial purposes, consists of a double- 
compartment oscillating tank or bucket, which measures out the 
hard water, and which actuates valves admitting a definite 
dosage of the chemical reagent solution at each tipping of the 
tank. The chemical solution is kept in a semi-cylindrical vessel 
directly above the tipping tank. The hard water flows first into 
one compartment of the tipping bucket, which, when full, tips 418 
WATER PURIFICATION 
and discharges the water it contained. This movement brings 
the other compartment under the supply pipe and the operation 
is repeated. The movement of the bucket in tipping raises by 
means of a cam a stopper valve in the bottom of the chemical 
supply tank. This valve is so adjusted that it allows the re- 
quired amount of chemical solution to pass before it closes again, 
being assisted in so doing by a weight and spring. The milk of 
lime soda ash solutions are kept mixed by a stirrer actuated by 
the motion of the tipping bucket. The movement of the bucket 
has also been used to operate chemical feed pumps. 
Mixing the Chemicals with the Hard Water.-Since the inti- 
mate mixing of the chemical reagents with the water is necessary 
to produce the desired exchange of acids and bases, this phase of 
the process is of the greatest importance. This is accomplished 
in two ways, namely, by passing the treated water through highly 
baffled compartments, or by mechanical agitation by means of 
stirrers. In a few plants agitation is effected by using compressed 
air. The baffling method is the only one available where large 
volumes of water are handled, and is exemplified by the plants at 
Columbus, Ohio, Grand Rapids, Mich., and New Orleans, La. 
Mechanical agitation is employed at Owensboro, Ky., where 
there are foui mixing chambers containing four sets of agitators. 
A mixing period of about 1 hr. is used at several of the large 
plants, while in small water-softening machines of the continuous 
type only a few minutes is allowed. The best period for agitation 
doubtless varies with different waters, and considerations of cost 
for storage in tanks or reservoirs will usually set the limit beyond 
which it is not practical to go. The colloidal character of the 
precipitates that are formed requires more or less agitation for 
their complete deposition. A short period of violent agitation 
is probably better than a longer period of less active agitation. 
Mixing of previously formed sludge deposits with freshly treated 
water, by pumping the latter into the bottom of a settling tank 
through a perforated pipe lying in the sludge, has also been used 
to facilitate the deposit of the colloidal precipitates produced by 
the reactions. 
The Settling of the Precipitates Formed.-The finely divided 
character of the precipitates, due largely to their colloidal char- 
acter, makes their deposition a rather slow process. In large 
plants, where a continuous flow of water must be maintained, 
elaborately baffled basins are used. These basins have been REMOVAL OF DISSOLVED MINERAL MATTER 
419 
described and their essential features discussed in previous 
chapters dealing with coagulation and settling in connection with 
the municipal water purification plants at St. Louis, Mo., New 
Orleans, La., and Columbus, Ohio. In smaller plants, chiefly for 
industrial purposes, two types of water-softening apparatus are 
Fig. 134.-Intermittent water softener for railroad water station, plan. 
5 in.Wash Pipe to connect with 
8 in.Vit Drain Pipe to River 
Siding 
Main Track 
PLAN 
found, namely, those that are intermittent and those that are 
continuous. 
Intermittent Water Softeners.-In apparatus of this type the 
water to be treated is pumped into tanks, treated with the chem- 
ical reagents, sometimes agitated mechanically to facilitate the 420 
WATER PURIFICATION 
Note:5 in. Wash Pipe to connect 
with 8 in.Vit.Pipe to River 
ELEVATION & SECTION 
8 in.Vit.Drain to River 
Fig. 135.-Intermittent water softener for railroad water station, elevation 
and section. 
- 3 x 17 Joists one 
between each Tank 
to tie Bl'd.together, 
Main Line C.H. 4 D.Ry. 
Fig. 136.-Intermittent water softener for railroad water station, section. 
SECTION REMOVAL OF DISSOLVED MINERAL MATTER 
421 
chemical reactions, and finally permitted to stand quiescent 
until the precipitate is deposited. As from 4 to 6 hr. must be 
allowed for settling, and as to this must be added the time of filling, 
emptying and cleaning the tank, considerable tank capacity is 
required. The chemical solutions are prepared at the ground 
level of the tanks, and are pumped into the latter; or the solid 
chemicals are hoisted to the top of the tank and dissolved and 
applied at this point. Agitation of the treated water is effected 
by paddles turned by an electric motor or water motor, or by 
compressed air forced in through a grid of perforated pipes on 
the bottom of the tank. 
The softened water is usually withdrawn from the surface by 
means of a floating pipe. While one tank is being drawn down, 
another must be filled, treated with chemicals, and allowed to 
stand for a period sufficient to produce a fairly clear water, so 
that it may be ready to supply softened water when the first tank 
has been emptied. 
Continuous Water Softeners.-The intermittent system re- 
quires considerable space for the tanks, and must receive quite a 
little attention if properly operated. Continuous water softeners 
have been designed, therefore, to meet these objections. They 
are all more or less similar in their general type of construction, 
and meet the demand for economy of space and automatic opera- 
tion. Their chief differences are found in the special devices used 
to measure the water and the applied chemical solutions, and in 
the means employed to facilitate the deposition of the precipitate 
in the water which is moving slowly but continuously through 
the compartments of the softener. 
The water enters softeners of this type generally at the top of 
a tall tank (Fig. 137). As it discharges, it is made to turn a 
waterwheel, the power from which is utilized to operate agitators 
in the softening and chemical tanks, and to perform other mechan- 
ical work, such as the operation of pumps for handling and pre- 
paring the chemical solutions. 
The devices used for proportioning the chemical solutions to 
the flow of water are varied in design. In general, they consist of 
weirs or orifices over or through which the flow of the chemical 
solution is made proportional to the flow of hard water into the 
tank. This is usually effected by a float in the hard-water 
receiving box. The float rising and falling with the fluctuations 
in flow operates cutoffs in weirs under constant head, or moves 422 
WATER PURIFICATION 
orifices up or down in order to maintain heads that vary but 
still are proportional to the head over the hard-water orifice. 
The water after receiving its chemical reagents passes into the 
mixing compartment, and flows downward. This compartment 
is often made cone-shaped, so that the velocity of the flow will 
Fig. 137.-Reisert automatic water softener. 
lessen as it nears the bottom of the tank. In others it is purely 
a mixing compartment, being provided with mechanical agitators 
in the form of stirrers. 
After leaving the mixing or central compartment of the soft- 
ener, the water turns and rises slowly in the annular space around 
the mixing compartment. Baffling, consisting of inclined per- REMOVAL OF DISSOLVED MINERAL MATTER 
423 
Sectional Plan 
Sludge Valves 
Longitudinal Section 
Fig. 138.-Plan and section of municipal water-softening plant at 
Owensboro, Ky. 424 
WATER PURIFICATION 
forated plates, is sometimes used to facilitate the deposition of 
the sediment. In other apparatus more space is provided in 
Fig. 139.-Section through settling basins and mixing tanks of water-softening plant at Owensboro, Ky. 
Settling Basin / 
-Slaking Tank 
Down Take 
^'Pes ,5b 
Section in Chemical Tanks 
Slaking Tank- 
Settling Basm 2 
this settling chamber, so as to maintain lower velocities, but no 
baffles are used. 
For the final clarification of the softened water filters (Figs. 
138 and 139) are generally used. These are frequently made of REMOVAL OF DISSOLVED MINERAL MATTER 
425 
wood fiber, filtration being upward through a somewhat shallow 
layer placed between perforated plates. Filters of sand and of 
gravel are also used for the purpose of clarifying the water. 
Rapid sand filters of the usual type are employed in large 
plants that are softening water for municipal purposes. Sand 
and gravel filters, as well as those of fiber, must have their 
filtering material renewed from time to time, since the gradual 
accumulation of sediment within the beds finally renders them 
useless. In the case of wood-fiber filters, the fiber is thrown 
away after it becomes fouled. Sand filters may be partially 
cleaned by the usual methods of washing with a reverse flow ol 
clean water; but in time this method produces little effect on 
account of the cementing action of the deposited colloidal pre- 
cipitates on the grains of sand. Cleaning the sand with hydro- 
chloric acid, whereby the lime and magnesium compounds are 
dissolved and washed out, is sometimes resorted to. Usually a 
renewal of the sand is a cheaper method of reconstructing the 
filter bed, than to attempt to clean the old sand, unless very 
large quantities of sand are to be replaced. 
Messrs. Hoover and Scott1 state that the sand of the Columbus, 
Ohio, filters is composed practically of 2 parts of scale and 1 part 
of sand. The sand has become so badly coated that the effective 
size has increased from 0.41 to 0.62 mm. These sand grains 
cement themselves together, forming hard lumps often as large 
as bushel baskets, and so hard that they must be removed, from 
the bed to be broken up. It is necessary at this plant to shovel 
the sand from one bay of the filter to another about twice a year, 
and requires the labor of five men about 2 weeks' time. 
Filtration through sand of a water that has been subjected 
to the softening process will still farther reduce the residual 
hardness. A 30-in. sand layer at the Columbus, Ohio, plant will 
reduce the hardness on an average of 20 per cent., when the raw 
water is at its maximum hardness. Experimentally it was shown 
that a 40 per cent, reduction in the hardness could be produced 
with a bed of sand 6 ft. deep. The phenomenon is one of absorp- 
tion through the agency of physical forces, and is not a chemical 
one. 
Zeolite Filters.-The practical application of the principles 
involved in the softening of hard waters by natural or artificial 
zeolites is effected in ordinary filter tanks of the pressure or 
1 Monthly Bulletin of the Ohio State Board of Health, December, 1914. 426 
WATER PURIFICATION 
gravity type, and in which the zeolite takes the place of the usual 
filter bed. The zeolite, however, does not act as a filtering me- 
dium, and turbid waters must first be filtered in order to obtain 
the best results. Distributing layers of gravel above and below 
the zeolite layer are sometimes used. In one of the cuts shown 
(Figs. 140-141) the upper gravel layer rests upon a perforated 
plate. Between the bottom of the plate supporting the upper 
gravel layer and the top of the zeolite is an open space. The 
zeolite, which may range in thickness from 20 to 40 in., rests 
upon a gravel layer, which in turn is supported on another per- 
■Manhole 
AirVent 
\ Salt Valve 
Salt Solution Pipe-^ 
Soft Water 
Outlet 
Salt 
Solution 
Tank 
Water Main 
Valve v 
to Empty 
Outlet to Drain 
Fig. 140.-Zeolite water softener of the pressure type. 
forated plate. Suitable piping, valves and manholes must, of 
course, be provided for operating the softener. 
Brine solution tanks are necessary in this process, since the 
zeolite is regenerated with a 10 per cent, sodium chloride solution. 
These tanks are sometimes placed above the softening tanks, 
and sometimes at the same level as the softener, in which latter 
case the brine solution must be pumped or ejected into the soft- 
ener. Where the plant must be operated constantly, all appara- 
tus must be virtually in duplicate, since the regenerative period 
is practically as long as the operating period. REMOVAL OF DISSOLVED MINERAL MATTER 
427 
The rate at which the water passes through the zeolite depends 
upon its depth and on the hardness of the water. Time must be 
allowed for the penetration of the water into the interior of the 
zeolite grains. Wickware1 states the usual, rate of flow through 
the bed is from 10 to 16 ft. per hour. The extreme limits of 
speed are for water containing 0.01 per cent, of lime, approxi- 
mately 27 ft. per hour; with 0.02 per cent., 16 ft.; and with 0.03 
'Raw Water Inlet 
Flint Bed- 
-Rinsing Water Inlet 
Regenerating Valve 
Gage Cock- 
Permutit- 
Charge 
XOutlet for 
Treated Water 
-Flint Bed 
Outlet for 
'Salt Solution 
Rinsing Water Outlet 
Fig. 141.-Zeolite water softener of the gravity type. 
per cent., 10 ft. per hour. Hoover2 and Scott found the most 
efficient rates for a water having a total hardness of 392 parts 
per million to be 55,000,000 gal. per acre per day for a gross rate, 
and 25,000,000 gal. per acre per day for a net rate. This gross 
rate (55,000,000 gal. per acre per day) is practically equal to 
the passage of the water through the bed at a rate of 7 ft. per 
hour. The water experimented with by these investigators had 
a hardness in terms of calcium carbonate of 0.0392 per cent. 
1 Francis G. Wickware: "European Power Plant Practice." Practical 
Engineer, June 1, 1912. 
2 C. P. Hoover and R. D. Scott: "Water Softening by the Permutit 
Process." Ohio Public Health Journal, August, 1915. 428 
WATER PURIFICATION 
This rate would appear to be in general agreement with the 
optimum rates stated by Wickware. 
Mr. R. N. Kinniard {Engineering Record, Dec. 25, 1915) ex- 
perimented with a natural zeolite and found it to act similarly 
to the artificial zeolite prepared by Gans. Dr. Edward Bartow 
has also made experiments with this same material and obtained 
Fig. 142.-A Permutit (zeolite) water-softening plant in Germany. 
results similar to those obtained by Mr. Kinniard. In these 
experiments rates of filtration of 2 gal. per square foot per minute 
(approximately 16 ft. per hour) gave good results on a water 
having a carbonate hardness of about 300 parts per million (0.03 
per cent.). 
The amount of salt required to regenerate artificial zeolite per 
1,000 gal. of water treated for each 100 parts per million of hard- REMOVAL OF DISSOLVED MINERAL MATTER 
429 
ness was found by Hoover and Scott to be 5.92 lb. Kinniard's 
figures differ somewhat from these as he was able with the natural 
zeolite to regenerate with 3M lb. of salt per 1,000 gal. of water 
treated for each 100 parts of hardness (Fig. 142). 
Practical Results Obtained in Water Softening.-The practical 
benefits derived from a well-softened water used as a public 
supply lay chiefly in the reduction of the amount of soap used, and 
in the diminished cost for operating apparatus for the production 
of hot water and steam. For domestic purposes in bathing, in 
laundry work and in culinary operations, a soft water is much to 
be preferred to a hard water. Certain industries, such as paper 
making, tanning, dyeing and bleaching are able to produce a 
better product with soft than with hard waters. Mr. George C. 
Whipple has brought out in a striking manner in his book on 
"The Value of Pure Water," the depreciated value of a water due 
to its hardness. The following data are taken from one of the 
tables in the above-mentioned book. 
State 
City 
Source of supply 
Total hard- 
ness, parts 
per million 
Depreciation 
per 
million gallons 
Maine 
Waterville 
Messalonskee River 
15 
81.50 
Maine 
Augusta 
Kennebec River 
20 
2.00 
Massachusetts. 
Cambridge 
Storage Reservoir 
33 
3.30 
New York 
Albany 
Hudson River 
64 
6.40 
Pennsylvania. . 
Philadelphia 
Schuylkill River 
179 
17.90 
Ohio 
Toledo 
Maumee River 
200 
20.00 
Ohio 
Columbus 
Scioto River 
335 
33.50 
At Winnipeg, Manitoba, Mr. Whipple states that the hardness 
of the water supply before treatment was 580 parts per million. 
By softening the hardness was reduced 387 parts per million, or 
its value was increased $38.70 per million gallons. At Oberlin, 
Ohio, the hardness of the raw water was 170 parts per million, and 
by softening it was reduced to 48 parts per million, thereby in- 
creasing its value $12.20 per million gallons. These increased 
values relate only to estimates based upon the greater worth of 
the water for domestic purposes. If industrial uses were also 
considered, the value of the softened water would be still further 
increased. 
A summary of certain analytical data obtained in the operation 430 
WATER PURIFICATION 
of the water-softening plant at Columbus,' Ohio, is of interest in 
showing the practical results of softening. 
Year 
Chemicals applied, grains 
per gallon 
Total hardness, parts per 
million 
Percentage reduc- 
tion in hardness 
Lime 
Soda ash 
Coagu- 
lant 
River 
water 
Settled 
water 
Filtered 
water 
1909 
7.9 
3.8 
1.76 
253 
105 
93 
63.2 
1910 
7.6 
5.3 
1.02 
270 
95 
85 
68.5 
1911 
7.5 
4.3 
1.57 
245 
90 
84 
65.7 
1912 
7.0 
3.4 
1.90 
222 
90 
80 
63.9 
1913 
7.6 
6.0 
1.50 
271 
100 
88 
67.5 
It is worthy of comment that 0.5 to 1.0 grain per gallon of the 
coagulant (sulphate of aluminum) applied to the softened settled 
water at Columbus reduces the temporary hardness from 20 to 
30 parts per million, whereas the theoretical reduction should be 
only 8 parts per million. This is accounted for by the precipital 
condition of the basic carbonate of magnesia, which in its colloi- 
dation is dragged down by the aluminum hydroxide produced 
by the decomposition of the alum. 
Hard waters that are colored with vegetable stain are some- 
times encountered. The water treated by the Grand Rapids,2 
Mich., plant is of this character. The total hardness of the 
Grand Rapids' water supply averages about 217 parts per million, 
and ranges from 104 to 288 parts per million. This is accom- 
panied by an average color of 32 parts per million, ranging from 
16 to 55 parts per million. The turbidity ranges from 5 to 135 
parts and averages 20 parts per million. It was found that with 
a color of 30 or 40 parts, from 3 to 4}^ grains of alum were required 
to reduce the color to 10 parts per million. This large amount of 
alum increased the permanent hardness of the water, beside 
adding materially to the expense of treatment. 
By adding sufficient lime to neutralize all of the bicarbonates 
present, it was found that a water could be produced with a color 
of 10 parts or less, and with a total hardness ranging from 88 to 
100 parts per million. The alum was, therefore, reduced to 
grain per gallon, and the cost of treatment was also lessened. 
1 Eng. Record, Apr. 11, 1914. 
2 Eng. Record, Sept. 13, 1913. REMOVAL OF DISSOLVED MINERAL MATTER 
431 
Mr. S. A. Greeley1 has recently given some operating data on 
municipal water-softening plants, which are of interest in this 
connection. 
Plant 
Average 
hardness 
of raw 
water 
Popu- 
lation, 
1910 
Quantities 
Period for 
reaction and 
settling 
based on 
actual 
quantities, 
hours 
Remarks on after- 
precipitation 
Rated 
m.g.d. 
