COLLOIDS 
AND THE 
ULTRAMICROSCOPE 
A MANUAL OF 
Colloid Chemistry and Ultramicroscopy 
BY 
Dr RICHARD ZSIGMONDY 
Professor of Inorganic Chemistry i!? the University of Gottingen 
A UTHOR1ZED TRAN SLA TION 
BY 
JEROME ALEXANDER, M.Sc. 
FIRST 
FIRST THOUSAND 
NEW YORK 
JOHN WILEY & SONS 
London: CHAPMAN & HALL, Limited 
1909 Copyright, 1909 
BY 
JEROME ALEXANDER 
Ehr ©rirntific 
Kubrrt 0rnmtnnnb anb (Company 
New lurk PREFACE TO THE ENGLISH EDITION 
The expectation that this book would arouse the 
interest of a wider circle has been agreeably fulfilled. I 
am glad to be able to state that Mr. Jerome Alexander, 
of New York, has undertaken to translate the book 
because of interest in its contents, and has carried 
out the work carefully and conscientiously. I have 
myself read over the whole translation and found it in 
excellent accord with the original text. 
In so far as concerns the experimental basis for answer- 
ing the fundamental questions of Colloid Chemistry, 
there is but little to be added to the German edition. 
In but few places has it become necessary to amplify 
the statements and extend the text because of later 
publications. 
May the English edition receive the same friendly 
reception as the German edition. 
R. Zsigmondy. 
Gottingen, November 14, 1908. 
III TRANSLATOR'S PREFACE 
After discussing the basic principles governing the 
colloidal condition, and the classification of colloids, and 
reviewing the most important work already done in 
this field, the author describes the development (upon 
the principle originally conceived by himself) of the 
ultramicroscope, which carries our range of vision well 
towards molecular dimensions; and he furthermore 
gives a detailed account of his own valuable pioneer 
researches with the new instrument. 
Giving as it does an actual insight into a sphere here- 
tofore beyond the range of direct observation, the ultra- 
microscope has proven to be a powerful weapon with 
which to attack numerous problems confronting the 
chemist, the physicist, and the biologist; and it will be 
of special value in deciding many mooted questions in 
theoretical and in technical colloid chemistry. 
As the far-reaching ramifications of colloid chemistry 
are better understood, its importance and the applicability 
of its principles to a great variety of industrial problems, 
become more and more evident. There might, for ex- 
ample, be mentioned agriculture, tanning, dyeing; rubber, 
cement, ceramics; soaps, photography, sugar-in fact, 
almost every industry is directly or indirectly involved. 
Professor Zsigmondy's work will, therefore, be of vital 
interest not only to scientists concerned with theoretical 
V VI 
TRANSLATOR'S PREFACE 
questions, but also to chemists, engineers, and others 
controlling technical processes. To physiological chemists 
and physicians it is indispensible. 
I must express my sincere thanks to Prof. Zsigmondy 
and also to Alexander D. Ross, M.A., B.Sc. (lecturer 
on Natural Philosophy in the University of Glasgow), 
both of whom read over the original manuscript of the 
translation and made valuable suggestions. In the 
English edition besides some additions to the text, there 
are included two beautiful colored plates originally pub- 
lished in Professor Zsigmondy's paper on "Colloid 
Chemistry." 
Jerome Alexander. 
New York, February 24, 1909. PREFACE TO THE GERMAN EDITION 
Some of the conclusions and observations herein set 
forth were originally intended for the Zeitschrift fiir 
physikalische Chemie; however, as the method of render- 
ing visible ultramicroscopic particles which were developed 
by Siedentopf and myself, has awakened the interest 
of a larger circle, I have decided to make readily accessible 
to the representatives of other branches of science the 
results of my ultramicroscopic investigation of fluids, 
and the experiments associated therewith; all the more 
so because they may be of some use to other workers in 
the same field. Another reason that led me to take 
this step was the fact that the problem I attacked is of 
interest not only to the physicist and the chemist, but 
to scientists in general. 
It will be shown by a number of examples how the 
properties of a solid-metallic gold-change with its pro- 
gressive subdivision, especially if the subdivision be 
carried as far as possible, to a degree approaching molec- 
ular dimensions; and as near as may be the size and 
properties of the individual particles in each case are 
accurately given. 
As will be seen in the following chapters, the problem 
is intimately connected with the question of the nature 
of colloidal solutions, the latter in fact being the starting 
point of this investigation. 
VII VIII 
PREFACE TO THE GERMAN EDITION 
The paucity of our knowledge on this subject is shown 
by the discussion of the question whether colloidal solu- 
tions are homogeneous or heterogeneous, and whether 
there is or is not progressive transformation from colloidal 
to crystalloiclal solution. As a preliminary discussion 
shows that the words "solution" and "suspension" 
are by no means synonymous, it is necessary, in order to 
pave the way, to consider what these expressions really 
mean. 
The title of this book, "Zur Erkenntnis der Kolloide," 
might lead to the belief that I intended to go into the 
structure of hydrogels or jellies. Chapter XXI contains 
some remarks on this subject. When working in a wide 
field, however, it is best to proceed from the simple to 
the complex. Solutions of colloids are of a simpler 
nature than jellies; they contain to a certain extent the 
structural units of the latter, and wherever before coagu- 
lation begins, the appearance of ultramicroscopic parti- 
cles is observed, it may be taken for granted that these 
are constituents of the hydrogel. Finally, when we have 
acquired more intimate knowledge of the actual constitu- 
tion of such colloidal solutions, we can then successfully 
theorize regarding the ultimate structure of jellies which 
does not admit of direct observation. 
But the constituent particles of hydrosols, which I 
have just compared to structural units, show such varia- 
tion in size, constitution, and properties, that it is neces- 
sary at the outset to confine ourselves to one especially 
simple case. As already intimated, the colloidal solutions 
of metallic gold have been chosen. The minute particles 
of gold composing them, their formation, their relation to 
each other, and to the particles of other colloids, are 
all closely considered. Reference to earlier work will 
show that they are not indifferent to the action of other PREFACE TO THE GERMAN EDITION 
IX 
colloids, and that when they are mixed with other colloids, 
products are often produced which have been in the 
past frequently mistaken for chemical compounds. 
In addition I have given, in more or less detail, an 
account of my ultramicroscopic investigations on various 
hydrosols, and also of the work already published by 
other investigators. 
On seeing the variety of these elementary relations, it 
will at once be manifest that in the study of the colloids 
we have outlined an enormous scientific structure, whose 
foundation is as yet scarcely begun. 
R. Zsigmondy. 
Jena, May, 1905. CONTENTS 
INTRODUCTION 
PAGE 
Introduction 1 
CHAPTER I 
Limitation of the Field 11 
CHAPTER II 
Classification of Hydrosols according to Two Different 
Points of View 19 
CHAPTER III 
History of the Irreversible Colloids 30 
CHAPTER IV 
Facts Pointing to the Homogeneity of Gold Hydrosols. 
Development of Ultramicroscopy 95 
CHAPTER V 
Description of the Apparatus for Making Visible Ultra- 
microscopic Particles 103 
CHAPTER VI 
Certain Terms often Used Herein 109 
XI XII 
CONTENTS 
CHAPTER VII 
PAGE 
Principles of the Ultramicroscopic Investigation of 
Fluids Ill 
CHAPTER VIII 
Preparation of Colloidal Gold Solutions 124 
CHAPTER IX 
Ultramicroscopic Examination of the Solutions of Gold . . 129 
CHAPTER X 
Motion of the Gold Particles 134 
CHAPTER XI 
Size and Color of the Particles 141 
CHAPTER XII 
The Color Change of Colloidal Gold 144 
CHAPTER XIII 
The Precipitation and Protection of Colloidal Gold 147 
CHAPTER XIV 
Filtration Experiments 153 
CHAPTER XV 
The Size of the Gold Particles Compared with the Size 
of Other Bodies 157 CONTENTS 
XIII 
CHAPTER XVI 
PAGE 
Superior and Inferior Limits of the Size of the Par- 
ticles 160 
CHAPTER XVII 
Amicroscopic Nuclei in Colorless Ruby Cilass 165 
CHAPTER XVIII 
General Remarks Concerning Metal Hydrosols 175 
CHAPTER XIX 
Ultramicroscopic Examination of Certain Solutions and 
Suspensions , 188 
CHAPTER XX 
Ultramicroscopic Investigations from the Publications of 
Other Scientists 198 
CHAPTER XXI 
On the Formation of Hydrosol and Hydrogel 212 
Recapitulation 236 COLLOIDS AND THE ULTRAMICROSCOPE 
INTRODUCTION 
Regarding Solution and Homogeneity. The original 
conception of the word "solution," and the one still extant, 
does not completely coincide with that developed from 
the generalization of the laws of gases. 
As considerable misunderstanding may arise by using 
the term "solution" to express different ideas, a few 
words on this subject may not be amiss. 
Solutions are more easily described by limitation than 
by definition. Originally they included solutions not only 
of crystalloids, but also many of the earlier known colloids. 
Graham included for instance solutions of albumen, glue, 
rubber, ferric oxid, Prussian blue, etc., and because of its 
uniform appearance Berzelius considered the purple of 
Cassius to be a solution. Ordinarily we speak of dilute 
aqueous solutions of zinc chlorid and ferric chlorid. 
although these solutions are partially hydrolyzed and lack 
homogeneity. 
An example of an old definition of "solution" is found 
in Felhing's " Handworterbuch der Chemie": 
"Solution is the equal distribution of a body in a liquid, 
the resulting mass being in all parts homogeneous and 
fluid enough to form drops. According as the dissolved 
body is solid, fluid, or gaseous, the product is termed a 2 
INTRODUCTION 
solution (in the restricted sense of the word), a mixture, 
or an absorption." 
Ostwald and Nernst also define solution as a homo- 
geneous mingling,1 a physical mixture,2 a homogeneous 
phase.3 Consequently homogeneity is even now regarded 
as the distinguishing characteristic of solutions. 
Because of the brilliant results following the general 
application by van't Hoff of the laws of gases, physical 
chemists came to regard these laws as typical attributes 
of solutions, although as enunciated they apply only to 
solutions in the more limited sense, viz., those of a crys- 
talloidal nature. As characteristic properties of crystal- 
loid solutions, may be mentioned: diffusibility, osmotic 
pressure, the capability of the solute (in the case of solids) 
to crystallize from supersaturated solutions. 
Regarding the above as the most important character- 
istics of solutions,4 these scientists must have naturally 
considered the question whether those mixtures which 
completely or partially lack these characteristics (colloidal 
solutions or hydrosols) could be classed as solutions at all.5 
1 Ostwald, Grundriss der allg. Chern., 2d ed., p. 119. 
2 Nernst considers solutions to be physical mixtures, and defines 
them as complexes of different substances, in every part physically 
and chemically homogeneous. Theoret. Chem., 3d ed., p. 99. 
3 Ostwald, Grundriss, 3d ed., p. 313. 
4 Some scientists apparently failed to observe the fact that in addi- 
tion to the conception of solution as used by them, there exists another 
with a much wider meaning which is in general use. I do not there- 
fore consider it necessary to put the word "solution" in quotation 
marks when speaking of colloidal solutions. 
5 It might here be mentioned that research on this point has lead 
to the conclusion: typical colloidal solutions are not solutions. It 
will, of course, be seen that this only means that typical colloidal 
solutions are not crysialloidal solutions, which statement hardly 
expresses any new idea of importance. The physico-chemical inves- 
tigation of colloids however, has lead to most important results, and 
very largely extended our knowledge of them. INTRODUCTION 
3 
The definitions, however, are not based upon the proper- 
ties above referred to, but give homogeneity as the most 
important characteristic. Because of this state of affairs 
the moot question as to the nature of colloidal solutions 
hinges on the question of their homogeneity or heterogeneity. 
This very circumstance has lead to an important exten- 
sion of our knowledge. We must therefore- next consider 
the concept "homogeneity." 
In his book: 11 Heterogeneous Equilibrium," Bakhuis 
Roozeboom 1 states: 
"We call a system homogeneous if all its mechanically 
separable parts show the same chemical constitution and 
the same chemical and physical properties. 
"Gases or liquids which have been well mixed possess 
this homogeneity of constitution because of the smallness 
of molecules and the grossness of our means of observa- 
tion." 
We still are unable to observe directly particles the size 
of a crystalloid molecule. However, by means of the 
Faraday-Tyndall convergent beam of light, we are able 
in an indirect manner, with the aid of the Nichol prism, 
to demonstrate the existence of very small boundary 
planes, and therefore of particles of unusually small size. 
As Bredig 2 has already pointed out, Tyndall's method 
is the most sensitive for the recognition of the lack of 
optical homogeneity. By this method, as we shall see, 
we can detect smaller quantities of substances than by 
spectrum analysis; but just because of its enormous sensi- 
tiveness, the most extreme care must be taken in forming 
conclusions as to the nature of a solution, upon the basis 
of the results of an examination by it. 
1 B. Roozeboom, Die heterogenen Gleichgewichte, Part I, p. 9, 
Braunschweig, 1901. 
2 Bredig, Anorganische Fermente, Leipzig, 1901, p. 21. 4 
INTRODUCTION 
This I pointed out several yearsago.1 Recently a similar 
view was advanced by Konowalow,2 who, because of its 
enormous sensitiveness, denies the value of the Tyndall 
method, and regards solutions in the critical state, and 
colloidal solutions as well, to be homogeneous solutions, 
which owe their inhomogeneity to a partial separation of 
the substance otherwise homogeneously dissolved. 
Consequently Konowalow took the stand that there 
are homogeneous solutions in which, as compared with 
crystalloid solutions, the work necessary to separate sol- 
vent and solute is excessively small.3 
This standpoint coincides to some extent with the one 
I have formerly taken.4 Because of observations, to which 
we will later on return, I regarded optical inhomogeneity 
as an incidental characteristic of colloidal solutions, 
attributing their opalescence (Lichtzerstreuung) to the 
presence of a smaller number of larger particles, while I 
regarded the bulk of the colloid as being in homogeneous 
solution. 
From the following remarks it will be evident that the 
importance ascribed to the optical inhomogeneity of 
colloidal solutions by Faraday, Picton and Linder, Spring, 
Breclig, and others, is well founded. Furthermore, we 
will find confirmation for the compromise view of Lobry 
de Bruyn, that a particle the size of a starch molecule 
1 Zeitschr. f. physik. Chem., Vol. XXXIII, p. 64. 
2 Konowalow, Crude's Ann., Vol. X, p. 360-392, 1903, and Vol. 
XII, p. 1160-64, 1903. 
3 The cause of opalescence in liquids, according to Konowalow, is 
the "very fine dust," which is found everywhere. The reason it is so 
strongly marked in the critical state and in colloidal solutions, is due 
to the fact that in these cases the vapor pressure, as well as the osmotic 
pressure, is practically independent of the constitution. 
4 Zsigmondy, Liebigs Ann., Vol. CCCI, p. 53; Zeitschr. f. physik. 
Chem., Vol. XXXIII, p. 64. INTRODUCTION 
5 
can within its own mass diffract light of sufficient intensity 
to be observed. 
The method of Tyndall was further extended by means 
of a method developed by H. Siedentopf and myself, of 
rendering visible ultramicroscopic particles.1 With its 
aid we can readily detertnine whether the inhomogeneity 
recognized by the Tyndall method is due to larger or to 
smaller particles. It also gives us within certain limits 
an idea of the size and color of the particles, and infor- 
mation furthermore if they are at rest or in motion. By 
this method we can even determine the major size limit 
of individual particles, which are beyond the bounds of 
visibility, providing we first fix the limit of ultramicro- 
scopic visibility of the particles of the body in ques- 
tion. 
The research described herein, made according to this 
method, has shown, in conformity with prior publica- 
tions,2 that the optical inhomogeneity which appears as a 
distinct haze when the particles are about 100 /i/j. in size 
becomes less and less as the particles become smaller; 
that fluids containing particles measuring 20 /z/z or less 
appear clear by ordinary daylight,3 and that the optical 
inhomogeneity recognizable by the method of Tyndall 
is still to be seen with particles approaching the size 
which the kinetic theory of gases ascribes to the molecules 
of crystalloids, but vanishes upon further subdivision. 
Hence solutions of crystalloids, unless they contain very 
1 Siedentopf and Zsigmondy, Drude's Ann., Vol. X, pp. 1-39, 1903. 
See also Chapter VI. 
2 Zsigmondy, Zeitschr. f. Elektrochemie, Vol. VIII, pp. 684-G87, 
1902; Siedentopf and Zsigmondy, Drude's Annal., Vol. X, pp. 30-39, 
1903. H. Siedentopf, Journ. Roy. Microscop. Soc., pp. 573-578, 1903. 
3 These facts apply to gold; with other substances noticeable opal- 
escence vanishes much sooner. 6 
INTRODUCTION 
large molecules1 or are reduced to hydrosols by colloid 
formation,2 should appear optically homogeneous when 
examined with the convergent beam of light (Lichtkegel.) 
Since, therefore, the hypothetical molecules of crystal- 
loids of low molecular weight escape direct observation 
because of their small size, and the best optical instru- 
ments that we have at present are powerless against them; 
and since some crystalloid solutions appear perfectly 
homogeneous optically, the question naturally arises if we 
cannot by other means observe the lack of homogeneity 
which the molecular theory presumes to exist in every 
solution. 
A means to this end might be the use of light waves of 
the greatest specific intensity, or else of ultraviolet light 
for illumination, in conjunction with the photographic 
plate to observe the diffracted light. 
It may be that we can also make use of differences of 
density which escape direct observation, by the applica- 
tion of mechanical force to a separation of the constituents. 
Some recently published work along this line by van 
Calcar and Lobry de Bruyn should be of great interest. 
These two Dutch investigators, by the action of centri- 
fugal force on salt solutions (KI, KCyS, etc.), have been 
able to produce a considerable change of concentration, 
and in the case of a saturated solution of Glaubers salt, 
were even able to force three eighths of the dissolved salt 
to crystallize at the periphery of the rotating vessel.3 
Van Calcar and Lobry de Bruyn regard this fact as new 
1 Lobry de Bruyn and Wolff, Ree. Trav. chim. Pays-Bas., Vol. 
XXIII, p. 155, 1904. 
2 Spring, Bull, d l'Acad. roy. de Belgique, No. 4, pp. 300-315, 1899. 
3 v. Calcar and Lobry de Bruyn, Rec. Trav. chim. Pays-Bas., Vol. 
XXIII, pp. 218-223, 1904. Earlier work in this field was done by 
Th. des Coudres, Bredig, Colley, and Quincke. INTRODUCTION 
7 
evidence of the continuity of the transition from colloid 
to crystalloid solution, which idea had been repeatedly 
advanced by the latter as well as by Picton and Linder. 
To me these experiments seem to support the view that 
even the homogeneous solutions of the crystalloids are 
not in a strict sense homogeneous, for in no fluid which is 
physically and chemically uniform in all its parts, and 
which therefore has the same specific gravity in every 
conceivable little section, could such a difference of con- 
centration be produced, no matter how great the speed of 
rotation of the centrifuge. 
Finally, we must mention that, according to Lord Ray- 
leigh, part of the polarized blue light of the sky may be 
attributed to the diffraction of sunlight by the molecules 
of the atmosphere.1 There are, therefore, media which, 
although in small volume they may appear entirely 
homogeneous and optically clear, may nevertheless possess 
an inhomogeneity so small that it can be observed only 
by looking through a depth comparable to the height of 
the atmosphere. 
After these remarks, on returning to the definition of 
solution, we see that, just as other investigators have said, 
when solutions are spoken of as homogeneous distributions, 
mixtures, etc., it cannot be meant that they are absolutely 
homogeneous mixtures. If such great homogeneity is 
demanded of solutions that we can detect no inhomo- 
geneity in them by our most sensitive methods, we would 
thereby exclude altogether from this classification solutions2 
not only of many colloids, but also of numerous crystal- 
loids, for example, fuchsin, ferric chlorid, chromic chloricl,3 
1 Philos. Mag., Vol. XLVII, pp. 375-384, 1899. 
2 It need hardly be said that polarized diffraction is referred to, not 
fluorescence. 
3 Spring, loc. cit., Picton and Linder, Journ. Chern. Soc., Vol. LXI, 
pp. 114-136. 8 
INTRODUCTION 
saccharose, raffinose,1 and solutions in the critical state.2 
We would thus run a risk of reducing the sphere of solu- 
tions every time we increase the sensitiveness of our 
methods of investigation. This danger can be easily 
avoided if we use the word "solution" in its usual chemical 
acceptation, meaning thereby subdivisions which appear 
clear in ordinary daylight, and which cannot be separated 
into their constituents by the ordinary mechanical means 
of separation (filtration and decantation). The word solu- 
tion will be used here in this sense. The question of the 
heterogeneity of colloidal solutions seems rather to resolve 
itself into the question of the heterogeneity of solutions in 
general; and this standpoint may lead to many investi- 
gations of the highest importance. 
Regarding Suspension and Colloidal Solution. While 
most chemists regard suspension as a coarse me- 
chanical mixture, the solid or semi-solid parts of which 
can be separated from the fluid by decantation and 
filtration,3 some physicists-Faraday for example-went 
much further, and considered as suspensions the smallest 
particles that can by any manner of means be recognized 
optically, even if the individual particles cannot be seen. 
They knew not how far this brought them into the domain 
of the colloidal solutions of the chemist, the physiologist, 
and the physician. 
While several investigators differentiate between sus- 
pensions and emulsions, according as the subdivided 
substance is solid or fluid, others have made no such dis- 
1 Lobry de Bruyn and Wolff, Rec. Trav. chim. Pays-Bas, Vol. XXIII, 
p. 155, 1904. 
2 Konowalow, loc. cit. 
3 Examples: Suspended stannic acid, alumina, etc.; suspended 
ferric oxid in contrast to: colloidally dissolved stannic acid, alumina; 
solutions of colloidal ferric oxid. INTRODUCTION 
9 
tinction. According to Quincke,1 suspensions contain solid 
or fluid particles, while emulsions contain only fluid 
particles, or else solid floating particles that are coated 
with a thin oily skin. 
From what precedes,2 it is evident that there is no uni- 
form conception of "suspension," and that therefore, 
because the multifarious meaning of this word can easily 
lead to misunderstanding, even to the extent of consider- 
ing as suspensions only the coarse mechanical ones, we must 
hold as erroneous the definition in most general use, that 
colloidal solutions and also colloidal metals are suspen- 
sions.3 
I believe that all misunderstanding could be avoided 
if in this definition we use the word subdivision4 instead 
of suspension. No one could object to the conception of 
colloidal solutions or hydrosols as very fine subdivisions, 
especially as this involves no assumption regarding the 
nature of the aggregation of the material subdivided, or 
the forces affecting the individual particles, or their stereo- 
metric nature, or their size and limits. These are ques- 
tions that cannot be answered by general definition or by 
hypotheses, but each case must be differentiated by 
experimentation, which will make the investigation of 
1 Drude's Ann., Vol. IX, pp. 1009 and 1010, 1902. 
2 Here might be mentioned the hypothesis of Lemery (Cours de 
chimie, Leyden, 1716) which at present seems to us extremely odd, 
that gold chlorid solution is a suspension of gold particles carried by a 
network of aqua regia. (See Ostwald, Lehrb. der allgem. Chern., 
2d ed., Vol. II, 2, p. 5.) 
3 Cf. B. A. Muller, Zeitschr. f. anorg. Chern., Vol. XXXVI, p. 340, 
1903. 
4 The term "subdivision" is frequently misunderstood. It is a 
generic term, "suspension" and "solution" being specific terms 
included within it, and it does not express any definite degree of fine- 
ness, as is commonly supposed. 10 
INTRODUCTION 
colloids the theme of a new, far-reaching science, built 
upon the present promising foundation. 
We may further characterize colloidal solutions and 
distinguish them in an unobjectionable manner from 
crystalloidal solutions, by incorporating into the definition 
of the former the characteristic noted by Graham, who 
showed that colloidal solutions lack the property (or else 
possess it very slightly) of diffusing through parchment 
membranes.1 
A safe distinction between coarse mechanical suspensions 
and colloidal solutions is that the latter have a more 
complete and thorough mixture of the constituents, and 
possess greater homogeneity, as compared with the 
former. The claim that colloidal solutions must show 
osmotic pressure, small though it may be, in order to be 
called solutions at all, does not appear to me to be essen- 
tial, for it adds but little to the characteristics of the 
hydrosols, whether their solutions possess measurable 
osmotic pressure or not. 
In this book we will designate as suspensions only the 
coarser subdivisions of solids, such as fine powders which 
undergo sedimentation. Finer subdivisions which no 
longer settle, will on the other hand be called colloidal 
solutions if they exhibit the other essential characteristics 
of such solutions. 
1 There might also be incorporated into the definition, the criterion 
noted by Bredig (Anorganische Fermente) "that the work necessary 
to separate solvent and solute is extremely small in comparison with 
crystalloid solutions." CHAPTER I 
LIMITATION OF THE FIELD 
i. Limitation as to Crystalloidal Solutions 
The chief characteristics distinguishing colloidal from 
crystalloidal solutions, have already been given by Graham. 
According to him colloidal solutions are distinguished 
from crystalloidal solutions by their small diffusibility, 
and the concomitant inability of the majority of them 
to pass through parchment membranes or jellies. (See 
Chapter III.) 
Numerous other distinguishing characteristics were 
noted by Graham as well as by other investigators, and 
Bredig 1 has done us the service of compiling them and 
adding some new ones. 
According to Bredig solutions of crystalloids and col- 
loids may be distinguished by means of: (a) diffusibility; 
(6) the work necessary to remove the solvent; (c) elec- 
trical migration; (d) coagulation; (e) absorption; (/) 
irreversible changes of constitution and hysteresis; (g) im- 
permeability toother colloids; (h) optical inhomogeneity; 
(i) electrical formation of sols. 
It is evident from this brief resume that there are many 
ways of distinguishing colloids from crystalloids. Not- 
withstanding this, no sharp line of demarcation can be 
established, for there are numerous intermediaries between 
both kinds of solutions. 
1 Anorganische Fermente, loc. cit., pp. 10-12. 
11 12 
LIMITATION OF THE FIELD 
2. Limitation as to Suspensions 
But little has been said of the characteristics which 
distinguish colloidal solutions from undoubted suspensions; 
a few words on this topic are therefore added. 
In opposition it might be said that there is such a mani- 
fest difference between real suspensions, for instance, sedi- 
ments of powdered substances like quartz, starch, metal 
dust, suspensions of bacteria, etc., and clear, colloidal 
solutions, that no words need bo wasted on this subject. 
Nevertheless I do not regard it unnecessary to consider 
the matter, because here too there are intermediate forms 
which lead so gradually from one class to the other, that 
an observer, having in mind only the intermediate forms, 
is apt to miss the remarkable difference between the mem- 
bers of both groups. 
(1) In the first place there is an enormous difference in 
size between the particles in hydrosols and those found in 
real suspensions. A glance at Plate III, Chapter XV, 
will make this more evident than words. 
The following properties of colloids will serve in part 
to show the difference: 
(2) Irreversible change of condition. 
(3) Change of the total energy of the system by coagu- 
lation. 
(4) Absorption of the subdivided substance by porous 
bodies, such as charcoal. 
(5) The property of colloids to enter into reactions among 
themselves, which bear a deceptive resemblance to chemical 
reactions, etc. 
To avoid possible misunderstanding it might right here 
be stated that most of these differences between hydrosols 
and suspensions are dependent upon the degree of sub- LIMITATION AS TO SUSPENSIONS 
13 
division of the material, and they are therefore not very 
great if closely adjoining members of the series be com- 
pared; but they become enormous when we compare 
typical representatives of true suspensions with those of 
colloidal solutions. These differences justify the classifi- 
cation into colloidal solutions and suspensions long ago 
established by chemists, physiologists, and physicians. 
Let us now consider the differences (2) and (3) above 
noted. 
But little expenditure of mechanical energy is necessary 
to stir up the fine powder of a practically insoluble sub- 
stance into a suspension. If this is allowed to stand, 
gravity separates the powder from the bulk of the fluid. 
The sediment can again be stirred up and again allowed to 
settle. This process can be repeated as often as desired, 
the powder remaining unchanged, except that the dis- 
tances between the particles will be larger after stirring 
them up, and smaller when they have settled. 
In the case of coarse suspensions, the addition of elec- 
trolytes does not appreciably change the above described 
process; this can readily be shown by allowing a suspen- 
sion of potato starch or quartz whose particles measure 
0.06 mm. and over to settle first in pure water and then 
in a dilute solution of table salt. Finer suspensions, such 
as the turbidity due to clay, are precipitated by the addi- 
tion of electrolytes, just as are colloidal solutions, but even 
in this case, as we shall see, characteristic differences can 
be determined. 
If the suspended powder is completely separated by 
desiccation from the water in which it was distributed, 
the original subdivision can be reproduced by stirring, 
provided that certain colloids which tend to cement 
together the individual particles are absent. 
But if an attempt is made to separate a colloidal metal, 14 
LIMITATION OF THE FIELD 
gold for instance, from the medium in which it is dis- 
tributed (which can easily be done by evaporating the 
fluid or adding electrolytes), it undergoes an irreversible 
change of condition, or coagulates as we say. The metal 
thereby completely loses its capability of resuming its 
original condition of subdivision by shaking or stirring, 
and the coagulation involves profound changes in its 
relations to light waves, as evidenced by change of color, 
polarization, etc.1 
Regarding these changes produced by coagulation, we 
understand with certainty only one, that is that numerous 
small submicroscopic 2 particles, which I will call a-par- 
ticles, are combined into larger, but still submicroscopic 
particles (^-particles). Comparison with the flocculation 
of fine suspensions, carefully studied by many investi- 
gators,3 has led to the conclusion that the submicroscopic ,5- 
gold particles are grouped together like the flocks in 
kaolin, clay, etc.; but in order to avoid unwarranted con- 
clusions based upon analogy, see the relative sizes given 
in Plates III and IV, Chapter XV. To see the vast 
difference just imagine flocks made up of the clay par- 
ticles or anthrax bacilli shown in Plate III, and compare 
them with imaginary flocks made up of gold particles 
which, though magnified 10,000 times, appear as tiny 
points. It will at once be seen that a-gold particles in 
such a "gold flock," must be much closer together than 
the kaolin particles in a clay flock, and that the attractive 
forces which we suppose to exist between the smallest 
1 The difference between suspensions and irreversible colloidal 
systems was known to Hardy, although not sharply characterized. 
2 For definition, see Chapter VI. 
3 See Barus and Schneider, Zeitschr. physik. Chem., Vol. VIII, 
p. 278, 1891; Quincke, Drude's Ann., Vol. VII, p. 59, 1902; Bod- 
lander, Gott. Nachr. 1893. - LIMITATION AS TO SUSPENSIONS 
15 
particles of a substance (cohesion forces), will effect a 
much more powerful union of the a-particles than in 
ordinary flocks of coarser material. 
But even without this exposition, just by the aid of 
rough experiments, we can easily establish the difference 
between the flocculation of fine clay particles, and the 
coagulation of colloidal metals. 
As far back as 1870, Ch. Schlosing showed 1 that the 
turbidity due to clay could, after its precipitation by salts 
of calcium or magnesium, be once more suspended in 
water and again flocculated by the addition of salts.2 On 
the contrary, coagulated colloidal gold cannot be brought 
into colloidal solution by stirring it up in pure water, 
even if all salt solution has been previously washed out. 
There results only a suspension of /?-gold particles, which 
soon deposits its gold. 
In order to change ^-gold into a-gold we must make 
use of chemical or electrical energy, for instance solution 
in aqua regia and reduction of the gold chlorid to colloidal 
gold, or electric atomization by Bredig's method. 
Finally coagulation involves change, considerable at 
times, in the total energy of the system, which can be 
directly determined by the calorimeter. Thus Prange 
(Chapter III) found that one gram of colloidal silver on 
precipitation by ammonium citrate, developed 250.98 to 
126.73 calories, depending on whether a dilute or a con- 
centrated solution was precipitated. Every gram-atom 
of silver accordingly liberates 27,100 to 13,700 calories. 
Even this does not yield the ordinary metallic condition. 
The conversion of the hydrogel into the metal (by means 
of H2SO4) sets free more heat, 60 calories per gram of 
silver, according to Prange. 
1 Ch. Schlosing, Compt. rend., Vol. LXX, p. 1345, 1870. 
2 Sometimes this flocculation is erroneously called coagulation. 16 
LIMITATION OF THE FIELD 
Barus and Schneider, in commenting on Prange's obser- 
vations, say: 
"In his experiments on colloidal silver, Prange found 
that the transformation of the almost solid colloid into 
normal silver liberated about 60 calories. Prange attri- 
butes this to the change of colloidal into normal silver. 
But the liberation of heat can be explained in the following 
manner: Work is necessary to transform a quantity of 
silver, already finely subdivided (normal silver) into a 
considerably finer state of subdivision (colloid)-that is, 
heat is necessary. Reversing the process, the transforma- 
tion of colloidal silver into the precipitate of normal silver 
(consisting of larger particles of the substance) liberates 
the heat again." 
On this point I agree entirely with Barus and Schnei- 
der. The size of the coagulated silver particles, as 
figured by these authors, approximates the true value, 
and is rather too small than too large. 
On the other hand the difference between the colloidal 
solution and suspension containing the same percentage 
of the same substance is evidence to me of a remarkable 
difference between the two kinds of subdivision. It is 
to be noted that the energy difference between suspended 
and colloidal silver is greater per gram-atom than that 
between the allotropic modifications of phosphorous, 
carbon, sulphur, etc. 
Intermediate Forms between Suspensions and Colloidal 
Solutions. Between the coarse mechanical suspensions 
and colloidal solutions there are many intermediate 
forms. In general the finer a body is subdivided the 
more it assumes the properties of a colloid. 
Even with potato and wheat starch we can readily see 
the influence of subdivision. While the settlement of the 
former is scarcely influenced at all by the addition of an LIMITATION AS TO SUSPENSIONS 
17 
electrolyte, the finer particles of the latter exhibit a dis- 
tinct flocculation and accelerated settlement upon the 
addition of table salt. 
Turbidity due to clay, ultramarine, the much investi- 
gated mastic-turbidity, and others, form good examples of 
the intermediate forms between colloidal solutions and 
suspensions. 
By pointing out the relations between fine suspensions 
and colloidal solutions, and thus establishing a new point 
of view, Ebell, Ostwald, Barus and Schneider, Spring, 
Lottermoser and von Meyer, and particularly Bredig, 
have rendered a great service which has already proved 
useful in leading the supporters as well as the opponents 
of this view to a series of experiments and the establish- 
ment of important facts. 
Summary. Colloidal metal solutions are differentiated 
from true suspensions of pulverized substances, not only 
by the dissimilar subdivision of the solid body but also 
by the irreversible change of constitution invariably 
shown by colloidal solutions; and further by change of 
the total energy liberated upon the separation of the 
atomized substance from the medium. There are many 
intermediate forms between the two classes of subdivision, 
so that there is a continuous series, one leading to the 
other. 
The colloidal metal solutions have been purposely con- 
sidered, to establish the distinguishing characteristics 
between colloidal solutions or hydrosols and true suspen- 
sions, because they show a striking similarity to certain 
fine suspensions in their behavior with reagents. This 
similarity has led to the attempt on the part of some 
investigators to remove them from the group of colloidal 
solutions and classify them as suspensions. We must 
oppose such a conclusion, based as it is on on^-sided 18 
LIMITATION OF THE FIELD 
observation, because of the distinguishing characteristics 
before given. Real metal hydrosols rather tend to form 
a class of colloidal solutions, which is distinguished from 
the majority of the others, not by a lesser degree of sub- 
division but by the nature of the original substance. 
We will take this up more fully later. CHAPTER II 
CLASSIFICATION OF HYDROSOLS ACCORDING TO TWO 
DIFFERENT POINTS OF VIEW 1 
One of the most important tasks of a nascent science 
appears to me to be the arrangement of the available 
facts upon one single basis. 
If only an imperfect attempt in this direction can be 
made, it is because the point of view is so new, and the 
facts are so incompletely known. I consider it of import- 
ance, however, to direct the attention of investigators to 
this subject. 
Just as geologists distinguish, according to the size of 
their constituent particles, between boulders, rubble, 
pebbles, gravel, sand, and dust, so too in dealing with 
more finely subdivided bodies, a classification based on 
the size of the particles will be of corresponding value.2 
1 The best classification of colloids yet suggested is that of Hardy 
(Zeitschr. f. physik. Chem., Vol. XXXIII, p. 326). Recently Muller 
(Zeitschr. f. anorg. Chem., 1903, Vol. XXXVI, p. 340) suggested a 
classification with which I cannot agree, for reasons that may in part 
be gathered from the following chapters. Nevertheless I fully agree 
with Muller that colloidal chemistry must establish a classification 
of hydrosols. Independently of Hardy, the changes of condition of 
colloids have been divided by Wolfgang Pauli (Arch. f. d. ges. Phy- 
siologic, 1899, Vol. LXXVIII, p. 315), into those easily and those 
with difficulty reversible. 
2 A classification of the various kinds of sand, pebbles, and dust, 
according to the size of their granules, has been attempted by Dr. 
Albert Atterberg, of Kolmar (Schwed. lai.dw. Akad., 1903, and Chem. 
Ztg., 1905, Vol. XXIX, p. 195). The capillarity (the height in milli- 
19 20 
CLASSIFICATION OF HYDROSOLS ACCORDING TO 
Pieces of like size have certain properties in common; for 
instance, sand and dust are carried by the wind, while 
cobble-stones and pebbles are unmoved; it makes no 
difference of what material the sand, dust, or pebbles 
consist. Sand will pass through a 10 mm. mesh sieve, 
but cobblestones and pebbles will not. 
Weight has a noticeable effect on suspensions, pro- 
ducing the deposition of the suspended particles, and its 
influence can be traced down to particles as small as 
one tenth of a wave length of light. With particles smaller 
than 60 /zy,1 and even often with larger ones, its influence 
is masked by that of the electrical charge of the particles 
and by the influence of other kinds of energy. Further- 
more in the case of much finer subdivisions, which we will 
take up later, the size of the particles immediately deter- 
mines a whole series of properties. We would be anticipat- 
ing if we discussed here the change of properties dependent 
upon the size of the particles. Part of the experimental 
meters to which water will ascend in sand) as well as the perme- 
ability with respect to water, are intimately dependent upon the size 
of the granules, both of which facts have great significance in agricul- 
ture. Atterberg (a Swede) makes the following classification of the 
fragmental pieces: 
। Diameter 
Block (boulders) 2 m. to 20 cm. 
Klapper (pebbles) 20 cm. to 2 cm. 
Grus (gravel) 2 cm. to 2 mm. 
Sand (sand) 2 mm. to 0.2 mm. 
Mo (earth) 0.2 mm. to 0.02 mm. 
Lattler (loam) 0.02 mm. to 0.002 mm. 
Ler (clay) smaller than 0.002 mm. 
With each of these various sizes are associated definite physical 
properties. Loam shows great capillarity, but little permeability; 
sand 2 mm. in diameter has a capillarity of only 25 mm.; gravel and 
pebbles have no perceptible capillarity; earth (mo) allows water to 
soak through quickly. (Capillarity, 428-2000 mm.) 
1 1 ^ = 1/1,000,000 mm. TWO DIFFERENT POINTS OF VIEW 
21 
section of this book deals with this subject. Polarization, 
the color of the particles in many cases, their behavior 
under the action of gravity and during filtration, the 
energy-content of the system, and the movement of the 
individual particles-all these vary with their increasing 
or diminishing size. 
One change must be mentioned here. While suspensions 
of larger particles, for example, 1-10 //, always possess a 
particularly inhomogeneous appearance,1 fluids containing 
particles smaller than 20 appear clear by ordinary day- 
light, even if the indices of refraction of the medium and 
subdivided substance are as different as those of water 
and metallic gold. While particles of the former kind 
can be easily and completely separated from the fluid in 
which they float, by filtration through filter paper, the very 
actively moving particles of the latter cannot by this means 
be separated from the fluid. It has been customary for 
chemists to designate as solutions,2 subdivisions of such 
perfect homogeneity, that, like the latter class above 
1 It is here assumed that the particles and medium have different 
indices of refraction, as is generally the case.' 
2 These solutions with particles of about 20 yy and less are not without 
influence on light, although by ordinary daylight they appear clear 
and transparent. With the intense transverse illumination of direct 
sunlight their characteristic inhomogeneity is easily recognized by 
their characteristic scattering and polarization of light. Although 
this method has long been known to physicists, it was neglected by 
chemists; for the slight turbidity almost always present, even if visible 
without direct sunlight, was ascribed to contamination of the solution 
by larger suspended particles. Finally more recent investigations (Pic- 
ton and Linder, Spring, Lobry de Bruyn, and others) have made 
evident the universality of this phenomenon, and shown that a large 
majority of colloidal and crystalloidal solutions are optically inho- 
mogeneous But even this inhomogeneity, most conspicuous at 
20 yy, becomes less and less, and almost or entirely vanishes with all 
substances in the neighborhood of the molecular dimensions. (See 
Introduction and Chapter IX.) 22 
CLASSIFICATION OF HYDROSOLS ACCORDING TO 
described, they are clear, transparent, and do not settle. 