Actual 
m.g.d. 
St. Louis, Mo 
170 
687,029 
160 
100 
40-602 
No trouble since fil- 
ters have been in 
operation. 
Columbus, Ohio. . . . 
300 
181,511 
30 
18.4 
20.6 
Grand Rapids, Mich. 
235 
112,171 
20 
14.0 
3 to 3.5 
Requires careful op- 
eration and mixing. 
Alum used. 
McKeesport, Pa.... 
100-600 
42,694 
10 
3.6 
7^-20 
Incrustation of filter 
sand. No deposits 
in mains. 
Owensboro, Ky 
300 
16,011 
3 
1.5 
48 
Some precipitate. 
Not enough to cause 
trouble. 
Oberlin, Ohio 
300 
4,365 
0.75 
0.28 
48 
Incrustation on filter 
sand. 
Daytona, Fla 
360 
3,082 
0.30 
0.17 
4-18 
A little trouble when 
settling period is 
short. 
Hinsdale, Ill 
360 
2,451 
1.00 
0.30 
20-40 
Port Tampa, Fla.. . . 
640 
1,343 
0.24 
123 
Results Obtained in Softening with Zeolites.-It will probably 
have been noted that in softening waters with lime and soda ash, 
the residual hardness in the treated water is still considerable. 
By passing a hard water through a zeolite bed, it becomes pos- 
sible to reduce its hardness to zero. This is the only practical 
method of complete softening known. An equivalent of sodium 
carbonate is always left in waters softened by this process. 
Hoover and Scott in experimenting with the hard Scioto River 
water and a " permutit" (artificial zeolite) filter obtained the 
following results. 
These investigators regard a water containing so large a quan- 
tity of sodium carbonate as undesirable for a public water supply. 
They made further experiments by filtering the water through the 
"permutit" after having first reduced the total hardness with 
1 Jour. Am. Waler-works Assn., March, 1916. 
2 Based on rated capacity. 
3 Depends on basins in service. 432 
WATER PURIFICATION 
Raw water 
Softened water 
Parts pe 
r million1 
Total alkalinity 
180 
185 
Bicarbonate alkalinity 
180 
147 
Normal carbonate alkalinity 
0 
38 
Incrustants (permanent hardness) 
200 
Negative2 
Total hardness. 
380 
0 
lime and soda ash to about 91 parts per million, the object being 
to make use of the "permutit" for a finishing process. A sodium 
carbonate alkalinity in the water passed through the "permutit, " 
equal to 62 to 105 parts per million, was found, depending upon 
whether nearly exhausted or fresh "permutit" was used. These 
quantities are still large and are somewhat objectionable in a 
water to be used for boiler purposes, on account of the foaming 
which the sodium carbonate would possibly produce. Their con- 
clusions3 as to the merits of the "permutit" are as follows: 
"Advantages of the Permutit Process.-(1) It is the only practical 
process, known to the writers, by which water of a zero hardness can be 
produced on a large scale. 
"2. Only one chemical is needed (common salt). 
"3. Variations in the hardness of the raw water are automatically 
taken care of. 
"4. There is no sludge to be removed. 
"Disadvantages- (1) Cost of operation is higher than with lime and 
soda ash. 
"2. Water to be softened must be perfectly clear, for if it contains 
turbidity the pores of the permutit become clogged. 
"3. The permutit-softened water contains residual sodium bicar- 
bonate, and if used in boilers may cause trouble by foaming." 
Bacterial Effects of Water Softening.--Bacterial removal by 
coagulation and precipitation is usually very efficient in water- 
softening plants. This is commonly due to the gelatinous char- 
acter of the magnesium hydroxide produced when enough caustic 
lime is added to form this precipitate, and to the causticity of the 
water itself. The concentration of hydroxyl ions is sufficient to 
1 Expressed in terms of CaCOs. 
2 Excess of sodium carbonate, 182 parts. 
3 Ohio Public Health Journal, August, 1915. REMOVAL OF DISSOLVED MINERAL MATTER 
433 
effect a coagulation of the positively charged clay particles, which 
form masses that settle rapidly. 
Beside mechanically assisting in the removal of the bacteria, 
the causticity of the water appears to actually kill intestinal 
bacteria, thus acting as a germicide. This may not be a specific 
toxic effect, but rather a secondary effect arising from the removal 
of the carbonic acid. It is claimed that a water in which the 
free and half-bound carbonic acid are removed, but which is not 
caustic, will act similarly toward intestinal organisms. 
In the treatment of the water supply of London, England, Dr. 
A. C. Houston, Director of Water Examination of the Metropoli- 
tan Water Board, made some studies which led to the conclusion 
that treatment of the water with quicklime in an amount suffi- 
cient to cause an excess of calcium oxide equal to 0.007 per cent, 
caused the death of B. coli in from 5 to 24 hr. As a result of this 
work, Dr. Houston proposed to treat about 75 per cent, of the 
water supply with a slight excess of lime, and this excess to be 
neutralized by mixing with the remaining 25 per cent, of the 
water, previously purified by long storage, or by the use of some 
other disinfectant. In this connection Mr. W. H. Dittoe, Chief 
Engineer of the Ohio State Board of Health,1 states, that: 
"While this method of disinfection of water has been carefully studied, 
several important features of the treatment remain to be demonstrated. 
Its bacterial efficiency has been shown, but its universal applicability 
has not been proven. The principal objection which has been advanced 
relates to the cost of the treatment necessary to secure disinfection. 
In this connection, however, due credit should be given to the beneficial 
softening effect also produced by the treatment. Its use as an available 
method for purification of water supplies for private use has not been 
demonstrated." 
Mr. Russel D. Scott,2 assistant bacteriologist of the Ohio State 
Board of Health, has made some interesting studies on the bac- 
teriological effects of water softening as carried out on the Scioto 
River water at Columbus, Ohio. His conclusions are as follows: 
"1. The bacteriological character of the Scioto River water during 
periods of long storage varies from that during flood stages, in that: 
(a) the number of bacteria is much less; (6) the 37° agar count is a much 
higher proportion of the 20° count; (c) of the B. coli group, the sac- 
charose fermenting type are the most numerous. 
1 Ohio Public Health Journal, August, 1915. 
2 Ohio Public Health Journal, June, 1915. 434 
WA TER P U RIF IC A TION 
"2. The lime softened water varies bacteriologically from the raw 
water, in that: (a) the number of bacteria is much less; (b) the ratio of 
the number of B. coli to the total number of bacteria is less; (c) the 
saccharose-fermenting types of B. coli are much more numerous; (d) 
the proportion of dextrose fermenters which do not ferment lactose is 
much greater; (e) the ratio of the 37° to the 20° count is greater; (/) 
the ratio of the gelatine to the agar count is very much higher. 
"3. The proportion of organisms not fermenting dulcite is not affected 
either by storage or by lime-treatment. This is also true of those that 
do not form indol. 
"4. The members of the B. coli group form somewhat less than 4 
per cent, of the total number of bacteria at 37°C. This ratio is, however, 
quite irregular. 
"5. The proportion of organisms in the river water which ferment 
dextrose but not lactose is not affected by storage." 
Annual Cost of Operation and First Cost of Water-softening 
Plants.-Mr. S. A. Greeley has recently compiled some data upon 
the cost of operation and the first cost of water-softening plants. 
The following tables are taken from his article on "Water-soften- 
ing Practice" in the March number of the Journal of the Ameri- 
can Water-works Association (vol. 3, No. 1, 1916). 
Plant 
Popu- 
lation, 
1910 
Actual 
quantity, 
millions 
of gallons 
per year 
Average pounds chemicals 
used per million gallons 
Opera- 
tors 
re- 
quired, 
total 
Cost of 
opera- 
tion per 
million 
gallons 
Lime 
Soda 
ash 
Iron 
sulp. 
Alum 
sulp. 
Hy- 
po- 
chlor- 
ite of 
lime 
St. Louis, Mo 
687,029 
34,656 
800 
415 
1.6 
35 
$4.56i 
Columbus, Ohio 
181,511 
6,716 
1,384 
1,054 
260 
4.0 
17.46 
Grand Rapids, Mich. 
112,171 
4,506 
( 
1,250 
100 
100 
50 
120 
75 
16 
11.34 
McKeesport, Pa 
42,694 
1,314 f 
3,500 
6,000 
200 
400 
27.57 
Owensboro, Ky 
16,011 
547 
2,000 
75 
1.0 
2» 
10.05 
Oberlin, Ohio 
4,365 
102 
2,045 
735 
12 
13.49 
Daytona, Fla 
3,082 
62 
2,910 
250 
I2 
23.80 
Hinsdale, Ill 
2,451 
100 
4,091 
1,368 
Small 
amount 
I2 
30.00 
1 Does not include pumping. 
2 Part time of one man. 
The hardness of the St. Louis raw water is 170 parts per mil- 
lion. The raw water treated at Columbus has a hardness of 
about 300 parts, and at Oberlin and Owensboro it averages the 
same. At Grand Rapids the raw water averages about 235 parts REMOVAL OF DISSOLVED MINERAL MATTER 
435 
per million of hardness, and at Daytona and Hinsdale it is 360 
parts per million. The raw water at McKeesport ranges from 100 
to 600 parts of hardness. 
The cost of operation is naturally affected by local conditions 
other than the hardness of the water. The size of the plant and 
the cost of the chemicals in different parts of the country, are 
factors that must be considered in making comparisons of costs. 
The first cost of these water-softening plants is shown by the 
following table, which is also taken from Mr. Greeley's paper. 
Quantities Cost of construction 
Date ponu 
Plant lation, Remarks 
tion 1910 S^m^d Total Ion* 
m.g.d. m.g.d. rated 
capacity 
_______ । 
St. Louis, Mo... . 1915 687,029 160.00 100.00 $1,495,000 $9,344 Cost approximate.1 
Columbus, Ohio... 1906 181,511 30.00 18.40 590,780 19,693 First-class con- 
struction. 
Grand Rapids, 
Mich 1910 112,171 20.00 14.00 449,569 22,478 First-class con- 
struction. 
McKeesport, Pa.. 1908 42,694 10.00 3.60 250,000 25,000 First-class con- 
struction . 
Owensboro, Ky... 1906 16,011 3.00 1.50 30,000 10,000 Excelsior filter. 
Oberlin, Ohio. .. . 1903 4,365 0.75 0.28 12,000 13,300 Pressure filter. 
Daytona, Fla 3,082 0.30 0.17 4,304 14,347 No filters. 
Hinsdale, Ill 1915 2,451 1.00 0.30 18,000 18,000 Filters 18 in. deep. 
Port Tampa, Fla. 1914 1,343 0.24 .... 7,000 29,100 No filters. 
1 Existing work also used, but not included in figures. 
References 
1. Alex. Deussen and R. B. Dole: U. S. Geological Survey Paper on 
Water Supply, No. 375. 
2. W. C. Mendenhall, R. B. Dole and Herman Stabler: U. S. Geo- 
logical Survey Paper on Water Supply, No. 398, 1916. 
3. Proc. Am. Ry. Eng. & Maintenance of Way Assoc., vol. 6, 610, 1905; 
vol. 8, p. 601, 1907. 
4. C. P. Hoover and R. D. Scott: "Advantages of the Use of Lime in 
Water Purification." Monthly Bulletin Ohio State Board of Health, 
December, 1914. 
5. C. P. Hoover and R. D. Scott: "Water Softening by the Permutit 
System." The Ohio Public Health Journal, August, 1915. 
6. Francis G. Wickware: "European Power-plant Practice." Practical 
Engineer, June 1, 1912. 
7. Eng. Record, Sept. 13, 1913; and Apr. 11, 1914. 436 
WATER PURIFICATION 
8. R. N. Kinniard: "Experiment with Natural Water Softening Zeolite." 
Eng. Record, Dec. 25, 1915. 
9. C. P. Hoover: "Water Purification at Columbus, Ohio." The Ohio 
Public Health Journal, June, 1915. 
10. Samuel A. Greeley: "Water-softening Practice." J our. Am. Water- 
works Assn., March, 1916. 
11. "Water Softening, Youngstown Sheet and Tube Mill." Eng. Record, 
Nov. 20, 1915. 
12. J. C. Wm. Greth: "Water-purification Facts for Steam Users." 
Proc. Rochester (N. Y.) Engineers Club, Feb. 25, 1910. 
13. W. A. Pownall: "Treatment of Water for Locomotive Use." Jour. 
Am. Water-works Assn., June, 1915. 
14. D. D. Jackson: "Water Softening by Filtration through Artificial 
Zeolite." Jour. Am. Water-works Assn., June, 1916. CHAPTER XXXI 
THE REMOVAL OF DISSOLVED MINERAL MATTER 
FROM WATER (CONTINUED) 
REMOVAL OF IRON AND MANGANESE 
The use of ground waters for public water supplies is common, 
and undoubtedly possesses certain advantages as well as some 
disadvantages. Many ground waters are entirely suitable for 
domestic and industrial purposes, except for the iron and man- 
ganese which they contain. Waters which contain less than 0.2 
part per million of iron are not objectionable, and even as much 
as 0.5 part may give little trouble. Iron is generally more easily 
separated from a water than is manganese. 
Iron is commonly present in the ferrous condition, that is as 
ferrous hydrate, ferrous bicarbonate or ferrous sulphate. Or- 
ganic matter, also, in some cases plays an important part in hold- 
ing both iron and manganese in solution. Iron, if present as a 
bicarbonate, can usually be precipitated from a ground water 
by aeration, if the content of manganese and organic matter is 
low. 
By aeration the carbon dioxide is removed and the iron is ox- 
idized to the ferric condition, in which form it is quite insoluble 
in the water. Manganese reacts more slowly with oxygen, and 
its precipitation requires a much longer period than does the iron. 
Filtration through sand and gravel after aeration effects the re- 
moval of both the iron and manganese in suspension, and doubt- 
less also some that may be in the form of a colloidal suspension. 
In general it may be said, that the precipitation of iron is inter- 
fered with by carbon dioxide, organic matter and manganese. 
Weston1 has noted that the precipitation of aluminum is more 
difficult than that of iron, but less so than that of manganese. 
The metal which forms the heaviest oxide is first precipitated. 
The same investigator found that in three waters containing both 
1 R. S. Weston: "Some Recent Experiences in the Deferrization and 
Demanganization of Water." Jour. New England Water-works Assn.} 
February, 1914. 
437 438 
WATER PURIFICATION 
iron and manganese, one could not be aerated to saturation with- 
out causing some of the iron and color to escape removal; one 
required complete oxidation to effect a satisfactory removal of 
manganese; and one demanded long aeration and contact in a 
trickling filter (which it was necessary at times to operate sub- 
merged) to produce satisfactory results. Other things being 
equal, it is well to remove carbon dioxide as completely as prac- 
ticable in order to prevent its corrosive effect on the distribution 
piping. 
It is quite evident from what has been stated above that 
waters containing iron and manganese vary in character, and 
that methods especially adapted to their peculiar properties must 
be used to effect purification. The problem is not rendered any 
easier of solution by the fact that water from the same source 
is not always constant in its characteristics, and hence may need 
at times a change in the method of treatment in order to meet 
these changed conditions. It is probable that much of the phe- 
nomena connected with the purification of waters containing iron 
and manganese can be explained by the quantitative relations 
existing between the concentrations of the several components 
at equilibrium. 
Iron Removal in Presence of Carbon Dioxide.-When iron is 
present in a water as the bicarbonate, it is accompanied by a 
considerable amount of free carbon dioxide. Johnston1 has 
pointed out in a recent paper that in any solution containing a 
carbonate, "there is a readily attained equilibrium between the 
carbonate ion, CO^, the bicarbonate ion, HC07, and the carbonic 
acid, H2CO3, and in turn between the carbonic acid and the 
partial pressure of carbon dioxide above the solution; conse- 
quently, these molecular species can coexist only in definite pro- 
portions determined by the several equilibrium constants." 
By agitation a considerable portion of the CO2 may be removed 
from the water. The oxygen dissolved in the water is then able 
to oxidize the iron, and thereby render it insoluble. More or less 
absorption of oxygen from the air takes place during the agita- 
tion of the water, and the oxidation and resultant precipitation 
of the iron is thus hastened. The removal of carbon dioxide 
by adding to the water an alkali, such as lime or soda ash, pro- 
1 John Johnston: "The Determination of Carbonic Acid, Combined and 
Free in Solution, Particularly in Natural Waters." Jour. Am. Chern. Soc., 
May, 1916. REMOVAL OF DISSOLVED MINERAL MATTER 
439 
duces a like result. Rapid filtration of the aerated and settled 
water through sand is usually the final step in the process of 
purification. 
As a typical example of waters of this type, Mr. Frank E. Hale' 
gives the following results obtained in treating in the laboratory 
a Long Island (N. Y.) well water by aeration, and also with lime 
and with soda ash: 
Analysis 
Parts per 
million 
Free carbonic acid (CO2) 
31.2 
Iron (Fe) 
11.2 
Hardness (CaCOs) 
47.0 
Alkalinity (CaCO?) 
32.0 
Chlorine (Cl) 
64.0 
Nitrate (N) 
0.0 
Experimental treatment 
Iron after aeration for % hr. and filtered through paper 
Iron after neutralization of carbonic acid with soda ash and filtra- 
7.2 
tion through paper 
0.3 
Iron after neutralization of carbonic acid with calcium hydrate, 
and filtration through paper .. 
0.1 
Rates of filtration 
350 gal. per minute 
650 gal. per minute 
Free CO2 
Iron (Fe) 
Free CO2 
Iron (Fe) 
Parts per million 
Raw water 
55 
5.00 
49 
5.40 
After nine aerators 
After eleven aerators and sedimen- 
41 
5.00 
40 
5.40 
tation tanks (water to filter).... 
31 
3.50 
31 
4.20 
Effluent from filters 
19 
0.05 
22 
0.05 
N. B.-A rate of discharge of 350 gal. per minute was equal to a rate of 
69,000,000 gal. per acre per day, and a discharge of 650 gal. per minute was 
equal to a rate of 114,000,000 gal. per acre per day. 
1 "Iron Removal by Rapid Sand Filtration." Jour. Am. Water-works 
Assn., March, 1916. 440 
WATER PURIFICATION 
This same water treated in a purification plant in which aera- 
tion was effected by a series of cascades over which the water 
splashed and foamed before entering a sedimentation tank, gave 
the foregoing results shown in the preceding table. 
The reduction in the quantity of CO2 as well as of the iron 
should be particularly noted in these results. 
The so-called Anderson revolving purifier, used in a few places 
in Europe, consists of a rotating drum containing a quantity of 
iron turnings. The water passes through the drum and is agi- 
tated in contact with the iron turnings. By the action of the 
excess CO2 more iron is dissolved, and as a result of the agitation 
more or less CO2 is also probably removed. Oxidation of the 
iron to the ferric condition then follows, and by its precipitation 
Fig. 143.-Sketch showing arrangement of Cohasset, Mass., deferrization 
plant. 
acts as a coagulant, and assists in precipitating suspended matter 
and even colloidal suspensions such as vegetable coloring 
matter. 
Iron Removal in the Presence of Organic Matter.-A well 
water which contained considerable organic matter, carbon diox- 
ide, iron and some manganese, was experimented with by Mr. 
R. S. Weston at Cohasset, Mass., in 1913 (Fig. 143). The color 
of the water ranged from 35 to 50 parts per million, the iron from 
0.4 to 1.15 parts, the CO2 from 40 to 56 parts and the manganese 
from 0.15 to 0.4 part. 
By means of an aerator containing 6 in. of stone, a "trickier" 
filled with coke, subsiding basins and a mechanical filter, the 
following purification effects were produced experimentally: REMOVAL OF DISSOLVED MINERAL MATTER 
441 
Well water 
Effluent from 
rieseler 
Effluent from 
filter 
Parts per million 
Color 
50.00 
40.00 
35.00 
Iron 
0.90 
0.35 
0.28 
Carbon dioxide 
54.00 
8.60 
8.00 
Dissolved oxygen 
2.17 
10.39 
10.25 
Note.;:-"Trickier" or rieseler operated at rate of 50,000,000 gal. per 
acre per day, and the filter at 100,000,000 gal. per acre per day. 
Mr. Weston's comments on the experimental results are as 
follows: 
"It will be readily observed that the bulk of the work was done before 
filtration. The reason for this is that contact between the water and 
the rough surfaces of the stones or coke effects the displacement of the 
carbon dioxide by oxygen; the oxidation of the iron and the man- 
ganese ; and the removal of the color. The filter acts simply as a strainer. 
At the beginning of the experiments, the stones were clean, but in less 
than a month they showed a red coating. This, on analysis, was found 
to be iron rust mixed with some organic matter. In other words, 
notwithstanding the fact that the season was most unfavorable, on 
account of low temperature, for the establishment of contact action, 
the stones had begun to remove both iron and organic matter to a 
satisfactory degree." 
Iron Removal in Presence of Manganese.-A well water in 
which manganese interferes with the removal of the iron is found 
at Middleboro, Mass. Experiments by the Massachusetts State 
Board of Health in 1911-12, and by Mr. R. S. Weston in 1913, 
have shown the difficulties connected with purifying this type 
of water. In Weston's experimental apparatus the water was 
delivered to a spray aerator, which discharged through the air 
a distance of about 20 in. to a coke "trickier." The latter was 
operated at a rate of 75,000,000 gal. per acre per day. For a 
depth of 6.7 ft., the coke of the "trickier" was not submerged. 
Contact with submerged coke, however, was obtained in the lower 
portion of the "trickier." The filter consisted of 26 in. of sand 
with an effective size of 0.28 mm. It was operated at a rate of 
10,000,000 gal. per acre per day. 
The results of the operation of these devices on Mar. 27 and 
Apr. 23, 1913, are given in the following table: 442 
WATER PURIFICATION 
Well water 
Trickier 
effluent 
Basin 
effluent 
Filter 
effluent 
Mar. 27 
Apr. 23 
Mar. 27 
Apr. 23 
Mar. 27 
Apr. 23 
Mar. 27 
Apr. 23 
Parts per million 
Color 
45.00 
45.00 
40.00 
38.00 
12.00 
10.00 
10.00 
10.00 
Turbidity 
3.00 
2.00 
5.00 
5.00 
2.00 
2.00 
1.00 
1.00 
Iron 
1.20 
0.90 
1.90 
1.80 
0.35 
0.25 
0.15 
0.20 
Manganese 
0.75 
0.72 
0.95 
0.90 
0.75 
0.55 
0.38 
0.25 
Carbon dioxide 
44.00 
48.00 
5.00 
5.00 
5.50 
5.30 
5.00 
5.00 
Dissolved oxygen.... 
3.50 
4.21 
11.07 
11.42 
10.83 
11.51 
10.91 
11.26 
Middleboro Purification Plant.-The plant designed and built 
by Messrs. Weston and Sampson1 to treat the water at Middle- 
Fig. 144.-Middleboro, Mass., deferrization plant, general plan. 
boro, Mass., as a result of the experiments mentioned above, is 
described in some detail as typical of well-designed plants for 
treating iron and manganese waters. 
The following description, cuts and illustrations are taken 
1 Jour. New England Water-works Assn., vol. 28, No. 1, 1914. REMOVAL OF DISSOLVED MINERAL MATTER 
443 
from Mr. Weston's paper, "Some Recent Experiences in the 
Deferrization and Demanganization of Water" (Fig. 144). 
"General Description of Plant.-A DeLaval steam-turbine-driven, 
centrifugal pump, having a capacity of 1,000,000 gal. per day, takes 
water from the present well and discharges it on top of the pile of coke 
-a so-called trickier-through a number of sprinkling nozzles. The 
trickier is built of concrete. It is 30 ft. in diameter, outside measure- 
ments, and contains 10 ft. of coarse coke, supported upon a grill work 
above the true bottom. The water from the trickier falls upon its 
true bottom and flows through a pipe into the settling basin. 
"The settling basin is built of concrete and has a capacity of 40,000 
gal. It is covered with a concrete roof and earth to prevent freezing 
of the water. 
"From the settling basin the water flows into the two compartments 
of the filter, which has a total area of 0.1 acre. These filters are simply 
concrete basins with groined roofs, covered like the settling basin 
with earth to prevent the freezing of the water in winter. On the bottom 
of the filter are laid tile underdrains. These are covered with 12 in. 
of graded gravel. Above the latter layer and supported by it is a layer 
of sand 3 ft. thick. 
"From the two filters the water flows into a regulator house in which 
are located all of the valves and gages for operating and regulating the 
filters. From the regulator house the water flows into a filtered-water 
basin which holds 42,000 gal. This is large enough to enable the pumps 
and filters to operate regularly and to supply the hydrants during fires 
of an ordinary character. For fire purposes this supply is at the rate 
of at least 1,000,000 gal. in 24 hr., or twice this amount during 1 hr. 
The present consumption is about 335,000 gal. per day; and the 
maximum consumption for any one day has been 873,000 gal. 
"Of these various parts the aerator and trickier are alone worthy of 
special comment, for although the filters embody some special features 
these may be readily apprehended by inspecting the plates. 
"Aerator and Trickier.-The aerator consists of a system of piping 
connecting the discharge from the low-lift pump with thirty-seven 2-in. 
nipples. On each nipple is screwed a cap drilled with twenty-four 
%6"in. holes, and pressed to make the upper surface convex and the 
axes of the holes radial. These sprays discharge upward, throwing the 
water on the surface of a layer of coarse coke, 10 ft. deep, contained in 
a concrete tower 28.3 ft. in diameter (Fig. 145). 
"The coke is supported upon 2 in. by 6-in. reinforced-concrete beams 
spaced 2 in. apart, and supported upon a beam 6 in. wide extending 
around the inside of the wall of the trickier and also upon four cross- 
walls, 6 in. wide, resting on the floor. The floor slopes to a channel 
leading to the 18-in. cast-iron pipe, which connects with the subsiding 444 
WATER PURIFICATION 
basins and filters. A shear gate provides for the closing of this outlet, 
when the trickier may be filled; then the valve may be opened quickly 
and the excess of accumulation in the coke flushed out through the 
subsiding basin and into the drain. It is not expected that flushing 
will be required oftener than once a year. 
"Filters and Appurtenances.-The two filters have no doors and run- 
ways, but special sand boxes have been designed into which the sand 
removed by scraping is dumped and washed out into bins placed outside 
the filter. The sand used in the filters had an effective size of 0.31 
mm., and a uniformity coefficient of 1.8. It was natural bank sand 
which did not require washing (Figs. 146 and 147). 
SECTIONAL ELEVATION 
ON B-B 
PLAN OF SPRINKLER SYSTEM 
CONCRETE SLAT 
SECTIONAL ELEVTATIO.N 
SPRINKLER CAPS 
SECTIONAL ELEVATION 
ON A-A 
IRON LADDER 
Fig. 145.-Coke "rieseler" or trickier with aerator, Middleboro, Mass. 
"The regulating devices are of the orifice type, and indicating rate 
and loss of head gages are provided for each filter. 
"Cost.-The complete plant has cost about $18,000, including 
engineering. 
"Results of Operation.-The filters were started on Sept. 26, 1913. 
They were raked twice and scraped once prior to Jan. 12, 1914. The 
yield for this first run was 40,000,000 gal., or at the rate of 400,000,000 
gal. per acre. The average results for this first run, between the dates 
above mentioned, are as follows:" REMOVAL OF DISSOLVED MINERAL MATTER 
445 
TRANSVERSE SECTION ON B-B 
SECTION THROUGH MANHOLE 
MIDDLEBORO WATER WORK'S 
WATER PURIFICATION PLANT 
FILTERS AND SUBSIDING BASIN 
Fig. 146.-Filters and subsiding basins at Middleboro, Mass., plant. 
5 Courses of Brick laid dry 
above 8* Concrete Wall - 
% Galvanized Iron - 
Pipe to Pumping Sta. 
12"8plit Tile 
SECTION AT LOWER END SECTION AT UPPER END 
MAIN UNDERDRAIN 
SECTION AT LOWER END SECTION AT UPPER END 
LATERAL UNDERDRAIN 446 
WA TER P URIF IC A TI0N 
MIDDLEBORO WATER WORKS 
WATER PURIFICATION PLANT 
REGULATOR HOUSE SUBSTRUCTURE 
14*Hand Wheel 
Graduated with 10 
equally spaced marks 
Floor Reinforcement to be sq. 
Twisted Rod 25'e. to c. each way 
e'w.I. Float Pipe I 
Bottom at El. 96,75| 
12 Flanged 
Sluice Gate ] 
SECTION ON A-A 
Fig. 147.-Regulator house and substructure of Middleboro, Mass., plant. 
12 Hand Wheel 
'Rigid Pillow .Block 
8 Sluice 
Gate 
"%'m '^"Flat iron 
'Circular C.l. Plate 
"• Hole for Lifting 
^Sanitary Base 
X Round Roda to c._! 
Lj"Round Roda 181 o. to c.' 
DETAIL OF MANHOLE 
Horiaontal Reinforcement to be sq. Twisted Roda 12*c. to c, 
/Vertical Reinforcement to be aq. Twisted Roda 21 "c. to c>J 
12 C.l. Pipe ; 
to Filtered 
Water Basin 
Joint'd 
12*0.1. Pipe 
Elev. of Invert 96.2a 
SECTION ON B-B 
X Brass Plate, Standard 
Drilled. Orifice to be a • 
true circle • / 
7"ORIFICE 
PLATE 
SECTION OF 
FLOOR ON C-C 
Filter No. 2 
Rods 
8"c. to c. 
18 Self-Contained] 
Sluice Gate I 
PIPING PLAN 
All Reinforcement in Wells 
and Bottom to be ^."sq., _ 
J' w isted.Rods j Q-0- 
4 Concrete Floor 
-All Reinforcement 
to be^' round, rods 
FLOOR PLAN 
Filter No. 1 
8 Self-Contained 
✓ Sluice Gate / 
8 Tile Pipe to 
connect with 
Main Drain REMOVAL OF DISSOLVED MINERAL MATTER 
447 
Well water 
Settling basin 
effluent 
Filter effluent 
Parts per million 
Color 
48.00 
22.00 
5.00 
Turbidity 
5.00 
3.00 
1.00 
Iron 
1.62 
0.46 
0.17 
Manganese 
0.67 
0.36 
0.27 
Hardness1 
27.30 
23.40 
Oxygen consumed 
1.79 
1.53 
1.05 
Carbon dioxide1 
41.00 
4.20 
4.60 
Dissolved oxygen1 
2.95 
10.20 
9.55 
1 One determination. 
Manganese in Ground Waters.-It is usually found that 
waters which contain considerable amounts of iron will also con- 
tain manganese. In some instances, however, large quantities 
of manganese are found associated with quite small amounts of 
iron. A case in point is the water supply of Dresden, Germany, 
which contains from 0.6 to 1.0 part per million of manganese, 
but only about 0.1 part of iron. Where manganese occurs as 
the sulphate, its removal is difficult, the reason being that man- 
ganese salts of active acids are not oxidized by the air, as are the 
corresponding salts of iron. 
Manganese like iron is especially troublesome in waters that 
are to be used in laundries, bleacheries, dye houses and paper 
mills, causing stains or dulling the effect of the bleaching or dye- 
ing. Drinking water, which contains those amounts of mangan- 
ese that are usually found in water supplies, appears to have no 
injurious effect. The growth of the "iron bacteria" in distribu- 
tion pipe lines is not an uncommon feature of waters that con- 
tain iron and manganese. 
In a great many European water supplies (Figs. 148 and 149), 
the occurrence of manganese is common, and the amounts found 
range from 0.1 to 6.5 parts per million. Iron, as stated above, 
is usually found associated with the manganese. The experience 
of the City of Breslau on the Oder River with the pollution of its 
ground-water supply in 1906 by iron and manganese, as reported 
by Mr. S. B. Applebaum (Journal Industrial and Engineering 
Chemistry, February, 1916) is of interest, because of the unusual 
circumstances. 448 
WATER PURIFICATION 
The city had been using since 1871 Oder River water filtered 
through slow sand filters. As it was believed that a ground- 
water supply would be safer, it was decided to obtain water from 
wells, although it was known that the water would probably 
contain iron. A series of 300 6-in. wells were bored about 60 ft. 
apart, and the water pumped from them was purified of its iron 
and manganese in "tricklers" and sand filters. The yield of the 
wells proved to be only about 10,000,000 gal. of water daily, 
whereas 15,000,000 gal. had been expected. As the pumping of 
the wells continued, the iron content of the water rose from 6 to 
20 parts per million, but the purification plant was able to remove 
Well Water Main 
Main To Filter^ 
Ture Water 
Pure Water 
Pure Water 
Fig. 148.-Deferrization plant at Hamburg, Germany. 
it. Early in 1906 after a season of drought, it was noted that 
the water table fell steadily. On Mar. 28, 1906, a flood in the 
Oder River overflowed the ground on which the wells were located. 
The chemical composition of the water changed over night, the 
total solids trebled, and the iron content of the water rose to 
100 parts per million, and the manganese to 50 parts per million. 
The water was also found to be slightly acid. 
The purification plant was unable to purify the water. After 
a while most of the wells again began to yield a normal water, 
but the remainder were finally abandoned. This phenomenon 
was repeated again in September of this same year. REMOVAL OF DISSOLVED MINERAL MATTER 
449 
The probable explanation of this sudden pollution of the well 
waters is that the level of the ground water, being drawn so low, 
permitted the oxygen of the air to oxidize the iron and manganese 
sulphides in the soil to sulphates. The decomposition of organic 
matter furnished the hydrogen sulphide which had originally 
converted the iron and manganese to sulphides. When the flood 
in the Oder River came, it washed the sulphates, formed as 
described above, into the wells. 
Dr. H. Luhrig undertook experimental work to determine the 
best method of removing the iron and manganese from the water 
at Breslau, and also from the well waters of Glogau's proposed 
supply on the opposite side of the Oder River. The general out- 
line of the experiments and the conclusions reached by Luhrig, as 
reported by Applebaum, are as follows: 
"Aeration and Filtration.-This method is successful if considerable 
iron is present with the manganese, and where both are present as 
bicarbonates. Aeration may be effected in tricklers or under pressure 
in a closed shell. This method will not remove manganese in any 
appreciable amount if the content of iron is low, nor where the 
manganese is present as the sulphate; and under the most favorable 
conditions the manganese is not entirely removed. 
"Aeration and Filtration, Addition of Lime, Settlement and a Second 
Filtration.-The greater part of the iron and manganese were removed 
by aeration and filtration. A slight excess of lime water (2 to 4 per 
cent.) was then added. Settlement in basins followed by filtration 
through sand, completed the process. This method is a practical 
one. Its disadvantages at Breslau were found to be the care required 
in applying the lime water to a water constantly varying in its composi- 
tion, and the disagreeable taste imparted to the water by the excess 
of lime applied. This taste was eliminated by adding alum, free CO2, 
or by mixing the effluent with filtered river water. 
"Addition of a Permanganate.-Oxidation of the manganese was quite 
successful. The cost for removing 8 parts per million of manganese 
sulphate was $6 per million gallons, and was regarded as low. One 
of the objections to the use of a permanganate is that one of the products 
of the oxidation is sulphuric acid, which must be neutralized. Over- 
or underdosing with this reagent may give trouble. 
"Ozone or Electrolysis.-Both methods are able to produce satisfactory 
results, ozone being more effective as an oxidizing agent than a per- 
manganate. The oxides produced, however, are very finely divided, 
and filters must be operated at low rates in order to successfully remove 
them. 