Subdivisions of the former class arc, on the other hand, 
considered as sediments or suspensions. Accordingly, if 
we arbitrarily omit the intermediate members (between 
1000 and 20 pp) we can differentiate according to the 
size of the particles, two sharply defined groups of sub- 
divisions-coarse mechanical suspensions with particles 
from about 1 p and over, and apparently homogeneous 
solutions with particles of 20 pp and less. Between these 
two classes of subdivisions come the transition forms 
which, according to their properties, are classed sometimes 
as colloidal solutions, sometimes as fine suspensions. 
These remarks suffice to show the importance of a 
classification based on the size of the particles. 
Before taking this up, we must first consider the classifi- 
cation established by Graham, and also that of Hardy. 
Graham divides homogeneous looking subdivisions, that is 
solutions, according to their greater or lesser diffusibility, 
into solutions of crystalloids and colloids. This classifica- 
tion of Graham is an exceptionally happy one, for although 
it establishes no sharp boundary between the two groups,1 
it brings into prominence a scries of contrasts between the 
members of both classes of solutions. 
In general, crystalloidal solutions are regarded as the 
complete subdivision of the substance, and colloidal solu- 
tions as the less complete. But it cannot reasonably be 
held that all crystalloidal solutions without exception 
possess a higher degree of homogeneity than colloidal 
solutions. In the following chapters we will learn of one 
case of colloidal solution where the subdivision is evidently 
carried as far as in the case of a crystalloid solution, and 
is probably even more complete. 
1 Graham himself knew of the transition forms between crystalloid 
and colloid solutions. TWO DIFFERENT POINTS OF VIEW 
23 
(Compare Chapters IX, XVII, and XVIII (6). 
On the other hand, coarse subdivisions, such for ex- 
ample as are formed by washing out sulphids, and which 
pass through filter paper, would be classed as colloidal 
solutions, and it was just in the case of such subdivisions 
that the heterogeneity of hydrosols was first dicovered. 
Regarding these relative proportions, reference should be 
made to the diagram, Plate I, which is based upon the 
size of the particles. If the subdivisions are arranged in 
successive order according to the size of their particles, 
beginning above, having the largest at the top and the 
smallest at the bottom, the relative sizes may then be 
seen at a glance. 
Several investigators have expressed the opinion that 
those colloidal solutions (such as the metal hydrosols) 
which, upon the addition of electrolytes, act as do sus- 
pended clay particles, should be considered as suspensions, 
and not as colloidal solutions at all. From this it might 
be assumed that, in the diagram, the metal hydrosols 
would lie next to or be included in with the suspensions; 
while those hydrosols which are not precipitated by traces 
of electrolytes would be nearer the solutions--that is, 
they would be in a finer state of subdivision. Such 
is not the case, however. Another classification than 
the preceding, one which, wp a certain point, is 
independent of the size of the particles, would be 
made by studying the effect of the addition of 
electrolytes, and of changes of temperature on the 
subdivisions. 
Such a classification was first successfully introduced by 
Hardy. He divided the colloids into those reversible and 
those irreversible, according as the change from sol to gel 
could or could not be made retrogressive by reversing the 
conditions of its production. The conditions studied by 24 
CLASSIFICATION OF HYDROSOLS ACCORDING TO 
Hardy1 as causing hydrogel formation were chiefly 
changes of temperature and the addition of electrolytes. 
But difficulties arise on attempting a classification of 
colloids in general, according to Hardy's method. Only 
in individual cases do ordinary changes of temperature lead 
to the formation of gels, as in the case of gelatin solutions 
which were closely studied by Hardy. Extreme changes 
of temperature (complete freezing, for instance), almost 
always lead to gel formation, but they sometimes cause 
irreversible change of constitution in the case of colloids 
classed by Hardy as reversible (gelatin, for example). 
If the changes produced by the addition of electrolytes 
are studied, it will be found that a certain hydrosol under- 
goes sometimes a reversible and sometimes an irreversible 
change of condition, according to the nature of the elec- 
trolyte added, so that it may appear doubtful to which 
class the hydrosol belongs.2 Thus the hydrosol of stannic 
acid is converted into a hydrogel by the addition of the 
slightest trace of most electrolytes. However, while acids 
and many salts give a hydrogel which remains insoluble 
after the elimination of the precipitant,3 other precipitants 
for example table salt or caustic alkalis, yield precipitates 
which redissolve in water upon the elimination of the 
reagent. 
In order therefore to make practical a classification 
along the lines laid down by Hardy, it is necessary to con- 
sider only such influences on the hydrosol as always lead 
to formation of gels or of solid colloids, and which, as far 
1 Hardy, Zeitschr. f. physik. Chem., 1900, Vol. XXXIII, pp. 326 and 
385. 
2 Pauli first referred to this; loc. cit. 
3 Lottermoser calls water-soluble colloids solid hydrosols. According 
to Hardy, however, we must regard jellies and solid colloids as hydro- 
gels, even if they are soluble in water. The word hydrogel is used 
here in this sense. TWO DIFFERENT POINTS OF VIEW 
25 
as possible, are independent of chance or outside influences, 
as well as of the personal equation. 
Evaporation is a method of treatment which in every 
case leads to the formation of a gel or of a solid residue. 
It is sufficient for our purpose to class as reversible those 
colloids whose solutions leave a residue soluble in water, 
when desiccated at ordinary temperatures;1 and to consider 
as irreversible those colloids which, under the same cir- 
cumstances, yield a residue insoluble in water. 
According to this classification, dextrin, gum arabic, 
most albumens, moly belie oxid, moly belie acid, etc., are 
reversible colloids, while, on the other hand, stannic acid 
and many other colloidal oxids and sulphids, as well as 
the pure colloidal metals, etc., are irreversible colloids. 
While the former are frequently insensitive to the addi- 
tion of electrolytes, the latter for the most part are ex- 
tremely sensitive, and are easily coagulated by the slightest 
addition of electrolytes. If solutions of reversible are 
added to those of the irreversible colloids, the latter may 
become reversible, in which case they are often protected 
against precipitation by salts (Schutzwirkung). Fre- 
quently a very small addition of a reversible colloid is 
sufficient to produce these effects in a striking manner. 
Carey Lea's colloidal silver, which can never be prepared 
free from foreign colloids, commercial argentum Crede, 
coilargol, Paal's colloidal metals, Mohlau's colloidal indigo, 
1 As will be seen, Hardy's basis of classification is adhered to; but 
instead of numerous conditions which lead to gel formation, here only 
one is taken into consideration. Evaporating the hydrosol is a very 
efficient means of causing gel formation; on increasing concentration 
many colloids are transformed into insoluble hydrogels, even though 
wet, while others lose their solubility on further evaporation. The 
reversible hydrosols, on the other hand, even after evaporation to 
dryness, retain the property of distributing themselves in water 
in their original condition. 26 
CLASSIFICATION OF HYDROSOLS 
all owe their solubility in water to the presence of pro- 
tective colloids.1 
The accompanying diagram, Plate I, shows how colloidal 
solutions and suspensions may be grouped according to 
the size of their particles. Above, in the microscopic field 
and below, even extending into the ultramicroscopic field, 
are to be found true suspensions, whose sphere extends 
much further up than is shown in the diagram. 
Colloidal solutions or hydrosols all contain ultramicro- 
scopic particles. No sharp line of demarcation can be 
drawn between suspensions and colloidal solutions; their 
spheres mutually invade each other; that of the hydrosols 
being here extended as far upwards as possible, to include 
everything which has formerly been classed as a colloidal 
solution. Several typical hydrosols are mentioned in the 
table. A perpendicular through the cross-hatching lines 
divides the whole area into two sections; on the right 
are the irreversible, on the left the reversible colloids. 
This division simplifies the diagram. It will be seen that 
on the right are those hydrosols which are distinguished 
by especial sensitiveness to electrolytes. This does not 
mean, however, that colloidal metals (and sulphids, etc.), 
more nearly approach suspensions than do reversible 
1 By the classification here followed two groups will be distinguished, 
whose typical representatives differ from each other in several import- 
ant particulars. As everywhere in nature there are also transitions 
here; nevertheless a sharp separation could be made if the conditions 
upon desiccation, such as temperature, moisture-content of the gas- 
phase, etc., are predetermined. An advantage of this classification 
is that outside influences which might act chemically upon the colloid, 
such as increase of temperature, chemical action of foreign electrolytes, 
etc., are as far as possible avoided. On the other hand, possible 
chemical changes upon desiccation of the colloid, are left out of con- 
sideration, for it is almost impossible to determine such changes 
satisfactorily and to separate them clearly from the physical changes 
that always occur. PLATE I 
Ultra microsco pic Field. Microscopic Field. 
Classification of Colloidal Solutions 
according to the size of the particles contained in them and according 
to their behavior upon dessication. 
I7'o face page 26.] TWO DIFFERENT POINTS OF VIEW 
27 
hydrosols (glycogen, albumen, etc.), as has often been 
maintained, but that they tend of themselves to assume 
the stable form-that is, the solid compact form of the 
substance (metal, sulphid, or oxid). This tendency is 
realized upon evaporation, as well as upon the addition 
of electrolytes. Such addition hastens the process, which 
in most cases takes place of itself.1 The very finest sub- 
divisions of metals are sometimes the most sensitive and 
coagulate very readily. 
It must, however, be noted that the behavior of hydro- 
sols upon evaporation cannot always be foretold from their 
reaction with electrolytes. Most reversible colloids-for 
instance, molybclic acid, tungstic acid, gum, albumen, 
gelatin-are not sensitive to small additions of electrolytes. 
On such addition either they are not precipitated or else 
they undergo a reversible precipitation, which, upon 
elimination of the precipitant, can be made retrogressive. 
(On this subject see W. Pauli.2) But there are also col- 
loids, reversible according to the above classification, which 
are extremely sensitive to the addition of electrolytes of 
every kind, Graham's caramel, for example. (Chapter 
III). In this case the specific influence of the subdivided 
substance or of its constituents, is distinctly changed. 
In the table sulphids, oxids, etc., are placed with the 
colloidal metals, while glycogen, albumen, etc., are grouped 
with the reversible colloids. The names written vertically 
indicate that these fluids may contain particles of the 
most diverse sizes, but not, however, that any one fluid 
must necessarily contain all sizes. 
Linder and Picton have already differentiated various 
1 Electrolytes are not, however, to be here considered as catalysers. 
(See Chapter XVIII, also W. Pauli, Ergebnisse der Physiologic, 1904, 
p. 157). 
2 W. Pauli, Pflugers Archiv., 1899, Vol. LXXVIII, p. 315. 28 
CLASSIFICATION OF HYDROSOLS ACCORDING TO 
kinds of arsenic sulphid according to their method of 
preparation, which they called arsenic sulphid a, fl, y, 
and d, and they have shown that the fluid As2S 3a contains 
the largest, while As2S35 contains the smallest particles.1 
Ultramicroscopy enables us to fix the approximate sizes 
when the particles can be rendered visible, and when these 
are known, the diagram, which is only outlined here, can 
be then partially filled in. As an illustration several 
colloidal gold solutions described in detail in the experi- 
mental section of this book (see Chapter IX), are arranged 
according to the sizes of their particles. 
At the bottom following the fluids Au73a, Au92, etc., are 
gold solutions with amicroscopic particles, which con- 
clude the series beginning with Bredig's colloidal gold, 
having particles from 20 to 80 /zy in size. Above is the 
gold suspension c which deposits its gold; above this 
should be placed the well-known suspensions of gold, 
obtained by precipitating a solution of gold chlorid with 
FeSO4 or oxalic acid. 
By way of further explanation of the diagram, it might 
be said that the individual particles of most colloidal inor- 
ganic oxids cannot be seen, for they, as well as most organic 
colloids, are not so easily rendered visible as are colloidal 
metals, and therefore I have formed no idea as to the 
size of their particles. Colloidal stannic acid contains, 
according to its method of preparation, submicrons 2 or 
amicrons.2 The same is the case with the purple of 
Cassius and colloidal silicic acid. 
Among the reversible colloids are included: Mohlau's 
1 Picton and Linder (loc. cit. in Chapter III) based their conclusions 
regarding the difference of the sizes of the particles, upon the behavior 
of the fluids when filtering through porcelain and upon diffusion 
experiments. 
2 Definition, see Chapter VI. TWO DIFFERENT POINTS OF VIEW 
29 
colloidal indigo and Paal's blue colloidal gold, which 
contain quite large copper-red particles 1; Lea's colloidal 
silver with particles of various size; and further, on the 
basis of the work of Raehlmann, Much, Romer and Sie- 
bert, Biltz and Gatin-Gruzewska, as well as Michaelis 
(Chapter XX), glycogen, several kinds of albumen, dyes 
with particles which are ultramicroscopically visible, but 
whose size is not yet determined. 
Lower down are soluble starch and crystallized albumen 
with question marks attached, because, although they show 
a light-cone (lichtkegel), their individual particles cannot 
be distinguished, but probably are about 5 yy. Finally, 
at the side are dextrin, many dyes such as Congo red, etc., 
and molybdic acid, etc., which from their general proper- 
ties are all to be regarded as intermediate between crystal- 
loidal and colloidal solutions, or else, like soluble starch, 
as crystalloid solutions with high molecular weight. If a 
3 per cent solution of soluble starch is allowed to stand, 
in course of time particles are formed in it, which may 
grow up to 100 yy (see Chapter XXI). 
At the very bottom of the reversible colloids I have 
placed the crystalloidal solutions and gas mixtures, not 
because the sizes of their particles can as yet be demon- 
strated experimentally by any method independent of 
theory, but just for the sake of completeness. In the 
present state of our knowledge, the hypothetical molecules 
of the crystalloids will serve quite well and be unobjection- 
able for the purpose of this table. 
1 The respective sizes of these particles are not yet determined, 
their appropriate place in the table is therefore still uncertain. CHAPTER III 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
i. Graham's Investigations 
Since the time when chemists commenced the examina- 
tion and description of animal and vegetable materials, 
colloidal substances have been investigated and described, 
and that in much greater detail in the older than in the 
modern text books of organic chemistry. In addition, 
many isolated observations on inorganic colloids have been 
made and recorded in recent years. 
To Thomas Graham is due the credit of having recog- 
nized and described in detail the similarity in the various 
phenomena, and of having first proposed a classification 
that gave an insight into the essential differences between 
the two separate classes of solutions. Here, as in many 
other instances, progress followed the discovery of a new 
method. While working on diffusion experiments, Graham 
found that those substances which easily crystallize from 
their solutions, would readily diffuse through jelly-like 
membranes, while on the contrary, amorphous substances 
either lacked this property or possessed it to only a very 
limited degree. Membranes of parchment paper are par- 
ticularly suitable for this work. By their aid Graham 
could not only easily separate crystalloids from colloids, 
but could also prepare a large number of the latter in a 
sufficient state of purity. He also introduced the expres- 
sions now in general use; colloid, crystalloid, hydrogel, 
30 LIQUID DIFFUSION APPLIED TO ANALYSIS 
31 
hydrosol, etc. Graham's work must therefore be dis- 
cussed in detail, especially as it is not well enough known.1 
A considerable part will be given in Graham's own words :2 
Liquid Diffusion Applied to Analysis 
By Thomas Graham, F.R.S., Master of the Mint 
(Received May 8, Read June 13, 1861) 
"The property of volatility, possessed in various degrees 
by so many substances, affords invaluable means of separa- 
tion, as is seen in the ever-recurring processes of evapora- 
tion and distillation. So similar in character to volatility 
is the diffusive power possessed by all liquid substances, 
that we may fairly reckon upon a class of analogous 
analytical resources to arise from it. The range also in the 
degree of diffusive mobility exhibited by different sub- 
stances appears to be as wide as the scale of vapor ten- 
sions. Thus hydrate of potash may be said to possess 
double the velocity of diffusion of sulphate of potash, and 
sulphate of potash again double the velocity of sugar, 
alcohol, and sulphate of magnesia. But the substances 
named, belong all, as regards diffusion, to the more "vola- 
1 In contradistinction to this fundamental work, other publications 
regarding irreversible hydrosols are rather briefly treated here. In 
the quotations there is no claim to completeness; much important 
work is passed over, the main object being to direct attention to 
certain points. The complete literature of the subject, in so far as 
not here given, can be found in A. Lottermoser, "Uber anoganische 
Kolloide," Stuttgart, 1901, and A. Muller, "Die Bibliographic der 
Kolloide" (Zeitschr. f. anorg. Chemie, 1904, Vol. XXXIX, p. 121). 
Work concerning the theory of colloids is referred to only in so far 
as germane. See also Bredig, " Anorganische Fermente"; A. Miiller, 
"Die Theorie der Kolloide," Leipzig and Vienna, 1903; Billitzer 
(Zeitschr. f. physikal. Chemie, 1905, Vol. LI, pp. 129-166). 
2 Thomas Graham, Philosophical Transactions of the Royal Society 
of London, 1861, Vol. 151, pp. 183-224. (Also Liebig's Annalen, 1862, 
Vol. CXXI, pp. 1-77). 32 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
tile" class. The comparatively "fixed" class, as regards 
diffusion, is represented by a different order of chemical 
substances, marked out by the absence of the power to 
crystallize, which are slow in the extreme. Among the 
latter are hydrated silicic acid, hydrated alumina, and 
other metallic peroxids of the aluminous class, when they 
exist in the soluble form; with starch, dextrin, and the 
gums, caramel, tannin, albumen, gelatin, vegetable, and 
animal extractive matters. Low diffusibility is not the 
only property which the bodies last enumerated possess in 
common. They are distinguished by the gelatinous 
character of their hydrates. Although often largely 
soluble in water, they are held in solution by a most feeble 
force. They appear singularly inert in the capacity of 
acids and bases, and in all the ordinary chemical relations. 
But, on the other hand, their peculiar physical aggregation 
with the chemical indifference referred to, appears to be 
required in substances that can intervene in the organic 
processes of life. The plastic elements of the animal body 
are found in this class. As gelatin appears to be its 
type, it is proposed to designate substances of the class 
as colloids, and to speak of their peculiar form of aggrega- 
tion as the colloidal condition of matter. Opposed to the 
colloidal is the crystalline condition. Substances affecting 
the latter form will be classed as crystalloids. The dis- 
tinction is no doubt one of intimate molecular constitution. 
"Although chemically inert in the ordinary sense, 
colloids possess a compensating activity of their own, 
arising out of their physical properties. While the rigidity 
of the crystalline tructure shuts out external impressions, 
the softness of the gelatinous colloid partakes of fluidity, 
and enables the colloid to become a medium for liquid 
diffusion, like water itself. The same penetrability appears 
to take the form of cementation in such colloids as can LIQUID DIFFUSION APPLIED TO ANALYSIS 
33 
exist at a high temperature. Hence a wide sensibility on 
the part of colloids to external agents. Another and emi- 
nently characteristic quality of colloids, is their mutability. 
Their existence is a continued metastasis. A colloid may 
be compared in this respect to water, while existing liquid 
at a temperature under its usual freezing-point, or to a 
supersaturated saline solution. Fluid colloids appear to 
have always a pectous modification; and they often pass 
under the slightest influences from the first into the second 
condition. The solution of hydrated silicic acid, for 
instance, is easily obtained in a state of purity, but it 
cannot be preserved. It may remain fluid for days or 
weeks in a sealed tube, but is sure to gelatinize and become 
insoluble at last. Nor does the change of this colloid 
appear to stop at that point. For the mineral forms of 
silicic acid, deposited from water, such as flint, are often 
found to have passed, during the geological ages of their 
existence, from the vitreous or colloidal into the crystalline 
condition. (H. Rose). The colloidal is, in fact, a dynam- 
ical state of matter; the crystalloidal being the statical 
condition. The colloid possesses Energia. It may be 
looked upon as the probable primary source of the force 
appearing in the phenomena of vitality. To the gradual 
manner in which colloidal changes take place (for they 
always demand time as an element), may the character- 
istic protraction of chemico-organic changes also be 
referred." 1 (Pp. 183-184). 
1 Some of Graham's general statements are at the present day open 
to criticism. Thus it is not quite right to state that colloids are 
incapable of passing over into the crystalline condition. Not all colloids 
give jellies; van Bemmelen has also shown that in the case of the 
hydrogels there is hardly a real chemical hydrate. Chemical inactivity 
and the slowness of reaction often referred to by Graham are not 
always characteristic of colloids. Objection might also be made to 
many other statements, but it should, however, be remembered that 34 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
Graham then describes the simple jar diffusion experi- 
ment and remarks that separation by diffusion is due to 
the property possessed by crystalloids of easily passing 
through jelly-like masses of mucus, starch, gelose, etc., 
which offer great resistance to or completely prevent the 
passage of colloids. For instance, with a mixture of sugar 
and gum arabic, thin French letter paper will practically 
allow only the former to pass through. This separation is 
analogous to that of carbonic acid and hydrogen by a thin 
film of water. He then continues: 
"It may perhaps be allowed to me to apply the con- 
venient term dialysis to the method of separation by 
diffusion through a septum of gelatinous matter. The 
most suitable of all substances for the dialytic septum 
appears to be the commercial material known as vegetable 
parchment, or parchment-paper, which was first pro- 
duced by M. Gaine, and is now successfully manufactured 
by Messrs. De la Rue." (P. 186.) 
The experiments made by the jar-diffusion method, 
the importance of which in physical chemistry has long 
been known, are described in detail on pages 186-199. 
Graham then takes up dialysis proper, describes the 
dialyzer and discusses the precautionary measures to 
be taken in carrying out the process. (Pp. 199-204.) 
Preparation of Colloid Substances by Dialysis 
"The purification of many colloid substances may be 
effected with great advantage by placing them on the 
dialyzer. Accompanying crystalloids are eliminated, and 
the colloid is left behind in a state of purity.1 The purifi- 
Graham was the first to recognize the similarity underlying the great 
variety of colloids, and also that his papers were written more than 
forty years ago. 
1 This should not be regarded as meaning that by dialysis a com- LIQUID DIFFUSION APPLIED TO ANALYSIS 
35 
cation of soluble colloids can rarely be effected by any 
other known means, and dialysis is evidently the appro- 
priate mode of preparing such substances free from 
crystalloids. 
"Soluble Silicic Acid. A solution of silica is obtained 
by pouring silicate of soda into diluted hydrochloric 
acid, the acid being maintained in large excess. But in 
addition to hydrochloric acid, such a solution contains 
chlorid of sodium, a salt which causes the silica to gela- 
tinize when the solution is heated, and otherwise modifies 
its properties. Now such soluble silica, placed for twenty- 
four hours in a dialyzer of parchment-paper, to the usual 
depth of 10 millimetres, was found to lose in that time 
5 per cent of its silicic acid and 86 per cent of its hydro- 
chloric acid. After four days on the dialyser, the liquid 
ceased to be disturbed by nitrate of silver. All the 
chlorids were gone, with no further loss of silica. In 
another experiment 112 grams of silicate of soda, 67.2 
grams of dry hydrochloric acid, and 1000 cc. of water were 
brought together, and the solution placed upon a hoop 
dialyzer, 10 inches in diameter. After four days the solu- 
tion had increased to 1235 cc, by the action of osmose, 
colloid bodies being generally highly osmotic. The solution 
now gave no precipitate with nitrate of silver, and con- 
tained 60.5 grams of silica, 6.7 grams of that substance 
having been lost. The solution contained 4.9 per cent 
of silicic acid.1 
plete elimination of crystalloids is always secured, although the puri- 
fication is so perfect that the impurities often fall below the limit of 
analytical determination (Zs). 
1 Jordis and Kanter (Zeitschr. f. anorg. Chemie., 1903, Vol. XXXV, 
p. 18), have found, contrary to Graham, that in the dialysis of colloidal 
silicic acid, the chlofin reaction of the contents of the dialyser taken 
from one to three weeks to disappear. Variations in the quality of the 
parchment membranes and in the conditions of the dialysis influence 36 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
"The pure solution of silicic acid so obtained may be 
boiled in a flask and considerably concentrated, without 
change; but when heated in an open vessel a ring of 
insoluble silica is apt to form round the margin of the 
liquid, and soon causes the whole to gelatinize. The pure 
solution of hydrated silicic acid is limpid and colorless, and 
not in the least degree viscous, even with 14 per cent of 
silicic acid. The solution is the more durable the longer 
it has been dialyzed and the purer it is. But this solution 
is not easily preserved beyond a few days, unless consider- 
ably diluted. It soon appears slightly opalescent, and 
after a time the whole becomes pectous somewhat rapidly, 
forming a solid jelly, transparent and colorless, or slightly 
opalescent and no longer soluble in water. This jelly 
undergoes a contraction after a few days, even in a close 
vessel, and pure water separates from it. The coagulation 
of the silicic acid is effected in a few minutes by a solu- 
tion containing 1/lO,OOOth part of any alkaline or earthy 
carbonate, but not by caustic ammonia, nor by neutral or 
acid salts. Sulphuric, nitric, and acetic acids do not 
coagulate silicic acid, but a few bubbles of carbonic acid 
passed through the solution produce that effect after the 
lapse of a certain time. Alcohol and sugar, in large quan- 
tity even, do not act as precipitants; but neither do they 
protect silicic acid from the action of alkaline carbonates, 
nor from the effect of time in pectizing the fluid colloid. 
Hydrochloric acid gives stability to the solution; so docs 
.a small addition of caustic potash or soda. 
"This pure water-glass is precipitated on the surface 
its course in a high degree, and perhaps this is the cause of their lack 
of agreement with Graham. Neither in the English nor German text 
of Graham's work does it appear that he tested for chlorin only the 
exterior water and not the dialyser contents, as Jordis and Kanter 
seem to assume. LIQUID DIFFUSION APPLIED TO ANALYSIS 
37 
of a calcareous stone without penetrating, apparently 
from the coagulating action of soluble lime-salts. The 
hydrated silicic acid then forms a varnish, which is apt 
to scale off on drying. The solution of hydrated silicic 
acid has an acid reaction somewhat greater than that of 
carbonic acid. It appears to be really tasteless (like 
most colloids), although it occasions a disagreeable per- 
sistent sensation in the mouth after a time, probably from 
precipitation. 
"Soluble hydrated silicic acid, when dried in the air- 
pump receiver, at 15°, formed a transparent glassy mass 
of great luster, which was no longer soluble in water. It 
retained 21.99 per cent of water after being kept two 
days over sulphuric acid. 
"The colloidal solution of silicic acid is precipitated by 
certain other soluble colloids, such as gelatin, alumina, 
and peroxid of iron, but not by gum nor caramel. As 
hydrated silicic acid, after once gelatinizing, cannot be 
made soluble again by either water or acids, it appears 
necessary to admit the existence of two allotropic modifi- 
cations of that substance, namely, soluble hydrated silicic 
acid, and insoluble hydrated silicic acid, the fluid and 
pectous forms of this colloid." (Pp. 204-205). 
Ordinary sodium silicate is not a colloid; it diffuses 
through membranes. Colloidal silicic acid has an acid 
reaction which can be neutralized by alkalis.1 
"The acid reaction of 100 parts of soluble silicic acid 
is neutralized by 1.85 part of oxid of potassium, and by 
corresponding proportions of soda and ammonia. The 
coZZAsilicates or co-silicates thus formed are soluble and 
more durable than fluid silicic acid, but they are pectized 
by carbonic acid or by an alkaline carbonate, after standing 
for a few minutes. The co-silicate of potash forms a 
1 See Jordis (Zeitschr. f. Electrochemie, 1902, No. 3G.) 38 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
transparent hydrated film on drying in vacuo, which is not 
decomposed by water, and appears to require about ten 
thousand parts of water to dissolve it. The silicate of 
soda which Forchhammer obtained by boiling freshly 
precipitated silicic acid with carbonate of soda, and 
collecting the precipitate which falls on cooling, contains 
2.74 per cent of soda, and is represented by NaO + 36SiO2 
(Gmelin). This silicate is probably a co-silicate of soda in 
the pectous condition. Soluble silicic acid produces a 
gelatinous precipitate in lime-water, containing six per cent 
and upwards of the basic earth. This and the other 
insoluble earthy co-silicates appear not to be easily obtained 
in a definite state. They gave out a more basic silicate to 
water on washing. The composition of these salts and that 
also of the co-silicate of gelatin, were found to vary 
according as the mode of preparation was modified. When 
a solution of gelatin was poured into silicic acid in excess, 
the co-silicate of gelatin formed gave, upon analysis, 
100 silicic acid with 56 gelatin, or a little more than half 
the gelatin stated above as found in that compound pre- 
pared with the mode of mixing the solutions reversed. 
The gallo-tannate of gelatin is known to offer the same 
variability in composition." (P. 206). 
" Soluble Alumina. We are indebted to Mr. Walter 
Crum for the interesting discovery that alumina may be 
held in solution by water alone in the absence of any acid. 
But two soluble modifications of alumina appear to exist, 
alumina and metalumina. The latter is Mr. Crum's 
substance. 
"A solution of the neutral chlorid of aluminium (A12C13), 
placed on the dialyzer, appears to diffuse away without 
decomposition. But when an excess of hydrated alumina 
is previously dissolved in the chlorid, the latter salt is 
found to escape by diffusion in a gradual manner, and the LIQUID DIFFUSION APPLIED TO ANALYSIS 
39 
hydrated alumina, retaining little or no acid, to remain 
behind in a soluble state. A solution of alumina in chlorid 
of aluminium, consisting at first of 52 parts of alumina 
to 48 of hydrochloric acid after a dialysis of six days, con- 
tained 66.5 per cent of alumina; after eleven days 76.5 per 
cent; after seventeen days 92.4 per cent; and after twenty- 
five days the alumina appeared to be as nearly as possible 
free from acid, as traces only of hydrochloric acid were 
indicated by an acid solution of nitrate of silver. But in 
such experiments the alumina often pectizes in the dialyzer 
before the hydrochloric acid has entirely escaped. 
■'Acetate of alumina with an excess of alumina gave 
similar results. The alumina remained fluid in the dialyzer 
for twenty-one days, and when it pectized was found to 
retain 3.4 per cent of acetic acid, which is in the propor- 
tion of one equivalent of acid to 28.2 equivalents of alumina. 
"Soluble alumina is one of the most unstable of sub- 
stances, a circumstance which fully accounts for the 
difficulty of preparing it in a state of purity. It is coagu- 
lated or pectized by portions, so minute as to be scarcely 
appreciable, of sulphate of potash and, I believe, by all 
other salts; and also by ammonia. A solution containing 
two or three per cent of alumina was coagulated by a few 
drops of well-water, and could not be transferred from one 
glass to another, unless the glass was repeatedly washed 
out by distilled water, without gelatinizing. Acids in 
small quantity also cause coagulation; but the precipi- 
tated alumina readily dissolves in an excess of the acid. 
The colloids gum and caramel also act as precipitants. 
"This alumina is a mqrdant, and possesses indeed all 
the properties of the base of alum and the ordinary alu- 
minous salts. A solution containing one half per cent of 
alumina may be boiled without gelatinizing, but when 
concentrated to half its bulk it suddenly coagulated. 40 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
Soluble alumina gelatinizes when placed upon red litmus 
paper and forms a faint blue ring about the drop, showing 
a feeble alkaline reaction. Soluble alumina is not pre- 
cipitated by alcohol nor by sugar. No pure solution of 
alumina, although dilute, remained fluid for more than a 
few days. 
"Like hydrated silicic acid, then, the colloid alumina 
may exist either fluid or pectous, or it has a soluble and 
insoluble form, the latter being the gelatinous alumina as 
precipitated by bases. It is evident that the extraordinary 
coagulating action of salts upon hydrated alumina must 
prevent the latter substance from ever appearing in a 
soluble state when liberated from combination by means 
of a base." (P. 207.) 
"Soluble metalumina" can be produced, according to 
Crum's method, by boiling off the acetic acid from a solu- 
tion of aluminium acetate, or according to Graham by 
dialyzing a solution of aluminium acetate altered by pro- 
longed boiling. After six days the contents of the dialyzer 
still contains some acetic acid. 
"The alumina exists in an allotropic condition, being no 
longer a mordant; and forming, when precipitated, a jelly 
that is not dissolved by an excess of acid. Metalumina 
resembles alumina in being coagulated by minute propor- 
tions of acids, bases, and of most salts. Mr. Crum found 
the solution of metalumina to require larger quantities of 
acetates, nitrates, and chlorids to produce coagulation 
than of the former substances. The solution of metalu- 
mina is tasteless, and entirely neutral to test-paper, 
according to my own observation. 
"Like alumina, the present colloid has therefore a fluid 
and a pectous form, the liquid soluble metalumina, and 
the gelatinous insoluble metalumina." (P. 208.) 
Graham then describes soluble ferric oxid and sesqui- LIQUID DIFFUSION APPLIED TO ANALYSIS 
41 
oxid, which are analogous to the corresponding aluminium 
colloids. The hydrosol of copper ferrocyanid is formed by 
mixing very dilute solutions of potassium ferrocyanid with 
copper sulphate (1:3000) and then dialyzing. Precipi- 
tated copper ferrocyanid is insoluble in potassium oxalate 
and oxalic acid, but is soluble in a quarter of its weight of 
neutral ammonium oxalate. A solution containing 3-4 
per cent of copper ferrocyanid is dark red-brown in color, 
transparent to transmitted light, but cloudy to reflected 
light. Iron sesquioxid has the same appearance. On 
dialysis a little ammonium salt is strongly retained. 
The solutions of copper ferrocyanid can be heated, but 
are easily made pectous by foreign substances, such as 
nitric, hydrochloric, or sulphuric acid (and also after 
gentle heating with oxalic and tartaric acid). Acetic 
acid does not have this effect. Metallic salts coagulate 
the solution. The colloidal solution of Prussian blue is 
made pectous by zinc sulphate, etc., but requires a large 
quantity of alkali salts for precipitation. Graham further 
describes soluble compounds of sugar with copper oxid, 
iron oxid, uranium oxid, and of sugar with lime. The 
latter upon heating forms a solid coagulum. The solution 
of copper tartarate in alkali is also colloidal. (Pp. 208- 
212.) 
Soluble chromium oxid (p. 212) is produced similarly to 
soluble alumina; after thirty-eight days the solution in 
the dialyzer was partly gelatinous and contained 1.5 parts 
acid to 98.5 oxid or one equivalent of acid to 31.2 equiva- 
lents of chromium oxid. Traces of salts make it gela- 
tinous, and then it is insoluble even in hot water. 
"It appears, then, that the hydrated peroxids of the 
aluminous type, when free, are colloid bodies; that two 
species of each of these hydrated oxids exist, of which 
alumina and metalumina are the types; one derived from 42 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
an unchanged salt, and the other from the heated acetate 
of the base; further, that each of these species has two 
forms, one soluble and the other insoluble or coagulated. 
This last species of duality should be well distinguished 
from the preceding allotropic variability of the same 
peroxid. The possession of a soluble and an insoluble 
(fluid and pectous) modification is not confined to hydrated 
silicic acid and the aluminous oxids, but appears to be 
very general, if not universal, among colloid substances. 
The double form is typified in the fibrin of blood." 
(Pp. 212, 213.) 
The ammoniacal solutions of copper and zinc oxids 
diffuse through a colloidal septum and are therefore not 
to be considered colloids. 
Dialysis of organic colloid substances (pp. 213-217). 
Tannic acid passes through parchment paper about 200 
times slower than sodium chlorid; gum arabic 400 times 
slower. 
"The separation of colloids from crystalloids by dialysis 
is, in consequence, generally more complete than might 
be expected from the relative diffusibility of the two 
classes of substances." (P. 214.) 
Vegetable gum, according to Fremy, is a gummate of 
lime, and after the addition of HC1, can be freed from lime 
by dialysis. One hundred parts of the remaining gummic 
acid was neutralized by 2.85 parts of potash. This quan- 
tity of potash is about equivalent to the lime originally 
present in the gum. The gummic acid, if dried at 100°, 
becomes insoluble in water. Gummic acid and glue give 
oily drops, which upon standing yield a colorless 
jelly, which becomes fluid at as low a temperature as 
25° C. 
Dextrin is colloidal, but slowly diffuses through animal 
mucus. LIQUID DIFFUSION APPLIED TO ANALYSIS 
43 
Caramel. Crude caramel, produced by heating raw 
sugar to 210-220°, when dialyzed allows a colored substance 
to pass through, while the substance richest in carbon 
remains behind. A ten per cent solution of this substance 
is gum-like and forms a weak jelly completely soluble in 
water. On evaporation in a vacuum, it dries to a black, 
shining mass, which still contains water and is tough and 
elastic. When thoroughly dry it can be heated to 120° 
and still remains completely soluble. If, however, the 
first solution is evaporated to dryness on the water-bath, 
it becomes insoluble. Soluble and insoluble caramel have 
the same constitution, represented by the empirical 
formula, C24H15O15. Liquid caramel is quite tasteless, 
neutral in reaction, and is extremely sensitive to crystal- 
loid reagents. Traces of mineral acids, alkali salts, and 
alcohol make it pectous; the precipitated caramel yields a 
brownish-black powdery substance insoluble in hot or 
cold water. "The presence of sugar and of the inter- 
mediate brown substances protects the liquid caramel in 
a remarkable degree from the action of crystalloids and 
accounts for the preceding properties not being observed 
in crude caramel."1 Pectous caramel is easily rendered 
soluble again by dilute potash, in which it becomes gela- 
tinous and then dissolves on heating. By acetic acid 
and dialysis, the potash can be removed once more, and 
pure, soluble caramel remains in the dialyzer. Caramel is 
600 times less dialyzable than sodium chlorid and 200 
times less than sugar. Graham refers to the analogy 
between caramel and anthracite coal: "Caramelization 
1 This is the first example of a protective action, recognized as such, 
and it is worthy of remark that substances capable of diffusion pro- 
duce the same effect. I should here state that caramel acts similarly 
to colloidal gold, from which it differs, among other points, in the 
water-solubility of its dry residue (Zs). 44 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
appears the first step in that direction-the beginning 
of a colloidal transformation to be consummated in the 
slow lapse of geological ages." (P. 216.) 
Albumen. Acetic acid was added to a solution of 
egg albumen, and the solution dialyzed; entirely ash- 
free albumen remained behind, which had a weak acid 
reaction. Pure albumen prepared according to Wurtz's 
method is 1000 times less dialyzable than sodium chloric!; 
if to it is added its weight of sodium hydrate, the lat- 
ter alone will dialyze from the solution; and its separa- 
tion from the albumen is complete. A film of albumen 
coagulated by heat is totally impermeable to albumen 
solution. 
Neither gelatinized starch, nor an aqueous solution of 
animal gelatin, nor meat extract, diffuses through colloid 
septa. 
Graham thereafter describes the well-known separa- 
tion of arsenious acid from colloidal fluids by dialysis. 
(Pp. 217-219.) 
"Colloidal Condition of Matter. I may be allowed to 
advert again to the radical distinction assumed in this 
paper to exist between colloids and crystalloids in their 
intimate molecular constitution. Every physical and 
chemical property is characteristically modified in each 
class. They appear like different worlds of matter, and 
give occasion to a corresponding division of chemical 
science. The distinction between these kinds of matter 
is that subsisting between the material of a mineral 
and the material of an organized mass. 
"The colloidal character is not obliterated by lique- 
faction, and is therefore more than a modification of the 
physical condition of solids. Some colloids are soluble in 
water, as gelatin and gum-arabic; and some arc insoluble, 
like gum-tragacanth. Some colloids, again, form solid LIQUID DIFFUSION APPLIED TO ANALYSIS 
45 
compounds with water, as gelatin and gum-tragacanth, 
while others, like tannin, do not. In such points the 
colloids exhibit as great a diversity of property as the 
crystalloids. A certain parallelism is maintained between 
the two classes, notwithstanding their differences. 
"The phenomena of the solution of a salt or crystalloid 
probably all appear in the solution of a colloid, but 
greatly reduced in degree. The process becomes slow; 
time, indeed, appearing essential to all colloidal changes. 
The change of temperature, usually occuring in the act 
of solution, becomes barely perceptible. The liquid is 
always sensibly gummy or viscous when concentrated.' 
The colloid, although often dissolved in a large propor- 
tion by its solvent, is held in solution by a singularly 
feeble force. Hence colloids are generally displaced and 
precipitated by the addition to their solution of any sub- 
stance from the other class. Of all the properties of 
liquid colloids, their slow diffusion in water, and their 
arrest by colloidal septa, are the most serviceable in dis- 
tinguishing them from crystalloids. Colloids have feeble 
chemical reactions, but they exhibit at the same time a 
very general sensibility to liquid reagents, as has already 
been explained. 
"While soluble crystalloids are always highly sapid, 
soluble colloids are singularly insipid. It may be ques- 
tioned whether a colloid, when tasted, ever reaches the 
sentient extremities of the nerves of the palate, as the 
latter are probably protected by a colloidal membrane, 
impermeable to soluble substances of the same physical 
constitution. (P. 220.) 