"Filtration through "Manganese Permutit."-This reagent like the 450 
WA TER P U RIF IC A TION 
"sodium permutit" described in a preceding chapter, is an artificial 
zeolite. The sodium "permutit" is treated with a dilute solution of 
manganous chloride, and forms "manganese permutit" by an exchange 
of the manganese for the sodium in the zeolite. The "manganese 
permutit" is then oxidized by a 2 to 3 per cent, solution of sodium per- 
manganate. It is presumed that higher oxides of manganese coat the 
grains of the reformed "sodium permutit." If the formula of the 
"permutit" is 2SiO2-Al2O3-Na2O, and is denoted by the formula Na2O 
Pmt, the reactions may be shown as follows: 
1. Na2OPmt + MnCl2 = MnOPmt + 2NaCl. 
2. MnOPmt + 2NaMnO< = Na20Pmt -J- MnO,Mn2O7. 
" The manganese oxides exert a strong oxidizing action on the man- 
ganese dissolved in the water, and the "sodium permutit" acts pre- 
sumably only as a carrier. By passing the water to be treated through 
a bed of this artificial zeolite, prepared as above described, the following 
reaction probably ensues: 
1. Na2OPmt + MnO, Mn2O7 + 2Mn(HCO3)2 = Na2OPmt + 
5MnO2 + 4CO2 + 2H2O. 
" When the higher oxides of manganese on the surface of the "sodium 
permutit" are gone, the filter bed becomes inactive, and must be re- 
generated with sodium permanganate. About 77,200 gal. of water 
were passed through the bed per pound of sodium permanganate used 
for regeneration; the rate of filtration being 9.2 gal. per square foot of 
bed per minute. About 1 per cent, of wash water was used to wash 
out the deposited oxides of iron and manganese." 
The experimental "manganese permutit" plant at Glogau con- 
sisted of closed filters 3.5 ft. in diameter and 6 ft. high, and con- 
tained a bed of zeolite 2 ft. deep. The capacity of one filter was 
supposed to be 6,000 gal. per hour. The actual rate of filtration 
at Glogau was 7.5 gal. per square foot per minute, although at 
Dresden filters were operated at rates as high as 20 gal. per square 
foot per minute. 
The estimated cost of treatment, based upon a price of 10 cts. 
per pound for sodium permanganate was from $1 to $2 per million 
gallons, depending upon the amount of manganese in the water. 
The high rates of filtration, absence of any continuous chemical 
feed, and low cost would appear to make it a valuable process 
for the removal of both manganese and iron. 
1. R. S. Weston: "Some Recent Experiences in the Deferrization and 
Demanganization of Water." Jour. New England Water-works Assn., 
No. 1, 1914. 
References REMOVAL OF DISSOLVED MINERAL MATTER 
451 
2. S. B. Applebaum: "Manganese in Ground Waters and Its Removal." 
Jour. Ind. Eng. Chern., February, 1916. 
3. F. E. Hale: "Iron Removal by Rapid Sand Filters." Jour. Am. 
Water-works Assn., March, 1916. 
4. Report Mass. State Board of Health, 1899, p. 549. 
5. Jour. New England Water-works Assn., 11, 277. 
6. Trans. Am. Soc. C. E., 64, 182. 
7. F. C. Amsbary: "Further Development of Iron Removal Plant and 
Storage." Jour. Am. Water-works Assn., June, 1916. CHAPTER XXXII 
THE CONTROL OF WATER-PURIFICATION PROCESSES 
The problems involved in water purification are obviously 
dependent upon a number of the arts and sciences for their suc- 
cessful solution. A knowledge of engineering in several of its 
phases, of chemistry and of bacteriology, is essential to one who 
desires to operate a purification plant intelligently, and to obtain 
from it the best results. No two plants, even of the same general 
type and purifying the same class of water, can be handled by 
exactly similar methods. Constant observation and study of 
the many details of plant operation are always needed, no matter 
how varied an operator's experience may be. While the sanitary 
purity and physical attractiveness of the purified water should 
be the first care of an operator of a plant, nevertheless the eco- 
nomic side should not be forgotten. 
Care of Plant and Equipment.-The various apparatus now 
commonly installed in water-purification plants for assisting the 
operator in handling the plant have been discussed in more or 
less detail in preceding chapters. It is desired here only to point 
out that unless this apparatus is kept in the best working order 
by constant care and attention, and unless it is properly used, 
it is worse than useless. Inaccurate gages, alarm bells that do 
not ring, tell-tales that never move, orifices that never deliver 
their indicated volumes of solution, controllers that do not con- 
trol, and safety devices that never prevent an accident, are evi- 
dences of gross incompetency, if not of actual criminal negligence. 
Pumps, motors, steam boilers, engines, electric generators, 
waterwheels and blowers need periodic overhauling and renewal 
of worn-out parts. The electrical machinery and apparatus 
especially must have the best of care to be of real service. Elec- 
trical transmission lines, pipe lines and valves should not be for- 
gotten because they are frequently out of sight underground. 
Buildings, reservoirs and tanks need more or less repair work 
done on them from time to time. The care of grounds, ditches 
and drains should not be forgotten even though they may only 
indirectly affect the operation of the plant. 
Detailed advice as to just how a plant should be kept in the 
452 WATER-PURIFICATION PROCESSES 
453 
best condition would require too much space. It is desired only 
to emphasize the painstaking effort that is needed to maintain the 
physical equipment of the plant at its highest efficiency; and to 
point out that the impairment of this efficiency by carelessness 
and neglect causes the plant to become an actual menace to the 
community to which it is supposed to be supplying a purified 
water. 
Supplies for the Plant.-It usually is the duty of the manager 
of a plant to keep it supplied with those materials that are con- 
stantly required for its operation. In those plants in which 
chemical coagulants are used, the uninterrupted supply of chem- 
icals is of the utmost importance. Inadequate storage facilities 
for chemicals often make the plant dependent upon uniform de- 
livery of these supplies. Where shipments are made from a con- 
siderable distance, orders should be made sufficiently in advance 
to guard against possible delays in transit. It is not always 
possible to foresee a change in the character of the water being 
treated that will require a suddenly increased demand for chem- 
icals, and, in consequence, a more rapid depletion of the stock in 
hand than was anticipated. However, a study of the meteorolog- 
ical data available may show the times when fluctuations in the 
quality of the raw water are likely to occur; and from this infor- 
mation one may be able to make provision in advance by in- 
creasing the stock of chemicals. 
The supply of other materials should be as carefully forecasted 
as possible. Repair parts should always be kept in stock. For 
all large plants, a stock room is a necessity; and if the plant is 
isolated, a small machine shop fitted with a lathe, drill press, 
grinder, pipe-cutting machine, and the necessary work benches, 
will be found of the greatest value in cases of emergency, and of 
real assistance in the routine operation of the plant. 
Records of Operation.-Unless the principal operation data 
are consistently recorded, and as constantly watched and checked 
by those in charge, the results of operation are liable to be irregu- 
lar and, at times, extremely unsatisfactory. A system of record 
sheets, which may be filled out without too much trouble by fore- 
men, and which afterward can be given a more permanent form 
of record in the office of the plant should be worked out for each 
plant. General forms may assist one, but usually forms must be 
adapted to each particular plant and to the special conditions 
involved. 454 
WATER PURIFICATION 
Some of the printed forms which should be provided are for 
the rate of flow of water through the plant, for the elevation of the 
water in the various reservoirs, basins and tanks, for the initial 
and final losses of head on individual filters, for the length of the 
period of service of the filters, for the depth to which sand is 
scraped, for the water filtered between cleanings, for the length 
of time required to wash a filter, for the quantities of chemicals 
used, for the setting of orifices, for the record of the periodic 
cleaning of tanks, and for many other operations about which 
information is desired. How these records shall be grouped, 
depends upon the particular plant involved, and on its type. 
Further elaboration of this data for permanent records should 
be made and kept in loose-leaf files and ledgers in the office of the 
plant. From these records the monthly and annual reports may 
be compiled. 
Laboratory data should, of course, be carefully recorded and 
compiled in the same manner as the usual operation data. The 
study of both sets of records by themselves, and in relation to 
each other, is the duty of an intelligent and careful operator. 
The needless piling up of data is not advisable, although more in- 
formation will probably be collected during the first year or two 
of operation than is necessary. The really essential information 
can be quickly separated from the unessential after a year or 
two of operation, and the records for future operation modified 
accordingly. 
Laboratory.-No purification plant can be said to be really 
equipped unless it is provided with a laboratory. This is true of 
any type of plant. Bacteriological data are absolutely essential 
for guidance in the intelligent supervision of the plant, and where 
chemicals are used, a certain amount of chemical data are also 
required. Physical tests of the appearance of the water, such as 
for color and turbidity, are generally needed, and sometimes tests 
for the odor of the water are useful. Chemical tests for total hard- 
ness, carbonate and caustic alkalinities, sulphates, chlorides, free 
chlorine and oxygen consumed are quite commonly required. In 
plants handling waters containing iron and manganese, these two 
elements must of course be tested for. In waters containing con- 
siderable organic matter, tests for nitrogen in its various forms 
may throw some light on the action of the plant. 
The arrangement and equipment of the laboratory need not be 
elaborate, although certain details and some essential apparatus WA TER-P U RIF IC A TION PROCESSES 
455 
must be installed if the laboratory is to be of real service. Incu- 
bators for making the bacterial examinations, and an ice chest for 
holding samples can not be omitted. A steam bath, hot plate, 
oven, steam-sterilizing apparatus, glassware for chemical and 
bacteriological work, and a moderate amount of laboratory chem- 
icals and supplies, constitute the practical minimum of equipment. 
Interpretation of Data.-The significance of the strictly engi- 
neering data obtained in operating the plant is usually quite 
apparent. Faulty operation of mechanical devices, such as in- 
correct gage and meter readings, or stoppage of flow in pipe lines, 
or leaking tanks, are usually not hard to trace and to correct. A 
careful scrutiny of this data each day will often enable the opera- 
tor to check serious trouble in its early stages, and thus prevent 
any interference in the continuous operation of the plant. 
The significance of the chemical data obtained in operating a 
water-purification plant is generally apparent from a careful 
study of the chemical characteristics of the water before and after 
treatment, in conjunction with the particular processes employed. 
The effect of chemical coagulants, subsidence, filtration and dis- 
infection can generally be directly traced with the aid of a com- 
paratively few chemical determinations. Excess treatment with 
chemicals must, of course, always be guarded against, and the 
chemical laboratory furnishes the information by which this 
treatment of the water may be adjusted. 
The removal of color, sediment, and sometimes odor, are readily 
detected by the standard methods of examination now commonly 
in use, and the inferences to be drawn from the results obtained 
are obvious. 
The significance of the bacteriological data obtained in the 
examination of a public water supply is not as clear as could be 
wished, and the same statement is practically true, even if the 
supply has been subjected to purification processes. Since the 
present technique of water bacteriology does not enable the pres- 
ence or absence of pathogenic organisms to be directly demon- 
strated by any practicable routine procedure, the results obtained 
are indicative only, and require information from other sources in 
order to arrive at any reasonable interpretation of their intrinsic 
value. 
Bacterial counts in themselves afford but little evidence of the 
purity of the water. A pure water may show a high bacterial 
count and a polluted water a comparatively low bacterial count. 456 
WATER PURIFICATION 
However, the number of bacteria in an untreated water compared 
with the number found in the same water after it has been puri- 
fied gives a fair measure of the degree of improvement effected. 
This is, of course, based on the assumption that the proportion 
of pathogenic organisms removed by the process is in the same 
ratio as are the total number removed. It is apparent that the 
hygienic character of a water can not be definitely fixed by merely 
a knowledge of the number of bacteria it may contain, and for 
this reason a reliable indicator of pollution would be of great 
value. 
Mr. George W. Fuller in his paper on the ''Biochemical and 
Engineering Aspects of Sanitary Water Supply"1 briefly and 
clearly states the difficulty of establishing the value of an indi- 
cator of pollution, as follows: 
"After eliminating as a satisfactory indicator of pollution the total 
count of bacteria, which may originate anywhere at all, the next natural 
step is to restrict our investigation to those particular bacteria which 
seem likely to originate in the human body. The normal rule seems 
to be that only human diseases are transmitted to other human beings, 
and that bacteria, in order to cause infection, must originate in man. 
Infectious pollution in water supplies arises almost wholly from dis- 
charges from the human intestines, and any bacteria which can be 
assumed to originate in the human intestines would be a fair measure 
of such human pollution which potentially might be dangerous or 
infectious. 
"Unfortunately, there are no such bacteria which are a positive 
indicator of human pollution. The nearest thing to it are the coli 
bacilli. Thus B. coli are always found in any discharges from the human 
intestines. They are also found, however, in the discharges from the 
intestines of animals, birds, fishes and on cereals, grains and many other 
places. They are even found in the air, the dust, and, because of their 
widespread occurrence, have often been termed 'ubiquitous.' Because 
of these conditions it is not a simple matter to say that any water supply 
which shows the presence of B. coli necessarily possesses any specific 
danger of human pollution or of infection which might at any time have 
been present. It merely suggests a potential danger in such a water." 
It is, of course, true that the same reasoning with reference to 
the significance of the presence of B. coli holds good as in the 
case of the total number of bacteria, namely, that so far as re- 
moval of B. coli is effected, it is probably in the same ratio as are 
the pathogenic forms. Where disinfection is practised there may 
1 Journal of the Franklin Institute, July, 1915. WATER-PURIFICATION PROCESSES 
457 
be some selective action on the bacteria; but if this does occur, 
it would seem probable that disease-producing organisms are 
likely to be fully as susceptible to the toxic effects of the disin- 
fectant, if not more so, as are the ordinary forms of water bacteria. 
The purity of a water supply is, in the final analysis, deter- 
mined by its effect upon the people who drink the water. Ab- 
sence of known water-borne diseases, such as typhoid fever, dys- 
entery and other intestinal disturbances in a community are 
good evidence of the quality of the water supply. However, 
since such diseases are known to be transmitted through other 
avenues, such as milk, fruit, fresh vegetables, shell fish, and so 
forth, the prevalence of these diseases in a community does not 
necessarily point to an infected water supply. 
Dr. W. H. Frost, Past Assistant Surgeon of the United 
Public Health Service, in a paper on "Some Considerations in 
Estimating the Sanitary Quality of Water Supplies,"1 has set 
forth the relative value of the evidence obtained in the laboratory 
and by sanitary surveys, as follows: 
"Of the laboratory examinations applicable to determining the nature 
and extent of pollution of a water supply, bacteriological examinations 
have the most direct bearing upon sanitary quality, which is a question 
of bacterial pollution. The most specific of the bacteriological examina- 
tions in general use are quantitative tests for bacteria of the B. coli 
group, since these tests afford a direct measure of the numbers of intes- 
tinal bacteria present, and since typhoid bacilli are found only in asso- 
ciation with intestinal discharges. Nevertheless, such tests, however 
accurate and specific they may be, show only the extent of pollution 
with intestinal discharges in general; they do not distinguish between 
pollution with intestinal discharges from lower animals which are not 
subject to infection with typhoid bacilli, and the much more dangerous 
pollution from human sources. They still further fail to distinguish 
between human discharges actually containing typhoid bacilli and 
discharges free from this specific infection." 
Speaking further of the correlation of a sanitary survey and 
bacteriological examinations, Dr. Frost states, that: 
"Sanitary surveys and bacteriological examinations give only indirect 
and inferential knowledge as to the probable presence and numbers of 
typhoid bacilli in a water supply. Even were this knowledge much 
more direct and exact, it would still fall short of being all that is needed 
for an estimate of the sanitary quality of the water. The second 
1 Jour. Amer. Water-works Assn., December, 1915. 458 
WATER PURIFICATION 
requisite is an equally exact knowledge of the effects which the known 
pollution will produce." 
He concisely sums up the values that we may legitimately 
attach to the available evidence on the quality of a water supply 
as follows: 
"To recapitulate: bacteriological examinations, combined with a 
careful survey of the sources of pollution and of the safeguards against 
them, give accurate knowledge of the extent of pollution of a water 
supply in terms of intestinal bacteria from all sources. They give only 
indirect and inferential knowledge of specifically dangerous pollution 
with typhoid bacilli. To make the knowledge acquired by bacterio- 
logical examinations and sanitary surveys significant with respect to 
the sanitary quality of water, it is necessary to know the effects produced 
by a given amount of pollution. Such effects can be definitely ascer- 
tained and measured in the case of epidemic outbreaks, proven to be 
due to the use of polluted water supplies. The effects can be definitely 
recognized but not accurately measured in the case of highly polluted 
water supplies causing high rates of endemic typhoid prevalence. In 
the case of slightly polluted water supplies, it is not philosophically 
possible to probe by present methods whether or not they may cause a 
relatively small incidence of typhoid fever, so small as to be obscured 
by other more prominent factors. 
"This paper is not intended to present a pessimistic point of view 
or to deny the possibility of forming a reasonably accurate estimate of 
the sanitary quality of a water supply. On the contrary, the writer 
wishes to express his entire confidence in the reliability of an expert 
opinion, formed after careful study from all angles, and stated conserva- 
tively, with an understanding of the limitations of the evidence. Unless 
these limitations are remembered, however, there is always the danger 
of drawing too sweeping conclusions from evidence bearing solely upon 
the extent of pollution of a water supply. 
"As regards establishing a firmer basis for opinions in the future, 
this must obviously be accomplished by improving and extending not 
only laboratory studies of the quality of water supplies, but equally 
epidemiologic studies of their relation to typhoid prevalence. Methods 
of epidemiologic study, like bacteriological and chemical methods, must 
be so standardized that results can be summarized into a general fund 
of knowledge. Routine bacteriological examinations of water sup- 
plies, closely coordinated with careful, long-continued epidemiologic 
studies may be confidently expected to increase, perhaps to an extent 
not now foreseen, our ultimate knowledge of the sanitary safety of 
water supplies." WATER-PURIFICATION PROCESSES 
459 
The Human Factor.-An enterprise to be successful must be 
directed and operated with intelligence and skill. Unless the 
manager of a purification plant is properly qualified for his work, 
and unless he is given competent assistance to operate the plant, 
it may become an actual menace to the community. Dr Allen 
J. McLaughlin1 well states the situation, when he says: 
"With a polluted source the mere installation of a purification plant 
does not guarantee safe water. Even if perfect in design and con- 
struction, unless efficiently operated and controlled, a safe effluent need 
not be expected. The writer has seen filter plants designed by our best 
engineers-perfect mechanisms, which if properly operated, would 
produce safe water-placed in the hands of an assassin who was a 
promoted stoker, and absolutely ignorant of bacteriology or chemistry." 
The responsibility for mal-administration of water-purification 
plants is frequently found in an uneducated public, which, if the 
plant is privately owned, is too indolent to enforce its contract 
rights, or if publicly owned is too careless of its own obligations 
to see that the plant is placed in the hands of technically trained 
men, who are competent to produce the proper results. Fortu- 
nately, there are communities fully awake to their responsibili- 
ties in conserving the public health by a safe water supply, and 
in such the operator of a plant finds his efforts to maintain an 
efficient force well supported. 
The manager of a plant who is able to instill into his assistants 
the need for that careful attention to details, without which no 
plant can be properly kept up, will be the most successful in 
maintaining an esprit de corps, which will be reflected in the gen- 
eral appearance of the plant, and in the quality of its output. 
Economy of operation, consistent with maintaining a high stand- 
ard of purity for the effluent of the plant, must be carefully and 
continuously studied. The safety of the supply, however, must 
always be kept in mind, for "eternal vigilance is the price of pure 
water." 
1 "Sewage Pollution of Boundary Waters." Jour. Am. Water-works 
Assn., May, 1914. APPENDIX A 
ON THE FLOW OF WATER THROUGH RAPID SAND 
FILTERS 
BY 
C. N. MILLER, ASSOC. M. AM. SOC. C.E. 
In this discussion it is assumed that the walls of the filter tank and all 
channels and piping are tight, so that no air can enter, and that the lower 
end of the outlet pipe is submerged or trapped so that no air can enter also 
at this point. 
The notation used is as follows: 
Let ha = the pressure of the atmosphere. 
Es = the elevation, in feet above some assumed datum, of the surface 
of the water in the filter tank. 
Eo - the elevation of the surface of the water in the effluent chamber. 
En = the elevation of any intermediate horizontal plane. 
va = the velocity of the water at the surface in the filter, equals zero. 
v0 = the velocity of the water in the outlet pipe. 
vn = the velocity of the water at any intermediate horizontal plane. 
ps = the absolute pressure at the surface of the water in the filter. 
po = the absolute pressure at the surface of the water in the effluent 
chamber. 
pn = the absolute pressure at any intermediate horizontal plane. 
hf = any one of the frictional resistances overcome in the passage of 
the water through the filter. 
All pressures are to be measured in feet of water. Absolute pressures are 
to be measured above zero pressure. 
The flow of the water being uniform, we have by Bernoulli's theorem: 
Ea + Ps + 2" ~ En + Pn + + En hf 
where 
Enhf - the sum of the frictional resistances between the surface of the 
water in the filter and the plane n. 
But vs = 0, and ps = ha, therefore, we have 
Eg + ha = En + pn 4- 9- + En hf 
4g 
or 
ha~Pn=(~+ K hf} - (E, - En) (1) 
Also 
|»2 
En + Pn + V - E0 + Po + 9" + 'S" hf 
"9 4g 
461 462 
WATER PURIFICATION 
but p0 = ha, therefore, we have 
En + Pn + 9T = Eo + ha + hf 
or 
(1) \ 
9^-9"+^"^ (2) 
We also have 
Es + ps + = Eo + Po + 2^ Eo hf 
but vs = 0, and p, = p0 = ha, therefore, 
Eq - Eo = g-j 4- hf (3) 
The velocity heads are quite small as compared with the frictional resist- 
ances, and for practical purposes may be neglected, in which case the above 
equations may be written: 
ha - Pn = E: hf - {Eq - E„) (4) 
ha - Pn = {En - Eq) - Sghf (5) 
Es -Eo= Sohf (6) 
From equation (4) it is seen that when the frictional resistance encountered 
down to any intermediate plane is less than (Es - En), that is the static head 
of water on the plane then the term ha - pn is negative or the absolute 
pressure at the plane is greater than atmospheric pressure; when the fric- 
tional resistance is equal to the static head the term ha - pn = 0, or the 
absolute pressure is equal to the atmospheric pressure; and when the 
frictional resistance is greater than the static head, the term ha - pn is 
positive or the absolute pressure is less than atmospheric pressure. 
From equation (5) it is seen that when the frictional resistance below any 
intermediate plane is less than En - Eo, that is the portion of the static 
head below the plane, the term ha - pn is positive, or the absolute pressure 
is less than atmospheric; when the frictional resistance is equal to En - Eo, 
the term ha - pn = 0, or the absolute pressure is equal to the atmospheric; 
and when the frictional resistance is greater than En - Eo, the term ha - pn 
is negative, or the absolute pressure is greater than atmospheric. 
Equations (4) and (5) also show that, when the frictional resistance 
above any intermediate plane is less than the static head on the plane, the 
frictional resistance below the plane must be greater than the remaining 
portion of the total static head by an equal amount. The converse of the 
above statement is also true. That this must be the case appears from 
equation (6), which shows that the total available head is equal to the total 
frictional resistance. 
When the absolute pressure at any intermediate plane is less than atmos- 
pheric pressure, a condition exists which is commonly described by stating 
that the filter is operating under a "negative head." From the above dis- 
cussion it is seen that this condition is only a particular case of the general 
problem of the flow of water through a filter. APPENDIX A 
463 
The accompanying sketch (Fig. 149) may illustrate some of the points 
brought out in the foregoing discussion. The arrangement of the filter, 
effluent chamber and appurtenances shown, is that of one of the Harris- 
burg, Pa., filters, according to information taken from an article in the 
Engineering Record of Mar. 9, 1907. The arrangement of the trapped tubes 
for indicating the losses of head at the various planes passing through the 
points at which the tubes penetrate the wall of the filter is adapted from the 
discussion by Mr. Fuertes, of his paper on "Water Purification at Steelton, 
Pa.," appearing in the Transactions of the American Society of Civil Engi- 
neers for March, 1910. The locations of the points at which the tubes pass 
Surface of Water 
Observations 
at 3.48 P.M. 
7.00 P.M. 
„El.En 
10.48 P.M. 
Top of Sand 
Bottom of GraveK 
Top of Gravely 
Fig. 149.-Diagrammatic sketch to illustrate flow of water through a filter. 
through the wall of the filter and the distances which the water in the tubes 
stand below the level of the water in the filter, as well as the relative loca- 
tions of the top of gravel, top of sand, etc., are scaled from Mr. Fuertes' 
diagram for the run of June 7, 1909. The arrangement of the tubes showing 
the trapping is diagrammatic. The level of the water in the filter varied 
slightly during the period in which the observations were taken, but for 
simplicity has been taken at an approximate average value as shown on the 
sketch. 
The horizontal distances of the vertical axes of the tubes B, C, D, E and 
F from the line OM are respectively equal to the distances of the points 
where the tubes pierce the side of the filter below the surface of the water in 464 
WATER PURIFICATION 
the same. The distance OG is equal to the distance between the level of the 
water in the filter and that in the effluent chamber. The line ON drawn at 
45° with the horizontal evidently cuts the vertical axis of each tube at a 
point on a level with that where the tube pierces the side of the filter, except 
the axis of tube G which is cut at the level of the surface of the water in the 
effluent chamber. 
Consider tube D and let the horizontal plane passing through the point 
where the tube pierces the side of the filter be taken as plane n. The nota- 
tions made on the diagram in regard to tube D are self-evident. The total 
head Es - Eo is divided by the water level in the tube into two parts, the 
upper one being equal to the frictional resistance above the plane n or to 
Snhf, and the lower one to the frictional resistance below plane n or to S^hf. 
We then have 
ha - pn = S'nhf - (E s - En) or equation (4) 
and 
ha - pn = (En - E^ - S^hf or equation (5) 
« 
The curves shown in the sketch are those given by the equation 
tan a Df 
, / X \ 
» -hr [dP 
where Df is equal to the distance from the top of the sand bed to the plane n 
passing through the point where the tube F pierces the side of the filter, hp 
is equal to the distance from the surface of the water in the filter to the level 
of the water in the tube F, that is to the head or force required to maintain 
the flow from above down to the plane n, and tan a is equal to the tangent of 
dy 
the angle the curve makes with the horizontal when x = Df, that is to 
Tangent a is evidently equal to the head required to force the water through 
a unit thickness of material such as that at the level of the plane n, and its 
value is taken as 0.34 for each of the three curves, it being assumed that the 
frictional resistance does not vary for a layer at this level during the run, 
which is practically the case. Each curve corresponds to a set of observa- 
tions taken simultaneously in all the tubes. 
By reference to the diagram it will be evident that as long as the curve 
connecting the water levels in the tubes remains above the line ON, the 
pressures at the various planes in the filter will be greater than atmospheric. 
When the curve cuts the line ON it indicates that a zone will exist in the 
filter in which pressures will be less than atmospheric. 
Since preparing the above diagram similar curves have been published 
by H. Malcolm Pirnie in the discussion of a paper by Joseph W. Ellms and 
John S. Gettrust, which appeared in vol. 80, page 1342 (1916) of the Transac- 
tions of the American Society of Civil Engineers. APPENDIX B 
AN APPROXIMATE FORMULA FOR CALCULATING THE 
DISCHARGING CAPACITY OF RAPID SAND 
FILTER WASH WATER TROUGHS 
BY 
C. N. MILLER, ASSOC. M. AM. SOC. C.E. 
Let ABCD represent a trough of rectangular cross-section receiving 
water uniformly along the level edges BC. 
Fig. 150.-Diagram to illustrate flow of water in a wash-water trough. 
Let Qi = the total quantity of water received by the trough in second-feet. 
Q = the quantity passing any section as at F in second-feet. 
v = the horizontal velocity at section F. 
v2 = the horizontal velocity at the lower end of the trough. 
a = the angle the bottom of the trough makes with the horizontal. 
O = the origin of coordinates. 
EFG = the surface curve of the water. 
I = the length of the trough. 
b = the width of the trough. - 
yi = the depth of water at the upper end of trough. 
yz = the depth of water at the lower end of the trough. 
w = the weight of a cubic foot of water. 
x and y = the coordinates of the surface curve at section F. 
The quantity of water Q falling through the height dy generates an 
increment of kinetic energy 
wi\2g) 
and overcomes the frictional resistance in a length dx, noting that Q and v 
both vary. Assuming that the work done in overcoming the frictional 
resistance may be expressed by 
. , (Qv2\ 
465 466 
WA TER P U RIF IC A TION 
we have the equation 
- wQdy = wd + wtd y 2g / or' dividing by w, 
-Qdy = {l+^d(^ (1) 
But Q = Qi xy therefore, 
I 
-Q'Xdy = (1 (2) 
Integrating equation (2) throughout the length of the trough, we have, 
n * 
Q^dy=(l+^^ (3) 
+1 tan a L 
Cyt 
But - I xdy = the area of the figure EFGH, which may 
J Ui + l tan a 
be expressed by m(yi + I tan a - y^l, where m equals a coefficient whose 
value would be % if the surface curve were a parabola. We have then by 
making this substitution 
w V (l/i + Han a - y2}l = (1 + f) 
or 
y. + Han a - y2 = (^) (4) 
But v2 = whence 
by2 
, i, 1+f /QA*1 ... 
y. +1 tan a-y,- - - (5) 
From equation (5) it can be seen that Qi equals 0 when y2 equals yx + I 
tan a, that is when there is no fall, and also when y2 equals 0. If yx is 
considered constant, there will be some value of y2 between those above 
mentioned, which will render Qi a maximum. Assume then that the levels 
will so adjust themselves that Qi will be a maximum. By differentiating 
equation (5) with respect to y2, considering y^ constant, we have 
M n \ 
_i = i±i2m (6) 
2mg \by2/ b \ y2 / 
In equation (6) put = 0, and we have 
i 1 + 
, mg b*y* 
whence 
Eliminate Qi between equations (5) and (7), and we have 
1/2 = %(yi + I tan a) (8) 
From equations (7) and (8), we have TABLES 
467 
Qi = + nana)% (9) 
By applying equation (9) to some experimental data obtained by Mr. 
J. W. Ellms at the Cincinnati filtration plant, a value of £ of 0.75 was ob- 
tained by assuming a value for m of %. Substituting these values of m and 
f in equation (9) we have finally 
Qi = 1.91b(?/i -|- Han a)% 1 (10) 
In the case of a trough whose cross-section is not rectangular, application 
of the above formulae may be made by assuming a rectangular section of 
equivalent area to the section of the trough under consideration. 
The value of y2 as given by equation (8) applies to the case where the dis- 
charge is into a chamber in which the water level is at or near the level of 
the water in the lower end of the trough. In the case of a free fall from the 
end of the trough, owing to weir action, the actual value of y2 will be less 
than that given by the equation. 
Table 1.-Approximate Equivalents of Various Measures of Rate 
of Filtration 
Million 
U. S. gal- 
lons per 
acre per 
24 hr. 
Million 
British 
gallons 
per acre 
per 24 hr. 
U. S. gallons 
per square 
foot per 
hour 
British 
gallons per 
square foot 
per hour 
1,000,000 U. S. gal. per acre per 24 hr 
1.000 
0.833 
0.96 
0.80 
1,000,000 British gal. per acre per 24 hr.... 
1.200 
1.000 
1.15 
0.96 
1 U. S. gal. per square foot per hour 
1.045 
0.870 
1.00 
0.83 
1 British gal. per square foot per hour 
1.255 
1.045 
1.20 
1.00 
1 cu. ft. per square yard per hour 
0.869 
0.724 
0.83 
0.69 
1 linear in., vertical velocity, per hour 
0.652 
0.543 
0.62 
0.52 
100 linealmm., vertical velocity, per hour... 
1 lineal meter, vertical velocity, per 24 hr. 
2.566 
2.139 
2.46 
2.05 
= 1 cu. meter per square meter per 24 hr .. 
1.069 
0.891 
1.02 
0.85 
Cubic 
feet per 
square 
yard per 
hour 
Vertical 
velocity 
in inches 
per hour 
Vertical 
velocity in 
millimeters 
per hour 
Vertical veloc- 
ity in meters 
per 24 hr. Or 
cubic meters 
per square 
meter per 24 hr. 
1,000,000 U. S. gal. per acre per 24 hr. . . . 
1.15 
1.53 
39.0 
0.935 
1,000,000 British gal. per acre per 24 hr. 
1.38 
1.84 
46.8 
1.122 
1 U. S. gal. per square foot per hour 
1.20 
1.60 
40.7 
0.978 
1 British gal. per square foot per hour. . . . 
1.44 
1.92 
48.9 
1.174 
1 cu. ft. per square yard per hour 
1.00 
1.33 
33.9 
0.813 
1 linear in., vertical velocity, per hour. ... 
0.75 
1.00 
25.4 
0.610 
100 lineal mm., vertical velocity, per hour. 
1 lineal meter, vertical velocity, per 24 hr. 
2.95 
3.94 
100.0 
2.400 
- 1 cu. meter per square meter per 24 hr. 
1.23 
1.64 
41.7 
1.000 
From Report on Water Purification at Louisville. 468 
WA TER P U RIF IC A TION 
Table 2.-Conversion of Statements of Chemical Composition 
Grains per 
U. S. gallon 
(231 cu. in.) 
Grains per 
British 
gallon 
(277 cu. in.) 
Parts per 
hundred 
thousand 
Parts per 
million 
1 grain per U. S. gal- 
lon 
1.000 
1.20 
1.71 
17.1 
1 grain per British gal- 
lon 
0.830 
1.00 
1.43 
14.3 
1 part per 100,000 
0.580 
0.70 
1.00 
10.0 
1 part per 1,000,000. . . 
0.056 
0.07 
0.10 
1.0 
Table 3.-Conversion Table of Hardness from English, French and 
German Degrees to Parts per Million 
Parts per 
million 
Clark 
degrees 
French 
degrees 
German 
degrees 
Parts per million 
Clark or English de- 
1.0 
0.07 
0.10 
0.56 
grees 
14.3 
1.00 
1.43 
0.80 
French degrees 
10.0 
0.70 
1.00 
0.56 
German degrees 
17.8 
1.24 
1.78 
1.00 
Note.-English degrees of hardness, Clark scale, are equivalent to grains 
of calcium carbonate per imperial gallon, and are multiplied by 14.3 to give 
parts per million. French degrees of hardness represent parts per hundred 
thousand of calcium carbonate, and are multiplied by 10 to give parts per 
million. German degrees of hardness represent parts per hundred thousand 
of calcium oxide, and are multiplied by 17.8 to give parts per million of 
calcium carbonate. TABLES 
469 
Metric system 
Length 
1 millimeter = 0.0394 inches 
1 centimeter = 0.3937 inches 
U. S. standard 
Length 
1 inch = 2.5309 centimeters 
1 foot = 30.4794 centimeters 
1 yard = 0.9143 meters 
1 mile = 1.6093 kilometers 
Metric system 
Weight 
1 gram = 15.4323 grains 
1 kilogram = 2.2046 lb. 
1 tonneau = 2204.55 lb. 
1 lb. 
1 cwt. 
1 ton 
U. S. standard 
W eight 
= 0.4536 kilos 
= 50.8024 kilos 
= 1016.0483 kilos 
1 meter 
1 kilometer 
= 39.3708 inches 
= 0.6214 miles 
Area 
1 sq. centimeter = 0.1549 sq. in. 
Area 
1 sq. in. = 6.4513 sq. centimeters 
Dry measure 
1 centiliter = 0.0181 pints 
1 pint 
Dry measure 
= 55.0661 centiliters 
1 sq. meter 
= 10.7631 sq. ft. 
1 sq. ft. = 0.0929 sq. meters 
1 liter = 0.908 quarts 
1 quart 
= 1.1013 liters 
1 are 
= 119.5894 sq. yd. 
1 sq. yd. = 0.8361 sq. meters 
1 hectoliter = 2.837 bushels 
1 bushel 
= 35.2416 liters 
1 hectare 
1 cubic meter 
= 2.4711 acres 
Volume 
= 35.3166 cu. ft. 
1 acre = 0.4047 hectares 
Volume 
1 cubic foot = 0.02831 cubic meters 
Liquid measure 
; 1 centiliter = 0.0211 pints 
1 liter = 1.0567 quarts 
1 hectoliter = 26.4176 gallons 
1 pint 
1 quart 
1 gallon 
Liquid measure 
= 47.3171 centiliters 
= 0.9563 liters 
= 3.7854 liters 
U. S. gallon 
Imperial 
gallon 
Cubic inch 
Cubic foot 
Pound 
Cwt. 