"A tendency to spontaneous change, which is observed 
occasionally in crystalloids, appears to be general in the 
other class. The fluid colloid becomes pectous and 
insoluble by contact with certain other substances, with- 46 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
out combining with these substances, and often under the 
influence of time alone. The pectizing substance appears 
to hasten merely an impending change. Even while 
fluid a colloid may alter sensibly, from colorless becoming 
opalescent; and while pectous the degree of hydration 
may become reduced from internal change. The gradual 
progress of alteration in the colloid effected by the 
agency of time, is an investigation yet to be entered 
upon. 
"The equivalent of a colloid appears to be always high, 
although the ratio between the elements of the substance 
may be simple. Gummic acid, for instance, may be 
represented by C12H11O11, but judging from the small 
proportions of lime and potash which suffice to neutralize 
this acid, the true numbers of its formula must be sev- 
eral times greater. It is difficult to avoid associating 
the inertness of colloids with their high equivalents, par- 
ticularly where the high number appears to be attained 
by the repetition of a smaller number. The inquiry 
suggests itself Whether the colloid molecule may not be 
constituted by the grouping together of a number of 
smaller crystalloid molecules, and whether the basis of 
colloidality may not really be this composite character 
of the molecule. 
"With silicic acid, which can exist in combination 
both as a crystalloid and colloid, we have two series of 
compounds, silicates and cosilicates, the acid of the 
latter appearing to have an equivalent much greater 
(thirty-six times greater in one salt) than the acid of the 
former. The apparently small proportion of acid in a 
variety of metallic salts, such as certain red salts of iron, 
is accounted for by the high colloidal equivalent of their 
bases. The effect of such an insoluble colloid as Prus- 
sian blue in carrying down small proportions of the pre- ON THE PROPERTIES OF SILICIC ACID 
47 
cipitating salts, may admit of a similar explanation. (P. 
221). 
"The hardness of the crystalloid, with its crystalline 
planes and angles, is replaced in the colloid by a degree 
of softness, with a more or less rounded outline. The 
water of crystallization is represented by the water of 
gelatination. The water in gelatinous hydrates is aptly 
described by M. Chevreul as retained by 1 capillary affin- 
ity/ that is, by an attraction partaking both of the phy- 
sical and chemical character." (P. 222). 
Graham, at the conclusion of his important work, 
gives an explanation of osmosis, which in the main agrees 
with the views of Liebig, published some time previously. 
The contents of another valuble work of Graham will 
for the most part be given in the author's own words. 
" On the Properties of Silicic Acid and other Analogous 
Colloidal Substances " 
By Thomas Graham, F.R.S., Master of the Mint 
(Received June 16, 1864) 1 
PRELIMINARY NOTICE 
"The prevalent notions respecting solubility have been 
derived chiefly from observations on crystalline salts, and 
are very imperfectly applicable to the class of colloidal 
substances. Hydrated silicic acid, for instance, when in 
the soluble condition, is, properly speaking, a liquid body, 
like alcohol, miscible with water in all proportions. We 
have no degrees of solubility to speak of with respect to 
silicic acid, like the degrees of solubility of a salt, unless 
it be with reference to silicic acid in the gelatinous con- 
1 Thomas Graham, Proceedings of the Royal Society, June 16, 
1864. (Also Pogg, Ann., 1864, Vol. CXXIII, pp. 529-541.) 48 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
dition, which is usually looked upon as destitute of solu- 
bility. The jelly of silicic acid may be more or less rich 
in combined water, as it is first prepared, and it appears 
to be soluble in proportion to the extent of its hydration. 
A jelly containing 1 per cent of silicic acid, gives with 
cold water a solution containing about 1 of silicic acid 
in 5000 water; a jelly containing 5 per cent of silicic 
acid gives a solution containing about 1 part of acid in 
10,000 water. A less hydrated jelly than the last men- 
tioned is still less soluble; and finally, when the jelly is 
rendered anhydrous, it gives gummy-looking, white 
masses, which appear to be -absolutely insoluble, like 
the light dusty silicic acid obtained by drying a jelly 
charged with salts, in the ordinary analysis of a silicate. 
"The liquidity of silicic acid is only affected by a 
change, which is permanent (namely, coagulation or pec- 
tization), by which the acid is converted into the gela- 
tinous or pectous form, and loses its miscibility with 
water. The liquidity is permanent in proportion to the 
degree of dilution of silicic acid, and appears to be favored 
by a low temperature. It is opposed, on the contrary, 
by concentration, and by elevation of temperature. A 
liquid silicic acid of 10 or 12 per cent pectizes spontane- 
ously in a few hours at the ordinary temperature, and 
immediately when heated. A liquid of 5 per cent may 
be preserved for five or six days; a liquid of 2 per cent 
for two or three months; and a liquid of 1 per cent has 
not pectized after two years. Dilute solutions of 0.1 per 
cent or less are no doubt practically unalterable by time, 
and hence the possibility of soluble silicic acid existing 
in nature. I may add, however, that no solution, weak 
or strong, of silicic acid in water has shown any disposi- 
tion to deposit crystals, but always appears on drying as 
a colloidal glassy hyalite. The formation of quartz cry- ON THE PROPERTIES OF SILICIC ACID 
49 
stals at a low temperature, of so frequent occurrence in 
nature, remains still a mystery. I can only imagine that 
such crystals are formed at an inconceivably slow rate, 
and from solutions of silicic acid which are extremely 
dilute. Dilution no doubt weakens the colloidal char- 
acter of substances, and may therefore allow their crys- 
tallizing tendency to gain ground and develop itself, 
particularly where the crystal once formed is completely 
insoluble, as with quartz. 
"The pectization of liquid silicic acid is expedited by 
contact with solid matter in the form of powder. By 
contact with pounded graphite, which is chemically 
inactive, the pectization of a 5 per cent silicic acid is 
brought about in an hour or two, and that of a 2 per 
cent silicic acid in two days. A rise of temperature of 
l°.l C. was observed during the formation of the 5 per 
cent jelly. 
"The ultimate pectization of silicic acid is preceded 
by a gradual thickening in the liquid itself. The flow 
of liquid colloids through a capillary tube is always slow 
compared with the flow of crystalloid solutions, so that a 
liquid-transpiration-tube may be employed as a colloido- 
scope. With a colloidal liquid alterable in viscosity, 
such as silicic acid, the increased resistance to passage 
through the colloidoscope is obvious from day to day. 
Just before gelatinizing, silicic acid flows like an oil. 
"A dominating quality of colloids is the tendency of 
their particles to adhere, aggregate, and contract. This 
idio-attraction is obvious in the gradual thickening of 
the liquid, and when it advances leads to pectization. In 
the jelly itself, the specific contraction in question, or 
synaeresis, still proceeds, causing separation of water, 
with the division into a clot and serum; and ending in 
the production of a hard stony mass, of vitreous struc- 50 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
ture, which may be anhydrous, or nearly so, when the 
water is allowed to escape by evaporation. The intense 
synaeresis of isinglass dried in a glass dish over sulphuric 
acid in vacuo enables the contracting gelatin to tear up 
the surface of the glass. Glass itself is a colloid, and 
the adhesion of colloid to colloid appears to be more 
powerful than that of colloid to crystalloid. The gelatin, 
when dried in the manner described upon plates of calc- 
spar and mica, did not adhere to the crystalline surface, 
but detached itself on drying. Polished plates of glass 
must not be left in contact, as is well known, owing to 
the risk of permanent adhesion between their surfaces. 
The adhesion of broken masses of glacial phosphoric acid 
to each other is an old illustration of colloidal synaeresis. 
"Bearing in mind that the colloidal phasis of matter 
is the result of a peculiar attraction and aggregation of 
molecules, properties never entirely absent from matter, 
but more greatly developed in some substances than in 
others, it is not surprising that colloidal characters 
spread on both sides into the liquid and solid conditions." 
(Pp. 335, 336.) 
Graham goes on to say that certain fluid substances 
which exercise no pectizing influence on colloids, can 
totally or partially displace the combined water of col- 
loidal silicic acid, without causing the coagulation of the 
colloid. Hydrochloric, nitric, acetic, and tartaric acids, 
sugar syrup, glycerin, and alcohol are in this class. He 
then describes the production of such a solution of silicic 
acid in alcohol and the alcogel. 
"The alcogel, or solid compound, is readily prepared 
by placing masses of gelatinous silicic acid, containing 
8 or 10 per cent of the dry acid, in absolute alcohol, and 
changing the latter repeatedly till the water of the hydro- 
gel is fully replaced by alcohol. The alcogel is generally ON THE PROPERTIES OF SILICIC ACID 
51 
slightly opalescent, and is similar in aspect to the hydro- 
gel, preserving very nearly its original bulk. The follow- 
ing is the composition of an alcogel carefully prepared 
from a hydrogel which contained 9.33 per cent of silicic 
acid: 
Alcohol 88.13 
Water 0.23 
Silicic acid 11.64 
100.00 
"Placed in water, the alcogel is gradually decomposed- 
alcohol diffusing out and water entering instead, so that 
a hydrogel is reproduced. (Pp. 337, 338.) 
"The compound of sulphuric acid, sulphagel, is also 
interesting from the facility of its formation, and the 
complete manner in which the water of the original hydro- 
gel is removed. A mass of hydrated silicic acid may be 
preserved unbroken if it is first placed in sulphuric acid 
diluted with two or three volumes of water, and then 
transferred gradually to stronger acids, till at last it is 
placed in concentrated oil of vitriol. The sulphagel 
sinks in the latter fluid, and may be distilled with an 
excess of it for hours without losing its transparency or 
gelatinous character. It is always somewhat less in 
bulk than the primary hydrogel, but not more, to the 
eye, than one fifth or one sixth part of the original vol- 
ume. This sulphagel is transparent and colorless. When 
a sulphagel is heated strongly in an open vessel, the last 
portions of the monohydrated sulphuric acid in com- 
bination are found to require a higher temperature for 
their expulsion than the boiling-point of the acid. The 
whole silicic acid remains behind, forming a white, opaque, 
porous mass, like pumice. A sulphagel placed in water 52 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
is soon decomposed, and the original hydrogel repro- 
duced. No permanent compound of sulphuric and 
silicic acids, of the nature of a salt, appears to be formed 
in any circumstances. A sulphagel placed in alcohol 
gives ultimately a pure alcogel. Similar jellies of silicic 
acid may readily be formed with the monohydrates of 
nitric, acetic, and formic acids, and are all perfectly 
transparent. 
"The production of the compounds of silicic acid now 
described indicates the possession of a wider range of 
affinity1 by a colloid than could well be anticipated. 
The organic colloids are no doubt invested with similar 
wide powers of combination, which may become of 
interest to the physiologist. The capacity of a mass of 
gelatinous silicic acid to assume alcohol, or even olein, 
in the place of water of combination, without disinte- 
gration or alteration of form, may perhaps afford a clue 
to the penetration of the albuminous matter of mem- 
brane by fatty and other insoluble bodies, which seems 
to occur in the digestion of food. Still more remarkable 
and suggestive are the fluid compounds of silicic acid. 
The fluid alcohol or olein compound favors the possi- 
bility of the existence of a compound of the colloid albu- 
men with olein, soluble also and capable of circulating 
with the blood. 
"The feebleness of the force which holds together two 
substances belonging to different physical classes, one 
being a colloid and the other a crystalloid, is a subject 
deserving notice. When such a compound is placed in 
a fluid the superior diffusive energy of the crystalloid may 
cause its separation from the colloid. Thus, of hydrated 
1 This is not an instance of chemical affinity and chemical combina- 
tion in the now adopted meaning of the word. See further on in this 
Chapter (Zs). ON THE PROPERTIES OF SILICIC ACID 
53 
silicic acid, the combined water (a crystalloid) leaves 
the acid (a colloid) to diffuse into alcohol; and if the 
alcohol be repeatedly changed, the entire water is thus 
removed, alcohol (another crystalloid) at the same time 
taking the place of water in combination1 with the silicic 
acid. The liquid in excess (here the alcohol) gains 
entire possession of the silicic acid. The process is 
reversed if an alcogel be placed in a considerable volume 
of water. Then alcohol separates from combination, in 
consequence of the opportunity it possesses to diffuse 
into water; and water, which is now the liquid present 
in excess, recovers possession of the silicic acid. Such 
changes illustrate the predominating influence of mass. 
"Even the compounds of silicic acid with alkalies yield 
to the decomposing force of diffusion. The compound 
of silicic acid with one or two per cent of soda is a col- 
loidal solution, and, when placed in a dialyzer over water 
in vacuo to exclude carbonic acid, suffers gradual decom- 
position. The soda diffuses off slowly in the caustic 
state, and gives the usual brown oxicl of silver when 
tested with the nitrate of that base. 
"The pectization of liquid silicic acid and many other 
liquid colloids is effected by contact with minute quan- 
tities of salts in a way which is not understood. On the 
other hand, the gelatinous acid may again be liquefied, 
and have its energy restored by contact with a very 
moderate amount of alkali. The latter change is grad- 
ual, 1 part of caustic soda, dissolved in 10,000 water, 
liquefying 200 parts of silicic acid (estimated dry), in 
60 minutes at 100° C. Gelatinous stannic acid also is 
easily liquefied by a small proportion of alkali, even at 
the ordinary temperature. The alkali, too, after lique- 
fying the gelatinous colloid, may be separated again 
1 See note on p. 52. 54 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
from it by diffusion into water upon a dialyzer. The 
solution of these colloids, in such circumstances, may be 
looked upon as analogous to the solution of insoluble 
organic colloids witnessed in animal digestion, with the 
difference that the solvent fluid here is not acid, but 
alkaline. Liquid silicic acid may be represented as the 
'peptone ' of gelatinous silicic acid; and the liquefac- 
tion of the latter by a trace of alkali may be spoken of as 
the peptization of the jelly. The pure jellies of alumina, 
peroxicl of iron, and titanic acid, prepared by dialysis, 
are assimilated more closely to albumen, being peptized 
by minute quantities of hydrochloric acid. 
"Liquid Stannic and Metastannic Acids. Liquid stan- 
nic acid is prepared by dialyzing the bichlorid of tin 
with an addition of alkali, or by dialyzing the stannate 
of soda with an addition of hydrochloric acid. In both 
cases a jelly is first formed on the dialyzer; but, as the 
salts diffuse away, the jelly is again peptized by the 
small proportion of free alkali remaining; the alkali 
itself may be removed by continued diffusion, a drop 
or two of the tincture of ioclin facilitating the separation. 
The liquid stannic acid is converted on heating it into 
liquid metastannic acid. Both liquid acids are remark- 
able for the facility with which they are pectized by a 
minute addition of hydrochloric acid, as well as by salts. 
"Liquid Titanic Acid is prepared by dissolving gela- 
tinous titanic acid in a small quantity of hydrochloric 
acid, without heat, and placing the liquid upon a dialyzer 
for several days. The liquid must not contain more 
than one per cent of titanic acid, otherwise it spontane- 
ously gelatinizes, but it appears more stable when dilute. 
Both titanic and the two stannic acids afford the same 
classes of compounds with alcohol, etc., as are obtained 
with silicic acid. COLLOIDAL SULPHIDS AND METALS 
55 
11 Liquid Tungstic Acid. The obscurity which has so 
long hung over tungstic acid is removed by a dialytic 
examination. It is, in fact, a remarkable colloid, of 
which the pectous form alone has hitherto been known. 
Liquid tungstic acid is prepared by adding dilute hydro- 
chloric acid carefully to a five per cent solution of tungstate 
of soda, in sufficient proportion to neutralize the alkali, 
and then placing the resulting liquid on a dialyzer. In 
about three days the acid is found pure, with the loss of 
about twenty per cent, the salts having diffused entirely 
away. It is remarkable that the purified acid is not 
pectized by acids or salts even at the boiling temperature. 
Evaporated to dryness, it forms vitreous scales, like 
gum or gelatin, which sometimes adhere so strongly to 
the surface of the evaporating dish as to detach portions 
of it. It may be heated to 200° C. without losing its 
solubility or passing into the pectous state, but at a 
temperature near redness it undergoes a molecular 
change, losing at the same time 2.42 per cent of water. 
When water is added to unchanged tungstic acid it 
becomes pasty and adhesive like gum; and it forms a 
liquid with about one fourth its weight of water, which 
is so dense as to float glass. The solution effervesces 
with carbonate of soda, and tungstic acid is evidently 
associated with silicic and molybdic acids. The taste 
of tungstic acid dissolved in water is not metallic or 
acid, but rather bitter and astringent. Solutions of 
tungstic acid containing 5, 20, 50, 66.5, and 79.8 per cent 
of dry acid possess the following densities at 19°: 1.0475, 
1.2168, 1.8001, 2.396, and 3.243. Evaporated in vacuo 
liquid tungstic acid is colorless, but becomes green in 
air from the deoxidating action of organic matter. 
Liquid silicic acid is protected from pectizing when 
mixed with tungstic acid, a circumstance probably 56 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
connected with the formation of the double com- 
pounds. 
11 Molybdic Acid has hitherto been known (like tungstic 
acid) only in the insoluble form. Crystallized molybdate 
of soda dissolved in water is decomposed by the gradual 
addition of hydrochloric acid in excess without any 
immediate precipitation. The acid liquid thrown upon 
a dialyzer may gelatinize after a few hours, but again 
liquefies spontaneously, when the salts diffuse away. 
After a diffusion of three days, about 60 per cent of the 
molybdic acid remains behind in a pure condition. The 
solution of pure molybdic acid is yellow, astringent to 
the taste, acid to test-paper, and possesses much sta- 
bility. The acid may be dried at 100°, and then heated 
to 200° without losing its solubility. Soluble molybdic 
acid has the same gummy aspect as soluble tungstic 
acid, and deliquesces slightly when exposed to damp 
air. Both acids lose their colloidality when digested 
with soda, for a short time, and give a variety of crystal- 
lizable salts." (P. 338-341.) 
2. Colloidal Sulphids and Metals: Order of Their Discovery 
If, in our present state of knowledge we consider the 
chief moot question as to the nature of colloidal solu- 
tions, it will at once be seen that the conflict of opinions 
had its origin in the irreversible colloids, and is in part 
due to a mutual misunderstanding. As stated in Chapter 
I, chemists designated as suspensions, only coarse, het- 
erogeneous subdivisions, which settle of their own 
accord, but considered clear non-settling subdivisions as 
solutions; while some physicists called subdivisions of 
this latter class suspensions. Soon, however, fluids were 
found which were intermediate between the suspensions COLLOIDAL SULPHIDS AND METALS 
57 
and the solutions of the chemists. These formed the 
subject of differences of opinion, and at the same time 
were the starting-point of investigations which afforded 
a partial, but still very incomplete, insight into the 
nature of hydrosols. 
Thus Berzelius regarded the hydrosol of arsenic sul- 
phid as a suspension, because of its property of settling 
after standing a while. 
Colloidal Sulphids. TI. Schulze, whom we must 
thank for the first thorough investigation on colloidal 
sulphids (1882), concluded from his observations on 
colloidal arsenic sulphid that this fluid is by no means 
identical with ordinary levigated arsenic sulphid; but 
rather that its action and properties coincide in many 
important points with the hydrosols of ferric oxid, 
aluminium, etc., described by Graham, and therefore had 
no hesitation in calling them solutions. The marked 
cloudiness of his fluid, Schulze considered fluorescence; 
for he should have been able to see suspended particles 
of arsenic sulphid under the microscope, while his fluid 
by microscopic examination appeared perfectly homo- 
geneous.1 In order to explain the existence of solutions 
of the insoluble As2S3, Schulze (just as had Graham), 
assumed the existence of allotropic modifications. 
Schulze's communication on the colloidal sulphids of 
arsenic and antimony were soon followed by further 
work on colloidal sulphids, especially that of Spring 
(1883), and later by that of Winssinger. Spring's article 
on colloidal copper sulphid led Ebcll (1883)2 to point 
1 Schulze cannot be blamed for not differentiating between diffuse 
dispersion and fluorescence, for the method of differentiating these 
two was practically unknown to the chemical fraternity. 
2 Berichte d. Deutschen chem. Gesell., 1883, Vol. XVI, pp. 1142-43; 
also Spring and de Boeck, Bull, de la Soc. chim., 1887 [2], Vol. XLVI1I, 
p. 165. 58 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
out the analogy between colloidal sulphids and the finest 
levigated ultramarine, which no longer settles.1 Ebell 
fortunately had at his disposal a fluid which may be 
considered as an intermediary between hydrosols and 
suspensions. This fluid contained particles which were 
still visible under the microscope, but which upon long 
standing settled only partially, or not at all. In layers 
about 2 cm. thick the fluid was transparent, and appeared 
clear to transmitted light. As is the case with many 
irreversible hydrosols, it left, when evaporated on the 
water bath, a shining varnish which could not be levi- 
gated in water to the original state of fineness; having 
in mind that its particles are microscopically visible, its 
relation to suspensions is indicated by the fact that pre- 
cipitates produced by salts assume their original state 
of subdivision in pure water. (This fluid should be further 
examined ultramicroscopically for the presence of finer 
particles.) Remembering that this fluid is obtained by 
the levigation of the crude product in the second step in 
ultramarine manufacture, and that in the first step clay 
is for a long time roasted with alkali sulphate and car- 
bon, it will not appear improbable that we have here a 
partially coagulated hydrosol, and that the particles 
visible under the microscope are already flocks, formed 
by the clotting together of incomparably much smaller 
particles under the influence of salts, which are at once 
brought into solution upon leaching the melt. Further 
evidence of this is its slow settling and the transparency 
of the fluid, a property common to levigated hydrogels. 
Spring, who originally held with Schulze that colloidal 
sulphids were especial water soluble modifications, later 
agreed with the opinion of Ebell and other investigators, 
that in this case there was an extreme subdivision of the 
1 Berichte d. Deutschen chem. Gesell., 1883, Vol. XVI, pp. 2429-2432. COLLOIDAL SULPHIDS AND METALS 
59 
material; but in addition he made the most important 
observation,1 based upon the work of Picton and Linder, 
and his own work on optically clear water, that many 
other colloidal solutions and also solutions of many crys- 
talloids show the same lack of 'optical homogeneity as 
do the irreversible hydrosols, which were chiefly sus- 
pected of being heterogeneous. 
Colloidal Silver. An excellent experimental investiga- 
tion of the water-soluble modifications of silver was worked 
out by Carey Lea, and at that time it attracted general 
attention. In 1889 Lea succeeded in producing a water- 
soluble substance containing over 95 per cent of silver. 
It was natural to regard this water-soluble form as an 
allotropic modification of the metal. The name chosen 
by Lea assumed as much. 
Prior to Lea, in 1887, 0. Loew 2 and Muthmann 3 had 
prepared ■ colloidal silver, and Muthmann had observed 
that the substances assumed by Wohler and others to 
contain silver suboxid, were mixtures of silver oxids with 
metallic silver. He had furthermore dissolved out with 
ammonia the colloidal silver contained in such reduc- 
tion products, established its constitution, and described 
its properties. 
Muthmann also observed that certain other amorphous 
bodies, such as ferric hydroxid and molybdic dioxid 
hydrate had similar properties to the finely divided sil- 
ver described by him, and he considered it not improb- 
able that the various modifications of silver which he 
noticed betokened variations in molecular condition. 
The connection of this work with Graham's appears to 
1 W. Spring, Bull, de l'Acad. roy. de Belgique, 1899 (Cl. de. sc.), 
No. 4, pp. 300-315. 
2 Loew, Berichte, 1883, Vol. XVI, p. 2707. 
3 Muthmann, Berichte, 1887, Vol. XX, p. 983. 60 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
have escaped Muthmann, for he makes no reference to it 
in his paper, and besides, the only correct terminology 
for his fluids "colloidal solutions or hydrosols" is not 
used by Muthmann. 
In the year 1890 Prange observed that the allotropic 
silver of Carey Lea could be regarded as a water-soluble 
colloidal form of this metal, and gave a series of proofs 
as to the colloidal nature of its solution. Since Prange's 
publication Lea's silver has been generally and properly 
called colloidal silver. 
Conflicting Views Regarding the Nature of Irreversible 
Hydrosols. Barus and Schneider1 in a most thorough 
paper, oppose the idea that colloidal silver is an allotropic 
modification of the metal. Without being quite able to 
disprove the presence of an allotropic modification, botli 
investigators pointed out the analogy between the 
properties of the colloidal silver and those of suspended 
clay particles, and adduced much evidence to show 
that the silver is present in a fine state of subdivision. 
Barus and Schneider regard colloidal solutions as true 
suspensions, in which the sedimentation can be calcu- 
lated from the mechanical laws of falling bodies based 
upon the radius of the particles, viscosity of the medium, 
difference in density between both substances, and the 
acceleration due to gravity. 
They came to the following conclusion: 
"Taking all facts into consideration, we think we may 
conclude that the view that colloidal silver consists of 
extremely finely divided little particles of normal silver 
which may be considered as continuing to float, because 
of the viscosity of the solvent, is in no way contrary to 
the observed properties of colloidal solutions. Inas- 
much as there are hardly any good reasons for the assump- 
1 Zeitschr. f. physikal. Chemie, 1891, Vol. VIII, p. 297. COLLOIDAL SULPHIDS AND METALS 
61 
tion of an allotropic molecule, it is simpler to adhere to 
the normal molecule. The same applies to colloidal 
solutions in general." 1 
From a little known criticism of this work by two 
English scientists, Picton and Linder (1892), who, 
among other things, deserved credit for having shown 
that by continued subdivision of sulphids, colloidal 
solutions exhibiting osmotic pressure and diffusibility 
nearly approaching those of the lower colloids can be 
obtained from mechanical suspensions, we quote only 
one part 2 having reference to the work of Barus and 
Schneider: 
"The evidence upon which most stress is laid in the 
above quoted paper (of Barus and Schneider) seems to 
us to be decidedly inconclusive. It is also noteworthy 
that throughout this paper there is an assumption that 
if the colloidal solutions contain very finely divided 
particles, the solution is a fact of mere mechanical sus- 
pension and nothing more. This is evident, too, from 
their method of calculating the size of the suspended 
particles. An assumption of this kind seems to us quite 
irreconcilable with fact." 
Although Picton and Linder completely agree with 
Barus and Schneider that in colloidal solutions there is 
the most extreme subdivision of the substance, and in 
their beautiful experimental work give much evidence 
therefor, they by no means favor the explanation given 
by the latter. They oppose the conception that col- 
loidal solutions are mere mechanical suspensions, and 
conceive the particles contained in hydrosols to be 
1 Wernicke has also expressed the same idea. Wied. Ann., 1894, 
N. F., Vol. LU, p. 515. 
2 Picton and Linder, Solutions and Pseudo-Solutions, J. of Chem. 
Society, 1892, p. 148. 62 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
large molecular aggregations. (Picton and Linder, p. 
169.) 
Picton and Linder produced four kinds of arsenic 
sulphid: As2S3 a, p, 8, of which a is the largest parti- 
cle (still visible microscopically) and £ is the smallest. 
AS2S3 a, p and f are retained upon filtration through 
a porcelain cell, while 8 is not. In this case no adsorp- 
tion takes place, for if AS2S3 were adsorbed by porce- 
lain, £ would also be retained by it; the particles of 
the first three hydrosols are much too large to pass 
through the filter. 
While hydrosols a and /? do not diffuse, hydrosols 
and 8 exhibit distinct diffusibility. A four per cent 
solution of arsenic sulphid had an osmotic pressure 
equivalent to 17 mm. of water. 
Picton and Linder are- of the opinion that the parti- 
cles which diffuse and produce osmotic pressure are 
identical with those which, upon Tyndall's test, betray 
their existence by polarization. This has an important 
bearing on the objections which can be urged (see this 
chapter further below) against the applicability of the 
osmotic method for the determination of molecular 
weight. 
If diffusibility were due to the presence of crystalloids 
or else to a partial crystalloid solubility of the arsenic 
sulphid, then the same cause would also operate in the 
case of the coarser hydrosols, which, however, show no 
trace of diffusion and no osmotic pressure. 
That in As2S3<1 the whole mass of the arsenic sulpid 
consists of smaller particles than the others, is self- 
evident from the fact that it passes through a clay cell, 
while the other three hydrosols are held back. 
It should be observed that Linder and Picton, even 
in their first communication, pointed out that all col- COLLOIDAL SULPHIDS AND METALS 
63 
loidal sulphids contain a certain quantity of sulphur- 
reted hydrogen which can be partially driven off by 
the addition of acid. Linder and Picton regard this 
sulphureted hydrogen as chemically combined, and 
consider the colloidal sulphids to . be complex chemical 
compounds of sulphid with H2S. Thus on the basis of 
their analyses they assign to cupric sulphid the formula 
9CuS,HS or 22CuS,HaS; to zinc sulphid the formula 
7ZnS,H2O or 12ZnS,H2S; to mercuric sulphid 31HgS,H2S 
and 62HgS,H2S. Further, colloidal arsenic sulphid, for 
which above the abbreviated formula AS2S3 has been 
used, contains combined sulphureted hydrogen in the 
proportion of 8 (or 16) As2S3:1H2S. The solubility of 
some sulphids in sulphureted hydrogen is attributed to 
the formation of complex hydrosulphids. 
Jordis, as is well known, recently adopted a quite 
similar view concerning colloidal silicic acid (and col- 
loids in general), in that he regards, for instance, the 
silicic acid hydrosol to be not a colloidal solution of 
silicic acid, but a solution of a chemical compound of 
the same.1 Besides, even Graham regarded solutions of 
colloidal silicic acid in a little alkali as silicates, and 
called them colli-silicates or co-silicates. 
New Hydrosols of the sulphid group have been made in 
large numbers by Winssinger (almost all of them colloidal 
sulphid solutions), by E. A. Schneider (colloidal gold 
sulphid and tin sulphid among others), colloidal oxids 
by Grimaux, Spring, Biltz, and others. 
Hydrosols of Metals have been produced by Faraday, 
Lea, 0. Loew, Muthmann, Lobry de Bruyn, Zsigmondy, 
Lottermoser (Hg, Bi, Cu, Pt, Pd, Rh), and by Bredig, by 
the very original method of atomizing metal by means 
1 Jordis, loc. cit.. Chapter III above and Chapter XXL 64 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
of an electric arc under water (Pt, Au, Pt, Ag, Cd, etc.). 
The numerous metallic hydrosols which Bredig has pre- 
pared in this manner were used by him in a very valuable 
chemico-contact investigation, the object of which was 
to throw new light on a series of hitherto unexplained 
phenomena which had been observed with organic fer- 
ments and enzymes. The analogy is so close that many 
of the same substances which are known as powerful 
blood poisons, and inhibit the catalytic action of the 
blood, also act as powerful "poisons" in hindering the 
catalytic action of colloidal platinum or gold. 
An important method for the manufacture of high 
percentage and reversible hydrosols is that of Paal,1 
who succeeded in producing among other things a col- 
loidal gold soluble in alkali and containing more than 
93 per cent Au. 
Recent investigators, Gutbier, Kiispert, Henrich, Bil- 
litzer, Blake, Donau, and others, by extending known 
methods, have largely increased the number of methods 
for the production of metallic hydrosols, and among 
other things have shown that numerous reducing agents 
are suitable for this purpose, for example, any energetic 
reducing agent which docs not of itself coagulate the 
hydrosol, or which exercises a protective action (Schutz- 
wirkung), is adaptable for the production of hydrosols 
of the noble metals or analogues of the purple of Cassius, 
providing the essential well-known conditions are observed: 
sufficient dilution and extreme purity of the reagents, 
or the use of protective colloids which prevent the indi- 
vidual particles from coalescing. 
Worthy of note is the method discovered by Donau 
for the production of colloidal gold by bubbling carbon 
1 Paal, Berichte, 1902, Vol. XXXV, p. 223G. COLLOIDAL SULPHIDS AND METALS 
65 
monoxicl through a dilute solution of chlorid of gold, in 
which case no protective colloids can be formed.1 
Purple of Cassius. A whole literature exists regarding 
the ammonia-soluble deep red precipitate known as the 
Purple of Cassius, which is obtained by mixing dilute 
solutions of chlorid of gold and stannous chlorid. As 
this substance has acquired significance because it gives 
an insight into many colloids, I must devote some space 
to it, even if I can mention only the most important 
points.2 
Even in the time of Berzelius there existed two differ- 
ent ideas regarding the nature of this substance. Berze- 
lius himself writes regarding it in his "Text Book of 
Chemistry ":3 
"It is not yet established what is the condition of 
the gold in the purple of Cassius. Some chemists con- 
sider it to be metallic and only mixed with the tin oxid. 
When, however, a gold salt which is mixed with any 
powder is reduced, the mixture gives a brick reel powder 
which does not appear metallic and does not, further- 
more, have the clear, transparent color of the purple; 
and if the purple is roasted in a small retort, it acquires 
the same brick-red color, indicating that the gold, upon 
roasting, has again assumed the metallic form." 
And further (page 245): 
"That the purple contains neither metallic gold nor 
1 The resemblance of this gold solution to those here described 
(Chapter VIII) is very great. The somewhat greater sensitiveness of 
Donau's gold soluton might be ascribed to its larger content of hydrogen 
ions. Donau, Wien. Akad. Ber., Vol. CXIV. Abt. Ila, 1905. 
3 A more detailed discussion will be found in my article: The Chemi- 
cal Nature of the Purple of Cassius. Liebig's Ann., 1898, Vol. CCCI, 
pp. 362-387. 
3 Second Edition, translated by Bldde and Palmstedt, 1823, Vol. II, 
p. 244. 66 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
oxid is evident from the fact that it dissolves in ammo- 
nium hydrate, which would leave undissolved the admixed 
metal, or would form fulminate of gold with the oxid. 
The ammonia solution is a dark red fluid, from which 
the purple precipitates as a jelly upon spontaneous 
evaporation of the ammonia." 
On the basis of his investigations, Berzelius regarded 
the purple as a chemical combination of tin sesquioxid 
(zinnoxyduloxid) with purplish oxid of gold. As may 
be seen, Berzelius laid great stress upon the homogeneity 
of the purple, and just because of the clear, transparent 
color of the precipitate and its solubility in ammonia, 
he was led to hold that the purple is not a mixture of 
finely divided gold and stannic acid, the color of which 
was known to him. 
It is of the utmost importance to bear in mind that 
a greater part of inorganic chemistry was developed at 
a time when such ideas held sway, and a considerable 
number of substances which were called chemical com- 
pounds, would to-day be considered as colloidal mix- 
tures or compounds, or as colloidal solutions, or as 
absorption compounds. 
Later, many workers took up the moot question as 
to the nature of the purple of Cassius-I mention only 
Gay-Lussac, Debray, Golfier Besseyre, Muller-without 
reaching any final conclusion, for the chief point of Ber- 
zelius, the solubility of the purple in ammonia, was not 
cleared up. E. A. Schneider was the first correctly to 
explain the purple of Cassius as a mixture of the hydro- 
sols of gold and stannic acid, and quite independently I 
was able to prove that the purple is not a chemical com- 
bination, by synthesizing it from its constituents. 
By simple mixture of the hydrosols of gold and stan- 
nic acid I produced the purple of Cassius with all of its COLLOIDAL SULPHIDS AND METALS 
67 
usual properties. That the stannic acid in this case 
docs not enter into a chemical reaction with the gold, 
is evident, firstly, from the indifference Qf metallic gold 
to colloidal stannic acid and similar bodies, and secondly, 
from the fact that by the presence of the stannic acid 
neither the absorption spectrum nor the intensity of 
the color of the colloidal gold changed in the least. 
These changes would have appeared, if the gold had 
entered into chemical combination (as by treatment 
with chlorin, potassium cyanid, etc., whereby the red 
color vanishes in a short time). Nevertheless the pres- 
ence of stannic acid has effected a profound change in 
the behavior of the gold toward reagents; it suffers no 
color change1 upon the addition of electrolytes, but 
precipitates together with the stannic acid as a purple- 
red ammonia-soluble gelatinous precipitate, and in this 
respect acts as if it were chemically combined with the 
stannic acid. 
Colloidal Mixtures which have Simulated Chemical 
Compounds. The extensive experiments which led to 
the result above set forth, were begun with the belief 
that they were to settle a principle of general applica- 
tion. When, therefore, it was possible to prove beyond 
doubt that the purple of Cassius is not to be considered 
a chemical compound, as the most brilliant chemist of 
his time had assumed, it at once became questionable 
whether a very large number of substances which had 
been called chemical compounds really were such. At 
the beginning of my investigations, I myself considered 
it probable that this was a case of a chemical compound 
1 As is well known, pure red gold hydrosols are immediately coagu- 
lated by the addition of electrolytes with the production of a blue 
color, and subsequently deposit metallic gold; the purple of Cassius 
acts quite differently (see pp. 13, 14, and 78). 68 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
of acid character, which owed its solubility in ammonia 
to the formation of a soluble salt; but I was led to the 
conclusion that it was an intimate mixture of colloidal 
gold and colloidal stannic acid, or a "colloid-compound " 
of gold and stannic acid, owing its solubility to the pep- 
tisizing action of the alkali. Making use of the language 
of the old chemists, I expressed the generalization of 
these facts in the following sentence: A mixture of col- 
loidal substances can, under certain conditions, act like 
a chemical compound, and the properties 1 of the one 
substance in such a mixture can be masked by the other.2 
I chose the expression "colloidal mixture" in order 
to lay particular stress on the fact that this is not a case 
of combination in a chemical sense. Later, in conjunc- 
tion with van Bemmelen's expression "absorption com- 
pound," I used the expression colloid-compound3 for 
1 By "properties" is here meant chiefly behavior with reagents. 
2 Bredig (Anorganische Fermente, 1st ed., p. 20) has taxed me 
with basing my definition upon a loose conception of chemical com- 
bination. In reply I might simply remark, that a long-established 
idea cannot be overturned by the mere dictum of a new definition. 
A chemist whom it was desired to convince, upon the basis of a defi- 
nition, that a substance described as a chemical compound by the 
highest authorities, really was not one, would rather doubt the accu- 
racy of the definition in question than assume that the authorities 
were in error, especially when the proof was not based upon experi- 
mental grounds. An old-established means of discrimination between 
a mixture and a chemical compound, and one used by Berzelius in 
this case, is that the constituents of the former can be easily separated 
by means of solvents, while those of the latter cannot. I recall only 
the well-known example of iron and sulphur: from the mixture of 
both substances the sulphur (or the iron) may be removed by a variety 
of solvents, which is not the case with the combination FeS. That 
in the case of colloids every characteristic cannot be applied to dis- 
tinguish between mixtures and chemical compounds will be shown by 
the substance of my work, and is indirectly expressed by the above 
definition. 
3 Zsigmondy, Verh. d. Gesell. Deutscher Naturforcher u. Aerzte 
(Hamburg), 1902, pp. 168-172. COLLOIDAL SULPHIDS AND METALS 
69 
colloidal mixtures in which both constituents are col- 
loids, in order to express the intimate union of both 
colloidal substances. 
It must here be noted, however, that not only have 
colloid compounds or colloidal mixtures, in which two 
colloids are united, been erroneously described as chemi- 
cal compounds, but so also have mixtures or absorption 
compounds of crystalloids with colloids, as well as col- 
loidal solutions of various substances, especially of 
metals in fused salts. 
Thus Kirchhoff and Bunsen,1 and later H.Rose,2 describe 
a blue potassium subchlorid K2C1, which, upon stirring 
in water, decomposes into KC1 and KOH with the evo- 
lution of hydrogen. It is very probable that this is a 
case of a colloidal solution of potassium in potassium 
chlorid, all the more so as Bronn 3 has shown that in all 
probability the red and blue solutions of potassium, 
sodium, etc., in ammonia arc most probably colloidal. 
The same applies to sodium subchlorid Na2Cl, described 
by H. Rose, and to rubidium and caesium subchlorids 
(Kirchhoff and Bunsen). These subchlorids interfere 
with the electrolysis of fused salts, and the metals dis- 
solve in the molten mass producing the color. It is 
worthy of notice that Bunsen and Kirchhoff (as well as 
H. Rose) were led by the homogeneous appearance of 
their molten salts, "which neither to the naked eye nor 
under the microscope show the slightest trace of metal- 
lic substance," to assume that they were chemical com- 
pounds. 
Recently R. Lorenz4 has pointed out the analogy 
1 Kirchoff and Bunsen, Pogg. Ann., 1861, Vol. CXIII, p. 345. 
2 H. Rose, Pogg. Ann., 1863, Vol. CXX, p. 1. 
3 Broun, Drude's Ann., 1905, Vol. XVI, p. 166. 
4 R. Lorenz. A complete collection of the observations of R. Lorenz 70 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
between these "compounds" and the "metal fog" 
(which also interrupts the electrolysis of fused salts), 
and it can be considered as almost proved that in this 
case too there is a colloidal solution of the metal in the 
fused salt. 
From the above it seems probable that with a number 
of substances which have been described as chemical 
individuals, the chemical compounds in question do not 
exist. In part, the proof of this has already been fur- 
nished by means of individual substances, by various 
investigators-each independent of the other. 
As examples may be mentioned: 1 
(1) All substances in which one constituent is a metal, 
and the other a colloidal oxid, a colloidal salt or a pro- 
tective colloid: 
Purple of Cassius as a colloidal mixture of colloidal 
gold and colloidal stannic acid. 