Ton 
Liter 
Cubic 
meter 
U. S. gallon 
1.0 
0.83 
231.0 
0.133 
8.33 
0.07455 
0.00372 
3.8 
0.00378 
Imperial gallon 
1.2 
1.0 
277.274 
0.16 
10.0 
0.0892 
0.00446 
4.537 
0.00454 
Cubic inch 
0.004329 
0.003607 
1.0 
0.00058 
0.03607 
0.0163 
Cubic foot 
7.48 
6.23 
1728.0 
1.0 
62.35 
0.557 
0.028 
28.375 
0.0283 
Pound 
0.0S3 
0.10 
27.72 
0.016 
1.0 
112.0 
2240.0 
Cwt 
13.44 
11.2 
1.8 
1.0 
20.0 
Ton 
268.8 
224.0 
35.9 
0.05 
1.0 
1000.0 
1.0 
Liter 
0.264 
0.22 
61.0 
0.0353 
0.001 
1.0 
0.001 
Cubic meter 
264.0 
220.0 
61028.0 
35.31 
1.0 
1000.0 
1.0 । 
1U. S. ton of shipping = 40 cubic feet = 32,143 U. S. bushels = 1,1326 cubic meters 
Equivalents of Various Weights and Measures 
Table 4.-Weights and Measures 
(From Slocum's " Elements of Hydraulics ") 470 
WATER PURIFICATION 
Table 5.-Head and Pressure Equivalents 
(From Slocum's "Elements of Hydraulics") 
Head of Water in Feet and Equivalent Pressure in Pounds per 
Square Inch 
Feet 
head 
Pounds per 
sq. in. 
Feet 
head 
Pounds per 
sq. in. 
Feet 
head 
Pounds per 
sq. in. 
1 
0.43 
55 
23.82 
190 
82.29 
2 
0.87 
60 
25.99 
200 
86.62 
3 
1.30 
65 
28.15 
225 
97.45 
4 
1.73 
70 
30.32 
250 
108.27 
5 
2.17 
75 
32.48 
275 
119.10 
6 
2.60 
80 
34.65 
300 
129.93 
7 
3.03 
85 
36.81 
325 
140.75 
8 
3.40 
90 
38.98 
350 
151.58 
9 
3.90 
95 
41.14 
375 
162.41 
10 
4.33 
100 
43.31 
400 
173.24 
15 
6.50 
110 
47.64 
500 
216.55 
20 
8.66 
120 
51.97 
600 
259.85 
25 
10 83 
130 
56.30 
700 
303.16 
30 
12.99 
140 
60.63 
800 
346.47 
35 
15.16 
150 
64.96 
900 
389.78 
40 
17.32 
160 
69.29 
1,000 
433.09 
45 
19.49 
170 
73.63 
50 
21.65 
180 
77.96 
Pressure in Pounds per Square Inch and Equivalent Head of Water in Feet 
Pounds per 
sq. in. 
Feet 
head 
Pounds per 
sq. in. 
F eet 
head 
Pounds per 
sq. in. 
Feet 
, head 
1 
2.31 
55 
126.99 
180 
415.61 
2 
4.62 
60 
138.54 
190 
438.90 
3 
6.93 
65 
150.08 
200 
461.78 
4 
9.24 
70 
161.63 
225 
519.51 
5 
11.54 
75 
173.17 
250 
577.24 
6 
13.85 
80 
184.72 
275 
643.03 
7 
16.16 
85 
196.26 
300 
692.69 
8 
18.47 
90 
207.81 
325 
750.41 
9 
20.78 
95 
219.35 
350 
808.13 
10 
23.09 
100 
230.90 
375 
865.89 
15 
34.63 
110 
253.98 
400 
922.58 
20 
46.18 
120 
277.07 
500 
1,154.48 
25 
57.72 
125 
288.62 
30 
69.27 
130 
300.16 
35 
80.81 
140 
323.25 
40 
92.36 
150 
346.34 
45 
103.90 
160 
369.43 
50 
115.45 
170 
392.52 TABLES 
471 
Table 6.-Discharge Equivalents 
(From Slocum's "Elements of Hydraulics") 
Gallons 
per 
min. 
Cubic 
feet per 
sec. 
Cubic feet 
per 
min. 
Gallons 
per 
hour 
Gallons 
per 24 
hours 
Bbls, per 
minute, 
42 gal. 
bbl. 
Bbls, per 
hour, 42 
gal. bbl. 
Bbls, per 
24 hours, 42 
gal. bbl. 
10 
1.3368 
600 
14,400 
0.24 
14.28 
342 8 
12 
1.6042 
720 
17,280 
0.29 
17.14 
411 4 
15 
2.0052 
900 
21,600 
0.36 
21.43 
514 3 
18 
2.4063 
1,080 
25,920 
0.43 
25.71 
617 1 
20 
2.6733 
1,200 
28,800 
0.48 
28.57 
685 7 
25 
3.342 
1,500 
36,000 
0.59 
35.71 
857 0 
27 
3.609 
1,620 
38,880 
0.64 
38.57 
925 0 
30 
4.001 
1,800 
43,200 
0.71 
42.85 
1 028 0 
35 
4.678 
2,100 
50,400 
0.83 
50.0 
1 200 0 
36 
4.812 
2,160 
51,840 
0.86 
51.43 
1 234 0 
40 
5.348 
2,400 
57,600 
0.95 
57.14 
1 371 0 
45 
0.1 
6.015 
2,700 
64,800 
1.07 
64.28 
1,543.0 
50 
6.684 
3,000 
72,000 
1.19 
71.43 
1 714 0 
60 
8.021 
3,600 
86,400 
1.43 
85.71 
2 057 0 
70 
9.357 
4,200 
100,800 
1.66 
100.0 
2 400 0 
75 
10.026 
4,500 
108,000 
1.78 
107.14 
2 570 0 
80 
10.694 
4,800 
115,200 
1.90 
114.28 
2 742 0 
90 
0.2 
12.031 
5,400 
129,600 
2.14 
128.5 
3,085.0 
100 
13.368 
6,000 
144,000 
2.39 
142.8 
3 428 0 
125 
16.710 
7,500 
180,000 
2.98 
178.6 
4 286 0 
135 
0.3 
18.046 
8,100 
194,400 
3.21 
192.8 
4,628.0 
150 
20.052 
9,000 
216,000 
3.57 
214.3 
5 143 0 
175 
23.394 
10,500 
252,000 
4.16 
250.0 
6 000 0 
180 
0.4 
24.062 
10,800 
259,200 
4.28 
257.0 
6,171.0 
200 
26.736 
12,000 
288,000 
4.76 
285.7 
6,857 0 
225 
0.5 
30.079 
13,500 
324,000 
5.35 
321.4 
7,714.0 
250 
33.421 
15,000 
360,000 
5.95 
357.1 
8,570 0 
270 
0.6 
36.093 
16,200 
388,800 
6.43 
385.7 
9,257.0 
300 
40.104 
18,000 
432,000 
7.14 
428.5 
10 284 0 
315 
42.109 
18,900 
453,600 
7.5 
450.0 
10,800 0 
360 
0.8 
48.125 
21,600 
518,400 
8.57 
514.3 
12,342.0 
400 
53.472 
24,000 
576,000 
9.52 
571.8 
13,723 0 
450 
1.0 
60.158 
27,000 
648,000 
10.7 
642.8 
15,428.0 
500 
66.842 
30,000 
720,000 
11.9 
714.3 
17 143 0 
540 
1.2 
72.186 
32,400 
777,600 
12.8 
771.3 
18,512.0 
600 
80.208 
36,000 
864,000 
14.3 
857.1 
20 570 0 
630 
1.4 
84.218 
37,800 
907,200 
15.0 
900.0 
21,600.0 
675 
1.5 
90.234 
40,500 
972,000 
16.0 
964.0 
23,143.0 
720 
1.6 
96.25 
43,200 
1,036,800 
17.0 
1,028.0 
24,685.0 
800 
106.94 
48,000 
1,152,000 
19.05 
1,142.0 
27 387 0 
900 
2.0 
120.31 
54,000 
1,296,000 
21.43 
1,285.0 
30,857.0 
1,000 
133.68 
60,000 
1,440,000 
23.8 
1,428. 0 
34 284 0 
1,125 
2.5 
150.39 
67,500 
1,620,000 
26.78 
1,607.0 
38,571.0 
1,200 
160.42 
72,000 
1,728,000 
28.57 
1,714.0 
41,143 0 
1,350 
3.0 
180.46 
81,000 
1,944,000 
32.14 
1,928.0 
46,085.0 
1,500 
200.52 
90,000 
2,160,000 
35.71 
2 142 0 
51 427 0 
1,575 
3.5 
210.54 
94,500 
2,268,000 
37.5 
2,250.0 
54,000.0 
1,800 
4.0 
240.62 
108,000 
2,592,000 
42.85 
2,571.0 
61,710.0 
2,000 
267.36 
120,000 
2,880,000 
47 64 
2 857 0 
68,568 0 
2,025 
4.5 
270.70 
121,500 
2,916,000 
48.21 
2,892.0 
69,425.0 
2,250 
300.78 
135,000 
3,240,000 
53.57 
3,214 0 
77 143 0 
2,500 
334.21 
150,000 
3,600,000 
59.52 
3 571 0 
85 704 0 
2,700 
6.0 
360.93 
162,000 
3,880,000 
64.3 
3,857.0, 
92,572.0 
3,000 
401.04 
180,000 
4,320,000 
71.43 
4,285.0 
102,840.0 472 
WATER PURIFICATION 
Table 7.1-Efflux Coefficients for Circular Orifice 
Values of efflux coefficient K in Eq. (32), Par. 55, Q = ^KbX^2g (H3'3 - h3'2), for cir- 
cular, vertical orifices, with sharp edges, full contraction and free discharge 
in air. For heads over 100 ft., use K = 0.592 
Head 
Diameter of orifice in feet 
on cen- 
ter of 
I 
orifice 
0.02 
0.03 
0.04 
0.05 
0.07 
0.10 
0.12 
0.15 
0.20 
0.40 
1.0 
in feet 
0.60 | 0.80 
0.3 
0.637 
0.628 
0.621 
0.613 
0.608 
0.4 
0.637 
0.631 
0.624 
0.618 
0.612 
0.606 
0.5 
0.643 
0.633 0.627 
0.621 
0.615 
0.610 
0.605 
0.600 
0.596 
0.592 
0.6 
0.655 
0.640 
0.630 
0.624 
0.618 
0.613 
0.609 
0.605 
0.601 
0.596 
0.593 0.590 
0.7 
0.651 
0.637 
0.628 
0.622 
0.616 
0.611 
0.607 
0.604 
0.601 
0.597 
0.594 0.591 
0.590 
0.8 
0.648 
0.634 
0.626 
0.620 
0.615 
0.610 
0.606 
0.603 
0.601 
0.597 
0.594^0.592 
0.591 
0.9 
0.646 
0.632 
0.624 
0.618 
0.613 
0.609 
0.605 
0.603 
0.601 
0.598 
0.5950.593 
0.591 
1.0 
0.644 
0.631 
0.623 
0.617 
0.612 
0.608 
0.605 
0.603 
0.600 
0.598 
0.595 0.593 
0.591 
1.2 
0.641 
0.628 
0.620 
0.615 
0.610 
0.606 
0.604 
0.602 
0.600 
0.598 
0.596 0.594 
0.592 
1.4 
0.638 
0.625 
0.618 
0.613 
0.609 
0.605 
0.603 
0.601 
0.600 
0.599 
0.596^0.594 
0.593 
1.6 
0.636 
0.624 
0.617 
0.612 
0.608 
0.605 
0.602 
0.601 
0.600 
0.599 
0.597 0.595 
0.594 
1.8 
0.634 
0.622 
0.615 
0.611 
0.607 
0.604 
0.602 
0.601 
0.599 
0.599 
0.597*0.595 
0.595 
2.0 
0.632 
0.621 
0.614 
0.610 
0.607 
0.604 
0.601 
0.600 
0.599 
0.599 
0.597 0.596 
0.595 
2.5 
0.629 
0.619 
0.612 
0.608 
0.605 
0.603 
0.601 
0.600 
0.599 
0.599 
0.598 0.597 
0.596 
3.0 
0.627 
0.617 
0.611 
0.606 
0.604 
0.603 
0.601 
0.600 
0.599 
0.599 
0.598 0.597 
0.597 
3.5 
0.625 
0.616 
0.610 
0.606 
0.604 
0.602 
0.601 
0.600 
0.599 
0.599 
o.sgs^.sg? 
0.596 
4.0 
0.623 
0.614 
0.609 
0.605 
0.603 
0.602 
0.600 
0.599 
0.599 
0.598 
0.597 0.597 
0.596 
5.0 
0.621 
0.613 
0.608 
0.605 
0.603 
0.601 
0.599 
0.599 
0.598 
0.598 
0.597 0.596 0.596 
6.0 
0.618 
0.611 
0.607 
0.604 
0.602 
0.600 
0.599 
0.599 
0.598 
0.598 
0.597 0.596 
0.596 
7.0 
0.616 
0.609 
0.606 
0.603 
0.601 
0.600 
0.599 
0.599 
0.598 
0.958 
0.597 0.596 
1 
0.596 
8.0 
0.614 
0.608 
0.605 
0.603 
0.601 
0.600 
0.599 
0.598 
0.598 
0.597 
0.596 0.596 
0.596 
?9.0 
0.613 
0.607 
0.604 
0.602 
0.600 
0.599 
0.599 
0.598 
0.597 
0.597 
0.596 0.596 
0.595 
10.0 
0.611 
0.606 
0.603 
0.601 
0.599 
0.598 
0.598 
0.597 
0.597 
0.597 
0.596*0.596 
0.595 
20.0 
0.601 
0.600 
0.599 
0.598 
0.597 
0.596 
0.596 
0.596 
0.596 
0.596 
0.596 0.595 
0.594 
50.0 
0.596 
0.596 
0.595 
0.595 
0.594 
0.594 
0.594 
0.594 
0.594 
0.594 
0 594 0.593 
0.593 
100.0 
0.593 
0.593 
0.592 
0.592 
0.592 
0.592 
0.592 0.592 
0.592 
0.592 
0.592 0.592 
0.592 
1 From Hamilton Smith's ••Hydraulics." TABLES 
473 
Table 8.1-Efflux Coefficients for Square Orifice 
Values of efHux coefficient K in Eq. (32), Par. 55, Q = ^iKbx^2g( - A3/2), for square, 
vertical orifices, with sharp edges, full contraction, and free discharge in air. 
For heads over 100 ft. use K = 0.598 
Head 
on cen- 
ter of 
orifice 
in feet 
Side of square in feet 
0.02 
0.03 
0.04 
0.05 
0.07 
0.10 
0.12 
0.15 
0.20 
0.40 
0.69 
0.80 
1.0 
0.3 
0.642 
0.63210 624 
0.617 
0.612 
0.4 
0.643 
0.637 
0.628 
0.621 
0.616 
0.611 
0.5 
0.648 
0.639 
0.633 
0.625 
0.619 
0.614 
0.610 
0.605 
0.601 
0.597 
0.6 
0.660 
0.645 
0.636 
0.630 
0.623 
0.617 
0.613 
0.610 
0.605 
0.601 
0.598 
0.596 
0.7 
0.656 
0.642 
0.633 
0.628 
0.621 
0.616 
0.612 
0.609 
0.605 
0.602 
0.599 
0.598 
0.596 
0.8 
0.652 
0.639 
0.631 
0.625 
0.620 
0.615 
0.611 
0.608 
0.605 
0.602 
0.600 
0.598 
0.597 
0.9 
0.650 
0.637 
0.629 
0.623 
0.619 
0.614 
0.610 
0.608 
0.605 
0.603 
0.601 0.599 
0.598 
1.0 
0.648 
0.636 
0.628 
0.622 
0.618 
0.613 
0.610 
0.608 
0.605 
0.603 
0.601 
0.600 
0.599 
1.2 
0.644 
0.632 
0.625 
0.620 
0.616 
0.611 
0.609 
0.607 
0.605 
0.604 
0.602 
0.601 
0.600 
1.4 
0.642 
0.630 
0.623 
0.618 
0.614 
0.610 
0.608 
0.606 
0.605 
0.604 
0.602 
0.601 
0.601 
1.6 
0.640 
0.628 
0.621 
0.617 
0.613 
0.609 
0.607 
0.606 
0.605 
0.605 
0.603 
0.602 
0.601 
1.8 
0.638 
0.627 
0.620 
0.616 
0.612 
0.609 
0.607 
0.606 
0.605 
0.605 
0.603 0.602 
0.602 
2.0 
0.637 
0.626 
0.619 
0.615 
0.612 
0.608 
0.606 
0.606 
0.605 
0.605 
0.604 
0.602 
0.602 
2.5 
0.634 
0.624 
0.617 
0.613 
0.610 
0.607 
0.606 
0.606 
0.605 
0.605 
0.604 
0.603 
0.602 
3.0 
0.632 
0.622 
0.616 
0.612 
0.609 
0.607 
0.606 
0.606 
0.605 
0.605 
0.604 
0.603 
0.603 
3.5 
0.630 
0.621 
0.615 0.611 
0.609 
0.607 
0.606 
0.605 
0.605 
0.605 
0.604 
0.603 
0.602 
4.0 
0.628 
0.619 
0.614 
0.610 
0.608 
0.606 
0.606 
0.605 
0.605 
0.605 
0.603 
0.603 
0.602 
5.0 
0.626 
0.617 
0.613 
0.610 
0.607 
0.606 
0.605 
0.605 
0.604 
0.604 
0.603 
0.602 
0.602 
6.0 
0.623 
0.616 
0.612 
0.609 
0.607 
0.605 
0.605 
0.605 
0.604 
0.604 
0.603 
0.602 
0.602 
7.0 
0.621 
0.615 
0.611 
0.608 
0.607 
0.605 
0.605 
0.604 
0.604 
0.604 
0.603 
0.602 
0.602 
8.0 
0.619 
0.613 
0.610 
0.608 
0.606 
0.605 
0.604 
0.604 
0.604 
0.603 
0.603 
0.602 
0.602 
9.0 
0.618 
0.612 
0.609 
0.607 
0.606 
0.604 
0.604 
0.604 
0.603 
0.603 
0.602 
0.602 
0.601 
10.0 
0.616 
0.611 
0.608 
0.606 
0.605 
0.604 
0.604 
0.603 
0.603 
0.603 
0.602 
0.602 
0.601 
20.0 
0.606 
0.605 
0.604 
0.603 
0.602 
0.602 
0.602 
0.602 
0.602 
0.601 
0.601 
0.601 
0.600 
50.0 
0.602 
0.601 
0.601 
0.601 
0.601 
0.600 
0.600 
0.600 
0.600 
0.600 
0.599 
0.599 
0.599 
100.0 
0.599 
0.598 
0.598 0.598 0.598 
0.598 
0.598 
0.598 
0.598 
0.598 
0.598 
0.598 
0,598 
1 From Hamilton Smith's "Hydraulics." 474 
WATER PURIFICATION 
Table 9.-Discharge per Foot of Length over Rectangular Notch 
Weirs 
Discharge over sharp crested, vertical, rectangular notch weirs in cubic feet per second 
per foot of length. Computed from Eq. (41), Par. 66: 
Q = 3.3bh3/2 for 6 = 1 ft. 