The purplish gold oxid and related compounds (Berze- 
lius, Buchner), as colloidal gold with impurities. 
The solution of aurous sulphite (observed by Kriiss and 
Clemens Winkler), as colloidal gold reduced by SO2 
(Zsigmondy). 
Silver-purple, as a mixture of colloidal silver and col- 
loidal stannic acid, produced synthetically by Lotter- 
moser. 
Argentous oxid and its salts, described by Wohler, 
v. d. Pfordten and others; explained by Muthmann as 
mixtures which owe their color to colloidal silver. 
To this group of colloid compounds belong the many 
analogues of the purple of Cassius, which on account of 
upon metal fogs, is found in his monograph, "Electrolysis of Molten 
Salts," Part II, published by W. Knapp, Halle a. S. (See p. 57 of his 
book.) 
1 The references to the literature are given partly in my article 
above cited, partly in Lottermoser "On Inorganic Colloids," Stuttgart. COLLOIDAL SULPHIDS AND METALS 
71 
their high metal content may be called colloidal metals. 
For example: 
The colloidal silver of Carey Lea, the colloidal gold, 
silver, platinum, palladium, etc., of Paal; the colloidal 
mercury of Lottermoser and others. 
(2) Many substances formed by the union of two col- 
loidal oxids, and sometimes regarded as sesquioxids or 
as chemical compounds of an acid oxid with a basic one. 
Thus Berzelius describes tin sesquioxid, and considers 
its solubility in ammonia as evidence of the existence of 
a chemical compound.1 This substance can, however, be 
regarded as purple of Cassius in which the gold is replaced 
by stannic oxid.2 
There are, besides, a whole series of stannic acids- 
intermediaries between ortho- and metastannic acids, 
each with its peculiar reaction, which may be regarded 
as colloidal mixtures or colloid compounds of the two 
extreme members of this group.3 The number of ap- 
parent chemical compounds would be indefinitely in- 
creased if every such body were described as a special 
hydrate or as a special allotropic modification of stannic 
acid, and even to-day this is often done with other col- 
loidal precipitates. 
The characteristic behavior of tin towards nitric acid 
containing ferric nitrate, described by Lepez and Storch, 
can be explained on the assumption of the presence of a 
colloidal mixture of ferric oxid and stannic acid in solu- 
tion. 
By mixing solutions of positive and negative hydro- 
sols, as Picton and Linder, and Lottermoser, have shown, 
precipitates can be obtained which, when one of the 
1 Stannous oxid does not dissolve in ammonia. 
2 Zsigmondy, Liebig's Ann., 1898, Vol. CCCI, p. 386 
3 Zsigmondy, Ibid, p. 372. 72 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
constituents contains an acid, and the other a basic oxid, 
may easily be mistaken for a salt-like chemical combina- 
tion of the two-which, indeed, has often been the case. 
This point in particular has been elucidated by the 
research of W. Biltz,1 who demonstrated the generality 
of these reactions. As Biltz very justly observes, the 
assumption of a salt-like compound is excluded if the 
negative constituent is metallic gold, which gives pre- 
cipitates when mixed with positive colloids just the 
same as do other negatively conducted colloids. 
The most important results of Biltz are the following: 
oppositely conducted hydrosols mutually precipitate each 
other without the addition of electrolytes; similarly 
conducted ones do not (with some exceptions in the 
case of negatively conducted colloidal solutions, of selen- 
ium and gold, for example, in which case Biltz assumes 
the existence of chemical reactions). 
Biltz further found that in the mutual precipitation 
of oppositely conducted colloids, an optimum of the 
precipitative action is to be observed. If the favorable 
conditions of precipitation are exceeded on either side, 
no precipitation at all takes place. (Protective action.) 
(3) Here belong the organic compounds of oxicls with 
albuminoid substances and the like. Thus Baal2 has 
shown that the alkaline solutions of the salts of heavy 
metals in protein substances contain the heavy metal, 
not, as has been assumed, in so-called organic combina- 
tion, but as colloidal oxids, which are held in solution by 
the protective action of the albuminoid or its decom- 
position products. 
All these determinations and facts serve to round out 
our knowledge of the absorption compounds, which is 
1 W. Biltz, Berichte, 1904, Vol. XXXVII, p. 1111. 
2 Paal, Berichte, 1902, Vol. XXXV, p. 2205. COLLOIDAL SULPHIDS AND METALS 
73 
based upon the extended and basic research of van 
Bemmelen,1 the pertinent results of which we will try 
to express in two sentences: 
Most of the hydrogels contain their water, not chemi- 
cally combined, but absorbed (or "adsorbed"); hydro- 
gels of ferric oxid, alumina, silicic acid, stannic acid, etc., 
are not hydroxids of these elements, but "absorption 
compounds " of oxid and water in varying proportions. 
By absorption, hydrogels may take up and hold acids, 
alkalis, salts, etc., dissolved in water, in such quantity 
sometimes that the resulting products may be mistaken 
for chemical compounds. 
Following the investigations of van Bemmelen, Biltz 
and Behre have recently shown 2 that the compound of 
arsenic acid with ferric oxid, which Bunsen considered 
basic ferric arsenite, 4Fe2O3, As2O3, 5H2O, is as a matter 
of fact an absorption3 compound in van Bemmelen's 
sense. 
W. Biltz has added to this result some interesting 
observations on the action of toxins and antitoxins, and 
has shown that these reactions may be considered as 
absorption phenomena. 
Further W. Pauli, developing the basic work of Ehr- 
lich, makes evident the fact4 that reaction between 
1 van Bemmelen, Landw. Vers. Stat., 1888, Vol. XXXV, pp. 69-136; 
Rec. Trav. Chim. Pays-Bas, 1888, Vol. VII, pp. 37-118; Zeit. f. 
anorgan. Chemie, 1894, Vol. V, p. 466; Ibid, 1896, Vol. XIII, p. 283; 
1898, Vol. XVIII, pp. 14 and 98; 1899, Vol. XX, p. 185; 1900, Vol. 
XXIII, pp. Hl and 321; 1903, Vol. XXXVI, p. 380; 1904, Vol. 
XLII, p. 265. 
2 Berichte, 1904, Vol. XXXVII, p. 3138, communication from W. 
Biltz. 
3 Biltz and other investigators use for the most part the word 
"adsorption"; in order to honor van Bemmelen who has attained 
special prominence in this field, I use the customary term "absorp- 
tion" chosed by him. 
4 Wolfgang Pauli, Wandlungen in der Pathologic durch die Fort- 74 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
toxin and antitoxin may be traced to the mutual influence 
of different kinds of colloids, and that this opens up a 
wide field for colloid chemistry. 
According to the above research the number of apparent 
chemical compounds, in which colloids play a role, is 
very large. We may here distinguish between: 
(1) Colloidal solutions of metals in crystalloids (for 
example, "sodium subchlorid," salts of "silver suboxid," 
"aurous sulphite," etc.); 
(2) Absorption compounds, like the hydrogels them- 
selves, or those in which a crystalloid is taken up by a 
colloid, which in particular have been carefully studied 
by van Bemmelen; 
(3) Colloid compounds, in which two colloids combine 
to form a new mixture which cannot be separated by 
solvents without destroying the colloidal character. 
Incidentally we might here remark that many, though 
by no means all of the phenomena of dyeing may be 
attributed to just such reactions of colloids with each 
other, or between colloids and crystalloids.1 Color lakes 
are to be considered, some as colloid compounds,- some 
as absorption compounds; only rarely is it the case 
that real chemical compounds or even solid solutions 
are formed. 
Colloidal Gold. In the year 1857 Michael Faraday2 
described various colored liquids which he had pro- 
duced by reducing gold chloric! solutions with phos- 
phorus, liquids which to-day would be called colloidal so- 
lutions or hydrosols of gold, terms that Faiaday could 
schritte der allgemeinen Chemie. Jubilee Address. Vienna, 1905, M 
Perles. 
1 Compare for example Krafft, Berichte, 1899, Vol. XXXII, p. 1008; 
Zacharias, Zeitsch. f. phys. Chern., 1902, Vol. XXXIX, p. 468; 
Biltz, Gott. Nachr., 1904, No 1. • 
2 Faraday, Phil. Trans., 1857, p. 154. COLLOIDAL SULPH1DS AND METALS 
75 
not use since the idea "colloidal solution" did not yet 
exist. 
His method was very crude, and yielded liquids appearing 
sometimes red, sometimes blue or violet, liquids which 
mostly deposited their gold content; but sometimes very 
stable hydrosols. Upon boiling, all these liquids under- 
went a marked change; they became darker, the red 
gave place to violet, the violet to blue; their cloudiness 
increased, and their permanence was correspondingly 
lessened; and a little while after boiling they all gave 
a deposit. To this deposit Faraday directed his par- 
ticular attention; he proved in many cases that whether 
it was obtained from a boiled liquid or not, whether it 
showed a blue, violet, or red color, it was metallic gold, 
notwithstanding its lack of resemblance to the noble 
metal. These investigations led Faraday to the con- 
clusion that the liquids owed their color to the extremely 
finely subdivided gold contained in them. He further 
showed by direct experiment (concentrating the sun's 
rays with a burning glass) the existence of a diffuse 
dispersion, which as often as not showed a gold 
color. 
Faraday found excellent confirmation for the correct- 
ness of his conclusion in the fact, brought out indepen- 
dently of the above research, that pure metallic gold- 
which he produced as a thin coating on quartz and 
glass by the "electrical evaporation " of a gold wire 
in hydrogen, etc.-showed to transmitted light the 
same colors as the above described gold-containing 
liquids. 
To that group of chemists for whom it would have 
had particular interest, Faraday's pregnant work has 
apparently remained entirely unknown-until 1898. I 
have read no paper on gold ruby glass or purple of Cas- 76 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
sius, in which Faraday was quoted.1 For instance, a 
quotation may be given from the paper of E. A. Schnei- 
der 2 on the purple of Cassius: 
"Of itself the hydrosol of gold is probably not perma- 
nent for more than a few minutes. Up to the present 
time all attempts to produce it direct, or from the purple 
solution, have been fruitless. Thus it was thought that 
by treating the purple solution with concentrated hydro- 
chloric acid or concentrated sulphuric acid, and then 
dialyzing, that the stannic acid would be separated from 
the gold. As a matter of fact, this did take place, but 
the gold in the dialyzer was found to be in a coagulated 
condition, after the acid and resulting tin compound had 
diffused out. The presence of the tin hydrosol is there- 
fore absolutely necessary for the existence of the gold 
hydrosol." 
This was written by E. A. Schneider, who not only 
showed how to prepare colloidal silver in a particularly 
pure form, but also was very ingenious in producing 
other colloidal solutions. He discovered the hydrosols of 
several sulphids of the noble metals, organosols of silver, 
of gold sulphid, colloidal iron phosphate, etc. 
From this it will be seen that it was by no means easy, 
as long as the essential conditions were unknown, to 
prepare the hydrosol of gold free from stannic acid, and 
as I did not know of Faraday's work, and was working 
with less energetic reducing agents than the phosphorus 
used by him, I was obliged to surmount all the difficul- 
ties which stood in the way of the production of a stable, 
sufficiently pure and homogeneous red colloidal gold 
solution. 
1 Just as little was he quoted by those investigators who investi- 
gated the nature of "silver suboxid" and similar substances. 
2 Zeitschr. f. anorg. Chem., 1894, Vol. V, p. 82. COLLOIDAL SULPHIDS AND METALS 
77 
A gold solution of such properties was absolutely 
necessary in order to finally settle the question of the 
chemical nature of the purple of Cassius. 
It was only by a thorough search of the literature, 
which I undertook after the conclusion of my experi- 
mental work on this subject, but before its publication,1 
that I found Faraday's work, which I rescued from the 
oblivion into which it had fallen. 
My colloidal gold solutions exhibited the necessary 
properties in the way of homogeneity and permanence, 
which chemists are wont to attribute to a solution in 
the usual acceptation of the term (see Introduction). 
The experiments which led to their production also 
taught me to know the influence exerted by certain 
impurities in the water on the properties of the gold 
hydrosol. 
While certain colloids favor the formation of the red 
gold hydrosol remarkably, others, even if present only 
in traces, make it almost impossible to obtain a usable 
colloidal gold solution. To the former belong colloidal 
stannic acid, for example; because of its presence, the 
gold formed by reducing gold chlorid solution with stan- 
nous chlorid, yields, not a brown or black precipitate, 
but a red precipitate, the purple of Cassius. Another 
colloid which protects the nascent colloidal gold was dis- 
covered by Faraday, and called by him "jelly." Lobry 
de Bruyn 2 (1898), characterized gelatin jelly as a pro- 
tective colloid (Schutzkolloid). Paal (1900) showed that 
lysalbinic acid and protalbinic acid possess the same 
protective properties. 
For my purpose the absence of foreign colloids was 
necessary; it was only when I had secured a gold hydro- 
1 Liebig's Annalen, 1898, Vol. CCCI, p. 29. 
2 Rec. des Trav. Chim. d. Pays-Bas, 1900, Vol. XIX, p. 236. 78 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
sol free from detectable quantities of foreign colloids, 
that I could synthesize the purple of Cassius, and show 
how the reactions of colloidal gold solutions could be 
influenced by the presence of other colloids. 
By the exclusion of protective colloids, however, the 
sensitiveness of the solution was increased to such an 
extent that I was obliged carefully to keep out the col- 
loids present in most kinds of distilled water, which 
would destroy the hydrosol as soon as it was formed. 
(See also Chapter VIII.) This I did by repeated redis- 
tillation, using a silver condenser, or by freezing the 
distilled water several times. 
The colloidal gold solution obtained by carefully fol- 
lowing my directions, is bright red and very stable; in 
contradistinction to Faraday's fluids, it stands heating 
to the boiling-point without being thereby altered, and 
it remains practically unchanged for months, even years. 
Upon the addition of most electrolytes 1 it quickly 
changes to a blue color, and then deposits gold as an 
extremely fine powder. The presence of foreign colloids 
of certain kinds prevents the color change as well as the 
precipitation of gold. Very minute quantities some- 
times answer, 0.0001 per cent of gelatin, for instance, 
which can by no means alter the viscosity of the fluid 
in any appreciable manner. Thus was disproved a 
correlated assumption which was thought to follow from 
certain observations of Faraday2 or Lobry de Bruyn, 
1 Exceptions to this rule are ammonia, and potassium cyanid and 
ferrocyanid. 
2 Faraday had allowed gold chlorid to dry out with "jelly," and 
found that the jelly colored red by metallic gold showed no color 
change with salt in contradistincton to the "ruby fluids" made without 
jelly. Faraday, however, has not stated from what colloid the jelly 
was made. The fact also escaped him that the same action exerted 
by the stiff jelly can also be exerted by minute traces of a glue-like 
substance. COLLOIDAL SULPHIDS AND METALS 
79 
that the added colloid works by increasing the viscosity 
of the medium, and preventing the sedimentation of the 
gold; for it is not the sedimentation but the coagula- 
tion of the metallic gold that is prevented. In this case 
the gold and the protective colloid exercise on each other 
a reciprocal influence, to which I shall later recur.1 
Protective Action of Many Colloids and the Gold Figure. 
By looking further into these reactions I was able to 
bring to light an almost unknown general property of 
many colloids, of preventing the coagulation of typical 
irreversible hydrosols; and further to establish a rela- 
tive measure of this protective action on bright red col- 
loidal gold solutions, by fixing the gold figure.2 
It must here be mentioned, however, that A. Lotter- 
moser and E. von Meyer had previously found that egg 
albumen and blood serum could completely or partially pre- 
vent the precipitation of colloidal silver by sodium chloric!;3 
1 Facts in this field of work have long been known (for example, the 
solubility in alkali of albuminates of the heavy metals; the property 
possessed by gelatin, of preventing the precipitation by hydrogen 
sulphid of sulphids of the heavy metals), but such protective action 
had for the most part either been designated as chemical reaction, 
or else it had been assumed that the viscosity of the colloid in ques- 
tion held in. suspension the precipitate produced. That we are here 
dealing with a general property of colloids, for which a quantitative 
expression can be found, was, as far as my knowledge goes, first brought 
to light in the work referred to. 
2 Zeitsch. f. analyt. Chem., 1901, Vol. XL, p. 697. Not all col- 
loidal gold solutions are suitable for this purpose. That obtained by 
reducing gold chlorid with phosphorus is not, for reasons to which I 
will later refer. It gives very different gold figures. 
3 Journal f. prak. Chem, 1897, N. F., Vol. LVI, p. 242. On this 
subject the authors say: "The cause of this fact we cannot yet state; 
the formation of silver albuminate seems to be excluded." In a 
reference to this paper in the Zeitsch. f. physik. Chem., 1898, Vol. 
XXVI, p. 368, signed with the initials W. O., it is stated as follows: 
"It is a medical fact that colloidal silver in salve form, etc., is taken 
up by the organism, although the salts present in the body-fluids 80 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
and that Lottermoser has followed up this reaction 
further.1 
By the "gold figure" is meant the number of milli- 
grams of colloid which are just insufficient to prevent 
the change to violet of 10 com. of bright red colloidal 
gold solution, by the addition of 1 cm. of a ten per cent 
solution of sodium chlorid.2 
There is thus made evident an enormous difference 
between the individual colloids, as may be seen from the 
table on the next page. 
In the paper referred to are given the conditions which 
must be followed in order to secure for any particular 
protective colloid, by the method mentioned, concordant 
gold figures. The influence exercised upon the gold 
figure by dilution, temperature, and length of the experi- 
ment, is stated. It is also shown that all solutions of 
protective colloids undergo change with the lapse of 
time, so that their effectiveness diminishes, in some cases 
so much that the protective effect vanishes. 
Fr. N. Schulz and myself established the gold figures 
for several constituents of egg albumen obtained by 
fractional precipitation according to the practice of 
physiological chemists.3 It appeared that the gold figure 
well serves the purpose of characterizing more exactly 
precipitate the solutions of colloidal silver. The authors first state," 
etc., and W. O. in his remarks on this point says: "In this case there 
is probably a formation of complex silver compounds similar to the 
silver compounds with the imids; to such substances silver does not 
act like a noble metal, finally dissolving with ease under the influence 
of free oxygen." 
1 Journal f. prak. Chem., 1905, N. F., Vol. LXXI, p. 296. 
2 The protective action may be better expressed by the reciprocal 
of the gold figures. Thus carrageen would have a protective action 
of 1-2, gelatin 100-200, raw sugar 0. 
3 Hofmeister's Beitrage zur chem. Physiol, u. Pathol., Vol. Ill, pp 
138-160 (1902). COLLOIDAL SULPHIDS AND METALS 
81 
TABLE I.-GOLD FIGURES OF SOME COMMERCIAL 
COLLOIDS 
(According to Zsigmondy) 
Colloid. 
Gold Figures. 
Remarks. 
Gelatin 
Russian glue 
Cologne glue 
Bone glue 
Isinglass 
Casein 
Egg albumen 
Gum arabic, la 
" " , Ila 
" " , Illa 
Irish moss 
Tragacanth 
Dextrin 
Wheat starch 
Potato starch 
Sodium stearate 
C C IC 
Sodium oleate 
Old stannic acid solution .. 
Urea | 
Raw sugar 
Gum arabic, selected in 
pieces 
0.005-0.01 
0.005-0.01 
0.005-0.01 
0.005-0.01 
0.01 -0.02 
0.011 
/ 0.15-0.251 
10.1-0.2 f 
0.15-0.25 
0.1 
0.5-4 
0.5-1 
about 2 
( 6-121 
1 10-20 J 
about 4-63 
about 253 
10 
0.01 
0.4-1 
oo 
00 
less than 500 
00 
0.4-0.6 
Aqueous solution made 
with a few drops of NH3. 
Two different commercial 
varieties. 
Two commercial varieties. 
At about 60° C. 
At boiling point. 
At ordinary temperature. 
At boiling point. 
the globulin, albumen, etc., and of testing for the pres- 
ence of certain impurities. The great difference in the 
value of the gold figure of crystallized and amorphous 
albumen, will be seen from the table on the following page.2 
Very marked was the action of a substance usually 
obtained in large quantity in the second albumin frac- 
1 Only once determined, without control. 
2 The addition of NaOH to albumins having low protective action 
increases it very considerably. 82 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
tion and called by us "impurity" This substance alone 
(without the presence of salts) colored the colloidal gold 
solution blue, and made it impossible to determine the 
gold figure of crystallized albumen, so long as the impurity 
was contained in it. 
TABLE II.-GOLD FIGURES OF SEVERAL ALBUMIN 
FRACTIONS AND ALBUMOSES 
Colloid. 
Gold Figure. 
Author. 
Remarks. 
Globulin 
0.02-0.05 
Schulz & Zsigmondy 
f c 
Obtained from 
Ovomucoid 
0.04-0.08 
white of egg 
by fractional 
precipitation 
Crystallized albumen 
Amorphous albumen 
2-8 
0.03-0.0G 
t i 
ll 
Fresh white of egg . 
Albumin from Merck 
0.08-0.15 
0.1-0.3 
Schulz & Zsigmondy 
I c 
Heteralbumose .... 
Protalbumose 
Synalbumose 1 
0.01-0.075 
1.60-3.36 
E. Zunz 
1I 
ll 
Primary albu- 
moses from 
Witte's pep- 
tone, isolated 
by Pick's 
method. 
Witte's peptone2. . . 
By the excellent work of Zunz 3 have been determined 
the gold figures of the primary albumoses (product of 
the peptic digestion of albumen, which, in addition to 
secondary albumoses and peptones, occur in Witte's 
peptone for example). Zunz determined with great 
1 Synalbumose does not protect colloidal gold but colors it blue. 
According to Zunz, 0.64-2.24 mg. of synalbumose is enough to turn 
10 ccm. of gold solution violet in the absence of salts. This reaction 
of synalbumose is prevented by the presence of protalbumose, casein, 
albumen, but not by heteralbumose despite its high protective action. 
2 Acts similarly to synalbumose, only it coagulates colloidal gold 
still more powerfully. 0.24-0.64 mg. of Witte's peptone have the 
same effect as 0.64-2.24 mg. of synalbumose. (Zunz.) 
3 Archives internat. de Physiologie, 1904, Vol. 1, p. 427. COLLOIDAL SULPHIDS AND METALS 
83 
accuracy the gold figures of protalbumose and heter- 
albumose, and macle the very remarkable discovery that 
synalbumose alone changed colloidal gold solutions blue, 
just like the "impurity" before described. 
Zunz has recently1 tested a series of pure albumoses and pep- 
tones as to their behavior with colloidal gold solutions. He shows 
that, just like synalbumose, they almost all possess the property of 
turning the colloidal gold violet without the addition of electrolytes, 
but that the quantity necessary to do this varies from substance to 
substance. 
The chief results are given in the following table: 
Quantity of the Substance in mg 
Substance. 
which suffices to Change to Vio- 
let 10 cc. of the Colloidal Gold. 
Thioalbumose 
2.60-4.00 
Albumose, A II 
2.24-3.20 
" BI 
0.08-0.32 
" B III. a 
0.20-0.80 
" B III. b 
0.50-1.40 
" B III. c 
0.40-1.20 
" B IV. b 
0.80-1 .60 
" B IV. c 
0.70-1.80 
" B/C 
0.80-2.80 
" CI 
1 .60-3.20 
Peptone, a . » 
0.24-0.52 
" Bb 
3.60-7.40 
" Pc 
4.4 -8.20 
Zunz has further found 2 that no relation exists between the action 
of albumoses and peptones upon colloidal gold, and the alteration by 
these substances of the surface tension of water. 
According to Traube and Bodlander,3 a 0.1 per cent solution of 
albumoses give 125.3 drops in the stalagmometer, while pure water 
gives 100, and a 2 per cent solution of egg albumen only 106.3 drops. 
1 E. Zunz, Bull. Soc. Roy. des sciences med. et nat., June 11, 1906. 
2 E. Zunz, ibid. 
3 Ueber die Unterscheidung von Eiweiss, Berichte d. D. Chem. 
Ges., 1886, Vol. XIX, p. 1871-6. 84 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
Whereas, therefore, albumen but slightly influences the surface tension, 
albumoses influence it very strongly, but to about the same extent 
among themselves; on the other hand their behavior with gold solu- 
tion varies greatly, as can be seen from the table. 
Substance. 
Gold Figure. 
Number of Drops in 
the Stalagmometer as 
Compared with 100 
Drops of Distilled 
Water. 
Heteroalbumose. . . 
Protalbumose 
Synalbumose 
0.01 to 1.075 
0.6 to 3.36 
Produces a change in color 
(0.64-2.24 mg.) 
114.4 
113.6 
113.7 
TABLE III 
Tn the additional paper 1 Zunz shows that there is no relation what- 
ever between the behavior of turbid mastic solution with albumoses, 
and that of colloidal gold with the latter. Thus heteroalbumose and 
synalbumose precipitate turbid mastic solution, whereas the other 
albumoses do not possess this property. As may be seen from what 
precedes, heteroalbumose does not exercise a precipitative action on 
colloidal gold, but rather a protective action. 
From these facts it is evident that the electric charges alone do 
not determine the reactions in question, for mastic and gold particles 
are both negatively charged. If the electric charges alone exercised 
the main influence, gold and mastic should react the same with the 
albumoses. But this is not the case, and we are therefore confronted 
with specific actions, whose nature is not yet completely understood.2 
W. Biltz 3 determined the gold figures of several inor- 
ganic colloids, and Biltz, Much, and Siebert, those of 
several sera and antitoxins.4 The authors use the gold 
E. Zunz, Arch, internal. de Physiol., 1907, Vol. V., p. 245. 
2 Neither is there any connection with the diffuse dispersion of light. 
According to Zunz all proteoses (albumoses) exhibit an intense light- 
cone, whereas the peptones show only a faint bluish cone, which 
vanishes upon the insertion of a yellow disk. 
3 Berichte, 1902, Vol. XXXV, p. 4437. 
4 Experimentelle Beitrage zu einer Absorptions-theorie der Toxin- COLLOIDAL SULPHIDS AND METALS 
85 
figure in order to determine approximately the loss of 
albumen content upon shaking with iron oxid. It appeared 
that albumen solutions containing salt gave up less albu- 
men to the iron oxid gel than did dialyzed ones. The 
physiological salt content, therefore, protects albumen to 
a certain extent against absorption by hydrogels. 
The authors further found that the gold figures of 
nutritive bouillon and toxin solutions could not be 
closely determined, because even without the addition 
of electrolytes, they immediately produce a blue color 
in colloidal gold solutions, a reaction which corresponds 
with that of synalbumose and that of Witte's peptone 
(see above). It was pointed out by Biltz, Much, and 
Siebert, that this reaction appears to be characteristic 
for peptones.1 The growth of bacteria changes the 
nutritive bouillon, so that it regains the property of a 
protective colloid. 
As previously mentioned above, W. Biltz observed 
that oppositely charged colloids mutually precipitate 
each other when they are mixed in certain proportions; 
if an excess of one or the other colloid is used, then this 
excess acts as a protective colloid. Similar facts have 
been determined by Bechhold,2 Neisser, and Friedmann,3 
in connection with the precipitation of mastic by posi- 
tive hydrosols, dye solutions, and the like. Protective 
action also appears in the mutual precipitation of acid 
and basic dye-stuffs, if there is used an excess of either 
dye-stuff over and above the quantity necessary for 
precipitation. 
neutralisierung. Behring's Beitrage zur experimentellen Therapie, 
1904, No. 10. 
^'Peptone" here refers to the commercial mixtures, not the pep- 
tones of the physiological chemist. 
2 Bechhold, Zeitschr. f. physik. Chem., 1904, Vol. XLVIII, p. 385. 
3 Neisser and Friedemann, Miichener med. Wochenschr., 1903, No. 11. 86 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
These facts lend interest to some observations I made 
several years ago, but which I have not been able as yet 
to study more closely. I found that colloidal gold and 
basic dyes like fuchsin (or methyl violet, methylene blue, 
etc.), sometimes precipitate each other quantitatively, 
but that the reaction between the two substances often 
failed to occur or ■ else took place very incompletely. 
I then attributed this to the protective action of impuri- 
ties, but it is very possible that in the unsuccessful 
experiments, I exceeded the optimum of precipitative 
action more accurately described by Biltz and other 
investigators. From the precipitate the dye can be 
extracted by alcohol (not, however, by water). There 
remained on the filter a gold colored powder which was 
not further investigated. 
Other investigators have shown that the protective 
action of some few colloids extends to a certain degree 
even to coarser subdivisions.1 Arthur Muller showed 
this in the case of a suspension of red phosphorus.2 That 
this is only superficially analogous to the protection in 
the case of gold, is at once evident when the two kinds of 
protective action are compared. While 0.025-0.05 mg. 
of gelatin completely protect 50 ccm. of colloidal gold, 
on the other hand, 25 mg. of the same colloid (that is 
500 to 1000 times as much) are insufficient to completely 
prevent the flocculation of 50 ccm. of phosphorus sus- 
pension by sodium chlorid. Furthermore, raw sugar 
prevents the precipitation of phosphorus, but not the 
color change of colloidal gold. 
Recently Arthur Muller and Paul Artmann3 deter- 
1 For instance Neisser, Friedemann, and Bechhold have determined 
protective actions in the case of suspensions of bacteria. 
2 Berichte, 1904, Vol. XXXVII, p. 11. 
3 Oestrr. Chemiker-Ztg., 1904, Vol. VII, pp. 149-151. COLLOIDAL SULPHIDS AND METALS 
87 
mined the smallest quantity of protective colloid neces- 
sary to prevent the precipitation of several irreversible 
hydrosols, AS2S3, CdS, etc. I reproduce their table. 
Metal Sulphid Solution. 
Smallest Quantity of Colloid in mg. 
Substance. 
Strength 
in 
Per Ct. 
Quan- 
tity 
used cc. 
Casein 
Gum, 
Arabic. 
Glue. 
Isin- 
glass. 
Albu- 
men. 
Dex- 
trin. 
As2S3 .... 
0.05 
5 
0.16 
0.32 
0.40 
0.55 
2.50 
50 
CdS 
0.10 
3 
0.65 
1.15 
1.80 
1.50 
8.00 
00 
Ag2S .... 
0.05 
5 
0.06 
0.13 
0.18 
0.18 
0.60 
00 
TABLE IV 
From the table it will be seen that the protective col- 
loids, gum, glue, dextrin, etc., exercise upon the colloidal 
sulphids a protective action similar to that exercised 
upon colloidal gold, but that the quantitative expression 
of the protective action varies with the colloid. As with 
gold, raw sugar exercises no protective action on sul- 
phids. Muller and Artmann further observed that under 
certain conditions glue solutions yield precipitates with 
colloidal solutions of cadmium and arsenic sulphids. 
Just as did Ebell in the case of the colloidal sulphids, 
and Barus and Schneider with colloidal silver, so Stockl 
and Vanino 1 have tried to prove that colloidal gold is 
an instance of very fine suspension. As Bredig and 
Coehn2 have already pointed out, and as may be seen 
from the above remarks, they have advanced no dis- 
tinctly new idea. I myself had considered my fluids as 
profound subdivisions of the metal, as may be seen from 
1 Zeitschr. f. physik. Chem., 1899, Vol. XXX, p. 98. 
2 Zeitschr. f. physik. Chem., 1900, Vol. XXXII, pp. 129-132. 88 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
my first paper on this subject.1 I must, however, object 
to the term suspension, for my fluids, because of their 
permanence and homogeneity, should be classed as col- 
loidal solutions, and not as "suspensions," as at present 
understood by chemists. 
The discussion between Stockl and Vanino and myself 
led Lobry de Bruyn to make some very noteworthy 
remarks 2 regarding colloidal solutions. From the molec- 
ular weight of soluble starch, Lobry de Bruyn calcu- 
lated the molecular diameter (approximately 5 /z/z) and 
called attention to the fact that, according to Lord Ray- 
leigh, particles of this size are large enough to polarize 
incident light. 
In concluding this section reference must be made to 
the valuable work of Schulze, Lottermoser and von 
Meyer, Picton and Linder, Coehn, Hardy, Spring, Bil- 
litzer, Quincke, Freundlich, and others, whose object 
was to determine the laws governing coagulation by 
electrolytes, and who by preference chose colloidal sul- 
phicls and metals as subjects of investigation. Taken in 
connection with the earlier work of Scheerer, F. Schulze 
Schloesing, Barus, and Bocllander, on the clearing of 
clay turbidity, these researches lead to a sharp distinction 
between flocculation and sedimentation,3 and estab- 
1 Zeitschr. f. Elektrochem., 1898, Vol. IV, p. 546: "Dr. Bredig 
exhibited to us yesterday a series of interesting properties of the 
electric arc. For example, by atomizing metals under water he 
obtained dark-colored fluids in which the metal was so finely sub- 
divided that it could be regarded as dissolved; but they are not 
solutions, because on long standing they lose part of their metal 
content. If the metal is still further subdivided fluids are obtained 
which no longer settle, that is, collodial metallic solutions. Up to the 
present Carey Lea's silver solution was the only one of this class 
known." 
2 Rec. des trav. chim. Pays-Bas., 1900, Vol. XIX, pp. 251-258. 
3 See for example, Spring, Bull. d. l'Acad. roy. de Belgique, 1900, 
No. 7, p. 468; see also note on p. 31. CONCERNING THE NATURE OF HYDROSOLS 
89 
lished the dividing line between the flocculation of clay 
turbidity and the coagulation of irreversible hydrosols 
(see Chapter I). That the two processes last referred 
to are not identical, has been shown in Chapter I. There 
is at present no unanimity of opinion as to the funda- 
mental causes of coagulation, other than what follows 
from comparing the views of Billitzer, Quincke, Freund- 
lich, and Jordis. 
In passing, it may be remarked that electrolytes not 
only cause pectization or coagulation, but also pepti- 
zation of colloids, and that a complete theory of col- 
loids cannot consider only one of these processes. In 
these very processes, moreover, the nature of the sub- 
divided substance and the presence of traces of foreign 
constituents (which may even arise from chemical 
action) influence to the greatest degree the behavior of 
the hydrosols, and as long as these conditions remain 
unexplained, generalizations can easily lead to error. 
3. Some Questions Concerning the Nature of Hydrosols 
The conviction has gradually gained more and more 
ground that colloidal solutions are in reality fine sub- 
divisions of an originally solid substance, and there is a 
diversity of opinion as to whether the individual parti- 
cles which polarize light are to be regarded as very large 
molecules in the sense of Picton and Linder and Lobry de 
Bruyn,1 or as minute particles of an originally solid sub- 
stance.2 
1 Compare also Nernst, Theoretische Chem., 3d. ed., p. 383. 
2 Another question as to whether the particles in hydrosols are 
solid or fluid, cannot at present be satisfactorily settled. Quincke 
assumes the latter condition. Compare Chapter XXI, wherein is 
briefly discussed the question propounded by Jordis, whether the 
same substance is present in hydrosols as in hydrogels. 90 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
In the first instance, the hydrosols in question would 
be considered as crystalloid solutions with very great 
molecular weight. In the latter classification, which is 
adopted by Bredig and others, they rank as heterogene- 
ous according to the phase rule, but then it must be 
borne in mind that even the homogeneous solutions of 
the crystalloids are, as a matter of fact, only apparently 
homogeneous (see Introduction, p. 7). 
Without doubt the question above referred to is theo- 
retically of high interest; but at present great difficulties 
stand in the way of its satisfactory solution. As Bredig 
has brought out,1 in the case of colloidal solutions the 
osmotic pressure (trennungsarbeit) can be referred to a 
partial crystalloid solubility of the colloid, to crystalloid 
impurities, to electric charge of the individual particles. 
So long as these influences are not individually deter- 
mined, the molecular weight of the dissolved colloid can- 
not be safely deduced from the results of the osmotic 
method. 
To decide the question whether some or all colloidal 
solutions possess large molecules, would be of particular 
importance for the kinetic theory; for it is just the 
very large molecules, the presence of which some inves- 
tigators assume in hydrosols, which would most facili- 
tate a purely experimental confirmation of important 
corallaries of that excellent theory. 
From the results of numerous physio-chemical investi- 
gations, and quite independent of any theory, it is evi- 
dent that the osmotic volume-energy 2 which, with crys- 
talloids, dominates part of the phenomena, becomes of 
1 Anorganische Fermente, p. 11. 
2 Regarding the expressons volume-energy, surface-energy, distance- 
energy, see Ostwald, Grundriss der allgemeinen Chem., 3d ed., pp. 53, 
247, 251. CONCERNING THE NATURE OF HYDROSOLS 
91 
less importance as compared to the influence of the elec- 
trical and chemical energy, or of 11 surface energy " 
("Oberflachenenergie ") 1 and perhaps of still other kinds 
of energy, as yet unknown. As the volume-energy2 di- 
minishes or disappears, the hydrosols correspond to real 
suspensions, but there is nevertheless an evident differ- 
ence between the two types of subdivision last referred 
to. While with the former the electrical energy and 
the surface energy (Oberflachenenergie) form an important 
part of the total energy, and render possible reactions 
which are deceptively similar to chemical reactions, with 
the latter their influence yields to that of gravity. (See 
Chapter I). The action of coarse, suspended powders 
is determined rather by purely mechanical influences 
and the speed with which their particles sink to the 
bottom, can be calculated according to the laws of falling 
bodies.3 
Before a comprehensive theory of colloidal solution can 
be promulgated, it will be necessary to determine in 
each special case by thorough experimental investiga- 
tion, the influence of these forms of energy which are 
most important in the case of hydrosols; a good begin- 
ning has already been made in this direction. I need 
only recall here the work of van Bemmelen, Spring, 
Paterno, Barus and Schneider, Picton and Linder, 
1 This expression is set in quotation marks simply because it might 
easily lead to the assumpton that only the surface of the particles 
contained in hydrosols comes into play. In this case the particles 
themselves act upon each other and the surrounding medium, and the 
existence of this action can be demonstrated by numerous observa- 
tions. Macroscopically considered, this same kind of energy comes 
into evidence as surface-energy. 
2 Which may be considered in the sense of Ostwald, Bredig, as well 
as in the sense of Lobry de Bruyn. 
3 See Barus and Schneider, loc. cit., p. 287. 92 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
Lobry de Bruyn, Coehn, Breclig, Bruni and Pappada, 
Lottermoser, Biltz, Billitzer, Quincke, Friedemann, and 
others, which has much advanced our knowledge of 
colloids and also suggested questions that deserve atten- 
tion. 
If we regard hydrosols as extremely fine subdivisions 
of substances originally solid, certain other questions 
immediately arise, some of which can be solved by the 
aid of ultramicroscopy. 
First. Are the particles contained in colloidal solu- 
tions independent of each other, or does the colloidal 
solution contain a network or honeycomb of threads or 
thin walls connected together? One would be led to the 
latter opinion by the assumption that the process of 
solution of a colloid is simply a matter of swelling up. 
We will see that this assumption is untenable, for as far 
as particles in hydrosols can be seen, they are freely 
movable. 
Second. Do the particles float quietly in the fluid, or 
fall to the bottom under the influence of gravity; or 
have they a more or less rapid motion of their own? In 
the latter case, what is the nature of the motion, oscilla- 
tory or translatory? What are the dimensions of the 
paths described relative to the size of the particles? Is 
the motion a temporary one brought about in the removal 
of differences of concentration, or is it a continuous one? 
Third. What color are the individual particles; can 
a relation be found between their size and color? 
Fourth. What is the size of the individual particles? 
Is it possible by careful preparation to produce colloidal 
solutions so homogeneous that the bulk of the subdivided 
substance cannot be recognized by the best optical 
methods, or can their heterogeneity in all cases be made 
evident by means of a beam of light? CONCERNING THE NATURE OF HYDROSOLS 
93 
Questions of this kind will, in certain definite cases, 
be answered in the experimental sections which follow, 
in which will also be given the answer to a question 
which led to this investigation: 
Faraday, as we know, observed in all gold hydrosols 
and ruby glasses which he investigated, the appearance 
of a diffuse dispersion. To him this was evidence that 
the gold in these fluids is not dissolved, but is present 
in fine subdivision. Evidence accumulated quite inde- 
pendently of Faraday, that the red colored substance 
in my fluids is metallic gold, led me to the same opinion 
reached by Faraday, namely, that we were dealing here 
with finely divided gold; nevertheless, upon the basis 
of my experience, I believed optical inhomogeneity to 
be an incidental characteristic of hydrosols, brought 
about by coarser particles in the otherwise homogene- 
ously subdivided (that is, dissolved)1 substances. Fara- 
day's view was afterwards espoused by various investi- 
gators, especially by Bredig,2 and extended to colloidal 
solutions in general. This very opposition led me to a 
searching investigation of the point in question. By 
means of ultramicroscopy I was able to decide the ques- 
tion in favor of the view held by Faraday, Bredig, and 
others It was apparent that the gold particles which 
occasion the red color of the colloidal solution, are the 
same ones that polarize light. 
On the other hand, it also became evident that the 
particles in the particularly clear colloidal gold solu- 
'"Dissolved" in the sense of the definition of the word solution. 