(From Slocum's "Elements of Hydraulics") 
Depth on crest 
in feet 
0.00 
0.01 
0.02 
0.03 
0.04 
0.05 
0.06 
0.07 
0.08 
0.09 
0.0 
0.000 
0.003 
0.009 
0.017 
0.026 
0.037 
0.049 
0.061 
0.075 
0.089 
0.1 
0.104 
0.120 
0.137 
0.155 
0.173 
0.192 
0.211 
0.231 
0.252 
0.273 
0.2 
0.295 
0.317 
0.341 
0.364 
0.388 
0.413 
0.438 
0.463 
0.489 
0.515 
0.3 
0.542 
0.570 
0.597 
0.626 
0.654 
0.683 
0.713 
0.743 
0.773 
0.804 
0.4 
0.835 
0.866 
0.898 
0.931 
0.963 
0.996 
1.030 
1.063 
1.098 
1.132 
0.5 
1.167 
1.202 
1.238 
1.273 
1.309 
1.346 
1.383 
1.420 
1.458 
1.496 
0.6 
1.538 
1.572 
1.611 
1.650 
1.690 
1.729 
1.769 
1.810 
1.850 
1.892 
0.7 
1.933 
1.974 
2.016 
2.058 
2.101 
2.143 
2.187 
2.230 
2.273 
2.317 
0.8 
2.361 
2.406 
2.450 
2.495 
2.541 
2.586 
2.632 
2.678 
2.724 
2.768 
0.9 
2.818 
2.865 
2.912 
2.960 
3.008 
3.055 
3.104 
3.152 
3.202 
3.251 
1.0 
3.300 
3.350 
3.399 
3.449 
3.501 
3.551 
3.600 
3.653 
3.703 
3.755 
1.1 
3.808 
3.858 
3.911 
3.963 
4.016 
4.069 
4.122 
4.178 
4.231 
4.283 
1.2 
4.340 
4.392 
4.448 
4.501 
4.557 
4.613 
4.666 
4.722 
4.778 
4.834 
1.3 
4.891 
4.947 
5.006 
5.062 
5.118 
5.178 
5.2.34 
5.293 
5.349 
5.409 
1.4 
5.468 
5.524 
5.584 
5.643 
5.702 
5.762 
5.82i 
5.881 
5.940 
6.003 
1.5 
6.062 
6.125 
6.184 
6.247 
6.306 
6.369 
6.428 
6.491 
6.554 
6.617 
1.6 
6.679 
6.742 
6.805 
6.867 
6.930 
6.993 
7.059 
7.121 
7.187 
7.250 
1.7 
7.316 
7.379 
7.445 
7.508 
7.573 
7.639 
7.706 
7.772 
7.838 
7.904 
1.8 
7.970 
8.036 
8.102 
8.171 
8.237 
8.303 
8.372 
8.438 
8.507 
8.573 
1.9 
8.643 
8.712 
8.778 
8.847 
8.917 
8.986 
9.055 
9.125 
9.194 
9.263 
2.0 
9.332 
9.405 
9.474 
9.544 
9.616 
9.686 
9.758 
9.827 
9.900 
9.969 
2.1 
10.042 
10.115 
10.187 
10.260 
10.332 
10.405 
10.478 
10.550 
10.623 
10.695 
2.2 
10.768 
10.841 
10.916 
10.989 
11.065 
11.138 
11.213 
11.286 
11.362 
11.435 
2.3 
11.510 
11.586 
11.662 
11.738 
11.814 
11.887 
11.966 
12.042 
12.118 
12.194 
2.4 
12.269 
12.345 
12.425 
12.500 
12.576 
12.656 
12.731 
12.811 
12.890 
12.966 
2.5 
12.935 
13.124 
13.200 
13.279 
13.358 
13.438 
13.517 
13.596 
13.675 
13.754 
2.6 
13.834 
13.916 
13.995 
14.075 
14.154 
14.236 
14.315 
14.398 
14.477 
14.560 
2.7 
14.642 
14.721 
14.804 
14.886 
14.969 
15.048 
15.131 
15.213 
15.296 
15.378 
2.8 
15.461 
15.543 
15.629 
15.711 
15.794 
15.876 
15.962 
16.045 
16.130 
16.213 
2.9 
16.299 
16.381 
16.467 
16.550 
16.635 
16.721 
16.807 
16.889 
16.975 
17.061 
3.0 
17.147 
17.233 
17.318 
17.404 
17.490 
17.579 
17.665 
17.751 
17.837 
17.926 
3.1 
18.011 
18.101 
18.186 
18.275 
18.361 
18.450 
18.536 
18.625 
18.714 
18.803 
3.2 
18.889 
18.978 
19.067 
19.157 
19.246 
19.335 
19.424 
19.513 
19.602 
19.694 
3.3 
19.784 
19.873 
19.962 
20.054 
20.143 
20.236 
20.325 
20.414 
20.506 
20.599 
3.4 
20.688 
20.780 
20.873 
20.962 
21.054 
21.146 
21.239 
21.331 
21.424 
21.516 
3.5 
21.608 
21.701 
21.793 
21.886 
21.978 
22.074 
22.166 
22.259 
22.354 
22.447 
3.6 
22.542 
22.635 
22.730 
22.823 
22.919 
23.011 
23.107 
23.202 
23.295 
23.390 
3.7 
23.486 
23.582 
23.678 
23.773 
23.869 
23.965 
24.060 
24.156 
24.252 
24.347 
3.8 
24.446 
24.542 
24.638 
24.734 
24.833 
24.928 
25.027 
25.123 
25.222 
25.318 
3.9 
25.417 
25.516 
25.611 
25.710 
25.809 
25.905 
26.004 
26.103 
26.202 
26.301 
4.0 
26.400 
26.499 
26.598 
26.697 
26.796 
26.895 
26.997 
27.096 
27.195 
27.298 TABLES 
475 
Table 10.1-Coefficients of Pipe Friction 
(From Slocum's "Elements of Hydraulics") 
Value of the friction coefficient f, in the formula 
- * J O 
d 2g 
Computed from the exponential formulas of Thrupp, Tutton and Unwin 
Material 
Diameter 
in inches 
Velocity oi now in feet per second 
2 
4 
6 
8 
10 
1 
0.032 
0.026 
0.024 
0.022 
0.021 
Lead pipe 
2 
0.030 
0.025 
0.023 
0.021 
0.020 
3 
0.029 
0.024 
0.022 
0.020 
0.019 
4 
0.028 
0.023 
0.021 
0.020 
0.019 
6 
0.034 
0.033 
0.032 
0.032 
12 
0.027 
0.027 
0.026 
0.026 
Wood pipe 
18 
0.024 
0.024 
0.023 
0.023 
24 
0.022 
0.022 
0.021 
0.021 
36 
0.020 
0.019 
0.019 
0.019 
48 
0.018 
0.018 
0.017 
0.017 
6 
0.026 
0.023 
0.022 
0.021 
0.020 
9 
0.025 
0.022 
0.021 
0.020 
0.019 
12 
0.024 
0.021 
0.020 
0.019 
0.019 
Asphalted pipe 
18 
0.023 
0.020 
0.019 
0.018 
0.018 
24 
0.022 
0.020 
0.018 
0.017 
0.017 
36 
0.021 
0.019 
0.017 
0.017 
0.016 
48 
0.020 
0.018 
0.017 
0.016 
0.015 
3 
0.024 
0.021 
0.019 
0.018 
0.017 
6 
0.022 
0.019 
0.017 
0.016 
0.016 
12 
0.019 
0.017 
0.015 
0.014 
0.014 
Bare wrought-iron 
24 
0.017 
0.015 
0.014 
0.013 
0.012 
pipe 
36 
0.016 
0.014 
0.013 
0.012 
0.011 
48 
0.015 
0.013 
0.012 
0.011 
0.011 
60 
0.015 
0.013 
0.012 
0.011 
0.010 
12 
0.025 
0.022 
0.021 
0.020 
0.019 
24 
0.020 
0.018 
0.017 
0.016 
0.016 
Riveted wrought-iron 
36 
0.017 
0.016 
0.015 
0.014 
0.014 
or steel pipe 
48 
0.016 
0.014 
0.014 
0.013 
0.013 
60 
0.015 
0.013 
0.013 
0.012 
0.012 
72 
0.014 
0.013 
0.012 
0.011 
0.011 
3 
0.028 
0.026 
0.025 
0.025 
6 
0.024 
0.022 
0.022 
0.021 
9 
0.021 
0.020 
0.020 
0.019 
New cast-iron pipe 
12 
0.020 
0.019 
0.018 
0.018 
18 
0.018 
0.017 
0.017 
0.016 
24 
0.017 
0.016 
0.016 
0.015 
36 
0.015 
0.015 
0.014 
0.014 
3 
0.059 
0.058 
0.058 
0.058 
6 
0.050 
0.050 
0.050 
0.049 
9 
0.046 
0.045 
0.045 
0.044 
Old cast-iron pipe 
12 
0.043 
0.042 
0.042 
0.042 
18 
0.039 
0.039 
0.038 
0.038 
24 
0.037 
0.036 
0.036 
0.036 
36 
0.033 
0.033 
0.033 
0.032 
1 Compiled from data in Gibson's "Hydraulics." 476 
WATER PURIFICATION 
Table 11.-Kutter's Values of Chezy's Coefficient 
(From Slocum's "Elements of Hydraulics") 
Values of the coefficient C in Chczy's formula v = Cx^rs according to Kutter's formula 
(Eq. (94) Par. 140): 
41.65 +0,00281 + L811 
s n 
C = 
i , 0.0028H __w_ 
■ 
Slope, 
s 
Coefficient of 
roughness, n 
Hydraulic radius r, 
in feet 
0.1 
0.2|0.4 
0.610.8 
1 
1.5 
2 
3 
4 
1 6 
8 
10 
1 15 
20 
0.009 
65 
87 
111 
127 
138148 
166 
179 
197 
209 
226 
238 
246 
262 
271 
0.010 
57 
75 
97 
112 
122 
131 
148 
160 
177 
188 
206 
216 
225 
240 
249 
g 
0.011 
50 
67 
87 
100 
109 
118 
133 
144 
160 
172 
188 
199 
207 
222 
231 
0.012 
44 
59 
78 
90 
99 
106 
121 
131 
147 
158 
174 
1S4 
192 
206 
215 
IO 
§ 
ft 
0.013 
40 
53 
70 
81 
90 
97 
111 
121 
135 
146 
161 
171 
179 
193 
202 
§ 
s 
o 
g 
.132 f 
0.017 
0.020 
28 
23 
38 
31 
51 
42 
60 
49 
66 
55 
72 
60 
83 
69 
91 
77 
103 
88 
113 
96 
126 
108 
135 
117 
112 
124 
155 
136 
164 
144 
6 
»-< 
o 
0.025 
17 
24 
32 
38 
43 
47 
55 
61 
70 
78 
88 
96 
102 
114 
121 
II 
II 
II 
0.030 
14 
19 
26 
31 
35 
38 
45 
50 
59 
65 
74 
82 
87 
98 
106 
0.035 
12 
16 
22 
26 
30 
32 
38 
43 
50 
56 
64 
71 
76 
86 
94 
0.009 
78 
100 
124 
139 
150 
158 
173 
184 
198 
207 
220 
228 
234 
244 
250 
Q 
0.010 
67 
87 
109 
122 
133 
140 
154 
164 
178 
187 
199 
206 
212 
220 
228 
g 
0.011 
59 
77 
97 
109 
119 
126 
139 
148 
161 
170 
182 
189 
195 
205 
211 
0.012 
52 
68 
88 
98 
107 
114 
126 
135 
148 
156 
168 
175 
181 
189 
196 
o 
s 
ft 
0.013 
47 
62 
79 
90 
98 
104 
116 
124 
136 
145 
156 
163 
169 
179 
184 
1 
o 
CM 
Tf< 
0.017 
33 
44 
57 
65 
71 
77 
87 
94 
104 
111 
122 
129 
134 
142 
149 
o 
.s 
0.020 
26 
35 
46 
53 
59 
64 
72 
79 
88 
95 
105 
111 
116 
125 
131 
o 
»-1 
6 
0.025 
20 
26 
35 
41 
46 
49 
57 
62 
71 
77 
85 
91 
96 
104 
110 
II 
II 
II 
0.030 
16 
21 
28 
33 
37 
40 
47 
51 
59 
64 
72 
78 
82 
90 
96 
0.035 
13 
18 
24 
28 
31 
34 
40 
44 
50 
56 
63 
68 
72 
79 
85 
0.009 
90 
112 
136 
149 
158 
166 
178 
187 
198 
206 
215 
221 
226 
233 
237 
r mih 
0.010 
78 
98 
119 
131 
140 
147 
159 
168 
178 
186 
195 
201 
205 
212 
216 
0.011 
68 
86 
106 
118 
126 
132 
144 
151 
162 
169 
178 
184 
188 
195 
200 
o 
ft 
0.012 
60 
76 
95 
105 
114 
120 
130 
138 
149 
155 
164 
170 
174 
181 
185 
$ 
o 
Uh 
0.013 
54 
69 
86 
96 
103 
109 
120 
127 
137 
143 
152 
158 
162 
169 
173 
X 
§ 
c 
IO 
0.017 
37 
48 
62 
70 
76 
81 
89 
96 
104 
111 
119 
124 
128 
135 
139 
6 
6 
0.020 
30 
39 
50 
57 
63 
67 
75 
81 
89 
94 
102 
107 
111 
118 
122 
II 
II 
II 
0.025 
22 
29 
38 
44 
48 
52 
59 
64 
71 
76 
84 
88 
92 
98 
102 
00 
0.030 
17 
23 
31 
35 
39 
42 
48 
53 
59 
64 
71 
75 
78 
85 
89 
0.035 
14 
19 
25 
30 
33 
35 
41 
45 
51 
55 
61 
66 
69 
75 
79 Table 12.-Sand Ejectors and Flow of the Water and Sand in 
Discharge Piping1 
Pounds 
pressure 
feed 
water 
Diameter 
jet, 
inches 
Best 
diameter 
for throat, 
inches 
Per cent, 
sand in 
discharge 
by volume 
Cu. yd. 
sand 
per 
hour 
Pressure 
of 
discharge 
in feetj 
Friction in feet per 
1,000 in discharge piping 
2W 
3" 
4" 
5" 
60 
0.5 
0.87 
20 
5.0 
28 
150 
140 
0.5 
1.01 
25 
7.2 
20 
176 
150 
0.5 
1.21 
30 
10.0 
14 
206 
168 
0.6 
1.04 
20 
7.2 
28 
178 
124 
90 
0.6 
1.21 
25 
10.3 
20 
222 
144 
110 
0.6 
1.46 
30 
14.3 
14 
275 
170 
120 
0.7 
1.06 
15 
6.5 
39 
200 
114 
70 
0.7 
1.21 
20 
9.7 
28 
250 
140 
88 
0.7 
1.41 
25 
14.0 
20 
325 
175 
102 
0.8 
1.21 
15 
8.5 
39 
298 
145 
71 
0.8 
1.39 
20 
12.7 
28 
370 
180 
89 
73 
0.8 
1.61 
25 
18.4 
20 
480 
240 
109 
82 
0.9 
1.20 
10 
6.5 
52 
350 
160 
61 
42 
0.9 
1.37 
15 
10.8 
39 
435 
202 
80 
56 
0.9 
1.56 
20 
16.1 
28 
540 
250 
102 
71 
80 
0.5 
0.87 
20 
5.7 
38 
154 
130 
0.5 
1.01 
25 
8.3 
27 
186 
145 
0.5 
1.21 
30 
11.5 
18 
225 
160 
0.6 
1.04 
20 
8.3 
38 
208 
128 
90 
0.6 
1.21 
25 
11.9 
27 
260 
153 
107 
0.6 
1.46 
30 
16.5 
18 
320 
187 
117 
0.7 
1.06 
15 
7.5 
52 
245 
127 
70 
0.7 
1.21 
20 
11.2 
38 
305 
158 
86 
0.7 
1.41 
25 
16.2 
27 
400 
204 
104 
71 
0.8 
1.21 
15 
9.8 
52 
370 
176 
76 
56 
0.8 
1.39 
20 
14.7 
38 
465 
222 
95 
71 
0.8 
1.61 
25 
21.2 
27 
600 
290 
115 
84 
0.9 
1.20 
10 
7.5 
70 
450 
200 
70 
43 
0.9 
1.36 
15 
12.4 
52 
550 
248 
91 
58 
0.9 
1.56 
20 
18.6 
38 
680 
315 
114 
73 
100 
0.5 
0.87 
20 
6.4 
47 
162 
126 
0.5 
1.01 
25 
9.2 
34 
200 
143 
0.5 
1.21 
30 
12.9 
23 
250 
164 
120 
0.6 
1.04 
20 
9.2 
47 
230 
133 
87 
0.6 
1.21 
25 
13.3 
34 
300 
165 
103 
0.6 
1.46 
30 
18.5 
23 
380 
210 
118 
105 
0.7 
1.06 
15 
8.4 
65 
292 
144 
71 
0.7 
1.21 
20 
12.6 
47 
360 
180 
88 
71 
0.7 
1.41 
25 
18.1 
34 
470 
235 
108 
83 
0.8 
1.21 
15 
11.0 
65 
450 
208 
82 
56 
0.8 
1.39 
20 
16.4 
47 
550 
260 
104 
71 
0.8 
1.61 
25 
23.7 
34 
340 
130 
86 
0.9 
1.20 
10 
8.3 
87 
... 