Formerly, in conformity with the quite generally accepted view, I 
regarded solutions as subdivisions completely homogeneous, optically 
speaking. The investigations of Spring and Lobry de Bruyn on this 
subject (see Introduction) were then unknown. 
' Bredig, Zeitschr. f. angew. Chem., 1898, No. 41. 94 
HISTORY OF THE IRREVERSIBLE COLLOIDS 
tions closely approach molecular dimensions, and the 
optical inhomogeneity of these same solutions is hardly 
greater than that of many crystalloid solutions,1 in both 
cases it is hardly observable. 
1 See Lobry de Bruyn, loc. cit. CHAPTER IV 
FACTS POINTING TO THE HOMOGENEITY OF GOLD 
HYDROSOLS. DEVELOPMENT OF ULTRAMICROSCOPY 
As the opinion just stated was based upon a long 
series of experiments which I never published in detail, 
and do not now need to publish, since the question has 
been settled in another manner, I may here briefly repeat 
the grounds upon which this opinion rests.1 
I do this for the particular purpose of showing the 
insufficiency of grounds of a similar nature, which may 
doubtless be urged in favor of the view that solutions in 
general are homogeneous. 
"To begin with, my view was supported by the obser- 
vation that although the majority of the hydrosols of gold 
(colloidal gold solutions), produced by me, for the most 
part showed a noticeable faint and diffuse scattering 
of incident light, they were sometimes very clear2 and 
sometimes very turbid. 
"Despite their varying appearance, all these fluids 
contained equal quantities of metallic gold, as I con- 
vinced myself by repeated analyses; they were red 
colored, and also showed approximately the same behav- 
ior toward reagents. 3 
1 Zsigmondy, VerhandL d. Deutschen physik. GeselL, 1903, Vol. V, 
p. 209. 
2 So clear that the light-cone is scarcely more perceptible than in 
distilled water. 
3 They were precipitated by electrolytes, with the development of 
a blue color. Upon quantitative test, however, they developed certain 
differences in their behavior with reagents, but not in their gold content. 
95 96 
DEVELOPMENT OF ULTRAMICROSCOPY 
"Since, then, these fluids showed in general the same 
mutual properties, and could in practice be differen- 
tiated from each other only by their greater or lesser 
turbidity, I assumed that the turbidity is not an essen- 
tial, but an incidental characteristic of colloidal gold 
solutions, occasioned by the presence of larger particles. 
"The turbidity always showed a more or less intense 
polarization, an evidence of the minuteness of the haze- 
producing particles even in very turbid fluids. 
"The above view received further support from a 
quantitative test of the sensitiveness of the recognition 
of subdivided gold by means of the convergent beam of 
light, which showed that in very cloudy fluids (upon 
considerable dilution) even less than 10-8 mg. of gold 
could in this manner be recognized with the naked eye 
by the visible path of the convergent beam,1 that is, a 
less quantity of substance than Kirchhoff and Bunsen 
could detect by means of spectrum analysis. My cloudy 
fluids containing 0.0005 per cent of gold were diluted 
100 to 1000 fold; the colorless dilute fluids still con- 
tinued to show a more intense light-cone than the 
undiluted, unclouded solutions containing 0.005 per 
cent. 
"The addition of a minute quantity of the very turbid 
fluid to the perfectly clear one sufficed to endow the 
latter with the polarized dispersion of the former, as 
evidenced by means of the light-cone; and this made 
evident of how small a quantity of coarse gold particles 
is necessary to produce, in a gold solution assumed to 
be homogeneous, the diffuse dispersion referred to. 
1 Since in using sun or arc light the diffuse light-cone shown by- 
distilled water exercises a detrimental effect, these tests were made 
with Welsbach light which produced no light-cone in the water I 
used for dilution. DEVELOPMENT OF ULTRAMICROSCOPY 
97 
"It must be acknowledged, however, that although 
the facts above referred to were favorable to my view, 
they gave no evidence of the existence of an optically 
clear gold solution. I hoped to be able to find out the 
truth by microscopical examination of the cone of 
light." 
Before I carry this discussion further it might be here 
stated that microscopical investigation led to the solu- 
tion of the question. This investigation showed that 
all the red-colored gold particles take part in the forma- 
tion of the polarized cone of light,1 and the greater or 
lesser cloudiness of the colloidal gold solution is due to 
the fact that the gold particles which produce the red 
color are larger in some cases and smaller in others. 
This result, which contradicts the view earlier advanced 
by me, is of great significance in judging colloidal solu- 
tions in general. 
By the observations of Fizeau2 and of Ambronn3 it 
was shown that very narrow light slits can still be seen 
even if their breadth is considerably less than the limit 
of microscopic resolvability. The narrower the slit of 
light the less the brightness of its microscopic image, 
whose breadth, however, from a certain point down- 
ward, no longer varies. 
It seemed to me probable that the larger particles 
assumed to be present in gold hydrosols, whose previ- 
ously mentioned property of dispersing transmitted light 
was known to me by the above experiments, could be 
made individually perceptible by a microscopic examina- 
1 But in the case of the almost homogeneous gold solutions, this 
conclusion cannot with certainty be reached; however, in this case 
too, the faint trace of a light-cone can be attributed to the gold present. 
2 Pogg. Ann., 1862, Vol. CXVI, p. 478. 
3 Wied. Ann., 1893, Vol. XLVIII, pp. 217-222. 98 
DEVELOPMENT OF ULTRAMICROSCOPY 
tion of the light-cone; for if the small particles reflected 
enough sunlight, even if their size was below the limit 
of microscopic resolvability, they would, just as were 
the narrow rays of light, be individually perceptible 
under the microscope and act to a certain extent as frag- 
ments of such light slits. 
The microscopical investigation should also allow me 
to determine if the space between the coarser individual 
particles was optically clear or was filled with smaller 
gold particles. 
As a matter of fact, in two very turbid fluids (AuP8 
and AU54)1, by the use of sunlight and a magnification 
of about one hundred diameters, I could recognize the 
presence of thousands of shining gold particles, whose 
size, as was shown by a rough calculation based upon 
the distance of the particles from each other and the 
amount of gold present, must have been less than the 
wave-length of light. With ordinary illumination,2 even 
with the best objectives, they were not perceptible. 
The arrangement of the apparatus assembled for this 
purpose was the following: 
The sun beams fell upon a mirror S (Fig. 1), were re- 
flected from this into the lens L, which brought them to the 
focus at 5, over which was arranged a low power micro- 
scope (Objective A, Huygens eyepiece 2 and 4).3 
As early as April, 1900, solutions of glue, gelatin, 
tragacanth, stannic acid, and also a silver hydrosol of 
1 AuP8 contained gold-glinting particles which settled in about 
eight days. Au54 was over half a year old and in this time had de- 
posited but little gold. The particles were green in color, the fluid 
bright purple-red. 
2 And even by use of the dark field illumination then in use. 
3 Quincke had used side illumination with lamp or sun-light to 
observe the electric migration of starch grains, etc., under a magnifi- 
cation of 30 diameters. (Pogg. Ann., 1861, Vol. CXIII, p. 568.) DEVELOPMENT OF ULTRAMICROSCOPY 
99 
coarser subdivision, were examined by this method. In 
all these fluids, besides a general illumination of light- 
cone, individual particles could also be perceived;1 but 
with the silver hydrosol there was visible a brilliant light- 
cone, which could be resolved into individual particles 
only upon extreme dilution. Of colloidal gold solutions, 
Au60 and AuP3 were also investigated. 
Fig. 1. 
The slightly turbid gold solution Auco (0.005 per cent 
Au) showed a green light-cone which vanished upon 
dilution without having revealed the presence of indi- 
vidual particles; the perfectly clear solution AuP3 (0.005 
per cent Au) only showed the presence of isolated gilded 
dust particles, while the space between them appeared 
optically clear. That this space still contained gold in 
an invisible form, was made evident by an experiment: 
upon addition of ordinary table salt the colloidal gold 
was coagulated, and then the fluid was thickly filled with 
brightly shining particles. 
x Some of these individual particles polarize light, others on the 
contrary do not. 100 
DEVELOPMENT OF ULTRAMICROSCOPY 
By these experiments were settled several questions of 
importance. Thus it was shown 
1. That gold of various degrees of subdivision can be 
present in one fluid without settling; 
2. That gold can be subdivided to the point of optical 
homogeneity.1 
It was further shown experimentally that by the 
microscopic examination of a solar image formed in a 
fluid, particles can be discerned which are beyond the 
range of ordinary microscopy. I was convinced that by 
the use of better objectives the individual particles could 
also be seen in Au60 and similar fluids, and I determined 
to attempt, with the assistance of a specialist, to render 
such particles visible also. An investigation of this sort 
seemed all the more promising, as the preliminary experi- 
ments had shown that the color, brightness, and polariza- 
tion of the individual particles show a great variation, 
which might serve more sharply to characterize them; 
furthermore, the colloidal gold solutions represented 
test objects which made it possible to ascertain approxi- 
mately the average mass of the individual particles, from 
the mean distance between them, and the known gold 
content. 
The perfected arrangement, bringing into use objec- 
tives of wider aperture, was later discovered by H. Sie- 
dentopf incidentally to our mutual work, on the basis of 
the principles developed by him for making visible ultra- 
microscopic particles. But first another problem had to 
1 Subsequent ultramicroscopic examination of gold ruby-glasses has 
shown, even in such homogeneous appearing subdivisions, the polarized 
light-cone can be seen by focusing objectives of higher aperture on 
the very outside layer of the preparation, and thus preventing, as 
far as possible, the absorption of the transmitted and reflected light. 
This precaution must be observed in the examination of apparently 
homogeneous colored solutions. DEVELOPMENT OF ULTRAMICROSCOPY 
101 
be solved. Because of the motion of their particles, 
colloidal gold solutions were not suitable for an extremely 
exact determination of the mass of the individual parti- 
cles. For this purpose it was necessary to prepare a 
graduated series of subdivisions of gold containing fixed,1 
and, so far as possible, changeless particles. Test objects 
for this purpose I had made for me in the glass factory 
of J. L. Schrieber, at Zombkowice, Russia, incidentally 
to my attempt to work out a method for the manufac- 
ture of gold ruby glass which can be molded. After 
this preparation could be begun in collaboration with 
H. Sieclentopf, an attempt to render visible finer ultra- 
microscopic particles, in which the work was so divided 
that H. Siedentopf carried out the construction and 
improvement of the apparatus for illumination, while I 
took charge of testing the apparatus constructed with 
my preparations, and of developing the method for the 
determination of size. The introduction of the bilateral 
slit, and the application of microscopic objectives for 
illumination by H. Siedentopf, were essential to render- 
ing visible the smaller particles. Siedentopf also con- 
tributed the theory of making ultramicroscopic particles 
visible, and calculated the probable limits of their visi- 
bility. 
It must be here noted that the limits of microscopic 
resolvability (limit of visible "separation ") determined 
by Abbe and Helmholtz, is often confounded with the 
limit of "visibility." That isolated particles, whose 
diameter is a fraction of a wave-length of light, can still 
be seen, is expressly stated by Abbe himself. He writes 
in this connection:2 "Such objects can be seen, no mat- 
1 For the purpose of making an exact count when determining the 
size of the gold particles. 
2 E. Abbe, Gesammelte Abhandlungen, Vol. I, p. 362. Jena, 1904, 
G. Fischer- 102 
DEVELOPMENT OF ULTRAMICROSCOPY 
ter how small they may be; it is only a question of the 
contrast of the light effect, good definition of the objec- 
tive, and sensitiveness of the retina." As may be seen, 
the results of our investigation are not at all, as has been 
often thought, in conflict with the statements of Abbe. 
We received splendid assistance in our work from the 
numerous means placed at our disposal by the firm of 
Zeiss.1 We were thereby enabled to work with the best 
objectives, and complete the mechanical development of 
the apparatus in a short time. Notwithstanding this, a 
year and a half of study was necessary to bring the appa- 
ratus to its present state of perfection. 
1 F. Auerbach states in his monograph: Das Zeisswerk und die Carl 
Zeiss-Stiftung in Jena 2d ed., Jena 1904. G. Fischer: 
"Recently in the workshop of Carl Zeiss, their scientific collaborator, 
Dr. Siedentopf, has at the instance of Zsigmondy worked out inde- 
pendently an idea previously conceived by Abbe, into a method 
which makes it possible to see minute particles by means of intense 
illumination." 
As this statement can easily lead to a misunderstanding that the 
idea conceived of by Abbe was known to Siedentopf or myself, it 
must be expressly stated that on the occasion of the conversation 
which preceded the support of our work by the firm of Zeiss, Professor 
Abbe made no mention of this idea, and that it is unknown to me when 
and where Abbe has expressed such an idea. Some verbal expression 
of opinion is probably referred to, which was not available to a wider 
circle. CHAPTER V 
DESCRIPTION OF THE APPARATUS FOR MAKING VISI- 
BLE ULTRAMICROSCOPIC PARTICLES 
It is easy to see why, with the earlier microscopic 
methods of observation, the individual particles in col- 
loidal solutions or in ruby glass could not be seen. Trans- 
mitted light1 was mostly used for observations, and the 
eye dazzled by the profusion of light, could not distin- 
guish the slight differences of brilliancy caused by the 
diffraction of the light due to very small particles, just 
as it is impossible to see the stars by daylight.2 
To render visible very small particles, there are neces- 
sary : 
1. The most intense possible illumination of the parti- 
cles, in such wise that no beam of the illumination enters 
direct the eye of the observer. 
2. A field of view as dark as possible. 
These requirements were met by the construction of 
an apparatus for illumination, the description of which 
by H. Siedentopf,3 I here repeat. Fig. 2 shows the appa- 
ratus natural size. 
1 The so-called dark field illumination, then very incompletely 
developed, has been relatively little used; I do not therefore need 
to take it up further. 
2 See the detailed discussion of the principles of making visible 
ultramicroscopic particles: Sidentopf and Zsigmondy, loc. cit., pp. 2-5, 
and Siedentopf, Berl. klin. Wochenschr., 1904, No. 32. 
3 Siedentopf and Zsigmondy, loc. cit., p. 8. See also the detailed 
description of the apparatus for making visible ultramicroscopic 
particles, by H. Siedentopf, Druckschr.-Verz. der optischen Werkstatte 
von C. Zeiss, Sign. M., 164, Jena, 1904. 
103 104 
DESCRIPTION OF THE APPARATUS, ETC. 
"The solar rays reflected from a heliostat, enter the 
darkened laboratory through an iris diaphragm. In the 
room is an optical bench about 1.50 meters (5 ft.) long, 
having a metal flange P, supported on an adjustable stand 
G, all made by C. Zeiss, Jena. 
"On this, by means of carefully adjusted brackets, are 
mounted the individual parts of the apparatus. The 
Fig. 2. 
light-rays first enter the telescope objective F, having a 
focal length of about 10 mm., which throws an image 
of the sun about 1 mm. in diameter on the finely adjust- 
able slit head S, which is modeled after Engelmann's 
microspectral-objective.1 By the horizontal bilateral slit, 
this image can be reduced to 5-50 hundredths of a milli- 
1 Th. W. Engelmann, Zeitschr. f. wissensch. Mikrosk., 1888, Vol. V, 
p. 289. H. Siedentopf, Sitzungber, D. k. Akad. d. Wissensch., zu 
Berlin, 1902, Vol. XXXII, p. 717. DESCRIPTION OF THE APPARATUS, ETC. 
105 
meter, as desired. The width of the slit may be read off 
from an index on the drum connected with the screw. 
The edges limiting the height of the slit are movable 
horizontally, and may be placed from to 2 mm. apart. 
A polarizer N may be placed behind the slit when desired. 
The iris diaphragm J shuts off any side-light which may 
be reflected from the edges of the slit. By means of the 
chisel-shaped diaphragm B, one-half of the beam of 
light may be cut off; this is necessary when immersion 
objectives are used, in order to prevent objectionable 
reflection from the mounting of the front lens, due to 
the closeness of the objective. A second telescope objec- 
tive F2 of 80 mm. focal length forms a quarter-size 
image of the slit at the focal plane E of the condenser K. 
By means of the microscope objective A A,1 used as a 
condenser, this picture E (reduced to one-ninth its size), 
is projected into the preparation. It should be seen 
that full use is made of the aperture of the condenser 
system K, by controlling the illumination of its rear 
focal plane. The upper half of the rays emerging from 
the objective A A are cut out, when the picture of the 
semi-diaphragm B, thrown by the telescope objective F2 
into the after-focal plane of A A, darkens the upper half 
of this plane. By means of two micrometer screws, 
working in a horizontal plane and perpendicularly to 
each other, the condenser-objective may be readily cen- 
tered in the optical axis of the microscope proper." 
In order to bring the desired part of a solid preparation 
into the axis of the illuminating beam, Siedentopf devised 
a metal prism having slides permitting the vertical micro- 
metric motion of a little plane table. 
For the examination of fluids I had made a small 
washable apparatus illustrated in Fig. 3, which makes it 
1 AA is the mark on the Zeiss objective referred to. 106 
DESCRIPTION OF THE APPARATUS, ETC. 
possible to examine numerous fluids, one after the other, 
without being obliged to rearrange the microscope for 
each case. 
A cell with a quartz window C, which serves as obser- 
vation chamber, is connected by tubes on the one side 
with the thistle-tube T for filling, and on the other side 
with the pinch-cock H for drawing off. In a few seconds 
one fluid can be run off and replaced by another after 
washing the apparatus, which remains ready in place for 
the next observation. 
H. Siedentopf has very cleverly attached this little 
Fig. 3. 
apparatus direct to the objective of the microscope, thus 
obviating the necessity of repeated focusing on one object 
plane; and the fine adjustment of the microscope tube 
may be used to move the cell vertically. 
Recently W. Biltz proposed a slight modification of 
this apparatus. In order to avoid the rubber tube con- 
nections Biltz fused a thistle-tube direct on to the cell. 
The danger of increased fragility has been avoided by 
H. Siedentopf, who made appropriate changes in the cell 
and the method of fastening it to the microscope objec- 
tive. The manner of handling the new apparatus is DESCRIPTION OF THE APPARATUS, ETC. 
107 
shown in Fig. 4. The cell, as a whole, is placed in a 
pocket under the microscope, and held there by the clips 
d, h, and s. 
Fig. 4. 
H. Siedentopf has also constructed a microscope for 
the examination of ultramicroscopic bacteria, which has 
a new arrangement for illumination, entirely different 
from the one here described.1 
1 H. Siedentopf, loc. cit., Chapter IV, and Journal of the Royal 
Microscopical Society, 1903, pp. 573-578. 108 
DESCRIPTION OF THE APPARATUS, ETC. 
Another arrangement preventing the rays in ultra- 
microscopic work from entering the microscope proper 
by total reflection from the cover-glass, has been described 
by Cotton and Mouton.1 
' Cotton and Mouton, Comp, rend., June 1903, Vol. CXXXVL 
pp. 1657-59. 
A detailed description with directions for the use of 
the ultramicroscope and its attachments, is issued in 
English and in German by the makers of the instrument, 
C. Zeiss, Jena, and may be had by writing for the cata- 
logs known as Micros 227, 228, 229. 
The observance of two precautions will save much time 
for beginners : 
1. Be sure the substance examined contains particles 
that can be rendered visible by the illumination usqd. 
2. Avoid vibration of the instrument, especially when 
examining fluids. 
J. A. CHAPTER VI 
CERTAIN TERMS OFTEN USED HEREIN 
E. von Behring has proposed to call the herein described 
apparatus for rendering visible ultramicroscopic parti- 
cles, 11 Ultraapparatus " for short, and to designate as 
"Ultramicroscope" the microscope constructed by H. 
Siedentopf for the examination of bacteria.1 The pre- 
liminary research was carried out exclusively with the 
first apparatus referred to above; and the name pro- 
posed by von Behring will be often used. I must observe, 
however, that other investigators 2 use the name "Ultra- 
microscope " for the instrument termed by von Behring 
"Ultraapparatus." It is therefore advisable to keep the 
expression Ultramicroscope for both kinds of apparatus, 
and in case it is necessary to distinguish between them, 
to call the former Ultraapparatus, and find some short 
name for the latter. 
H. Siedentopf has also proposed to establish a further 
classification of the so-called ultramicroscopic particles.3 
He terms ultramicroscopic a particle or dimension which 
is below the limit of microscopic resolvability (in prac- 
tice about | /i). According as the ultramicroscopic par- 
ticle can be rendered visible or not, it is designated as 
1 H. Siedentopf, see note 3, p. 103. 
2 Raehlmann, Zeitschr. f. arztl. Fortbildung, 1904, No. 5. W. Biltz, 
Gottinger Nachr., 1904, No. 4. 
3 H. Siedentopf, Berlin klin. Wochenschr., 1904, No. 32. 
109 110 
CERTAIN TERMS OFTEN USED HEREIN 
" submicroscopic " or "amicroscopic." These terms will 
often be used here. 
In order to avoid continual use of the term "particles," 
I will introduce an additional simplification, and ampli- 
fying Siedentopf's proposed nomenclature, I shall call 
ultramicroscopic particles "ultramicrons," the submicro- 
scopic particles "submicrons," or "hypomicrons," and 
the amicroscopic particles "amicrons." CHAPTER VII 
PRINCIPLES OF THE ULTRAMICROSCOPIC INVESTIGA- 
TION OF FLUIDS 
The manipulation of the glass apparatus, for the exami- 
nation of fluids, illustrated in Fig. 3, is very simple. The 
microscope tube is taken out from the stand, the space 
between the objective and the upper quartz surface of 
the cell is covered with the immersion fluid (in the case 
of objective D* with water, which may be squirted 
between the objective and the cell from a wash bottle), 
after which the quartz window in front is dried with a 
clean linen cloth. The funnel is then filled with a turbid 
fluid (diluted milk, a solution of gum gamboge, commer- 
cial colloidal silver, etc.), the pinch-cock H (Fig. 3) is 
opened until all air is expelled from the apparatus, which 
is sometimes effected by tapping or holding the tube 
slanting.1 Then the tube, together with the objective 
and the washable apparatus attached to it, is again slid 
into the stand; the microscope proper is moved so as to 
bring within the field of view the apex of the illuminated 
cone (which can be easily found in the turbid fluid), and 
care is taken that the narrowest constriction of the cone 
is brought to the center of the field.2 
1 Good cells fill up at once with the fluid and retain no air bubbles; 
the turbid fluid must be greatly diluted. 
2 Beginners will best learn the necessary manipulation by practice. 
Before using the apparatus it is necessary to see that the glass tubes 
connected by rubber tubing approach each other closely. 
111 112 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
When the apparatus is adjusted, the fluid used for this 
purpose is run off by opening the pinch-cock H (Fig. 3), 
and the cell rinsed out with water until all turbidity- 
producing particles are removed; the apparatus is then 
filled with the fluid to be examined, the first portion of 
it, in quantitative determinations, being used to wash 
out the water. 
I will now take up the description of examinations of 
a general character, and will begin with distilled water. 
i. Distilled Water 
As Spring has found, distilled water is not optically 
clear; it can be freed from the last traces of suspended 
or colloidally dissolved substances only by somewhat 
complicated operations. It is evident, therefore, that 
these impurities prejudice the use of ordinary distilled 
water in ultramicroscopic investigations to such a degree 
that it cannot be used to dissolve colloids, or to dilute col- 
loidal solutions. From experience, however, it has been 
learned that in the case of water which stood long enough 
protected from dust, the dust particles are present in 
relatively small number, so that only a few of them 
come into the field of view, and the space actually under 
observation is for the most part free of them. The use 
of such water for many investigations, therefore, seems 
permissible.1 In between the suspended dust particles 
which float perfectly quietly, the water either appears 
1 Stirring with glass rods and taking water from squirt-bottles, should 
both be avoided. Water is best drawn from a fixed vessel whose 
siphon tube does not reach the bottom, into a beaker previously rinsed 
out, and from this it can be poured into the thistle tube of the appa- 
ratus. It is advisable to set up several fixed vessels. The siphon is 
closed with a rubber tube and a pinch-cock. (See also the suggestions 
of Biltz and Gahl, mentioned in Chapter XX of this book.) COLLOIDAL SOLUTIONS 
113 
optically clear, or shows upon further opening the slit 
(whereby a light-cone of greater depth is seen) a weak, 
diffused shimmering of light, due to traces of colloidally 
dissolved substance,1 which do not disturb ordinary 
investigations, where a narrow opening of the slit is 
necessary. 
In further prefecting the method, and in the examina- 
tion of fluids whose particles approach the limit of visi- 
bility, there should of course be used only water which 
shows no light-cone in the ultraapparatus. By double 
distillation and prolonged standing water can be obtained 
that is suitable for refined work; in which case dust 
must be excluded as far as is possible. 
Recently W. Biltz has published a very good method 
for the production of clean water (see Chapter XX). 
2. Colloidal Solutions 
In colloidal solutions the microscopic form of the 
light-cone is distinctly different from that of distilled 
water. The distance between the particles in them is 
usually less than | and in such cases there appears a 
more or less intense homogeneous polarized light-cone; 
very often larger particles are present which shine 
brighter and give the light-cone a heterogeneous appear- 
ance. In order to form an idea as to the individual 
particles, extreme dilution (0.001-0.0001 per cent and 
less) is almost always necessary; then only does the 
distance between the particles become so great that its 
microscopic resolution is possible. 
If the light diffracted from the individual particles 
is intense enough to create an impression on the retina, 
1 Probably of silicates from the glass or of metal oxids from the 
cooling tube. 114 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
the microscopic pictures of the particles will appear as 
little reflecting disks; in the case of submicrons, with 
the most favorable illumination, this is always the case. 
If the diffracted light is too weak, the light-cone will 
gradually vanish with increasing dilution, and there is 
no optical resolution, the solution in such cases contain- 
ing amicrons. 
It is very important to distinguish the casual, sus- 
pended particles which, especially in concentrated solu- 
tions, are often present in large numbers, from the regular 
constituents of colloidal solutions. This can be accom- 
plished by extreme dilution; the spaces between the 
suspended particles then become so large that the light- 
cone which lies between them and is produced by the 
smaller particles, can easily be recognized. Upon still 
further dilution there is re-solution into individual parti- 
cles (submicrons), or else the light-cone vanishes (ami- 
crons). In the latter case the presence of extremely 
finely divided substance can be demonstrated by the 
addition of optically clear precipitants.1 By such addi- 
tion the colloidally dissolved particles can be forced to 
combine into larger ones, which frequently, then thickly 
fill the field. 
In this manner I demonstrated the presence of col- 
loidal gold and of albumen in almost homogeneous 
solutions. 
Those who later enter on this field of work will do well 
always to make such a test, for the uninitiated are very 
apt to be deceived by large particles, especially in the 
examination of concentrated solutions.2 
' Sometimes by boiling the fluid, as for example with albumen. 
2 In dilution experiments it is to be observed in addition, that dis- 
tilled water sometimes precipitates individual colloids, globulin for 
instance being precipitated by water. In such cases instead of water a COLLOIDAL SOLUTIONS 
115 
Polarization. Tyndall has already shown that light 
is polarized by particles which are very small as com- 
pared with its wave-length, whereas by larger particles 
this effect is produced only partially, or not at all. In 
connection with Rayleigh's theory, these facts were 
often investigated and Tyndall's observations confirmed.1 
I might here make mention of an observation which 
may seem very strange with reference to the foregoing, 
but wdrich has been repeatedly confirmed in the case of 
colloids; there are particles floating in the fluid, which, 
notwithstanding their relatively large size as compared 
with the wave-length of light, totally polarize it, and 
can therefore be rendered invisible by turning the Nicol. 
Particles of this kind I have observed, for instance, upon 
the precipitation of a dilute solution of CuSCU by Na2COs; 
the light diffusely scattered by the precipitated hydrogel 
of copper carbonate proved to be polarized, even if the 
flocks were so large that they could be seen with the 
naked eye. 
The same is the case with colloidal ferric oxid; its 
solutions, as well as its flocculent precipitate, plane 
polarize a beam of light in the same direction. 
This phenomenon appears to be quite common, and 
shows itself as soon as hydrogel particles float in the 
fluid, no matter whether they are large or small as com- 
pared with the wave-length of light. This is no excep- 
tion to the rule of Tyndall, above mentioned, which 
diluent must be chosen which exerts no precipitative action on the 
ultramicrons, providing it is desired to obtain the particles in their 
original condition. In other cases increasing dilution may work 
disruptively upon the individual particles, by transformation into 
crystalline solution, for example. Cases of this kind have not yet been 
carefully investigated. 
1 See for instance B. Pernter, Wiener Denkschriften der Akademie 
der Wissenschaften, 1901, Vol. LXXIII. 116 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
refers to little drops of fluid in steam-clouds, and also 
to such emulsions as the turbid mastic solution, the 
polarization of which was closely investigated by Pernter, 
who among other things concluded "that the presence 
of larger particles involves a decrease of the polarization, 
in proportion to the size and number of such particles 
present." 1 
The hydrogel particles must, however, be conceived as 
being thoroughly permeated with fluid. According to 
Butschli, hydrogels possess a fine, web-like structure, or 
according to Quincke, they consist of foam-cells, or foam- 
chambers with invisible walls. Some such structure, 
but much finer and perhaps otherwise constituted, must 
be assumed in the case of each hydrogel particle, even if 
it is smaller than a wave-length of light, that is so small 
that the examination of its structure is no longer possible. 
Returning to the ultramicroscopic examination of gel 
particles, a striking difference is at once evident between 
such an examination and that with the naked eye. 
While the rays diffracted by the flocculated particles 
of the Fe2O3 hydrogel, when macroscopically observed, 
do not appear much brighter than those of the light- 
cone in the original hydrosol, and can be almost com- 
pletely blotted out by turning the plane of polarization 
of the Nicol prism, there can, however, upon ultramicro- 
scopic examination of the same fluid, be seen in addition 
to numerous small particles, dazzlingly bright larger 
particles whose light, although weakened, cannot be 
entirely blotted out by any position of the Nicol prism. 
On the other hand the less brilliant particles partially or 
completely disappear with crossed Nicols, sometimes 
even if they have reached considerable size. 
An explanation of this phenomenon I will not here 
1 Loc. cit., p. 11 (p. 310). COLLOIDAL SOLUTIONS 
117 
attempt, but will merely call attention to the fact that 
practically no inference as to the size of hydrogel parti- 
cles can be drawn from the degree of polarization. For 
a determination of their size, or more properly the mass 
of the colloid-substance contained in them, recourse 
should be had to methods which will soon hereafter be 
described. 
Color. Ultramicroscopic particles are usually colored, 
the color of their tiny reflecting disks, when the particles 
are alike, being nearly complementary to the color of the 
mass of the fluid by transmitted light, as has already 
been mentioned.1 
So far as our present experience goes this relation is 
general; it indicates-in harmony with the behavior of 
turbid media-that the color of the fluid is produced by 
the preferential separation from the impinging beam of 
light, of those component rays which are scattered by 
the individual particles. It would seem reasonable to 
extend this view to the rays scattered by the molecules 
of dye-stuff solutions, but such an hypothesis is palpably 
incapable of demonstration because of the extreme weak- 
ness of the diffraction in particles of such small mass. 
In order to get a correct idea of the color of the indi- 
vidual particles, it is quite evidently necessary to dilute 
the fluid until a disk one quarter centimeter2 thick is 
practically colorless. 
3. Determination of the Size of Particles 
In the paper 3 frequently before referred to, two methods 
were described in detail for the determination of the size 
1 Siedentopf and Zsigmondy, loc. cit. 
2 The approximate length of the path of the light beam in the fluid. 
3 Siedentopf and Zsigmondy, loc. cit., pp. 16-29. 118 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
of ultramicroscopic particles. It was there pointed out 
that these methods give only the average mass of the 
A 
individual particles, according to the formula -, where A 
n 
is the mass of the substance subdivided, and n is the num- 
ber of particles in it. 
For the purpose of tabulation in the case of gold ruby- 
glass, the average linear dimension I of the individual 
particles was figured out on the assumption that the 
gold particles are cubical 1 and are of solid gold, having a 
specific gravity s = 20,2 that is, according to the formula 
First Method. According to this method a certain 
illuminated volume of glass or of fluid (V) is blocked out, 
Fig. 5. 
the size of this volume determined, and the particles 
contained in it counted. By means of the eye-piece 
micrometer a part of the cone of rays dd (see Fig. 5), 
may be sharply defined from side to side, whereby the 
length and breadth of the volume chosen may be known. 
1 That this assumption is not quite accurate is brought out on p. 36 
of the paper quoted. Regarding the detailed discussion of sources of 
error, I refer to pp. 21-29 of the same paper. 
2 Rose found the specific gravity of precipitated gold to be 19.55 to 
20.7. DETERMINATION OF THE SIZE OF PARTICLES 
119 
The depth of the illuminated volume V1 thus defined 
may be easily determined with the eye-piece micrometer 
upon the quarter (90°) of rotation of the graduated slit 
(see Fig. 2, Chapter V). 
The mass A of the subdivided substance contained in 
the volume, is found directly from the concentration, 
providing the total mass of the subdivided body is present 
in the form of ultramicroscopic particles, as in many 
colloidal gold solutions. When, however, one part of 
the subdivided substance is in homogeneous solution (a), 
and another part is present in the form of submicroscopic 
particles (/?), then since is unknown, we take, in place 
of the latter the sum and thus obtain a major 
limit for the size of the particles.2 
For full details for the determination of volume, as 
well as for the sources of error in this method, I may 
refer to the original article, merely remarking here that 
in exact determinations of the size of the particles, the 
error does not exceed twenty 'per cent of the linear dimen- 
sion, and is usually considerably less. 
Second Method. From the distance between the parti- 
cles, and using the same assumption, the linear dimen- 
sion I may be obtained by means of the formula 
7 
X S ' 
where r is the mean distance between the particles. 
1 For particulars see pp. 17, 21, and 22 of the paper referred to. 
It will be seen that the depth of the illuminated volume thus deter- 
mined can enter the calculation only if it is less than the depth of 
view of the microscope. 
2 Sometimes /? can be indirectly determined, as in the case of gold 
ruby glass, where we can determine /? by colorimetry. The major 
limit is not much different from the actual size of the particles, pro- 
vided that a does not constitute the bulk of the substance present. 120 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
In using this method it should be carefully noted that 
the mean distance between the particles can be correctly 
measured only when the depth of the layer under obser- 
vation is at least as large as this mean distance. The 
depth of the layer in question must therefore be exactly 
determined; for if this be not done, or if it is impossible, 
apparent distances are obtained which may sometimes 
differ in no small degree from the actual distances. 
As mentioned in the paper above referred to,1 the 
measurement of the distance between the particles in 
the case of fluids is made very difficult by their motion; 
numerous measurements must therefore be made on one 
and the same fluid in order to obtain an average of prac- 
tical value. The probable accuracy of the results 
obtained is considerably increased, if various degrees of 
dilution are examined. 
In the case of colloidal gold solutions I have for the 
most part worked as follows: Each fluid to be investi- 
gated was first reduced to a content of 0.005 per cent of 
metallic gold, and when possible, the first series of counts 
made with this. Thereafter the hydrosol was diluted 8, 
27, 125 times, or 10, 100, some times even 1000 times, 
and at each dilution the measurement of the distance 
between the particles again repeated. The mean of all 
the figures was taken, and from this was calculated the 
distance between the particles in the original concen- 
tration; and as a control of the value thus obtained, in 
many instances the number of particles contained in a 
given volume was determined by the first method. 
First Method Applied to Fluids. As the particles are 
in continuous motion, it is not possible to count off a 
large number of them; it is therefore necessary to so 
limit the volume of the fluid illuminated, or so dilute a 
1 Siedentopf and Zsigmondy, loc. cit., p. 29. DETERMINATION OF THE SIZE OF PARTICLES 
121 
given volume of the fluid, that on the average about 
from 2-5 particles are contained therein, which number 
may be determined at a glance. 
By frequent repetition of this same procedure in dif- 
ferent parts of the fluid, and with various degrees of 
dilution, there may be obtained a pretty good average 
value of the number of particles in a certain volume, 
which thus serves to control the value obtained by the 
other method.1 
The results of neither method, of course, have any 
great degree of accuracy, but they both satisfactorily 
serve the purpose of this book-to establish some kind 
of order in a field hitherto inaccessible to direct observa- 
tion. The brightness of the diffracting disks also fur- 
nishes a good control for the size of the particles, because 
1 With fluids which contain a variety of different sized particles, 
it is evident that the number of particles in a given volume will not 
be found proportional to the concentration of the colloidally dissolved 
substance, but that in the more concentrated ones it is too small. 
For example, with a silver hydrosol (Bredig's method) I obtained, as 
an average of about 20 separate measurements with a concentration, 
C = 0.0037 per cent: 
Concentration in 440 /? Calculated to concentration C 
C 3.50 particles 3.50 
C/2 2.28 " 4.56 
C/4 1.95 " 7.80 
C/8 1.62 " 12.90 
Upon greater dilution many more particles are therefore counted 
than with greater concentration, and the reason of this is chiefly because 
the smaller particles become distinctly visible only when by great 
dilution the larger particles are in part removed from the field of 
view. 
The average linear dimension, according to the first count was, 
found to be too large in the proportion of 12.9 : 3.5, that 
is about 1.5 times greater than by the last count. In such cases 
it will be better to report, not the average value, but the superior 
limit determined by various degrees of dilution. 122 
ULTRAMICROSCOPIC INVESTIGATION OF FLUIDS 
in general those particles of a certain substance which 
have been determined to be the largest, must naturally 
diffract most light. The brightness and color of the 
diffraction-pictures, as well as their motion, are such 
characteristic features of the particles that it is easy, in 
a mixture of two fluids of different properties, to deter- 
mine by ultramicroscopic observation of the particles, 
which belong to one fluid and which to the other. 
The results of the series of investigations hereafter 
given, thus mutually support each other. 
In addition it may here be stated that gold particles 
whose linear dimension was found to be 15 /z/z (15.10-6 
mm.), lie at the limit of visibility with illumination 
by the electric arc,1 and that smaller particles can be 
discerned only by the use of sunlight; while the smallest 
that can as yet be rendered visible, which are about as 
small as 5 /z/z, can be seen only with the brightest sun- 
light on very clear days and with efficient means for 
cutting off all extraneous light. 
As the index of refraction of metallic gold is very 
different from that of water, and this difference is favor- 
able for rendering visible ultramicroscopic particles, it 
stands to reason that particles of other substances which 
are smaller than 15 /z/z cannot be made visible by arc- 
light with the present apparatus, excepting those of some 
few metals and perhaps those of fluorescent dye-stuffs. 
I am obliged to state this here, because many divergent 
statements have been made by others about the limits 
of visibility of ultramicroscopic particles, and such state- 
1 This statement refers to our apparatus ordinarily used for examining 
fluids (objective AA [num. ap. = l/3] for illumination, and objec- 
tive D* [num. ap. = 3/4] for observation). According to Siedentopf 
the brightness of the diffraction disks is proportionate to square of 
the product of the numerical apertures of the objectives used for 
illumination and observation. DETERMINATION OF THE SIZE OF PARTICLES 
123 
merits might easily lead to an exaggerated idea of the 
method described by Siedentopf and myself. Without 
further remarks, it follows from what has been said that 
when colored particles are observed with the ultra- 
microscope in the solutions of dye-stuffs like acid fuchsin, 
naphthol yellow, etc., which dissolve as electrolytes, 
these particles cannot be molecules or ions of the dye- 
stuff in question, that is, they cannot be its smallest 
particles. CHAPTER VIII 
PREPARATION OF COLLOIDAL GOLD SOLUTIONS 
The gold hydrosols used for the experiments hereafter 
described, were prepared by two different methods. 
One I have described in Liebig's "Annallen," Vol. 301, 
p. 30; the other will be given here. 
First Method. A dilute solution of chlorid of gold is 
mixed with potassium carbonate and reduced at the 
boiling-point by formaldehyde. If bright red clear fluids 
are desired, which can be used as colloid reagents, the 
details described in the "Zeitschrift fur Analytische 
Chemie " 1 must be exactly followed. 
The chief essential is distilled water of sufficient purity. 
The absence of electrolytes is not so essential as the 
absence of colloids; traces of the former do no harm, 
seeing that a certain quantity of them is introduced 
into the water by the reagents themselves. On the other 
hand traces of colloids almost always present in all com- 
mercial distilled water completely prevent the formation 
of bright red gold hydrosols. With such water there 
are usually obtained blue, violet, purplish suspensions 
which, in a few days or weeks, deposit the greater part 
of their gold. 
For this reason I gave directions to redistil the dis- 
tilled water, using a silver worm for condensation.2 
1 Zeitschr. f. analyt. Chem., 1901, Vol. XL, p. 710. 
21 often obtained bright red fluids by using frozen water, and once 
124 PREPARATION OF COLLOIDAL'GOLD SOLUTIONS 
125 
Some impurities which I have found particularly 
injurious are: Phosphates of the alkaline earths, silicates 
from the glass itself, colloidally dissolved substances 
originating from the cooling worms ordinarily used, 
organic substances in commercial distilled water, and 
others of like kind. 
It is not difficult to see how these impurities act. The 
colloidally dissolved substances permeate the entire 
water, and in spite of their very small quantity are so 
thoroughly distributed that the distances between their 
particles are for the most part beyond the resolving 
power of the microscope. Therefore, as above described, 
they appear upon ultramicroscopic observation as an 
extremely faint homogeneous, polarized light-cone, which 
encloses much larger dust particles. The finest dust of 
Konowalow's1 fluids, which according to the publica- 
tions of this investigator play an important role in the 
turbidity of critical solutions, may, in so far as aqueous 
fluids are concerned, be the same as the colloidal impuri- 
ties here described. 
The mutual precipitation of colloids makes possible 
the formation of large particles capable of producing 
turbidity. As Picton and Linder,2 Lottermoser,3 and 
recently W. Biltz 4 have shown, colloid solutions with 
oppositely conducted particles mutually precipitate each 
other. When therefore reduced gold and the colloidal 
even by using water condensed in a tin cooler, but in any event only 
after prolonged distillation. The same was observed by W. Biltz, 
Gottinger Nachr. 1904, No. I. 
' Konowalow, Drude's Ann., 1903, Vol. X, pp. 360-392 and also 
Vol. XII, pp. 1160-64. The term "finest dust" is perhaps not a 
happy one, for ordinary dust particles are as a rule incomparably larger. 
2 Loc. cit 
s Ueber anorganische Kolloide, Stuttgart, 1901. 
4 Berichte, 1905, Vol. XXXVII, pp. 1095-1116. 126 
PREPARATION OF COLLOIDAL GOLD SOLUTIONS 
impurities are oppositely conducted, there is a possi- 
bility that a coarse turbidity may be found.1 
Other conditions may enter, but to discuss them would 
lead us too far. 
By the use of sufficiently pure water a quantitative 
reduction of gold may be obtained with formaldehyde in 
fluids which contain only 0.00005 per cent of gold, that 
is, one hundred less times gold than the bright red ones, 
and the fluids thus obtained may be evaporated without 
decomposition to one-fiftieth (1-50) of their volume, and 
still further. In such fluids the gold is amicroscopic, 
that is, in particles so small that they cannot be made 
visible. 
The second method for preparing almost homogeneous 
gold solutions may be regarded as a combination of my 
method with that of Faraday. This method has the 
advantage of making it relatively easy to obtain the 
4 Recently Jordis (Zeitschr. f. Electrochem., 1904, p. 510) has 
advanced the opinion that the impurities dissolved in water are neces- 
sary, as "sol-formers," to the formation of colloidal metals. Upon 
the basis of his observations on silica and also the observations of 
Lobry de Bruyn and Paal, Jordis infers that it is easier to obtain 
more concentrated solutions of colloidal gold in the presence of certain 
organic substances (protective colloids) than without them. It should 
not be overlooked that the colloids of Lobry de Bruyn and Paal are 
analogous to the purple of Cassius, showing a much different behavior 
with reagents and upon desiccation than the colloidal gold made by 
Bredig's method or my own, and being subject to different conditions 
of formation. To this I shall later return. A protective colloid would 
by all means facilitate the production of colloidal gold and act as a 
"sol-former"; the impurities contained in water, however, generally 
have the opposite action, and are not "sol-formers," but sol-destroyers. 
Therefore the advice to work with purest -water possible. The addi- 
tion of alkali or alkali carbonate produces, as Bredig has shown, a 
decrease in the concentration of the hydrogen ions, which also militate 
against the formation of metal hydrosols, or an increase in the concen- 
tration of hydroxyl ions which are favorable to the formation of gold 
hydrosols (see also pp. 64 and 77-79). PREPARATION OF COLLOIDAL GOLD SOLUTIONS 
127 
finest subdivisions even in more concentrated solutions 
(0.005 per cent Au). 
I use in exactly the same concentration the same mix- 
ture as with reduction by formaldehyde, but instead of 
heating with formal, reduce at room temperature with 
a few drops of an ethereal solution of phosphorus.1 (Pro- 
portions 120 cc. water condensed in a silver cooler, 15 mg. 
AuHC14.3H2O, 37 mg. K2CO3). Here, as in the other 
method, the use of colloid-free water is essential. The 
formation of the hydrosol takes place slowly, the fluid 
first becoming bright brownish red, and then gradually 
bright red, often with a tinge of brown-red, without show- 
ing the slightest turbidity to transmitted or reflected 
light. 
If a good solution has been prepared,2 it can be seen, 
upon testing with a lens and sunlight, that in it diffuse 
dispersion is almost entirely lacking, and that it is almost 
indistinguishable from a mixture of the reagents without 
gold. Ultramicroscopically there may be seen in it a few 
gilded dust particles, and among them an unresolvable, 
extremely faint light-cone, which soon disappears upon 
diluting the fluid with the purest kind of water, no indi- 
vidual particles being discernible. 
This fluid stands boiling without alteration, in contra- 
distinction to that prepared by Faraday, which was 
thereby rendered turbid, and deposited its gold. 
The contamination of 120 cc. of the purest water (con- 
densed in a silver cooler) with a few drops of water in 
which had been stirred up some powdered glass, or the 
1 It is best to take about one-half cubic centimeter of an ethereal 
solution of phosphorus, made by diluting a concentrated solution with 
five times its volume of ether. 
z From a larger number of separately prepared fluids the best must 
be selected in order to obtain fluids of the greatest homogeneity. 128 
PREPARATION OF COLLOIDAL GOLD SOLUTIONS 
use of commercial distilled water for the preparation of 
the gold solution, yielded sometimes very turbid fluids, 
which, like Faraday's, become violet and settle upon 
boiling, instead of being nearly homogeneous.1 
The influence upon the reduction of gold of the col- 
loidal impurities in distilled water was not known to 
Faraday;2 he obtained, therefore, upon reduction with 
phosphorus, a variety of fluids, mostly suspensions which 
settled soon after their preparation, but sometimes stable 
colloidal solutions, which, however, would not stand 
boiling. 
Third Method. Proceed as in the first method, but 
shortly before the addition of the formaldehyde add to 
the 120 cc. of water from 0.5 to 4 cc. of the nearly homo- 
geneous gold solution produced according to the second 
method. The tiny gold particles act as nuclei (see Chap- 
ter XVII, 3), and grow in the reducing mixture. Accord- 
ing to the quantity of gold fluid added, there are obtained 
gold solutions with larger or smaller particles.3 
1 In these experiments no alkali carbonate was added. 
2 He also knew nothing of the beneficial influence of the addition 
of alkali carbonates on the formation of finer subdivisions (for the 
purpose of neutralizing the acid formed.) Nevertheless very profound 
subdivisions may be obtained with phosphorous without K2CO3, 
providing the other conditions are observed. The age of the phos- 
phorous solution used and its degree of dilution also influence the 
properties of the gold solution. 
3 Zsigmondy, Z. phys. Chern., Vol. LVI, p. 65 (1906). CHAPTER IX 
ULTRAMICROSCOPIC EXAMINATION OF THE SOLUTIONS 
OF GOLD 
The figures given in the following table are the result 
of a very careful investigation. Apart from the fact that 
the work was several times interrupted and taken up 
again, it was further complicated because for the most 
part I needed direct, bright sunlight, which is somewhat 
rare in Jena; it often happened that just when all the 
preparations for an examination had been made, a cloud 
would pass before the sun. 
The hydrosol AU97 was most thoroughly investigated, 
an average value being obtained based upon more than 
three hundred individual measurements made in collabora- 
tion with Dr. F. Kirchner on three different concentra- 
tions. 
We found as the average value of the distance between 
the particles: 
Distance 
Observed. 
Distance Calculat- 
ed for Undiluted 
Fluid. 
With undiluted fluid 
0.972 p 
2.33 ft 
5.67 p 
1.16 H 
1.13 H 
Upon dilution, 1:8 
Upon dilution, 1:125 
Average of combined observations 1.13 fi. 
Size of particles calculated therefrom 15.25 pp. 
The values for the sizes of particles in fluids which 
were less carefully examined, will show considerably 
129 130 
EXAMINATION OF THE SOLUTIONS OF GOLD 
greater variation from the true mean, the error in some 
cases being perhaps thirty per cent. There is great 
uncertainty in the values for fluids about the limit of 
visibility. On the other hand the accuracy of the results 
are strongly supported by the practice which the eye 
gets in estimating the size of the particles from the bright- 
ness of the diffraction-images.1 
As often previously stated, our method yields only an 
average value for the mass of the particles; to make com- 
parison convenient, the linear dimensions wTere calcu- 
lated therefrom as before, on the assumption that the 
particles are cubical and completely filled with metallic 
gold. I must, however, expressly caution against assum- 
ing that this imaginary cubical or globular form cor- 
responds with the actual fact; many observations indi- 
cate that never, or only in isolated cases, do the metal 
particles have a similar shape on all sides. To this sub- 
ject I shall later return. 
In the first column of the following table are given 
the names of gold solutions, AuP designating the solu- 
tion obtained with phosphorus by the second method, 
while AU73, Auioi, etc., designate those obtained by the 
first method. With the exception of Au73 and Au73a, 
which were reduced from a 0.0005 per cent solution of 
gold chlorid, all the other liquids when they were made 
contained from 0.005 to 0.006 per cent of gold. For the 
purposes of examination they were diluted to 0.005 per 
cent Au, and fractions of this percentage. 
In the second and third columns are given the colors 
of the particles and the fluid, and when no individual 
particles were visible, that of the light-cone. 
The length of the path described by the submicro- 
1 As has been brought out in Drude's Ann., Vol. X, p. 29, the differ- 
ences in brilliancy are extremely great. EXAMINATION OF THE SOLUTIONS OF GOLD 
131 
scopic particles, designated by A in column V, serves 
merely for a preliminary orientation. The motion of 
translation is made conspicuous, while the much smaller 
motion of oscillation, to which every particle is subjected, 
is not given. As a rule-not without exception-it may 
be stated that the motion is more rapid the smaller the 
particles and the greater the number contained in unit 
volume. Upon dilution, therefore, the activity of the 
motion decreases somewhat. 
Very large particles of gold and gilded dust particles 
float, for the most part, quietly in the diluted fluid; in 
concentrated ones they are sometimes struck by the 
smaller particles, thus acquiring motion. 
In the table the fluids are arranged according to the 
size of the particles. 
Glancing at the table, there is seen at the top a very 
interesting fluid: AuPie. When just made, this is bright 
red, and completely clear both to reflected and trans- 
mitted light. The gold in it is subdivided to a degree 
approaching homogeneity, for outside of coarse yellow 
particles about 0.05 mm. apart (probably gilded dust), 
hardly a trace can be seen of the bulk of the gold, even 
in the ultraapparatus. 
The milliards of individual particles which fill every 
drop of the other fluids (for instance AU97 or Auioi) are 
here completely lacking, or more correctly, they escape 
observation on account of their small size. Their ability 
to diffract light has become so reduced that even in their 
totality, they do not diffract more light than is absorbed 
by the colored fluid, so that it is doubtful if any light- 
cone due to gold particles in the solution can be seen at 
all. That this is a colloidal solution of metallic gold, 
and not some red chemical compound of gold, can at once 
be proven by the addition of electrolytes. As the experi- 132 
EXAMINATION OF THE SOLUTIONS OF GOLD 
ment in the second section of the table shows, the well- 
known color-change to blue then takes place, and the 
light-cone-due to gold particles-then becomes visible. 
(For more in this connection see Chapter XII.) That 
reduction with phosphorus yields the metal, Faraday1 
has already shown. I have repeatedly tested the fluid 
AuP, and have reached the conclusion that, if my instruc- 
tions are adhered to, almost homogeneous gold solutions 
can be obtained. By reducing the gold with formaldehyde, 
I obtained, as already mentioned, similar fine subdivi- 
sions only when I had at my disposal water of especially 
good quality, and diluted the gold chlorid solution before 
reduction to from eighty to one hundred times its usual 
concentration, that is down to about 0.000005 per cent 
of gold. 
Between the finest subdivisions of gold just described, 
and those whose particles just become visible under most 
favorable conditions (Au73a, line 7), there are a whole 
series of transition forms which can at present be pro- 
duced, but whose detailed description would carry us 
too far. 
Upon examining the fluids in order of the size of their 
particles, there are seen in the ultraapparatus, with 
increasing size, the following phenomena: the light-cone 
is hardly discernible (line 1, AuPio); it becomes more 
distinct and brighter (line 3), but vanishes upon dilution; 
the color of the light-cone may be recognized (lines 4-6), 
it becomes still brighter, and acquires a heterogeneous 
appearance, but the individual particles are still amicro- 
scopic (AU73); upon sufficient dilution and with the 
brightest sunlight the individual particles become visible 
1 Phil. Trans, of the Roy. Soo. of London, 1857, Vol. CXLVII, p. 163. 
2 Slight deviations from the directions may result in very turbid 
fluids. TABLE V 
i. 
Designation 
of the 
Solution. 
n. 
Color 
of the Fluid. 
nr. 
Color of the 
Particles (P).1 
Color of the 
Light-cone (Lk). 
IV. 
Average 
Size of 
Particles, 
Calculated 
in jU/Z. 
A< in /z. 
VI. 
Date of 
Preparation. 
VII. 
Date of 
Observation. 
Remarks. 
1 
AuPi0 
bright red 
? 
? 
March 13, 1902 
Single yellow particles 40-50/t apart, with no 
light-cone visible between them, or at most 
only the suggestion of one. 
2 
AuP16 
+ 
NaN03 
blue 
Lk. and P. golden 
yellow 
? 
March 13, 1902 
The addition of NaNOs produces a golden 
yellow fog: gradually an immense number 
of small golden particles become visible. 
3 
AuP18 
red 
Lk. whitish 2 
April 10. 1902 
Bright particles, between them a homoge- 
neous light-cone, not resolvable even upon 
dilution. The light of the cone shows, 
linear polarization. 
4 
AuP18 
red with a 
tinge of 
violet 
Lk. yellowish 
April 17, 1902 
Light-cone not resolvable; upon dilution it 
becomes more and more indistinct until it 
vanishes. 
5 
AuP18 
violet red 
Lk. golden yellow3 
April 23, 1902 
Light-cone quite intense, but not resolvable 
even upon dilution. Can be made to vanish 
upon turning the Nichol. 
6 
Au73 
pink 
Lk. greenish 
July 6, 1901 
March 22, 1902 
Diffuse light-cone, which at times assumes a 
heterogeneous appearance. 
7 
Au73a 
pink 
Lk. greenish 
about 6 
over 10 
July 6, 1901 
April 7 and 15, 
1902 
Light-cone resolvable only by brightest sun- 
light: extremely lively motion. 
8 
Au82 
bright red 
P. green 
about 10 
3-10, 
some 20 
March '5, 1902 
April 23, 1902 
Fluid thickly filled with particles in extremely 
active motion; measurement, of distance 
between particles possible only upon ex- 
treme dilution. 
9 
Au97 
bright red 
green, few golden, 
yellow P. 
15.3 
March, 1903 
Observers: Kirchner and Zsigmondy. 
10 
Aujoi 
bright red 
P. green, 
some few yellow 
15.5 
Nov. 3, 9. and 13. 
1903 
Light-cone hardly resolvable by arc-light. 
Individual particles easily visible by sun- 
light. 
11 
Auas and egg 
albumen 
red 
green and yellow 
P. 
15.5 
2-15 
3-9 
Filtrate of a colloidal gold solution which had 
passed, without residue, through a Pukall 
filter. 
12 
Au92b 
bright red 
P. green 
17 
April 29, 1902 
Clear fluid. 
13 
Au90 
bright red 
P. green 
18 
2-15 
March 13, 1902 
March 14. 1902 
14 
Au9ia 
violet red 
P. yellow and green 
about 23 
1-3 
March 15, 1902 
April 4 and 6, 1902 
Particles possess a labored trembling motion. 
15 
Au 83a 
violet red 
P. red, green, 
yellow 
32 
0.7-1.5 
some 3-6 
Feb. 19. 1902 
March 13 and 21, 
1903 
Turbid solution, partially settled; particles 
of unequal size, the average value therefore 
not expressing the true proportions. 
16 
Au208a 
bright red 
green 
32 
Slightly turbid fluid. 
17 
Au70 
violet 
yellow, red, green 
35 
1-7 
July 6, 1901 
March 23, 1902 
In making this fluid ordinary distilled water 
was used; at first it was violet red, some- 
what turbid. 
18 
Au90 + N aCl 
+ glue 
dirty blue 
copper red 
30 
March 25, 1902 
Augo was diluted, NaCl added, and then glue. 
19 
Au90 + NaCl 
+ glue 
blue violet 
golden yellow 
54 
A<l/z 
Particles bright, in slow and labored motion. 
20 
Au2 
purple red 
yellow green P., 
very bright 
38 
1898 
Nov. 14, 1903 
This five-year old fluid, at the time of its 
preparation, was, upon the addition of a 
small quantity of gelatin, evaporated to 
about 1 per cent Au and put aside; it had 
repeatedly deposited some red, transparent 
sediment, which redissolved. 
21 
Gold 
suspension a 
violet red 
P. green, yellow 
red 
75 
2-3 
Settled suspension, shaken up again; the 
green particles very bright. 
22 
Gold 
suspension b 
bright violet 
yellow 
95 
Settled suspension, shaken up again. 
23 
Gold 
suspension c 
bluish 
yellow 
130 
Settled suspension, shaken up again. 
1 In the absence of any special statement, the bright red and pink fluids usually contain a small number of larger-sized yellow particles. 
2 If the light-cone is very faint, it generally appears white in colored media. This is chiefly due to the fact that, for the most part the very color most 
strongly diffracted is absorbed by the colored medium. Besides the eye cannot differentiate between very faint shades. In the case of ruby glasses the 
observation has been repeatedly made that the whitish color becomes a distinct green upon examining the outermost layer of the preparation. 
3 Here the color of the light-cone is not complementary to the transmitted light; this indicates that in addition to amicroscopic particles diffracting 
yellow light, still smaller green ones are present, whose light cannot be seen because of its faintness. 
4 Amplitude of the notion of translation. [To face page 132] EXAMINATION OF THE SOLUTIONS OF GOLD 
133 
(smallest submicrons, in fluid Au73a, for example); the 
individual particles then become visible even by arc- 
light (line 10 and the following). The particles become 
brighter and brighter, and shine with various colors 
(lines 14-20), their activity decreases; finally, they become 
so large that they sink to the bottom (lines 21, 22); they 
must, however, be still larger in order to be seen in the 
microscope with ordinary illumination. CHAPTER X 
MOTION OF THE GOLD PARTICLES 
As but few have as yet seen the motion of the smallest 
particles approaching the limit of visibility with sunlight, 
a few remarks on this subject are not out of place.1 
I here repeat the description which I had written for 
my Wurzburg lecture under the inexpressible impression 
of the first view of this rare phenomenon, a description 
of which I had stricken from the manuscript because, on 
re-reading it at that time, it seemed exaggerated. I had 
pointed out that large suspended gold particles float 
quietly in the fluid and sink slowly to the bottom, or 
else show only an unimportant Brownian movement, and 
that by conceiving colloidal solutions as suspensions, one 
would naturally imagine that the gold particles in these 
too would be as quiet as those in the real suspensions. I 
then added the description above referred to: 
"How entirely erroneous was this idea! The small 
gold particles no longer float, they move-and that with 
astonishing rapidity. A swarm of dancing gnats in a 
sunbeam will give one an idea of the motion of the gold 
particles in the hydrosol of gold! They hop, dance, 
1 For the demonstration of nltramicroscopic particles with arc light 
on the occasion of the exhibition of our apparatus at Jena, Berlin, 
Kassel, and Leipzig, Siedentopf and myself chose for the purpose of 
distinctness, coarser subdivisions of metals having particles of 20-40 up. 
134 MOTION OF THE GOLD PARTICLES 
135 
jump, dash together, and fly away from each other, so 
that it is difficult in the whirl to get one's bearings. 
"This motion gives an indication of the continuous 
mixing up of the fluid, and it lasts hours, weeks, months, 
and, if the fluid is stable, even years. 
"Sluggish and slow in comparison is the analogous 
Brownian movement of the larger gold particles in the 
fluid, which are the transition forms to ordinary gold 
that settles. 
"The smallest particles which can be seen in the hydro- 
sol of gold, show a combined motion, consisting of a motion 
of translation by which the particle moves from 100 to 
1000 times its own diameter in one sixth to one eighth of a 
second, and a motion of oscillation of a considerably 
shorter period, because of which the possibility of the 
presence of a motion of oscillation of a higher frequency 
and smaller amplitude could not be determined, but is 
probable." 
In Table 5 (page 132), column V, lines 7 and 8, etc., are 
given in y the numerical values for the amplitude of the 
motion of translation, designated by A.1 At the time of 
examination hydrosol AU92 was over a month old, and 
hydrosol Au73 three quarters of a year old; notwith- 
standing this they showed the most active motion. Even 
in still older solutions similar motions were evident, for 
instance in the fluid Auso (made October 12, 1898, exam- 
ined April 26, 1901).2 Frequently-especially in the case 
of less stable fluids-a decrease in the activity of the 
motion may be seen in the course of time, probably in 
connection with the commencement of the coagulation of 
the solution in question. 
1 See what is said on this subject in Chapter IX, p. 130. 
2 A fluid which had lost the larger part of its gold by being frozen 
in the winter of 1900. 136 
MOTION OF THE GOLD PARTICLES 
I must expressly call attention to the fact that this 
motion of the little gold particles just described, differs 
in many respects from the typical Brownian move- 
ment.1 
In the following table are arranged some of the data 
concerning the Brownian movement given by various 
investigators for comparison with those of the gold 
particles.2 
Kind of Particles. 
Diameter in /l. 
Free Path in /I. 
Observer. 
1. Gold (Au73) 
0.00G 
over 10 
Zsigmondy 
2. ' ' (Au90) 
0.01 | 
3-4, some-1 
times and । 
c c 
2a. " (Au70) 
0.035 
over J 
1-7 
c c 
2b. Dust particles in dis- 
tilled water 
less than 0.2 
imperceptible 
(c 
3. Little spheres 
0.5 
2.5 
Regnauld 
4. Gum gamboge 
1-2 
Zsigmondy 
Chr. Wiener 
4a. " " 
1-2 
5. 
1.1 
1.4 
Ramsay 
6. 
4 
0 
Exner 
TABLE VI 
The free paths referred to in the table are traversed in 
different times. The little gold particles, as far as may 
be observed, cover the distance designated, in £ to | of a 
1 Compare this with gum gamboge, Chapter XIX. Atterberg 
(see note 1, p. 17) found (in harmony with Exner) that sand grains 
less than 2/i exhibit lively Brownian movement, while grains coarser 
than 3y no longer move. 
2 The more lively motion (up to 8y) observed by Renard in iodin 
particles and turbid solutions of rosin in water and alcohol, may be 
explained according to Quincke's idea (Drude's Ann., 1902, Vol. VII, 
p. 67), as the consequence of the fluid stretching out in the little 
interstices to the surface of the particles. MOTION OF THE GOLD PARTICLES 
137 
second, and even less; larger particles take somewhat 
longer to traverse their shorter paths. 
The speed of the motion of the particles cannot there- 
fore be seen from the table; it is very difficult to observe, 
although the difference in motion may, however, be 
clearly recognized. Perhaps more accurate data may be 
made possible by the cinematograph (instantaneous photo- 
graphy); for the present we must be satisfied to obtain 
only an approximate picture of the motion. In order to 
make this idea clearer, there is illustrated in Fig. 1, 
Plate II, the motion of a gold particle 0.01 y in 
diameter,1 and for comparison the paths and diameters 
of the other particles referred to in the table, all 
enlarged 5000 times. And finally, in Fig. 6 is given 
(according to arbitrary scale), the typical Brownian 
movement of the fat globules in milk, according to O. 
Lehmann.2 
Concerning this motion, Lehmann writes: "Here we 
may wait in vain for such a drop to come to rest, 
for it moves continuously, describing an irregular 
zig-zag line to and fro, vibrating unsteadily about 
a mean position seldom reached; only when by 
chance it rests against the side of the vessel, or 
when several unite to a larger conglomerate, does the 
motion cease." 
In contradistinction to this typical Brownian move- 
ment about a mean position, the motion of the little 
gold particles is continuous, a quick moving gold 
particle for instance, after a series of speedily 
' From the great variety of paths observed I select one arbitrarily, 
so that the diagram has no claim to accuracy. 
2 Lehmann, Molekularphysik, Leipzig, 1888, Vol. I, p. 264. In 
this book there is also discussed in detail the work of other investi- 
gators (up to 1888), from which several dates are taken. 138 
MOTION OP THE GOLD PARTICLES 
executed zig-zag moves, rushing across the illumi- 
nated field of view, almost as if it were a living thing, 
and vanishing; whereas larger particles may for the 
most part be conveniently viewed for a long time before 
the slow current of the fluid removes them from the field 
of view. 
In accord with the observation of Exner, that larger 
particles show less active Brownian movement than 
smaller ones, I can also state that in general the 
motion of the gold particles becomes more sluggish 
and of greater period with increasing size, and in 
the case of quite large sized particles sometimes ceases. 
This latter rule is subject to exceptions, for I some- 
times found quite large submicroscopic particles in 
active motion. 
Furthermore, the kind of the subdivided substance (or 
its surface properties?), appear to influence the motion; 
thus the ultramicroscopic dust particles of distilled water 
show no appreciable movement. 
Regarding the Cause of the Motion. I must here oppose 
the natural inference that the warming of the illuminated 
layer of fluid or a one-sided illumination of the particles 
might have considerable influence on the motion described. 
The motion persists even if a water cell be interposed in 
the path of the rays of light,1 and it remains unchanged 
whether a certain part of the fluid be illuminated for a long 
or a short time. 
With the help of a micrometer screw on our apparatus, 
the light-cone can be moved horizontally. By suddenly 
turning the screw, I have repeatedly moved the cone 
1 With very bright particles which could be seen in the broader 
portions of the light-cone, I could see that the motion there was no 
less active than at the apex of the light cone, where existed the maxi- 
mum heat and light effect. Motion of Particles of Various Sizes and Kinds 
1. Gold particles 0.01/z m diameter. Particles 4^hi diameter (without motion). 
?• Part?cfeTumAmAe 6- Brownian molecular motion according to O. Lehmann (arbrtrary scale) 
O. X ell llLLtJo Ol . KAAA 
Linear magnification, 1. oOOO. [ro face paae 1381 
PLATE II MOTION OF THE GOI.D PARTICLES 
139 
into a distant, hitherto unilluminated portion of the 
field of view, and found the movement of the gold 
particles just as lively there as elsewhere, and simi- 
larly independent of the direction of the illuminating 
rays. 
In determining the motion of the particles it should 
be seen: 
1. That during the observation the particles are in a 
space entirely confined, so that there can occur no change 
of concentration by dilution, etc.; 
2. The motion is independent of the direction of the 
light rays, and independent of the time a given 
portion of the fluid has been illuminated, and as far 
as may yet be determined, of the intensity of the 
illumination; 
3. That in general the smaller gold particles exhibit a 
much more active motion than the larger ones; that 
sometimes larger particles are met with, having active 
motion; 
4. That some of the fluids investigated which showed 
lively motion, were several months, some even one and 
one half years old. 
5. That the particles appear to influence each other, 
and that for the most part the activity of the motion is 
somewhat decreased by the dilution of the gold solu- 
tion. 
In the case of very small gold particles there is a possi- 
bility that the continuous motion may be due to a cer- 
tain extent to the molecules contained in them; I con- 
sider more probable, however, the assumption that the 
electrically charged gold particles enter into an action of 
interchange with the ions of the fluid and with each other.1 
1 Meanwhile The. Svedberg has shown that the motion of the par- 
ticles is dependent upon their electric charges. 140 
MOTION OF THE GOLD PARTICLES 
It will require very careful study to elucidate with 
certainty the cause of this motion. Although the causes 
of this remarkable phenomenon may be manifold, it is 
the kinetic theory of fluids which appears to be of prime 
importance in explaining the motion, which persists 
uninterruptedly in the fluid for months and years.1 
1 These predictions have been most satisfactorily fulfilled. Einstein, 
and v. Smoluchowsky (Drude's Ann., 1906, Vol. 21, 756-780), worked 
out the kinetic theory of the Brownian movement; the numerical 
values of their calculations agree quite well for the most part with 
the amplitudes of the motion of the particles as directly determined 
by The Svedberg (Zeit. f. Elektrochem., 1906, pp. 853-860) and 
Ehrenhaft. CHAPTER XI 
SIZE AND COLOR OF THE PARTICLES 
The table in Chapter IX confirms-in agreement with 
Faraday-the statement1 heretofore often made, that 
there is no recognizable interdependence between the 
color and size of the gold particles in ruby-glass and those 
in gold hydrosols, and that therefore no inference as to 
the size of the particles can be drawn from the nature of 
the absorption curve of light rays.2 
AuPie is a bright red fluid whose gold is in almost 
homogeneous subdivision, while Au2osa is a fluid of the 
same color; both have absorption bands in the green, 
and contain bright green particles averaging 32/z/z, some 
larger, some smaller; and on the other hand Au73, 92, 101, 
etc., are various bright red fluids with green particles 
about 6-20y/z. 
Still larger green particles arc doubtless contained in 
1 Zsigmondy, Zeitschr. f. Elektrochemie., 1902, No. 36; Siedentopf 
and Zsigmondy, Drude's Ann., Vol. X, p. 35, and Verhandlungen d. 
Deutschen physikal Ges., Vol. V, p. 212. 
2 Stoekl and Vanino, and recently Ehrenhaft (Berichte d. Kais. 
Akad. d. Wiss., Vienna, 1903, Vol. CXII, p. 181) have calculated from 
the light absorption of gold sols, the size of the particles contained 
in them. Their conflict with the facts supported by numerous experi- 
ments, indicates that the suppositions made by these investigators 
are not correct. See also the theoretical treatment of the question of 
optical resonance of finely subdivided metals, by F. Pockels. Physikal. 
Zeitschr., 1904, Vol. V, No. 6, pp. 152-156. 
141 142 
SIZE AND COLOR OF THE PARTICLES 
gold suspension A, for the brightly shining green particles 
sank to the bottom just like the yellow and red ones. 
From the above may be seen, as has been previously 
noted in the case of ruby-glasses,1 that green gold particles 
may occur in various sizes. Every red fluid, no matter 
what the size of its green particles, changes upon the 
addition of electrolytes to a blue color, and then contains 
no (or only a few) green particles, but has instead yel- 
low or red ones, which are all larger than the green ones 
formerly present; and from the finest subdivision AuPi6 
may be obtained extremely small yellow ones, approach- 
ing the limit of visibility, whereas the red hydrosols with 
large green particles yield quite large brightly shining 
yellow or red particles. It is therefore within our power 
to produce both red and yellow particles of various sizes. 
From the preceding it necessarily follows that there 
are gold particles of various size (mass), which for the 
most part diffract green light, and which when spread 
through a fluid, endow it with a light absorption whose 
maximum is in the green; and on the other hand, that 
there are gold particles of various sizes which diffract 
chiefly yellow or red light, and also communicate to the 
fluid a corresponding light absorption. For this reason, 
and also because of the facts developed by the study of gold 
ruby-glasses, we may regard as ill conceived any attempt 
to calculate from the light absorption the size of the particles 
in metal hydrosols. Considerable significance must be 
attributed, without doubt, to a series of unknown factors 
(form and shape of the particles, substance of the sub- 
divided metal, and others besides). 
That the substance of the subdivided metal is not 
without influence upon the color of the diffracted light, 
is quite evident from the fact that in the case of gold we 
1 Siedentopf and Zsigmondy, p. 30. SIZE AND COLOR OF THE PARTICLES 
143 
always find red, green, and yellow particles, with palla- 
dium and platinum, mostly white or gray-white (with 
slight shading); and with silver the most variously col- 
ored particles with a brilliant play of colors, all of which 
will be later referred to (Chapter XIX). CHAPTER XII 
THE COLOR CHANGE OF COLLOIDAL GOLD 
If an electrolyte is added to a red colloidal gold solu- 
tion, the solution changes color from red to blue. This 
color change indicates an irreversible change of the col- 
loidal gold; if the gold is pure, under no circumstances 
can it be made retrogressive. If, however, protective 
colloids are present, the color change of gold may be 
made reversible. Faraday had already observed that 
"jelly," when dried out with gold chloric!, etc., became 
red or blue after drying, and that the blue jelly became 
red again upon moistening. Preparations exhibiting a 
similar striking color-play upon moistening or drying, 
had been obtained by me several years ago by drying 
bright reel colloidal gold solutions with very little gelatin. 
The thin skin of gold-gelatin remaining, became red when 
moistened and blue again when dried. This change of 
color may be repeated as often as desired, and recalls 
the color change of blue cobalt chlorid upon treatment 
with water. As Kirchner1 has shown, silver bromid- 
gelatin plates, made and developed according to Lipp- 
mann's method, in which colloidal silver is the color- 
producing constituent, show a similar color change upon 
moistening. 
In collaboration with F. Kirchner, there was under- 
1 F. Kirchner, Ber. d. math.-phys. Klasse d. Kgl. Sachs. Ges. d. 
Wissensch. zu Leipzig, 1902. Further Inaug.-Diss., Leipzig, 1903. 
144 THE COLOR CHANGE OF COLLOIDAL GOLD 
145 
taken an elaborate investigation of gold-gelatin mixtures.1 
The gold solution used for their preparation was the 
fluid Au97 (see Table V, Chapter IX), containing 0.005 
per cent of metallic gold. The average mass of the indi- 
vidual particles was 7.10-14 mg., corresponding to a 
diameter of 15^/z. The number of gold particles per 
cubic centimeter was about 0.7 milliards. Each fifty 
cubic centimeters of this gold solution was mixed with 
varying quantities of gelatin (from 0.5 to 10 parts of gela- 
tin to one part of gold), and the mixtures, after being 
dialyzed in a closed vessel, were poured out upon glass, 
etc. Upon drying there remained a blue or dark violet 
film, which when moistened showed the color change 
referred to, and when examined by transmitted light with 
microscope objectives of highest aperture, was resolved 
into a mass of extremely small, intensely-colored granules, 
imbedded in an apparently colorless matrix. Each indi- 
vidual granule contained several hundred (larger granules, 
even several thousand) submicroscopic particles. In them 
the a-particles (Chapter I) are not inseparably united; 
they (the /9-particles) consist of a mass of gold and gelatin 
very rich in gold (both apparently equally distributed), 
imbedded in colorless gelatin. On boiling'with water the 
original subdivision of the gold can be reproduced. Upon 
drying out, therefore, numerous gold particles have 
formed a little group, just as do the particles of a pure 
hydrosol upon the addition of salt. In each case the 
color changes from red to blue, and in each case the 
cause of the color change is the same. The original a- 
gold particles (see Chapter I), which we may consider as 
resonators,2 approach each other, and this approach 
1 Kirchner and Zsigmondy, Crude's Ann., 1904, Vol. XV, pp. 573- 
595. 
2 Metal particles had been previously conceived of as resonators by 146 
THE COLOR CHANGE OF COLLOIDAL GOLD 
influences their time of vibration. The chief cause for 
the formation of "flocks " in water and groups in gelatin, 
appears to us to be the attractive forces (cohesion forces) 
between mutually similar gold particles, which tend to 
unite them. The great influence of cohesion upon optical 
constants has been already pointed out by Wernicke.1 
M. Planck2 advanced a dispersion theory for isotropic 
dielectrics based upon the assumption that fixed resona- 
tors are distributed in the ether, and comes to the con- 
clusion that the crowding together of the resonators has 
as a result an increase as well as a broadening in the 
maximum of absorption, which takes place more quickly 
toward the red than toward the blue. These very 
changes in the maximum of absorption are clearly evi- 
dent in Kirchner's silver preparations, as well as our 
gold-gelatin preparations. The application of Planck's 
theory to the latter showed quite a complete arrange- 
ment between the calculated and empirically determined 
absorption curves, confirming the correctness of the 
assumption that the cause of the color change is to be 
sought for in the change of the distance between the 
particles. As to the differences between theory and experi- 
ment, and for all details, I must refer to the paper cited.3 
It might be remarked in passing that these experiments 
may have theoretical importance in regard to many 
solid colors, ultramarine among others. 
Wood, Kirchner, Kossonogow, Ehrenhaft. (Literature in the papers 
quoted.) 
1 W. Wernicke, Wied. Ann., 1894, Vol. LII, p. 515. 
2 Planck, Drude's Ann., 1900, Vol. I, p. 92. Sitzungber. d. Kgl. 
Akad. d. Wissensch. zu Berlin, May 1, 1902. 
1 Regarding the optical behavior of colloidal metals, see further 
Ehrenhaft, Ann. d. Phys., 1903, Vol. II, p. 489. Mention must also 
be made here of the detailed theoretical investigations as to the origin 
of the colors of metal hydrosols carried out by G. Mie, Drude's An- 
nalen, 1908, Vol. 25, p. 377, and by Maxwell Garnet, Phil. Trans., 
1904, Vol. 203, p. 385, and 1906, Vol. 205, p. 237. CHAPTER XIII 
THE PRECIPITATION AND PROTECTION OF COLLOIDAL 
GOLD 
The coagulation of colloidal gold by the addition of 
electrolytes is not easy to follow ultramicroscopically. 
It usually takes place so quickly that there can be seen 
only the result of the union of the small particles, not 
the actual process. Sometimes, however, it can be 
observed. 
I shall now describe Experiment No. 2 of Table V, page 
132. The light-cone becomes visible; wavy yellow nebu- 
losities are seen; the fog thickens still more, and then 
appear tiny individual particles having an active Brownian 
movement. The particles unite and turn about their 
common center of gravity. After their union, a "molec- 
ular motion" begins again, but a slower one than before. 
The process does not always proceed as just described; 
here too a great multiplicity of phenomena appear. If 
under the microscope the electrolyte solution is allowed 
to run into the gold solution, there is such a violent whirl- 
ing that the course of the coagulation cannot be seen; 
when the fluid comes to rest, the union of the particles 
has been completed. 
If a gold solution with visible particles is very largely 
diluted, the process of union is hindered, and there are 
usually seen larger gold-glinting particles, together with 
smaller green ones which attract each other, but for the 
most part separate without uniting. 
147 148 
PRECIPITATION OF COLLOIDAL GOLD 
By adding gelatin to the coagulating fluid, one can 
interrupt the course of the union of the particles at any 
desired moment. Nos. 18 and 19 of the table are two 
experiments of this kind. In Experiment 18 the gold 
particles of Auao have united into smaller flocks than in 
Experiment 19. 
Colloidal Gold and Fuchsin. As I discovered several 
years ago, colloidal gold and fuchsin (or instead of fuchsin 
methyl violet, Bismark brown, methylene blue), mutually 
precipitate each other, sometimes almost quantitatively 
so that the fluid is decolorized (see Chapter III). By 
means of alcohol, not water, however, a greater part of 
the dye-stuff can be removed from the precipitate. Before 
precipitation commences, the mixture of gold and fuchsin 
first changes to a dirty violet-reel; it then contains only 
large golden yellow particles, in which sometimes the 
process of union can be seen. 
Protective Action. The protective action against the 
addition of electrolytes, exerted on colloidal gold solu- 
tions by numerous colloids, has in many cases been quan- 
titatively determined by me alone or in collaboration 
with Fr. N. Schulz 1 (see Chapter III, page 79). This 
protective action can be explained, as I have already 
remarked, upon the assumption that a gold particle 
unites with few or many particles of the protecting col- 
loid, or upon the reverse assumption that one particle 
of the protecting colloid unites with several gold particles.2 
Protective action can also be explained by assuming 
that the "suspended" particles are encased in an ole- 
aginous skin of the other colloid, according to the idea3 
1 Schulz and Zsigmondy, Beitrage z. chem. Phys, and Path., Vol. Ill, 
p. 137 (1902); Zeitschr. f. analyt. Chem., Vol. XL. p. 697 (1901). 
2 Verh. Naturforscher-Vers., Hamburg, 1901, pp. 168-172. 
3 Or by being surrounded by a skin of easily movable fluid, such as 
Konowalow assumes to exist in turbid fluids in the critical state. PRECIPITATION OF COLLOIDAL GOLD 
149 
advanced by Quincke.1 This view, supported as it is 
by the analogy of the well-known action of surface energy, 
can be advantageously applied towards explaining the 
action exerted by protective colloids upon larger sus- 
pended particles.2 
I must not refrain, however, from pointing out certain 
considerations which are contrary to it, when the particles 
to be protected are very small, as in the case of gold solu- 
tions of the finest kind. 
If we consider the dimensions of the molecule in the 
light of the kinetic theory, the assumption that the 
smallest gold particles are homogeneously encased by the 
relatively large gelatin or albumen molecules (providing 
the amicrons contained in these hydrosols actually are 
molecules), is untenable.3 
There is another objection: some gelatin and some 
glycogen solutions, and many albumen solutions are 
found to be permeated with ultramicroscopic particles 
(see Chapters XIX and XX), whose presence must 
receive consideration even if perhaps they are not the 
smallest present in the solution and do not contain the 
real substance of the dissolved colloid. Particles of this 
kind as well as the smallest particles of the protective 
colloid can unite with the colloidal gold; even much 
larger hydrogel particles can fix the gold, as we will see 
in discussing the reaction between gold and alumina gel. 