240 
78 
45 
0.9 
1.36 
15 
13.9 
65 
300 
100 
60 
0.9 
1.56 
20 
20.8 
47 
370 
128 
76 
150 
0.5 
0.87 
20 
7.8 
71 
195 
124 
90 
0.5 
1.01 
25 
11.4 
51 
240 
150 
110 
0.5 
1.21 
30 
15.8 
35 
305 
180 
118 
0.6 
1.04 
20 
11.3 
71 
310 
160 
86 
75 
0.6 
1.21 
25 
16.3 
51 
400 
205 
104 
83 
0.6 
1.46 
30 
22.7 
35 
520 
260 
125 
100 
0.7 
1.06 
15 
10.3 
98 
410 
190 
78 
56 
0.7 
1.21 
20 
15.4 
71 
500 
236 
98 
71 
0.7 
1.41 
25 
22.2 
51 
640 
310 
120 
85 
0.8 
1.21 
15 
13.5 
98 
285 
99 
59 
0.8 
1.39 
20 
20.0 
71 
350 
123 
75 
0.8 
1.61 
25 
29.0 
51 
450 
160 
93 
0.9 
1.20 
10 
10.2 
131 
335 
100 
50 
0.9 
1.36 
15 
17.0 
98 
410 
129 
67 
0.9 
1.56 
20 
25.5 
71 
510 
163 
85 
1 Taken from American Civil Engineers' Pocket Book by permission of author, Mr. 
Allen Hazen and of publishers, John Wiley & Sons, Inc. 
477 INDEX 
A 
Acid mine waters, 17 
Air agitation, systems of, 212 
compressors, 298 
tanks, 298 
Algicides, 391 
Alkalinity, theoretical reduction of, 
by chemical coagulants, 
340, 341 
Aqueducts of ancient Rome, 3 
Automatic control of application of 
chemical solutions, by Ven- 
turi tubes, orifices and 
weirs, 282-286 
B 
Bacteria in water, 14 
Basins, coagulation, 66 
Cincinnati, O., 69, 70 
Grand Rapids, Mich., 71, 72 
settling, St. Louis, Mo., 68 
Blaisdell sand washing'' machine, 
143 
Bleaching powder, solution of, 268 
C 
Calcium and magnesium salts, re- 
moval of, 411 
carbonate, solubility of, in water, 
341-434 
Channels, mixing, 66 
velocity in, 66 
Chemical coagulants, loss of, by 
adsorption, 93 
solutions, application of meas- 
ured volumes of standard 
strength, 280 
preparation of, 286-289 
Chemicals, buildings for storage of, 
259 
Chemicals, crushing and conveying 
systems at St. Louis, Mo., 
259 
difficulties encountered in stor- 
ing and in operating con- 
veying equipment at St. 
Louis, Mo., 261 
mixing and storage tanks for 
solutions of, 266 
tanks for, 267 
piping for solutions of, 266, 
267 
unloading, elevating and con- 
veying in bulk, 258 
used in water purification, 256 
Chlorine, liquid, apparatus for apply- 
ing, 275-380 
and its compounds, application 
of, 373 
chemistry of the action of, in 
water, 371, 372 
for disinfection of water sup- 
plies, 368-370 
tastes and odors in water pro- 
duced by, 396, 397 
Cholera organisms, isolation of, 24 
Cincinnati settling reservoirs, re- 
moval of sediment by, 81 
water purification plant, re- 
moval of sediment and bac- 
teria, by, 86 
Classification of natural waters, 8 
Cleaning settling and coagulation 
basins, 72-74 
cost of, 75, 76 
Coagulated sediment, amount of, in 
water applied to filters, 
332 
Coagulation, basins, practical effi- 
ciencies of, 80 
chemical and physical changes 
produced by, 331 
reactions involved in, 335-338 
479 480 
INDEX 
Coagulation, effect of temperature 
on, 94, 95, 333, 334 
effects of chemicals used to pro- 
duce, on dissolved constitu- 
ents of water, 338-340 
time required for, 56 
with chemicals, 55 
Coefficients of pipe friction, table of, 
475 
Colloids, adsorption of chemicals on, 
57 
artificial, 58 
importance of, in water purifica- 
tion, 58 
natural, 57 
Columbus water purification plant, 
removal of sediment and 
bacteria by, 87 
Compressed air for agitating sand 
beds, 324, 325 
Contamination of water, dissolved 
impurities producing, 11 
from manufacturing wastes, 
16 
from sewage, 15 
natural impurities producing, 
10 
suspended impurities produc- 
ing, 12 
Control of water purification proc- 
esses, 452-459 
D 
Diarrhea, death rate in Cincinnati 
before and after purifica- 
tion of water supply, 40 
Discharge equivalents, table of, 471 
Disease producing organisms, vital- 
ity of, 22 
vitality of Bacillus typhosus, 
23 
vitality of Spirillum cholera;, 
23 
transmission of, through drink- 
ing water, 22 
Disinfectants, point of application 
and amounts used, 398 
selective action and effects of, 
396 
Disinfection, bacterial reduction pro- 
duced by, 399 
cost of apparatus for, 40G, 407 
of water supplies, chemicals 
used in, 368-371 
historical, 367, 368 
plants, cost of operation of, 408 
practical results obtained by 
chlorine, ozone and ultra- 
violet light, 402-405 
relative effects on various spe- 
cies of bacteria, 400 
Dissolved mineral matter, removal 
1 of, 410-451 
Dry feeding of chemicals, apparatus 
for, 274-276 
modification of method of, at 
St. Louis, Mo., and Cin- 
cinnati, O., 277-280 
Dysentery, death rate in Cincinnati 
before and after purifica- 
tion of water supply, 40 
E 
Earl rate controller, 231 
master rate controller, 236 
Eductors for chemical solutions, 272 
Efflux coefficients for circular ori- 
fices, table of, 472 
square orifices, table of, 473 
Enteritis, death rate from, in various 
cities, 39 
Equivalents of measure, table for 
the conversion of, 467 
F 
Filter, function of, 90 
structure of, 90 
Filtration, explanation of some well- 
known phenomena, 92 
properties of colloidal form of 
matter with reference to, 
91 
theories of, 89 
Fish, effect of copper sulphate on 
certain, 394 
Flow of water through rapid sand 
filters, 461-464 INDEX 
481 
G 
Gas for laboratories, 303 
Gravel, depth of, in filter beds of 
Cincinnati plant, 328 
Grit reservoirs, removal of sediment 
by New Orleans, 82 
H 
Hackensack filter plant, rate con- 
trollers for, 225 
Handling and storing of chemicals, 
equipment for, 256 
Hardness of river waters before and 
after coagulation and filtra- 
tion, 356 
table for the conversion of equi- 
valent measures of, 467 
Hazen's theorem, 40 
Head and pressure equivalents for 
water, table of, 470 
positive and negative, 317 
total, in rapid sand filters, 
317 
Hygienic effects of sterilization of 
water, 406 
I 
Infiltration galleries, ancient, 3 
Introduction, 1 
Iron and manganese, removal of, 
from water, 437 
removal of, in presence of car- 
bon dioxide, 438-440 
in presence of manganese, 
441 
in presence of organic matter, 
440 
removal purification plant at 
Middleboro, Mass., 442- 
447 
K 
Kutter's values of Chezy's coeffi- 
cient, table of, 476 
L 
Lime, handling of, at Columbus 
water purification plant, 
262-265 
slaking, 268 
Loss of head, 317, 318 
average yield of filtered water 
per foot, for different tur- 
bidities in applied water, 
352 
effect on quality of water by, 
132, 133 
gages, 246 
M 
Magnesium carbonate, solubility of, 
in water, 341-343 
hydroxide, solubility of, in 
water, 341-343 
Manganese in ground waters, 447- 
449 
methods for the removal of, 
449-451 
Measuring apparatus, 238 
Mechanical rakes for rapid sand 
filters, 211 
Methods of purification, natural, 16 
by filtration, 18 
by means of minute plant and 
animal organisms, 18, 19 
by precipitation, 17 
by sedimentation, 16 
effects of sunlight on, 16 
Microorganisms, cause of the des- 
truction of, by disinfectants, 
395 
killing of, by copper sulphate, 
394 
that produce odors and tastes 
in waters, 392, 393 
Microscopic plant and animal life 
in water, 14 
Mills-Reincke phenomenon, 40 
Mixing of sand and gravel in wash- 
ing rapid sand filters, 327 
Modern purification plants, develop- 
ment of, 4 482 
INDEX 
Motors, for centrifugal pumps, 297 
N 
Nichols sand washing machine, 143 
Notch weirs, rectangular, table of 
discharge per foot of length 
over, 474 
O 
Operating tables for rapid sand 
filters, 252 
Orifice feed tanks for chemical solu- 
tions, 280 
Oxygen dissolved in water, 345, 346 
Ozone, application of, 385 
formation of, 380 
quantity used for disinfection, 
386 
Ozonizers, modern, 381-383 
yield of, 383 
P 
Pittsburgh A frame baffles, 99-102 
settling basins, removal of sedi- 
ment by, 81 
Potomac River water, effect of 
coagulants on turbidity of, 
98 
Power plant for filtration plants, 
290-292 
Preliminary treatment of water for 
slow sand filters, results 
from, and costs of, 112-115 
Proportional feed pumps for chem- 
ical solutions, 281 
Puech-Chabal system of filtration, 
110, 111 
Pumping machinery, 292 
Pumps, centrifugal, 293 
for chemical solutions, 272 
pressure, 297 
wash water and flushing, 296 
Purification plants, care of, 452 
laboratory for, 454 
records of operation for, 453, 
454 
Purification plants, supplies for, 453 
the human factor in the opera- 
tion of, 459 
R 
Rapid sand filter plants, air piping 
and wash water piping, sys- 
tems of, 191 
arrangement of filters in, 177- 
180 
clear water reservoirs for, 181 
effluent piping and conduits 
for, 187-189 
general arrangement of, 173- 
177 
influent piping and conduit 
for, 185-187 
number and capacity of, in 
United States, 48 
number, size and kind of filter 
tanks, 183, 184 
pipe gallery of, 192 
waste water piping for, 189- 
191 
Rapid sand filters, agitation and 
washing of, 170, 321 
approximate cost per million 
gallons of filtering capacity, 
305 
bacterial purification by, 357, 
358 
cost of construction of, 305-311 
cost of operation of, 347, 361- 
365 
cost in detail of Columbus and 
Cincinnati water purifica- 
tion plants, 306, 307 
detailed cost of operation at 
Columbus, O., New Orleans, 
La., and Louisville, Ky., 
362-364 
development of, 159-164 
development of the methods for 
washing, 322 
early types of, 164 
effect of microscopic plant and 
animal life on periods of 
service, 350, 351 INDEX 
483 
Rapid sand filters, efficiency of oper- 
ation, 347 
efficiency in the removal of 
B. coli, 358-361 
essentials for efficient washing 
of, 321 
estimated costs of filter plants 
at St. Louis, Mo., Trenton, 
N. J. and Baltimore, Md., 
309-311 
fixed charges on, 365 
gravel layers in, 201 
gravity type, 167-172 
high velocity method of wash- 
ing, 325-327 
influence of variable and con- 
stant rates of filtration on 
periods of service, 348-350 
manifolds and types of strainer 
nozzles, 170-172 
maximum effective rate of fil- 
tration for different turbid- 
ities of applied water, 314, 
315 
preliminary, 103-105 
pressure type, 164-167 
rates of filtration through, 312, 
313 
types of filter bottoms, 202-205 
typical strainer and underdrain 
systems in various large 
plants, 206-210 
uniformity of rate of filtration, 
313-315 
velocity of wash water in feet 
per minute, 323 
waste water conduits and lateral 
gutters for, 184, 185 
Rate control, a simple method of, 221 
automatic, 222 
of filtration, approximate equiv- 
alents of various measures 
of, table of, 468 
of flow controlling apparatus, 
220, 315 
gages, 244 
Refrigerating plant, 303 
Reservoirs, grit, New Orleans, La., 
64, 65 
Reservoirs, impounding, 60 
settling, Albany, N. Y., 64 
first cost of, 78 
Louisville, Ky., 64 
storage and settling, Cincinnati, 
O., 61-63 
types of settling, 60 
Roughing filters, effectiveness of, 94 
Pittsburgh contact, 99 
S 
Sand and gravel, depth of, in rapid 
sand filters, 214-216 
ejectors and flow of the water 
and sand in discharge pip- 
ing, table of, 477 
mechanical analysis of, 125, 126 
methods and apparatus for 
cleaning, 136-146 
old methods of washing, 138 
present methods of transporta- 
tion and washing, 139-142 
surface washing of, 137 
washing machines for, 142-144 
Screens between gravel and sand 
layers, 216-218 
Sediment, coagulation of, 54, 55 
colloidal character of, 54 
physical and electrical proper- 
ties of colloidal suspen- 
sions, 54 
settlement after coagulation, 67 
Sedimentation, plain, 50 
theory of, 51 
velocities of falling particles in 
still water, 52 
Settling, cost of clearing water by, 77 
reservoirs, practical efficiencies 
of, 80 
Simplex rate controller, 230 
water meter, 243 
Slow sand filters, arrangement of 
sand courts and controll- 
ing chambers in, 128, 129 
cost of construction of, at 
Albany, N. Y., Washington, 
D. C. and Pittsburgh, Pa., 
147, 148 484 
INDEX 
Slow sand filters, cost of operation 
of, at Washington, D. C., 
Wilmington, Del. and 
Philadelphia, Pa., 155- 
158 
depth of water on, 129, 130 
efficiency of, 150 
effluent controlling devices 
for, 135, 136 
general form and construc- 
tion of, 117-120 
gravel layers, size and depth 
of, 121-123 
influent controlling devices 
for, 134 
loss of head in, 131 
methods for partial cleaning 
of, by raking, spading, 
scraping and piling, 144, 
145 
preliminary treatment of 
water for, 97 
removal of sediment and bac- 
teria by, 151-155 
resanding of, 145, 146 
sand in, size and depth of, 
123-127 
service periods and yields of, 
150, 151 
underdrain systems of, 121- 
123 
filtration, systems of, 116 
Sodium hypochlorite, apparatus to 
produce, 373, 374 
Softening of waters, 410, 411 
application of chemical reagents 
in, 417 
chemical reagents used in, 414- 
416 
chemistry of, 412 
mixing chemicals with hard 
waters, 418 
settling of precipitates formed, 
418 
Statistics of disinfection of wTater 
supplies in United States, 
48 
of purification plants, 45 
population supplied, 46 
Statistics of purfication plants, 
summary of, 47 
Steam boiler plants, 301 
St. Louis coagulation basins, relative 
efficiencies of, 83 
purification plant, diagrams 
showing characteristics of 
raw and treated water, 84, 
85 
Strainer systems of rapid sand filters, 
nozzles, perforated plates 
and pipes, 200-202 
Surface waters, flowing, 9 
impounded, 9 
Susquehanna River water, removal 
of bacteria from, 88 
T 
Torresdale preliminary filters, 105- 
110 
Typhoid fever, characteristics of 
outbreaks of, 29, 30 
death rates from, 28, 29 
death rates in cities changing 
from polluted to purified 
water supplies, 36-38 
death rates in cities using vari- 
ous classes of waters, 34r-35 
epidemics in various cities, 
31-32 
organism, isolation of, 24 
U 
Ultra-violet light, apparatus for pro- 
duction of, 388 
conditions necessary for effi- 
cient application of ,389-390 
effect of, on various species of 
bacteria, 401 
historical, 387 
Underdrain systems of rapid sand 
filters, 202-205 
Underground waters, 10 
V 
Vacuum, partial, in filter beds, 319- 
321 INDEX 
485 
Valves, 193 
electrically operated, 195-198 
hydraulically operated, 194 
Vegetable coloring matter, influence 
of, on reduction of alka- 
linity when chemical coagu- 
lants are used, 344 
in water, 343 
Velocity of flow of water through 
sand, strainers and under- 
drains, 218, 219 
Venturi meters, 239 
Vivian rate controller, 227 
W 
Washing rapid sand filters, flotation 
of sand in, 328, 329 
loss of head in, 328, 329 
Wash-water, percentage used at New 
Orleans, La., Cincinnati, 
O., Harrisburg, Pa. and 
Columbus, O., 349-353 
pumps, 295 
relation of turbidity of applied 
water to quantity of, 354, 
355 
tanks, 300 
troughs, approximate formula 
for calculating the dis- 
charging capacity of rapid 
sand filter, 465-467 
volume of, required, 351-353 
Water borne diseases, 25 
epidemics of, 26 
from cholera, 26 
from typhoid fever, 27 
Water level gages, 247 
Water purification, definition of, 1 
general methods of, 43-45 
history of, 2 
interpretation of data obtained 
in, 455-458 
objects and methods of, 43 
Water sampling and inspecting de- 
vices, 250 
softening, bacterial effects of, 
432, 433 
data on operation of munici- 
pal plants, 431 
plants, cost of and cost of 
operation of, 434, 435 
practical results obtained in, 
429-431 
results obtained in using 
zeolites, 431, 432 
softeners, continuous type, 421- 
425 
intermittent type, 419, 420 
supply, ancient systems of, 2 
Waters, classification of, 12 
Weights and measures, table of, 469 
table of equivalent of various, 
469 
Wells, ancient, 2 
Weston rate controller, 224 
Z 
Zeolite filters, 425 
Zeolites, reactions of natural and 
artificial, on hard waters, 
413