In the case of the explanation first advanced, there- 
fore, we must bear in mind the possibility that several 
1 Quincke in Drude's Ann., 1902, Vol. VII, p. 95. 
2 Neisser and Friedemann, Munchner, med. Wochenschr., 1903, 
No. 11. , 
3 Compare the linear dimension of the starch molecule (5 [i[i) calcu- 
lated by Lobry de Bruyn, and the graphic representation of the 
relative sizes of ultramicroscopic particles given further on (Chapter 
XV, Plate IV). 150 
PRECIPITATION OF COLLOIDAL GOLD 
gold particles can unite with one particle of the protec- 
tive colloid, as well as that one gold particle can unite 
with several or many particles of the protective colloid. 
Unfortunately the reaction between gold and protective 
colloid for the most part escapes direct observation in 
the ultraapparatus, at least so far as concerns the smallest 
gold particles whose behavior would just be the most 
interesting, for they are just as invisible as the amicrons 
of the protective colloid. In some cases the reactions 
between two colloids can be followed to a certain point 
ultramicroscopically. Two of these cases will be men- 
tioned here. A detailed discussion of the reactions con- 
cerned and of the literature on this subject, I shall reserve 
until later. 
Colloidal Gold and Gelatin. A mixture of Au9o (0.0005 
per cent Au) and gelatin (0.01 per cent), was placed in 
the ultraapparatus. The activity of the gold particles 
had somewhat diminished, everything else being appar- 
ently unchanged (as compared with a pure gold solution 
of the same concentration); in the presence of the gold, 
the much less luminous gelatin particles could not be 
recognized. With the fluid Au90 the particles continu- 
ously formed groups and then flew away from each other; 
the somewhat large addition of gelatin (about one hun- 
dred times as much as was necessary to produce a pro- 
tective effect), appeared to oppose group formation, with- 
out, however, destroying the free mobility of the gold 
particles. Even subsequent addition of NaNO3 altered 
the microscopic appearance but little. Larger additions 
of gelatin produced a further decrease in mobility, but 
by no means destroyed it. 
If, however, there is added to the gold just enough 
gelatin to produce the protective action, no change at 
all is seen in the mobility of the gold particles in the PRECIPITATION OF COLLOIDAL GOLD 
151 
ultraapparatus. This fact is of importance because 
Lobry de Bruyn1 has attributed the prevention of 
many precipitative reactions in the presence of colloids 
to decreased mobility (which was partially the case in 
the jellies used by him; see Chapter III, p. 78); and 
recently Muller2 has advanced the opinion that the 
protection of the gold is due to the natural increase in 
viscosity due to the addition of the colloid, which 
prevents the "suspended " gold from settling. That 
this latter view cannot be correct is evident from the 
fact that the slime of quince kernels, despite its vis- 
cosity exhibits no appreciable protective action on 
gold, while traces of glue (0.0001 per cent), which 
suffice for this purpose, do not, as we have seen, influence 
the mobility of the particles, and do not hinder the 
deposition of suspended gold particles if they are large 
enough to sink to the bottom in water. 
From what has been said it follows that under the 
conditions in question the gold particles remain iso- 
lated; the addition of gelatin does not diminish the 
number of their little reflecting disks, nor is their bright- 
ness increased. The union between the gold and gelatin 
which must on other grounds be assumed, takes place 
in such wise that each gold particle is united with one 
or several gelatin particles (which were present in large 
excess), whereby the gold is protected against precipita- 
tion by electrolytes. 
Colloidal Gold and the Hydrogel of Alumina. As I 
have already shown,3 freshly precipitated alumina takes 
the gold from a colloidal gold solution, and forms with 
it a red colored lake, just as carmine lake is formed by 
iBer. d. D. chem. Ges., Vol. XXXV, p. 3079 (1902). 
*lbid. Vol. XXXVII, p. 11 (1904). 
3 Verb., d. Ges. D. Naturf., Hamburg, 1901, pp. 172-198. 152 
PRECIPITATION OF COLLOIDAL GOLD 
shaking carmine solution with alumina gel. In order 
to follow the process of union, levigated alumina gel 
and colloidal gold were simultaneously introduced into 
the ultraapparatus; the process is not easy to follow, 
because it is almost completed before the fluid reaches 
the observation cell; besides, the dazzling brilliancy of 
the larger gold-laden hydrogel particles prevents one 
from seeing the individual grains. Sometimes one may 
see a bright shining gold particle attach itself to the 
groups. More often the individual gold submicrons 
unite with small gel particles which have already taken 
up gold, to form stationary or moving systems. 
Here, too, the union takes place in such a way that 
several gold particles are taken up by one gel particle. 
The gold united with the alumina is protected against 
the addition of salt, just the same as that which is 
combined with gelatin. CHAPTER XIV 
FILTRATION EXPERIMENTS 
In order to obtain some insight into the question so 
important to bacteriology, as to the size of the filter 
pores of clay filters and filter cells, I made a number of 
experiments in the filtration through such media of 
colloidal gold solutions having particles of known size. 
For this purpose three different types of filters were 
selected. 
1. Filter cell, candle-shaped, according to Dr. Maassen. 
2. Clay filter, bottle-shaped (ballon filter), according 
to PukalL 
3. Chamberland filter cell. 
As may be supposed from the speed with which water 
filters through, the first named has the largest pores, 
while the Chamberland cell has very fine ones. 
Without going into details, I may mention the important 
factors on which such experiments are judged. 
All three kinds of filters contain pores large enough 
to allow the passage of gold particles of about 30//« and 
less. The pores of a cell are of very different sizes, 
the Chamberland cell containing, for example, large 
pores, which allow the gold. particles to pass through, 
and others which retain most of them. The size of the 
pores is, however, not the sole criterion in filter experi- 
ments. It is of especial importance in coarser filters, 
whether the particles to be filtered are held to the 
153 154 
FILTRATION EXPERIMENTS 
surface of the cell by adhesion or "adsorption " (A), 
or not (B). 
(A) In the first instance the substance to be filtered 
gathers upon the outside surface (and to a certain 
extent in the deeper pores), and prevents the other 
particles from forcing their way through; first, because 
the pores are made smaller; second, because the particles 
held fast to the surface of the cell repel the freely moving 
particles following the course of the current.1 
(B) When adhesion or adsorption do not take place, 
all colloidally dissolved substances pass freely through 
the cell, providing the pores are large enough. 
This latter case occurs, for instance, when a gold 
hydrosol with particles 20-30 yy, and containing egg 
albumen, is filtered through a Pukall cell or a Maassen 
filter. All the gold particles pass quite smoothly 
through the cell without appreciable change of concen- 
tration. The experiments were made with colloidal 
gold containing egg albumen. The first of this kind 
were performed by Prof. Fr. N. Schulz at the Physio- 
logical Institute at Jena, and I have repeated the 
experiment several times with the same result. 
If the egg albumen or similar protective colloid is 
omitted, and the pure gold hydrosol used, then matters 
proceed as in case A, and very many gold particles are 
attached to the exterior surface of the cell (even in 
cells with large pores), and at once partially, and by 
prolonged use, totally prevent the passage of the other 
gold particles. Bredig first observed this in the Pukall 
filter. I obtained at first a slightly diluted, but finally 
a totally colorless filtrate; the outer fluid remained 
1 This action may be due to the well-known negative electric charge 
of the particles, which apparently also affects the adhering gold 
particles. FILTRATION EXPERIMENTS 
155 
unchanged at first, but when the surface of the cell 
had accumulated sufficient gold, it became richer in 
colloidal gold in proportion to the duration of the 
experiment.1 
The experiment was made with all three filters, 
yielding the same result, except that the very fine pored 
Chamberland filter promptly gave a very diluted 
filtrate. 
In the ultraapparatus the filtrates showed the same 
properties as the original fluid, but the dilution was 
considerable, fewer particles being in the filtrate; the 
concentrated exterior fluid contained, however, numer- 
ous larger gold particles in addition to the original 
ones. 
The fact that protected gold particles of 30 y/z and 
over easily pass through Maassen and Pukall filters, 
should be of interest to bacteriologists. The Chamber- 
land filter, too, contains, besides the very small pores 
chiefly present, others which permit the passage of 
particles of the size mentioned. 
In order to avoid misunderstanding I may repeat 
that this experiment can give no direct measure of the 
size of the pores, because the size of the gold particles 
has been calculated on the assumption that they are 
cubical, whereas we know nothing of their actual shape; 
nevertheless a gold solution with protective colloid 
added can be used as a relative measure of the size of 
the pores of various cells. 
Perhaps the above remarks may lead to the production 
of better filters with uniform pores, for it does not 
1 The course of filtration through Pukall cells observed by me, differs 
somewhat from that described by Bredig, who obtained, upon filtra- 
tion of his hydrosols, at first a colorless, then a colored, and finally, 
colorless filtrate again. 156 
FILTRATION EXPERIMENTS 
appear impossible to fill up the larger pores with a 
uniform fine-pored film permeable to fluids, by sucking 
through coarser subdivisions of gold, platinum, or 
silver. PLATE III 
Linear magnification, 1:10,000 
A. Human blood corpuscles (diameter 7.5 /i, thickness 1.6 fi). 
B. Fragment of rice starch granule (according to v. Hohnel.1) 3-8 
C. Particles in a kaolin suspension. 
E. Anthrax bacillus2 (length 4-15 y, width about 1 a). 
F. Cocci (diameter about. 0.5-1 /t, rarely 2 /t). 
g, h. Particles of colloidal gold solutions Au^, Au82, Au87 (0.006-0.015 (i) 
i, k, I. Particles from settled gold suspensions (0.075-0.2 p) 
1 Die Starke- und die Mehlprodukte, Kassel u. Berlin, 1882. 
2 W. Migula, Die Bakterien, Leipzig, 1906. 
]To face page 157 PLATE IV 
Linear magnification, 1 : 1,000,000 
a-d-Hypothetical Molecular Dimensions. 
a. Hydrogen molecule-dia. 0.1 yy (O. E. Meyer). 
b. Alcohol molecule-dia. 0.5 yy. 
c. Chloroform molecule-dia. 0.8 /l/i (Jager). 
d. Molecule of soluble starch-about 5 yy (Lobry de Bruyn). 
e-h-Goll Particles in Colloidal Gold Solutions. 
e. Gold particle in AuPw (so small that its size cannot be determined). 
eP " " " " (about 1.7 yy).1 
e2. " " " ( " 3.0 yy).1 
f- " " " Au73O ( " 6 /Vx)- 
g. " " " Au92 ( " 10 yy). 
h. " " " Auw ( " 15 yy). 
i. " " " settled gold suspension a. 
1 Zsigmondy, Z. physik. Chern. 1906, Vol. 56, p. 65; Z. f. Elektro- 
chem. 1906, p. 632. 
[To face page 157.) CHAPTER XV 
THE SIZE OF THE GOLD PARTICLES COMPARED WITH 
THE SIZE OF OTHER BODIES 
From the preceding sections it follows that the 
particles in colloidal gold solutions are much smaller 
than those of substances that have heretofore been 
susceptible of direct observation; further, that the 
size of the amicroscopic gold particles approximates the 
hypothetical molecular dimensions. In order to show 
this diagrammatically, I have prepared Plate III, com- 
paring the approximate sizes of several microscopic 
bodies1 with those of the gold particles in colloidal 
solutions-all enlarged 10,000 diameters; in Plate IV 
the latter are compared with the hypothetical dimensions 
of the molecule, magnified 1,000,000 times. In Plate IV, 
instead of the gold particles themselves, whose form is 
not known, is represented, enlarged a million times, 
the plane surface of the corresponding cube, which, 
when filled with metallic gold of twenty specific gravity, 
would contain the same mass as the gold particles in 
the fluids Au73, Au93, etc., according to Table V, Chapter 
IX. Even when enlarged 10,000 diameters, in Plate 
III, the gold particles f, g, and h in the colloidal solution 
appear as extremely minute points, while in the same 
1 The figures in the table are meant to express pictorially the rela- 
tive sizes, and by no means give a true illustration of the bodies in 
question. 
157 158 
THE SIZE OF THE GOLD PARTICLES 
magnification the particles of a blood, starch, or bacteria 
suspension, assume quite considerable dimensions.1 
Unfortunately I can as yet give no information as 
to the size of the particles in albumen and glue solutions, 
and in the solutions of colloidal oxicls and sulphids, for 
although in many of these individual particles can be 
made visible, we have as yet no satisfactory information 
as to their size. As work with the ultraapparatus is 
now being done in various quarters, it is to be hoped 
that the desired goal in this direction will soon be 
reached. 
Of great interest are the observations already noted 
by Raehlmann,2 that the individual particles in glycogen 
solutions can be made visible, and that their disappear- 
ance upon the addition of diastase can be followed in 
the ultraapparatus (see also Chapter XX); further, the 
analogous observation of Romer3 concerning solutions 
of true proteins. That the particles made visible in 
glycogen and protein solutions consist of the essential 
constituents of the solution in question, appears to be 
proven beyond reasonable doubt by their action toward 
specific ferments. That we are dealing here with rela- 
tively large particles is evident from their visibility by 
arc light. 
After this significant beginning we may watch with 
interest the development of this newly-discovered field. 
It would be a useful and important work for the scientists 
who make discoveries in this field to determine closely 
in each case the size of the particles they recognize. 
1 The starch grains of rice are the smallest. Potato starch has 
granules as large as 100 /i in diameter, and would therefore be one meter 
in diameter on the scale of Plate III. 
2 Munchner medizin Wochenschr., 1903, No. 48. 
3 Berlin, klin. Wochenschr., 1904, No. 9. THE SIZE OF THE GOLD PARTICLES 
159 
In many colloidal solutions the particles cannot be 
macle visible, which is the case with the colloidal solu- 
tions of many metallic oxids. From this it might be 
concluded that the particles in such solutions are smaller 
than those in colloidal gold solutions having visible 
particles. Such a conclusion would be very premature. 
For the limit of visibility of ultramicroscopic particles 
has been carefully determined experimentally only for 
gold; it will vary from substance to substance, and with 
most bodies be considerably higher than with gold. 
The nearer the indices of refraction and dispersion 
of the subdivided substance and the medium, the less 
light the particles will diffract, and the larger the particles 
must be in order to be ultramicroscopically visible. 
The very appearance of the subdivision clearly shows 
the difference between noble metals and oxids, etc. 
Whereas many fluids appear almost clear,1 despite the 
relatively large particles contained in them, gold sus- 
pensions with particles of 70-100 /zy show an intense 
turbidity. Gold and silver subdivided in water disperse 
light more intensely than do most other substances of 
the same particle-size. Subdivisions of both noble 
metals are therefore especially suitable as test objects 
in ultramicroscopy, because their individual particles 
can be seen down to extremely small dimensions; but 
even they, as they become smaller, gradually lose the 
property of diffracting light-this property almost vanishes 
as they approach, or when they reach dimensions which 
the molecular theory ascribes to molecules. 
1 For example, combined suspensions of alumina and iron oxid 
hydrogel, also of hydrogels of Prussian blue, Congo red, etc. CHAPTER XVI 
SUPERIOR AND INFERIOR LIMITS OF THE SIZE OF 
THE PARTICLES 
Superior Limit. The question as to the superior 
limit of the size of particles in colloidal solutions should 
be of some interest. I do not believe that this limit is 
constant; it will depend upon the specific gravity of 
the substance subdivided, upon the internal friction of 
the fluid, and also with many hydrosols upon the electric 
charge of the particles, and especially upon the manner 
in which their space is filled with the mass of the sub- 
divided substance. A densely filled particle will tend to 
sink to the bottom quicker than one which is distended 
in the filling. Bredig gives as the superior limit of the 
size of the particles in the metal hydrosols made by 
him, a linear dimension of 0.14 n, that is 140 /ty.1 
From Table V, Chapter IX, it appears, however, 
that gold particles larger than 75 /zy already begin to 
settle, and up to the present I have not come across 
any stable hydrosol whose average particle-size was 
over about 60 /z/z. On the other hand, in rare instances, 
I could observe that even bright red gold sols with 
particles about 30 /z/z showed, upon standing perfectly 
quiet in a room at 4-10° C., a discoloration of the super- 
natant fluid. 
Thus the superior limit for stable gold hydrosols can- 
1 Bredig, Anorganische Fermente, p. 21. 
160 SUPERIOR AND INFERIOR LIMITS 
161 
not be determined with accuracy; in any case it is 
somewhat lower than Bredig has stated. The chance 
observation above referred to, of the incipient separa- 
tion of a red gold sol, was of great significance to me, 
because it gave me the first indisputable proof that the 
ultramicroscopic green gold particles are identical with 
those which gave the fluid its red color; for the colorless 
fluid was practically free from ultramicroscopic particles, 
while the colored layer contained a large number. 
The inferior limits of the size of the particles in colloidal 
gold solutions cannot be experimentally determined 
with certainty. From Table V (p. 132), Example 1-8, 
it may be seen that there are a series of transition forms 
between Au73a having the smallest visible particles, 
and the gold solution AuPie which, when freshly pre- 
pared, is almost homogeneous; and it is evident that 
the particles of this transition group are much smaller 
than 6 pp. Therefore the size of the gold particles in 
AuPie surely approach the dimensions of the crystalloid 
molecule. 
Additional questions are also worth discussing: Have 
we, in gold solutions having the properties of the fluid 
AuPie, reached the limit of divisibility of the metal, 
or is there a still more complete solution of it? Color- 
less, optically clear gold ruby-glass contains such a 
solution.1 
1 This expression needs explanation. Spring had already demon- 
strated the presence in colorless ruby glass of gas bubbles which destroy 
its homogeneity. Besides there were frequently found ultramicro- 
scopically large yellow-gold particles in the glass, which had separated 
out during smelting if the glass was supersaturated. In some varieties, 
little submicrons can be seen even in the colorless glass. Between 
these incidental constituents, often scattered in widely separated sec- 
tions, many kinds of ruby glass show no further inhomogeneity. The 
space between them appears clear, and, overlooking these incidental 162 
SUPERIOR AND INFERIOR LIMITS 
The investigations of Golfier, Besseyre, Knapp, Muller, 
Ebell, Spring, and others, have macle evident, and Spring 1 
has also pointed out that in colorless gold ruby-glass 
we have to do with an optically homogeneous solution 
of metallic gold. To the objection which might perhaps 
be raised by those unfamiliar with this field, that we 
have here to do with the solution of an oxicl or a silicate 
.of gold, I may, upon the basis of my entire experience 
in the glass factory at Zombkowice, reply as follows: 
Even in the presence of a considerable excess of energetic 
reducing agents in glass, which certainly had removed 
all oxygen from the melt and even reduced oxygen 
present in the gold, upon sudden chilling colorless glasses 
are always obtained, which when reheated become red. 
I can therefore confirm the view held by Ebell, Spring, 
and others, that metallic gold can be dissolved in glass 
without either coloring it or detracting light. My efforts 
to obtain a stable, colorless gold solution in water, remain 
as yet fruitless; in glass, colorless gold solutions are 
sometimes realized. I do not consider these as colloidal, 
but essentially as supersaturated, crystalloid solutions 
of the metal, which view is supported by the observations 
on the solubility of metals in molten salts, made by 
R. Lorenz and his pupils. 
R. Lorenz2 first discovered this phenomenon in the 
constituents, we may speak of optically clear ruby glass. With other 
kinds of ruby glass, between the incidental constituents there may 
be seen an extremely faint whitish light-cone, which quite often is 
visible only upon the most intense illumination with sunlight (in 
June when the sun is in high altitude). We have here amicrons, 
which must be considerably smaller than about 4-7 yy; such glasses 
cannot of course be called optically clear. 
1 W. Spring, Bull, de l'Acad. roy. de Belgique (cl. d. sc.), 1900, 
No. 12, p. 1017 and 1021 (3d section). 
2 For details see R. Lorenz, Electrolysis of Molten Salts, Part II, 
Halle a. S., W. Knapp. SUPERIOR AND INFERIOR LIMITS 
163 
electrolysis of molten halogen salts of the heavy metals 
(ZnCL, PbCR, CdCL, etc.). In these cases there appears 
about the molten metal which separates out at the 
cathode, a cloud-effect which spreads through and 
finally completely fills the molten electrolyte. This 
phenomenon is easily produced by throwing into a 
molten salt a piece of the easily fusible metal in question.1 
The formation of metal fog is dependent upon the 
temperature, the fog appearing and disappearing in 
the molten mass as the temperature falls and rises. At 
higher temperatures it vanishes by solution, and the 
melt assumes a most characteristic color; upon cooling, 
on the contrary, it reappears. This solution and reap- 
pearance of the fog may be repeated as often as desired 
by alternately heating and cooling the melting pot. 
In collaboration with Hclfenstein,2 R. Lorenz has 
attempted to find out the relative, so-called solubility 
of metals in molten salts, and has obtained a series of 
numerical results. 
R. Lorenz3 also determined, in collaboration with 
G. Auerbach,4 that a melt of cadmium chloric! " satu- 
rated " with cadmium fog, after cooling and re-solution, 
leaves metallic cadmium in magnificent little crystals. 
According to R. Lorenz a far-reaching parallelism can 
be traced between the formation of metal fogs and the 
vapor pressure 5 of metals. 
The greater the vapor pressure of the metal, the 
greater should be its crystalloid solubility, and it may 
well be assumed that at the fusing temperature at which 
1 Zeitschr. f. Elektrochem., Vol. VII, p. 277 (1900). 
2 Zeitschr. f. anorg. Chern., Vol. XXIII, p. 255 (1900). 
3 Zeitschr. f. Elektrochem., Vol. II, p. 318 (1895). 
4 Zeitschr. f. anorg. Chem., Vol. XXVIII, p. 42 (1901). 
5 Zeitschr. f. anorg. Chem., Vol. XXIII, p. 97 (1900). 164 
SUPERIOR AND INFERIOR LIMITS 
glass melts (over 1400° C.), a small quantity of gold 
undergoes crystalloid solution. This solution can be 
very much overdone, for it is supersaturated 1 even at 
the working temperature of the glass and sometimes at 
ordinary temperature after quick chilling. 
That colorless ruby glass, too, is probably filled with 
amicrocsopic gold particles which act as centers of 
crystallization, will be brought out in the next chapter. 
1 At this temperature the gold separates out in the form of ultra- 
microscopic particles, often upon slow cooling, almost always upon 
reheating, but not upon quick cooling, CHAPTER XVII 
^MICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
If molten colorless gold ruby glass is allowed to cool 
slowly, according to its quality, it either becomes red 1 
upon cooling, or else remains colorless. If quickly 
chilled to a set, ruby glasses of all grades remain color- 
less. Colorless ruby glass, as Spring has shown, may 
be without effect upon light. The normal red color of 
the ruby glass is almost always brought out by reheating 
it to the softening point; the color-change thereby 
produced is technically known as ''coloring the ruby 
glass " (Anlaufen des Rubinglases). After coloring, part 
of the metallic gold which is homogeneously dissolved 
in the colorless glass, separates out in the form of ultra- 
microscopic particles which reflect green light.2 
i. Gold Particles in Ruby Glass 3 
The gold particles in ruby glass, like those in colloidal 
solutions, are either submicroscopic or amicroscopic, 
according to circumstances. Poorly made, spoiled ruby- 
glass, upon coloring, yields a blue or violet shade instead 
of red. Below are described the phenomena which may 
1 Glasses very rich in lead, or containing certain impurities, are 
thereby colored yellow or brown instead of red. 
2 See Siedentopf and Zsigmondy, pp. 19 and 30. 
3 For further information regarding the size, color, and polarization 
of gold particles in ruby glasses, see Siedentopf and Zsigmondy, loc. 
cit., pp. 30-38. 
165 166 
AMICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
be observed in good and in spoiled ruby glass, if both the 
glasses are heated at only one spot. WT7A, reference to 
the following observations it must be noted that the glasses 
under investigation zvere slowly cooled, and that they were 
heated up more quickly than they had previously been 
cooled. 
1. A piece of colorless ruby glass which had been 
very slowly cooled, was heated at one spot, so that ir 
began to melt at a (see Fig. 6), and still remained cold 
at d. At a the glass became intensely red, the red 
color decreasing toward c; at d the glass remained 
colorless. 
Fig. 6. 
2. Spoiled ruby glass becomes blue at a, the color 
decreasing toward c; at & it is violet, at b' bright red, 
and at d colorless. 
In the ultraapparatus using homogeneous immersion 
and brightest sunlight, the following was seen: 
Sample 1, good ruby glass. At a and b there were 
submicroscopic specks of green appearance and very 
close together; at b' a homogeneous green light-cone, 
becoming weaker as c is approached.1 This light-cone 
is produced by amicroscopic gold particles. In so far 
as the individual particles can be traced, their average 
distance from each other is the same as at a, only toward 
b' their brightness considerably diminishes. 
1 With insufficient illumination the light-cone appears homogeneous 
even at a. It is not alone sufficient that the distances between the 
particles be resolvable; individual particles must also be bright 
enough to be seen. ON SPONTANEOUS CRYSTALLIZATION 
167 
Sample II, spoiled ruby glass. The individual particles 
are much brighter and much further apart than in 
sample I; they are differently colored, being copper- 
red at a, shading off to yellow, and they are green where 
the glass is red to transmitted light. Individual parti- 
cles can be recognized even beyond c, and they are, on 
the average, about the same distance apart, that is, 
they are just as far apart at a as at b' and c. The bright- 
ness of the reflecting disks diminishes from a towards 
b', c, and d. At d faint specks of indefinite color may 
still be distinguished, but not at all points.1 
2. On Spontaneous Crystallization 
All these phenomena are easily explained, if one recalls 
the analogy existing between the formation of ruby 
glass and the devitrification of amorphous substances. 
Devitrification of a chilled melt is known to consist 
of an interior crystallization of the isotropic mass. It 
is a general rule that in supersaturated solutions crystals 
can be formed only if centers of crystallization (nuclei), 
which may either be added or spontaneously formed, 
are present in the solution. Lowitz, Gay-Lussac, Lowel, 
Violette, Gernez, and especially Ostwald, deserve credit 
for a series of excellent researches on this subject.2 
Among other things Ostwald determined the smallest 
quantity of crystals necessary to produce the crystalliza- 
tion of supersaturated solutions or supercooled melts, 
1 The above figure is diagrammatic and gives only an approximate 
idea of the entire process. In Example II the temperature gradient 
upon warming was greater. Therefore the distance a d, as may be 
imagined, is much smaller than in Example I. Certain imperfections 
in glasses, such as air bubbles, cloudiness, etc., are not described in 
order to avoid prolixity. 
2 Ostwald, Lehrb. d. allg. Chem., 2d ed., II, 2, pp. 704-784. 168 
AMICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
and found it to bo 10~9 to 10 ~12 gms. in the cases he 
investigated. 
Recently Tammann 1 has made some very interesting 
contributions to our knowledge of the process of devitri- 
fication. 
According to Tammann, the spontaneous crystalliza- 
tion of a supercooled melt depends upon two factors: 
First, its ability to crystallize spontaneously, as meas- 
ured by the number of centers of crystallization formed 
Fig. 7. 
in a unit mass of the fluid per unit of time; second, 
upon the speed of crystallization. 
Upon examining organic melts Tammann further 
found that the number of centers of crystallization per 
unit of time and weight increased at first as the super- 
cooling increased, but quickly diminished again at 
temperatures considerably below the melting point. 
But the number of centers of crystallization formed can 
be determined only when they have grown to visible 
size. In cases of excessive supercooling Tammann 
accomplished this by subsequently raising the tempera- 
ture high enough to increase the speed of crystallization. 
The process for making red ruby pressed glass is 
1 Zeitschr. f. Elektrochem., 1904, Vol. X, p. 532. SPONTANEOUS CRYSTALLIZATION IN RUBY GLASS 
169 
quite similar. The glass is first cooled and then heated 
to redness. 
In addition to the power to crystallize spontaneously 
(sp KV) and speed of crystallization (KG), Tammann 
holds that the viscosity (77) of the supercooled melt 
also influences its behavior. I reproduce from his paper 
a diagram (Fig. 7), illustrating the dependence of these 
three properties on the temperature. 
The relative position of the three curves is not always 
the same as here illustrated, for they differ with 
different substances; the diagram shows, however, that 
the speed of crystallization and the ability to crystallize 
spontaneously, increase with diminishing temperature 
and then decrease again, while the viscosity increases 
steadily. 
3. Spontaneous Crystallization in Ruby Glass 
The working temperature of colorless molten gold 
ruby glass is several hundred degrees lower than the 
temperature at which it is smelted. If we conceive it, 
at the working temperature, to be a supersaturated 
crystalloid solution of metallic gold, and the smallest 
amicroscopic particles of gold to be centers of crystal- 
lization, we can apply Tammann's results to the con- 
ditions existing in gold ruby glass. It will at once be 
seen why ruby glass sometimes remains colorless upon 
simple cooling, and only becomes red when heated to 
the softening point. In this case the optimum temperature 
for spontaneous crystallization is so low that the glass is 
very viscous and the speed of crystallization reduced to a 
minimum. The nuclei formed by cooling can no longer 
grow in the almost solidified glass and either remain 
hidden or lie at the limit of visibility. If by heating 
the glass acquires a certain mobility, the gold in solution 170 
AMICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
separates out upon the nuclei present, which by growth 
become submicrons visible in the ultraapparatus, the 
glass turning red or becoming darker in color. 
That none, or but few nuclei are formed at the 
higher temperatures at which these centers rapidly 
increase in size, is evident not only from the fact that 
the glass examined remained colorless and optically 
clear1 upon slow cooling, but also from the fact thaat 
the mean distance between the particles in the reheated 
glass was just as great at a as at b2 and c. If new 
nuclei had been formed during reheating at temperatures 
favorable to the development of the original nuclei 
into submicrons, these new nuclei would have developed 
into submicrons in the strongly heated portion of the 
glass (at a) and there would be more particles per unit 
volume at a than at b or c. But the number of particles 
per unit volume was approximately the same in all the 
variously heated parts of the glass. 
If, as in the case in question, the ruby glass is very 
slowly cooled and then quickly heated again, it is self- 
evident that most of the gold nuclei must have been 
already formed during the cooling. Colorless, slowly 
cooled gold ruby glass contains one part of its gold, 
therefore, in supersaturated solution, and the rest in 
the form of nuclei so small that they affect the homo- 
geneity of the glass only slightly or not at all. Upon 
1 The higher temperatures at which the crystals grow quickly are 
slowly passed by in cooling the molten glass. If nuclei had been 
formed at these temperatures they would have certainly developed 
into visible bodies. In the pieces examined this was only partially 
the case. With other ruby glasses the temperature for spontaneous 
formation of nuclei is much higher; such glasses color up red upon 
slow cooling. 
2 In Sample I this was only indistinctly recognizable; in Sample 
II, distinctly. SPONTANEOUS CRYSTALLIZATION IN RUBY GLASS 
171 
subsequent reheating these nuclei serve as centers of 
growth for separating out the gold dissolved in the glass. 
The hotter the glass is kept,1 the more speedily they 
grow at the expense of the gold in (crystalloid) solution. 
Therefore (the time of heating being the same) the 
largest particles are to be found at the places most 
strongly heated.2 The gold will continue to separate 
out until the supersaturation for that particular tem- 
perature and glass composition has disapperacd. 
Colorimetric tests have shown that in gold ruby glasses 
under ordinary working conditions, only part of the 
total gold (about half), separates out in the form of 
ultramicroscopic particles and serves as a color-producing 
constituent, the balance remaining in homogeneous 
solution. 
At still higher temperatures the crystalloid solubility 
of the gold is considerably greater, and by smelting at 
1350-1400° C. it is therefore easy to convert red ruby 
glass into the colorless form, which can again undergo 
the process of coloring. 
Application to Technical Observations (made in actual 
practice). The value of this view of the coloring of 
ruby glasses is shown incidentally by the fact that it 
simply and naturally explains certain heretofore incom- 
prehensible observations made in the course of their 
manufacture. 
Thus it was observed that the rim of a piece of pressed 
ruby glass remained colorless, while the middle became 
red. In the press the rim was more quickly cooled 
than the middle, and by subsequent reheating was 
heated more quickly and to a higher temperature. 
1 But only to a certain limit of temperature, as will soon be seen. 
1 This heating should not be carried so far that the glass becomes 
fluid, otherwise the separated gold will redissolve. 172 
AMICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
Investigation showed that the rim contained a much 
lesser number but considerably larger green gold particles 
than the rest of the glass. In the light of the above 
remarks these facts arc easily explained as follows: The 
rim of the glass had very quickly passed through the 
optimum temperature for the formation of nuclei, so 
that only few were formed; subsequent heating to a 
higher temperature therefore led to a speedy growth 
of these nuclei into particles of 110-145 pp. The middle 
section of the same glass had time enough to form a 
large number of nuclei, which, being less strongly 
heated, grew more slowly.1 
In spoiled ruby glass the formation as well as the 
growth of nuclei is disturbed; fewer nuclei are formed 
and these grow more slowly than in good ruby glass. 
In this case I think that instead of simple crystals either 
crystal-druses or sphero-crystals are formed; if every 
individual little crystal is conceived to act as a resonator 
for light waves, such crystals would be able to influence 
each other differently in the crystal-druse than in simple 
crystal formation, and thus by displacing the maximum 
of spectral resonance, produce a change of color in the 
glass and the diffracting disks. (See Chapter XII 2.) 
Difference between Devitrification and Ruby-glass 
Formation. From what is stated in the present chapter, 
it is evident that the phenomena which appear during 
the coloring of ruby glass can be explained by consider- 
ing the analogy between the process of coloring and 
that of devitrification, which was closely studied by 
1 The relations are not always as here given; we are considering 
here only the case where the temperature for the maximum value of 
sp KV. is lower than for the maximum value of KG. 
2 In many kinds of ruby glass the color change is due to a con- 
glomeration of the little submicrons into larger ones, in such cases the 
process is exactly the same as in the coagulation of colloidal gold. SPONTANEOUS CRYSTALLIZATION IN RUBY GLASS 
173 
Tammann. But the coloring of ruby glass should not he 
confounded with ordinary devitrification. 
The difference between the two is that in the formation 
of ruby glass several milliards of nuclei exist and grow 
in a cubic millimeter, and that the hypothetical little 
crystals, of whose form and structure we know nothing, 
are so small that their presence changes only the visible 
appearance and not the working properties of the glass; 
whereas, in the case of devitrification, relatively few 
nuclei are present and lead to the formation of quite 
large crystals, which prevent the usual working up of 
the glass. 
Regarding the size of the smallest gold nuclei hardly 
anything can at present be stated; for in order to apply 
the method used for determining their size, an idea 
must first be formed as to the mass per unit volume 
of the gold separated out into nuclei. That only a 
small fraction of the total gold can be present in this 
form is evident from the fact that most of the gold has 
the property of depositing upon the centers of growth; 
but how large this fraction is cannot yet be exactly 
determined. The next question is, if colorimetry is 
applicable here; its applicability must first be demon- 
strated.1 
Perhaps the intensity of the diffracted light will allow 
us to form an idea of the size of the particles, providing 
some relation may be found between their size and their 
diffracting power. 
Preliminarily I might state that the superior limit 
of the mass of the smallest submicroscopic gold particles 
is from 1.2 to 6.8 X10-15 mg. (indicating a linear dimension 
1 My recent work on amicroscopic gold nuclei give very promising 
indications of the applicability of colorimetric methods (Nov. 1, 1908). 174 
AMICROSCOPIC NUCLEI IN COLORLESS RUBY GLASS 
of from 4-7 and that they were found in the medium 
heated portion of a piece of glass which had been heated 
at one end, as described in the case of good ruby glass 
on page 166. In the less strongly heated portion of 
this glass no individual particles could be recognized, 
but the amicrons present give rise to a light-cone which 
became weaker as the slightly heated portions of the 
glass were approached, and finally vanished where the 
glass was colorless when the opening of the slit was small. 
Upon opening the slit wide enough and using the brightest 
sunlight, even here there could be seen a faint but dis- 
tinct whitish polarized light-cone. 
In the sphere of amicrons, therefore, there exist 
numerous transition forms between the smallest sub- 
microns and the smallest nuclei. If the former consist 
of several thousand atoms,2 it is probable that the 
latter contain only a small number of gold atoms. 
Recapitulation. The preceding remarks lead to the 
assumption that at ordinary temperatures colorless gold 
ruby glass contains gold in two different forms; most 
of it is present in supersaturated crystalloid solutions, 
but part exists as nuclei which at higher temperatures 
serve as centers of growth, and are so small that their 
presence influences the homogeneity of the glass slightly 
or not at all.3 
1 Siedentopf and Zsigmondy, loc. cit. 
2 The mass of 1.2 to 6.8 X 10-15 mg. indicates 8 to 42 thousand atoms, 
if the absolute atomic weight of gold is taken as = 196 X8.2X 10-22. 
This, however, is the major limit. 
3 In very great thicknesses, which cannot be practically realized, 
even colorless ruby glass would presumably show some color, just as 
colorless air appears colored in thick layers. (See p. 7 and Spring, 
loc. cit., note 1, p. 162). CHAPTER XVIII 
GENERAL REMARKS CONCERNING METAL HYDROSOLS 
i. Protective Action of Water 
Nascent gold reduced in water exists in homogeneous 
subdivision, just as vapor of gold does at first in a very 
hot tube. Neither in the tube nor in a beaker of water 
is this subdivision stable at ordinary temperatures; 
in each case crystallized masses form if the metal is in 
sufficient concentration. 
While in the case of gold vapor we have no way of 
maintaining the original state of subdivision, we do 
know of methods which, although not always satisfac- 
tory, produce approximately this effect if the gold is 
reduced in fluid media. 
One of these methods consists in the very extreme 
dilution 1 of the mixture employed for reduction. 
From concentrated solutions of gold chlorid well-formed 
macroscopic crystals can sometimes be obtained upon 
the reduction of the metal. Bearing in mind this ten- 
dency of gold towards crystallization, it is evident that 
the more dilute the fluid and the more centers of cry- 
stallization formed in the fluid, the smaller the crystals 
1 As brought out in Chapter VIII, the colloidal impurities of water 
usually tend to prevent the formation of metal solutions, and must 
therefore be removed as carefully as possible. The above considera- 
tions lead to the result that the formation of a metal hydrosol can 
also take place without the presence of foreign colloids. 
175 176 
REMARKS CONCERNING METAL HYDROSOLS 
will be. If the reduction is effected in very extreme 
dilution, so that the whole of the chloric! is almost 
instantaneously transformed into metallic gold, such 
centers would be simultaneously formed in enormous 
numbers, and the entire metal content of the fluid 
consumed either in their formation or in their growth; 
in the former case there would be obtained an approxi- 
mately homogeneous fluid with amicroscopic particles; 
in the latter, a fluid containing amicrons or submicrons, 
according to the size of the crystals formed. 
In fact, every drop of a colloidal gold solution of the 
latter variety contains milliards of individual particles, 
indicating the presence of an extremely large number 
of nuclei during reduction; still much larger must be 
the number of gold particles and nuclei in the gold 
solutions which are almost homogeneous. That these 
particles cannot settle out is evident from the fact that 
the commencement of sedimentation is observed where 
the particles are 30 yy; suspensions that settle completely 
contain particles 80-200 yy and still larger in size. (See 
Chapter XVI.) 
We can conceive the solutions of colloidal metals to 
be fluids teeming with such tiny hypothetical crystals.1 
But it would be a grave error to carry the comparison 
too far and to apply to these extremely fine subdivisions 
of the substance the results obtained with ordinary 
suspensions of gold crystals (which do not influence 
each other). I refer to Chapters I and II, wherein some 
of the differences are mentioned; I fully believe that 
11 must here expressly state that I lay no particular stress upon 
this idea, which would be abandoned if it could be replaced by a better 
one. It has, however, often been of use to me, and has enabled me 
to regard things, otherwise difficult of comprehension, from a broad 
point of view. (See Chapter XVII.) PROTECTIVE ACTION OF WATER 
177 
a world of wonders would be opened up to us if we could 
observe the actual structure of the particle itself, and 
its action upon its immediate surroundings. 
Graham's remark that in colloidal solutions there 
resides activity,1 is especially applicable * to colloidal 
metal solutions. More so than all others do they undergo 
irreversible changes of condition, changes irreversible 
in the fullest sense of the word, for all means which, with 
other colloids, can be successfully used to fluidify the 
hydrogel formed by coagulation, fail in the case of 
coagulated pure metal hydrosols. The greatest tendency 
of these tiny metal particles is to unite into larger 
complexes, and a variety of trivial causes are sufficient 
to bring about this union. .Upon such union there 
occurs a much more complete separation from the sur- 
rounding medium than with other colloids. This com- 
plete separation from the fluid and union of the particles 
with each other involves a considerable liberation of 
heat.2 No actual hydrogel is formed, but a metal- 
sponge or metallic powder. And even the relatively 
large particles, of which these latter consist, have not 
exhausted their tendency toward union. With gold, 
slight pressure with a polishing iron is sufficient to 
convert the dried powder into coherent metal, quite 
in contradistinction to dried hydrogels, which for the 
most part are friable. 
We have characterized the colloidal metal solutions 
as fluids which, among other things, differ from equally 
concentrated suspensions of the same metal by their 
larger energy content.3 
1 See Chapter III, p. 34. 
3 See Prange's experiment, p. 15. 
3 The development of heat upon coagulation is explained by the fact 
that the living force with which the individual particles hurl themselves 178 
REMARKS CONCERNING METAL HYDROSOLS 
It should be remarked that coagulation (in the absence 
of protective colloids) always begins if the individual 
particles are brought near enough to each other. It also 
commences upon the addition of electrolytes whether 
such addition increases the electropotential difference 
between the individual particles and the medium (accord- 
ing to Billitzer) or not. 
All these facts, together with the impossibility of 
preparing metal hydrosols from powdered metal in a 
purely mechanical way, are evidence that with con- 
centrated colloidal metal solutions the process of coagu- 
lation must be self-contained, involving the performance 
of a definite quantity of work, and therefore being asso- 
ciated with a decrease in the free energy. 
I consider the attractive forces between the individual 
particles to be the actuating cause of this process, 
forces of whose intimate nature we know nothing with 
certainty, but which are identical with those which oppose 
the mechanical separation of the united particles (forces 
of cohesion). 
If this idea is correct, the next question is what causes 
the relative permanence of the dilute metal hydrosols. 
It is safe to regard three causes as of importance in 
producing the relative permanence of such hydrosols: 
First, the proportionately large distance between the 
individual particles in dilute solutions, which is unfavor- 
able to their meeting each other. Second, the films 
of water which immediately surround the particles and 
are probably held by them with more or less force.1 
upon each other, becomes heat upon their conglomeration. In the 
absence of complications, we can calculate therefrom the mechanical 
work supplied on the average by one particle upon coagulation. 
1 Regarding the pressure of fluids which are absorbed by porous 
substances, see Lagergren, Zeitschr. f. anorg. Chern., 1900, Vol. XXIII, 
p. 323. Barus and Schneider also assume aqueous envelopes. PROTECTIVE ACTION OF WATER 
179 
Third, the electric charge of the ultramicrons, which 
Hardy already regarded as of importance in producing 
the stability of the irreversible hydrosols.1 
Finally, the addition of protective colloids materially 
increases the stability. 
The influence of the first cause above referred to is 
shown by the fact that the change in color of colloidal 
gold solutions indicative of coagulation and produced 
by the addition of salt, is retarded if the distances 
between the particles is increased.2 On the contrary, 
if the distances be decreased, coagulation begins spon- 
taneously. 
That attractive forces exist between the particles 
and water is evident from the fact that metals are 
1 Billitzer has already explained this on the sound basis that either 
anions are given up by the particles or cations are taken up by them. 
(Bredig had previously expressed a similar view.) If one considers 
that the formation of colloidal gold proceeds much more smoothly in 
neutral or faintly alkaline solutions than in slightly acid ones, and 
reflects that in the former class of solutions the number of hydroxyl 
ions is vastly greater than in the latter, the idea is easily formed that 
the charge of the particles is due to hydroxl ions being taken up by 
adsorption. The assumption that the alkali or another electrolyte, or 
else some trace of impurity in the water, is absolutely necessary for 
the formation of a hydrosol, does not seem to me to be of much sig- 
nificance; for the purest water contains in itself all the conditions 
necessary to give the hydrosol permanence. True protective action 
by electrolytes is unknown with colloidal metal solutions. On the 
contrary, the smallest quantities almost always produce coagulation. 
Furthermore the assumption of a "finest dust" for condensation 
nuclei appears to me to be superfluous, for if gold is reduced (or 
vaporized by the electric arc) in a medium in which, as a crystalloid, 
it is so difficultly soluble as in water, extreme dilution itself will yield 
all the conditions necessary to the formation of centers of crystalliza- 
tion. But this does not mean that many foreign substances in almost 
molecular subdivision may not act as nuclei. Perhaps phosphorous 
proposed by Faraday for reducing gold, or else some of the oxidation 
products of phosphorous, assume such a function. 
2 See Kirchner and Zsigmondy, Drude's Ann., 1904, Vol. XV, p. 591. 180 
REMARKS CONCERNING METAL HYDROSOLS 
wetted by water and also that pieces of metal take up 
water from the air, and thus increase in weight. The 
view that the particles of the surrounding fluid medium 
which are united with the metallic particles by attractive 
forces, oppose the union of the latter, is, among other 
things, supported by an experiment of Barus and Schneider, 
showing that at critical temperature colloidal silver 
precipitates from alcohol just as do clay particles from 
superheated water (Wiedem. Ann., 1893, Vol. XLVIII, 
p. 335). Thus when fluid and vapor become identical, 
the protection exercised by the fluid on the metal 
disappears. 
Freezing, the addition of electrolytes, etc., produce 
a gathering of the particles which cannot be made 
retrogressive. In short, any circumstance which removes 
or to a certain degree lessens one or several of the before- 
mentioned influences that give the metal sols their 
stability, will cause the coagulation of the metal hydrosol, 
and a complete theory of coagulation cannot therefore 
allow for only certain of these causes individually, but 
must take them all collectively into consideration. 
Regarding coagulation of hydrosols by electrolytes, 
it should also be noted that the addition of salt produces 
not only a gathering but also an electric discharge of 
the particles, so that according to Hardy the coagulum 
is isoelectric with the surrounding medium; further- 
more, according to Picton and Linder, Spring, Whitney 
and Ober, Billitzer, and others, there are taken up by 
the precipitated hydrogel a certain number of such ions 
of the precipitating salt as have an opposite charge to 
the particles. 
As Quincke has shown,1 we must also take into con- 
1 Quincke, D rude's Ann., 1902, Vol. VII, p. 95 (paragraph 5). Para 
graphs 3 and 4, advanced by Quincke as a basis for explaining the PROTECTIVE ACTION OF WATER 
181 
sideration the eddies produced in the fluid by the addition 
of electrolytes, for they cause the particles to dash 
together in a most powerful manner.1 This influence 
can be shown to be unimportant by the fact that alcohol 
and other non-electrolytes produce just as strong eddies 
and even increase or diminish the difference of electric 
potential between the particles and the medium, but 
effect no immediate coagulation of the metal hydrosol. 
On the other hand, it must be pointed out that such 
addition may exercise a protective action, slight or 
passing though it may be. Thus Blake 2 ascribes to 
ether a protective action on colloidal gold. 
According to what precedes a protective action3 can 
be also ascribed to pure fluid water, which prevents 
the metal particles from forming larger complexes and 
makes possible the production of colloidal metals. The for- 
mation of a colloidal metal solution can be conceived of as 
a condensation process interrupted in its very incipiency. 
It should be noted that electrolytes in general favor 
the separation of matter subdivided in water, not only 
bringing about the precipitation of metal hydrosols 
and of mechanical suspensions, but also diminishing 
the solubility of non-electrolytes (such as ether, phenyl 
thiocarbonate,4 etc.).5 
turbid solutions of mastic, clay, and oleic acid, are not well applicable 
to the case in question. 
1 Freundlich has found that solutions of arsenic sulphid are much 
more indifferent to a very slow addition of barium chlorid than to a 
sudden addition. Zeitschr. f. phys. Chem., 1903, Vol. XLIV, p. 143. 
2 Blake, Am. Jour, of Sci., 1903, IV (16), p. 435. 
3 No definite idea should be ascribed to the term "protective action," 
which simply expresses the fact that a union of the ultramicrons is 
prevented, which, without the presence of water, would doubtless take 
place at ordinary temperatures. 
4 Rothmund, Zeitschr. f. phys. Chem., 1900, Vol. XXXIII, p. 401; 
Biltz, Zeitschr. f. phys. Chem., 1903, Vol. XLIII, p. 41. 
6 This action of electrolytes, which extends to crystalloids as well 182 
REMARKS CONCERNING METAL HYDROSOLS 
2. The Finest Subdivisions of Gold 
In AuPie we have become acquainted with a gold 
solution with amicroscopic particles whose dimensions 
cannot be much larger than those which the kinetic 
theory of gases ascribes to the molecules of crystalloid 
solutions; in fact it is still an open question whether 
the discrete particles in gold solutions of the best kind 
must be necessarily larger than those assumed to exist 
in crystalloid solutions of substances having high molecular 
weight. 
That actual discrete particles are to be assumed in 
the latter follows unequivocally from the research 
referred to in the Introduction, especially that of Lobry 
de Bruyn and Wolff,1 and van Calcar and Lobry de 
Bruyn.2 
Barus and Schneider have expressly stated that it is 
not necessary to assume an allotropic modification of 
silver in colloidal solutions of this metal, and that it is 
simpler to hold to the normal molecular weight (which 
with metals is for the most part the same as the atomic 
weight).3 
On page 288 of their article they state: 
"The question therefore arises if, under favorable 
conditions, there may not exist suspensions in whose 
particles are united 1000, 100, 10, or even still fewer 
as to colloids, appears to have its cause not so much in the action of 
the ions in uniting the particles, but rather in a change in the rela- 
tions of the medium to the substance subdivided in it, a fact pointed 
out by Faraday. In the sense expressed by Donnan (Zeitschr. f. 
phys. Chem., 1901, Vol. XXXVII, p. 741) this may perhaps be ex- 
pressed by saying that r < 1. 
1 Note 1, p. 6. 
2 Note 3, p. 6. 
2 Chapter III, p. 59. THE FINEST SUBDIVISIONS OF GOLD 
183 
molecules. In extreme instances such mixtures would 
be difficult to distinguish from a true solution, and in 
colloidal solutions, we think, we have to do with such 
extremely fine subdivided sediments." 
Apart from the expression " suspensions " and " sedi- 
ments," which do not appear to be at all appropriate 
terms for such fine subdivisions,1 Barus and Schneider 
have expressed about the same idea that I have formed 
with regard to gold solutions with amicroscopic particles. 
If we imagine a metal solution dissolved in water 
consisting of an enormous number of metal particles, 
each of which consists of but few atoms, it is an open 
question whether, in their totality, these particles could 
exert osmotic pressure. If we assume that this property 
is lacking we would have before us an ideal case of col- 
loidal solution. Since it would have no osmotic pressure 
("the driving force ") 2 it would be incapable of diffusion 
and be devoid of the property of passing through mem- 
branes. Because of the high tendency toward union 
possessed by their particles, trivial causes would lead 
to their coagulation; in short, they would have the 
characteristic properties of typical, irreversible hydrosols. 
Such substances would be regarded as heterogeneous 
in the sense of the phase rule; their particles would 
grow in a supersaturated solution of the same metal, 
just like centers of crystallization. (Chapter XVII.) 
Furthermore, the particles in them would be hardly 
larger than those in crystalloid solutions; in both, the 
optical inhomogeneity would almost or entirely vanish. 
There would therefore be possible an ideal colloidal solu- 
1 Instead of the expression "true solution," used by Barus and 
Schneider, it would be better to say crystalloid solution. See also 
pp. 1-10. 
2 See Nernst, Theoretische Chem., 3d ed., p. 384. 184 
REMARKS CONCERNING METAL HYDROSOLS 
tion combining with its own essential characteristics the 
homogeneity of a crystalloid solution. It cannot with 
certainty be stated whether this ideal limiting case has 
been realized, but colloidal solutions of the type AuPie 
closely approach it, with the exception that the dimen- 
sions ascribed to molecules actually lie between 0.1 
and 1 pp. 
The study of the finest subdivisions lying on the 
border line between crystalloid and colloid solutions, 
would doubtless be of great interest. It might be antici- 
pated that upon the transition of the crystalloid solution 
into a colloidal one containing the smallest amicrons, 
there would be a sudden change in certain properties 
(such as osmotic pressure, behavior with reagents, etc.), 
whereas other properties (such as the optical homogeneity) 
would change only imperceptibly or not at all. 
Biltz and Gahl (Chapter XX), have investigated two 
analogous cases and observed a sudden change in optical 
homogeneity upon the separation of sulphur and selenium 
from their crystalloid solution. 
The cause of this, according to my experience, is that 
there had spontaneously formed in the crystalloid 
solutions of these elements, relatively fewer nuclei than 
in the finest subdivisions of gold; in the very much 
supersaturated solution these nuclei had grown with 
extreme rapidity. With gold solutions it is within our 
power to permit the simultaneous formation of an 
extremely large number of nuclei, which soon exhaust 
the supply of gold and remain extremely small. In such 
cases the ultraapparatus is incapable of showing the 
transition from crystalloid to colloid solution, because 
the optical homogeneity of the fluid is not perceptibly 
altered by this process. ACTION OF PROTECTIVE COLLOIDS 
185 
3. Action of Protective Colloids 
We must 'differentiate two kinds of action exercised 
by protective colloids: First, protective action upon 
finished metal hydrosols free from appreciable quantities 
of foreign colloids; second, protective action in the 
production of colloidal metals. Regarding the first 
kind of protective action, the most important facts have 
been stated in Chapter III and XIII;1 in addition it 
might here be mentioned that even protective colloids 
having the same electric charge as the. metal to be pro- 
tected, exercise a very efficient protection, and that 
the origin of the protection of the gold can be most 
simply explained by the assumption that specific attrac- 
tive forces bring about a union of the ultramicrons of 
metal and protective colloid, even if both carry like 
electric charges. That a union actually occurs I have 
been convinced of several years ago by experiments 
which cannot be discussed until later. 
It has also been stated that in observing certain 
relations in the "gold figure," we can obtain a relative 
measure of the protective action upon a certain metal 
hydrosol. 
The second type of protective action, which is closely 
related to but not identical with the former, comes into 
play in the production of water or alkali-soluble col- 
loidal metals. They make it possible to obtain very 
concentrated, water-soluble colloidal metals, which, 
because of their appreciable content of foreign colloids, 
may be regarded as analogues of the purple of Cassius, 
or else as colloid compounds. (Chapter III.) 
Here belong the processes for the production of the 
purple of Cassius, of colloidal metals according to Carey 
1 Pages 81-84 and pp. 150-152. 186 
REMARKS CONCERNING METAL HYDROSOLS 
Lea, Lottermoser and Paal, as well as Mohlau's process 
for making colloidal indigo. 
As to the action of protective colloids in the formation 
of metal hydrosols, only hypotheses can at present be 
advanced. 
It seems to me not improbable that the amicrons 
of the protective colloid take up by adsorption a part 
of the dissolved metallic salt, and that the reduced 
metal then remains united with the amicrons in such 
fine subdivision that the homogeneity of the fluid is 
not affected notwithstanding the reduction of the metal. 
(Some years back I was thus enabled, by reducing gold 
chlorid in the presence of stannic acid, to obtain purple 
of Cassius whose solution in ammonia diffracted just 
as little light as the ammoniacal solution of stannic 
acid by itself. A faint, polarized light-cone was dis- 
cernible in both cases). 
It can also be assumed that the metallic salt is not 
adsorbed, but that the reduced metal is taken up by 
the amicrons of the protective colloid. Finally it appears 
to be a general property of protective colloids to pre- 
vent or interfere with the growth of tiny little crystals. 
Other less probable assumptions are also possible, but 
it would take too long to recount them. 
Only one thing must be referred to: The protective 
action is not absolute; the protective colloid is incapable 
of preventing the incipient condensation of the metal, 
which would have to be the case in order to obtain 
stable colorless metal solutions analogous to quickly 
cooled ruby glass.1 The protection is quite often so 
perfect that the metal particles which separate out do 
1 In individual cases temporarily colorless gold solutions may indeed 
be formed; by indicative measurements the subsequent reduction 
could be easily determined. ACTION OF PROTECTIVE COLLOIDS 
187 
not appreciably increase the inhomogeneity of the 
solution. 
The reactions of metal solutions made in the presence 
of protective colloids, are chiefly determined by those 
of the latter; if, however, the metal is present in large 
excess, its tendency to combine into larger particles 
comes into evidence, influencing the reaction accordingly. CHAPTER XIX 
ULTRAMICROSCOPIC EXAMINATION OF CERTAIN 
SOLUTIONS AND SUSPENSIONS 
To supplement what has been said in the preceding 
chapters, there are here given some of my observations 
on various colloidal solutions and suspensions, and also 
on solutions of dyestuffs. 
Professor Bredig of Heidelberg was kind enough to 
place at my disposal the following four hydrosols of 
definite concentration prepared according to his method, 
for which I herewith tender him my sincere thanks. 
Bredig's Gold Hydrosol. The gold hydrosol examined 
was a fairly clear fluid, purple or violet-red to transmitted 
light, with a brownish diffuse dispersion. It contained 
0.0053 grm. of gold in 100 grms., and tAoo normal 
NaOH. The fluid contained red, yellow, and green 
particles; the yellow ones present in the smallest num- 
ber were usually the brightest, while the red and green 
ones, present in about equal numbers, exhibited various 
degrees of brilliancy. All of the particles, however, 
were considerably brighter than those in my bright 
red colloidal solutions, for instance Augo, Augg, etc., 
and despite their active motion had smaller free paths 
than those in the latter. The amplitude of the trans- 
latory motion was generally smaller than 4 y. Because 
of the great difference between the particles, their mean 
size (42 yy), determined by measuring the spaces between 
them, is naturally of little value. The very brightest 
188 CERTAIN SOLUTIONS AND SUSPENSIONS 
189 
particles could be separated by filtration through very 
heavy filter paper. By comparing their brightness with 
that of gold particles in ruby glasses and in gold hydrosols 
already examined, the size of the particles in Bredig's 
preparation was placed at 20-80 yy. 
Bredig's Colloidal Platinum. A yellow-brown fluid 
exhibiting slight turbidity to transmitted light (platinum 
content1 0.0034 grm. in 100 cc.). At least four different 
grades in the sizes of the particles were determined.2 
In contradistinction to the gold particles the platinum 
particles are not strongly colored, but are all white or 
rather gray-white, with tinges of yellow or blue. I 
could recognize no particles with decided red or green 
color. Although the smaller particles, far outnumbering 
the larger ones, were much fainter than those in the 
gold hydrosol previously described, a calculation based 
upon the distance between them gave as their mean 
size 44 yy, which indicates that in fine subdivision plati- 
num diffracts less light than gold, the size of the particles 
being about the same. Feb. 9, 1904. 
Bredig's Palladium Hydrosol. (0.0050 grm. Pd in 100 
grms.) After long standing (which yielded no sediment), 
this blackish-brown fluid appeared almost clear. By arc 
light could be seen single bright white particles with 
tinges of yellow, blue, etc., and a light-cone due to 
invisible particles. With the use of sunlight the smaller 
particles could also be seen; they were not in active 
motion, and showed no decided color. 
Bredig's Colloidal Silver Solution. (0.0038 grm. Ag in 
100 grms.) A yellowish-brown fluid showing a gray 
cloudiness to transmitted light; it can hardly be imagined 
1 The content of this and other colloidal metal solutions were 
kindly given me by Prof. Dr. Bredig. 
2 By comparing the brightness of the diffraction disks. 190 
ULTRAMICROSCOPIC EXAMINATION OF 
what a beautiful color play this unsightly fluid offers 
upon ultramicroscopic examination. Blue, violet, yellow, 
green, and reel particles of various shades and rare bril- 
liancy of color, move in endless array. One particle 
approaches another, circles it in a rapid zigzag motion, 
and is then dashed away again; sometimes one follows 
another almost in contact but without reaching it. 
Sometimes several group together and dance like gnats 
in a sunbeam, especially when, for the fraction of a 
second, one particle approaches the other. The fluid 
was examined in various degrees of dilution. Often the 
influence of one particle on another could be seen even 
at a distance of 2-4 /i. The linear dimensions of the 
individual particles was about 50-77 fi/i. 
Silver Hydrosol+ NaCl.. The introduction of sodium 
chloric! solution to the silver sol in the cell immediately 
produced a whirling together of the particles which 
made it impossible to observe the process of coagulation. 
Upon equalization of the differences of concentration by 
diffusion streams, the considerably larger particles can 
be seen quietly floating. Sometimes two of these 
approach each other, but rarely can the process of the 
union be observed. 
Colloidal Silver made According to Carey Lea. Dr. 
Lottermoser of Dresden was kind enough to send me 
some colloidal silver made according to the method of 
Carey Lea, a rather concentrated solution containing 
1.276 per cent Ag. Upon dilution this dark brown fluid 
showed the same splendidly colored picture as Bredig's 
hydrosol; brought to the same concentration, both 
fluids would present about the same appearance.1 
1 This statement refers to arc light. With sunlight there could be 
seen in Lottermoser's preparation still much smaller particles in very 
active motion. CERTAIN SOLUTIONS AND SUSPENSIONS 
191 
A solution of Argentum Crede (solid colloidal silver), 
obtained more than four years ago from the Heyden 
Chemical Works, contained somewhat larger and less 
variegated particles. 
Hydrosol of Mercury. The hydrosol of mercury which 
was kindly placed at my disposal by Dr. Lottermoser,1 
also contained numerous brightly shining, active particles. 
Unfortunately I neglected to make an immediate thor- 
ough examination of this preparation, which is interesting 
in several respects. As I was about to take up the com- 
plete examination after several weeks standing, I was 
sorry to find that it had coagulated. 
Colloidal Ferric Oxid. Two solutions of ferric oxid 
of different origin were examined; in concentrated 
solution (1-5 per cent) both showed a very intense 
polarized bluish light-cone, which became fainter upon 
further dilution, and in which (besides the diffused light) 
slow moving individual particles could also be recognized. 
One of the fluids (A) appeared heterogeneous in a con- 
centration of 0.001-0.0005 per cent; the other (B) 
remained apparently homogeneous until the light-cone 
vanished. 
According to this result it seems to me very probable 
that the solution of ferric oxid, in addition to large 
individual particles easily made visible, contain chiefly 
amicrons whose size should be between those of the gold 
solution AuPi32 and those of a gold solution of medium 
subdivision. 
As Biltz 3 has found, colloidal ferric oxid and colloidal 
'A. Lottermoser, "On Colloidal Mercury," Jour. f. pr. Chem., 
N. F., 1898, Vol. LVII, p. 484. 
2 In which the light cone was much less distinct than in the iron 
solution of like concentration. 
3 Biltz, Ber. d. chem. Ges., Vol. XXXVII, pp. 1095-1116. 192 
ULTRAMICROSCOPIC EXAMINATION OF 
gold mutually precipitate each other. I therefore mixed 
solution B with AuPx. Before mixing, individual parti- 
cles were not to be seen in either, but a short time after 
mixing there could be seen a large number of bluish- 
green individual particles, having a brilliancy equal to 
that of gold particles about 15 y in diameter, but a much 
less active motion. 
Hydrosol of Silver lodid. A solution of silver nitrate 
(0.2 mg. AgNOs in 100 cc.) was mixed with a very 
dilute solution of KI. The immediate result was a 
perfectly clear fluid in which could be seen a blue polar- 
ized light-cone. At first no individual particles were 
visible, but after a long time extremely small particles 
having active motion could be seen, appearing as blue 
diffraction-disks. April 7, 1902. 
After ten days the particles became distinctly visible 
by sunlight, their active motion persisting. Calculated 
from the measurement of the distance between them, 
the approximate mass of these individual particles was 
10~14 mg. 
Lottermoser1 has also succeeded in producing con- 
centrated and quite stable solutions of silver ioclid. Dr. 
Lottermoser was kind enough to send me some of his 
preparation for examination; in this yellowish-white, 
milky hydrosol, the particles have already grown to 
considerable size. They are so closely crowded in the 
concentrated fluid that it is a wonder that so many 
particles can continue to exist alongside of each other. 
The concentrated fluid, Lottermoser states, contained 
0.98 grms. silver iodid in 100 cc. In it were seen num- 
berless white particles, rushing past each other as in a 
surging throng. One extremely important observation 
should be noted: isolated dust particles and threads, 
1 Lottermoser, Jour. f. pr. Chem ,, 1903, Vol. LXVIII, p. 341. CERTAIN SOLUTIONS AND SUSPENSIONS 
193 
sometimes stellated, were surrounded by a ring-shaped 
(or spherical) space, optically transparent, into which 
no particles of silver iodid could force their way. 
Upon further dilution the color of the individual 
particles became evident; they were mostly bluish- 
white, some being yellow-white and some greenish-white. 
Average size of particles about 60 /z/z. April 7, 1902. 
lodin Suspension. An alcoholic solution of iodin was 
poured into water, until a permanent turbidity was 
produced. Extremely bright shining particles could be 
seen, some showing slight Brownian motion, others 
not. 
As soon as the particles reached the focus of the light- 
cone, they were dashed back from it as from a racket, 
in the direction of the light rays.1 This phenomenon 
may perhaps, like the negative photodromy of Quincke,2 
be attributed to the unilateral heating of the suspended 
iodin. April 2, 1902. 
Barium Sulphate Suspension. These were produced 
with dilute solutions of H2SO4 and BaC12 in varying 
proportions. 
Only larger particles were visible, floating quietly and 
shining brilliantly. 
The crystalloid solubility of BaSO4 appeared to be 
too great to permit of the preparation of a colloidal 
solution. Either larger crystals are formed at the very 
outset, or else perhaps certain crystals grow at the 
expense of the smaller ones incidentally formed, in analogy 
with the rule stated by Lord Kelvin,3 that in vapors the 
1 Upon repeating this experiment two years later I could no longer 
discover the conditions under which the iodin particles were dashed 
away. 
2 Quincke, Drude's Ann., 1902, Vol. VII, p. 86. 
3 Lord Kelvin, Proc. Roy. Soc. Edinburgh, 1870, Vol. VII, p. 63. 194 
ULTRAMICROSCOPIC EXAMINATION OF 
large drops grow at the expense of the smaller ones.1 
April 4, 1902. 
Silver Chlorid Suspension. Lottermoser, in his paper 
on colloidal silver halids,2 remarks that the hydrosol 
of silver chlorid cannot be obtained according to the 
method described by him for the preparation of silver 
halids: even when very dilute solutions are used, the 
resulting fluid is bluish colored to transmitted light, and 
deposits AgCl. This statement I can entirely confirm. 
The precipitate yielded by very dilute sodium chlorid 
with silver nitrate solution, showed a strongly polarized 
bluish light-cone produced by larger particles having 
slight Brownian movement. Several experiments made 
in April, 1902, gave the same result. In one instance 
the distance between the particles was about 10 y, the 
amplitude of the motion of translation about 1 «, A 
hydrosol can no more be obtained in this case than in 
the case of BaSO<, apparently for the same reason. 
April 7, 1902. 
Turbid Solution of Gamboge. The very bright particles 
exhibit the slow, trembling motion characteristic of the 
Brownian movement; and to show that the motion of 
the gold particles is of a quite different nature, it is best 
to mix the two fluids. While the gamboge particles 
have a free path of 1-2 y,3 that of the gold particles is 
10-20 y and over. April 29, 1902. 
Carmine. Levigated commercial carmine exhibited 
bright red shining particles with but little or no motion. 
Solution of Carmine Dye in Ammonia (0.2 per 
cent carmine). Much smaller particles 6-10 « apart 
1 According to Ostwald (Lehrbuch. d. allg. Chem., 2d ed., II, p. 757) 
a solution may be supersaturated as to larger crystals, but hypersatu- 
rated as to smaller ones. 
2 Jour, prakt. Chem., N. F., 1903, Vol. LXVIII, p. 343. 
3 In freshly prepared solutions the motion is somewhat more active. CERTAIN SOLUTIONS AND SUSPENSIONS 
195 
show Brownian movement: A = 1-2 /z (in this and all 
subsequent cases A denotes the amplitude of the motion 
in question). The light-cone between them cannot be 
resolved. The smallest particles are not visible, as I 
already stated in 1902.1 April 2, 1902. 
Gelatin Solution. A 0.2 per cent gelatin solution 
two days old appeared heterogeneous. The fluid was 
entirely filled with small whitish particles at the limit 
of visibility; along with them were visible much larger 
hydrogel particles. In a 0.01 per cent gelatin solution 
the distance between these free moving particles was 
less than 1 /z; upon further dilution the microscopic 
picture became indistinct because of the predominance 
of the diffuse light arising from still smaller particles.2 
March 26, 1902. 
Soluble Starch.3 A 0.01 per cent solution showed 
by sunlight a distinct but faint diffuse polarized light- 
cone, together wth few brighter particles. March 14, 
1902. 
Further remarks regarding soluble starch are to be 
found in Chapter XXI, 
Rice Starch. When levigated in water isolated 
starch grains, illuminated by the light rays, shine bril- 
liantly, but exhibit no appreciable Brownian movement. 
April 2, 1902. 
1 Zsigmondy, Zeitschr. f. Elektrochem., 1902, p. 686. 
2 Upon repeating this experiment in the spring of 1904, using a 
gelatine solution several days old which had been prepared at boiling- 
point, the light-cone appeared homogeneous, and the individual 
particles could not be seen. The solution of March, 1902, was prob- 
ably prepared at medium temperature, and not at boiling-point. As 
I have already shown (Zeitschr. f. analyt. Chem., 1901, Vol. XL, p. 714), 
under these conditions there is formed a less perfect colloidal solution 
of gelatine. 
3 This preparation was kindly given me by Prof. Lintner, of Munich, 
to whom my best thanks are due. 196 
ULTRAMICROSCOPE) EXAMINATION OF 
Starch Paste. In diluted starch paste were to be 
seen: First, very large and extremely brilliant clumps. 
Second, numerous smaller particles showing slight 
Brownian movement. Third, finally upon extreme dilu- 
tion a distinct light-cone which could be almost entirely 
blotted out by turning the Nicol. This light-cone 
became much more distinct after the large suspended 
particles settled, and upon further dilution assumed a 
heterogeneous appearance. lodin colored it blue with- 
out facilitating the view of the individual particles. 
April 2, 1902. 
Cigarette Smoke. Gases containing floating parti- 
cles can be examined in our apparatus just as well as 
fluids. Cigarette smoke, which can easily be blown 
into the apparatus through the funnel-tube, exhibits 
an especially beautiful appearance. By closing the 
pinch-cock the air enclosed is protected from the motion 
of the outside currents. The smoke, even when it was 
very thick, contained particles 10-20 apart, which 
danced about in the elastic medium with an extraordi- 
narily active motion, and as soon as they were struck 
by the direct sunlight shined most brilliantly. Even 
their oscillations had an amplitude of 10 y and over; 
their motion of translation was considerably more 
extensive. April 29, 1902. 
Fluorescent Dyestuffs. For the sake of complete- 
ness I must here mention some experiments made in 
collaboration with FI. Siedentopf, having for their 
object the examination of solutions of strongly fluor- 
escent dyestuffs in the ultraapparatus. 
Fluorescein. A solution of this dyestuff, diluted to 
1 :1,000,000, showed quite an intense light-cone.1 It 
was still visible when this solution was diluted a hundred 
1 The depth of the image of the slit was 1-2 /z, its breadth 3-6 /z. CERTAIN SOLUTIONS AND SUSPENSIONS 
197 
fold, and finally vanished gradually upon further dilu- 
tion. (As a matter of course a Nicol prism was added 
as an analyzer, whenever necessary, in order to observe 
the fluorescent cone alone without interference with the 
colloids in the water.) Sometimes a kind of cloud for- 
mation could be seen, but so indistinct that no con- 
clusion can be drawn from this phenomenon. Several 
repetitions of the experiment led to the same result, even 
when objectives made entirely of quartz and fluorspar 
were used for illumination. April 10, 19, and 23, 1902. 
The very strongly fluorescent aescorcein, which had 
been kindly placed at our disposal by Prof. 0. Lieber- 
mann 1 of Berlin, for which we here express to him our 
sincere thanks, showed under the same conditions a 
distinct light-cone with a content of 2.5. IO-7 mg. per 
cubic centimeter. Upon further dilution the light-cone 
disappeared. 
Tetraiodofluorescein behaved in the same manner; 
several other non-fluorescent dyestuffs were still less 
visible in solution. 
There is, therefore, but little prospect, with the present 
arrangement of apparatus (especially with the use of 
arc light), of being able to see the molecules of ordinary 
dyestuffs which are dissolved as electrolytes and diffuse, 
particularly if they exhibit slight or no fluorescence. 
On the other hand it is very likely that in the colloidal 
solution of dyestuffs2 the individual particles can be 
seen, if they are large enough; and it is not impossible 
that the molecules of fluorescent dyestuffs can be made 
visible, providing they are very large and sufficiently 
fluorescent. 
1 Liebermann and Wiedermann, Ber. d. D. chem. Ges., 1901, Vol. 
XXXIV, p. 2611. 
2 Benzopurpurin, for example. (See note, p. 206). CHAPTER XX 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
PUBLICATIONS OF OTHER SCIENTISTS 
Some Investigations of Raehlmann. I have above 
made mention of Raehlmann's 1 very interesting obser- 
vation that in the ultraapparatus a large number of 
submicrons can be seen in a solution of glycogen, also 
their disappearance under the influence of diastase. 
Raehlmann was also able to identify albumin particles 
in the urine of nephritic patients; in this connection 
he has pointed out that the distance between the indi- 
vidual particles in urine containing originally one per 
cent of albumin can be measured only after very extreme 
dilution (1 :40,000), and that at a dilution of 1:500,000 
individual particles can still be seen in the solution. 
Earlier publications of Raehlmann describe the ultra- 
microscopic picture of certain dyestuff suspensions and 
dyestuff solutions and their mixtures.2 Raehlmann con- 
siders the new method of illumination as of especial 
advantage because it permits particles to be seen in 
their own color, thereby facilitating the differentiation 
between particles arising from different dyestuffs. He 
examined incidentally suspensions of chrome yellow and 
ultramarine, also the colloidal solutions of Prussian 
1 Raehlmann, loc. cit. (Chap. XV) and Berl. klin. Wochenschr., 
1904, No. 8. 
2 Ophthalmolog. Klinik, 1903, No. 16; Ber. d. D. physik. Ges., 1903, 
Vol. V, pp. 330-339; Physikal. Zeitschr., 1903, Vol. IV, pp. 884-890. 
198 PUBLICATIONS OF OTHER SCIENTISTS 
199 
blue, carmine, etc., and described in detail the-ultra- 
microscopic picture of these fluids and their mixtures. 
Raehlmann also examined the solutions of the electro- 
lytes, methyl violet and naphthol yellow, and discovered 
ultramicroscopic particles even in them. 
Reference must here be made to the ultramicroscopic 
investigation of Raehlmann on the constituents of the 
blood.1 
Investigations of Much, Romer, and Siebert. Inter- 
esting ultramicroscopic investigations were conducted by 
Much, Romer, and Siebert at von Behring's Insti- 
tute.2 In order to obtain a comparative measure for 
the number of ultramicroscopic particles, these investi- 
gators diluted their fluids to such an extent that only 
3-4 ultramicroscopic particles were present; the figure 
expressing the dilution thus obtained was termed the 
ultra-value. In this manner were examined numerous 
sera, albumin solutions and wheys. 
Just as Raehlmann had been able to see a large number 
of individual particles in the diluted urine of nephritic 
patients,3 these investigators also found individual par- 
ticles in albuminous urine, their number varying with 
the severity of the illness. Here, as well as in the fol- 
lowing experiments, the ultra-value served as a com- 
parative measure of the number of submicrons. In 
Table VII which follows, arc arranged the results of the 
examination of the urine of healthy persons and of 
nephritic patients. 
The coincidence between the ultra-value and the 
results of the boiling test is in every case remarkable, 
1 Deutsche mediz. Wochenschr., 1904, No. 29. 
2 Much, Romer, and Siebert, Ultramicroscopic Investigations. 
Zeitschr. f. diat. u. physik. Therapie, 1904, Vol. VIII, pp. 19 and 94. 
3 Raehlmann, loc. cit. 200 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
and leads to the conclusion that the greater part, if 
perhaps not all, of the albumin in urine is in the form 
of ultramicroscopic particles. 
Sample. 
Reaction on Boiling. 
Ultra-value. 
I 
0 
0 
coarsely flocculent ppt. 
0 (diabetic urine) 
0 
slight turbidity 
coarsely flocculent ppt. 
0 (diabetic urine) 
opalescence 
coarsely flocculent ppt. 
< 4 
(I 
0 
0 
0 
1:25 
1:60 
1:5000 
1:10 bright light-cone 
1:25 
1:400 
1:10,000 
1:30 bright light-cone 
1:200 
1:15,000 ['After removal) 1:25 
1 • 7 000 1 of albumin by I . r 
i. Z,UOOS boiling with fl-6 
1:10,000 I acetic acid J 1:35 
1:50 
1:20 
1:20 
II 
Ill 
IV 
V 
VI 
VII 
VIII 
IX 
X 
XI 
XII 
XIII 
(Normal urine, ex- 
clusive of portion 
voided in morning) 
XIV 
(Same; 4 o'clock, 
p.m., after heavy 
meal) 
XV 
(Same; 11 o'clock, 
p.m., after light 
supper) 
TABLE VII 
With this new method of determining albumin care 
must be taken, because other constituents of the urine, 
for example the phosphates, which sometimes separate 
out, may also be present in a similar form.1 
Much, Romer, and Siebert have, in addition, deter- 
mined the ultra-value of different sera and colloidal 
solutions, which are repeated in Table VIII. 
Also of interest are the experiments on the digestion 
of globulin from horse serum by pepsin and hydro- 
1 The authors themselves have referred to the influence of the mucin 
content. PUBLICATIONS OF OTHER SCIENTISTS 
201 
2-4 particles were counted in one field of view of the 
Proportion- 
ultra-apparatus when the fluid was diluted 
ate Figure. 
1. Horse sera: 
Paul serum 
100,000 
1 
Dora serum 
80,000 
4/5 
Ida serum 
60,000 
3/5 
old tetanus serum. Ballon la 
350,000 
3.5 
2. Fresh antitoxic horse serum 
300,000 
3 
3. 10 per cent globulin solution from horse 
serum 
20,000 
1/5 
4. Milk serum 
800,000 
8 
5. 10 per cent solution of Witte's peptone.... 
2,000 
1/50 
6. 10 per cent atmidalbumose from horse 
serum 
2,000 
1/50 
7. Marburg nutritive bouillon 
250 
1/400 
8. Urine in slight albuminaria 
500 
1/200 
9. Urine in severe albuminaria 
20,000 
1/5 
10. 10 per Cent gelatin solution . 
4,000 
1/25 
11. 10 per cent agar solution 
10,000 
1/10 
12. 10 per cent solution of laundry soap 
200,000 
2 
13. 10 per cent mucin solution 
300,000 
3 
TABLE VIII 
chloric acid,1 and of fat-free milk (ultra-value 750,000), 
by hydrochloric acid pepsin and pancreatin. A series 
of remarkable experiments was inaugurated with milk 
freed from fat, and milk freed from casein by filtration 
(whey or lactoscrum). The wheys of a cow highly 
immunitized to tuberculosis exhibited bactericidal prop- 
erties against bacterium coli. These wheys were sub- 
jected to electrolysis and an examination made of the 
anode, cathode, and intermediate wheys. It appeared 
that the anode wheys showed very powerful bactericidal 
and agglutinative properties and a high ultra-value, 
whereas, as may be seen from the following table,2 the 
1 After incubating for one-half hour the ultra value decreased from 
100,000 to 100. 
2 The spore-destroying and agglutinative action referred to is indi- 
cated by + + +, when less effective by + +, when very slight by +, 
and when absent by 0. 202 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
cathode wheys were almost free from ultra-particles 
and entirely free of bactericidal properties. 
Bactericidal 
Properties. 
Agglutinative 
Properties. 
Ultramicroscopic 
Examination. 
Anode wheys 
+ + + 
+ + + 
1:4000 
Cathode wheys 
0 
0 
1:50 
Intermediate wheys 
+ + 
+ + 
1:3400 
Non-electrolyzed wheys . . 
+ -0 
+ 
1:200 
TABLE IX 
Some Remarks on the Preceding. I cannot here 
enter into details regarding the work, and return to the 
method of determining the ultra-value only in view of 
more recent work in this field. By determination of 
the ultra-value the authors obtained mutually com- 
parable figures which gave an approximate idea of the 
relative number of submicrons contained 1 in the fluid. 
Unfortunately in this work, which was certainly care- 
fully carried out, there are lacking the facts necessary 
to reach a conclusion as to the absolute number of 
particles in a unit volume, from which could be obtained 
an idea of the major limit of the size and the mass of 
the particles. Such information would certainly have 
been of general interest and would have made possible 
comparison with the work of other investigators. 
By "Held of view " is evidently meant only that part 
of the field of view directly illuminated, for only in this 
do the ultramicroscopic particles become visible. The 
dimensions of the illuminated space can, however, be 
arbitrarily chosen, and vary according as the light-cone 
is defined, from front to rear,2 and according to its breadth 
and its depth. 
1 Or perhaps formed by dilution. 
2 If the field of view of the eyepiece is taken as a front-to-rear limit, PUBLICATIONS OF OTHER SCIENTISTS 
203 
The value of all future work in this field would be 
considerably enhanced if the facts as to the number 
of particles in a given concentration (or else as to the 
ultra-value) were based upon a certain predetermined 
illuminated volume of the fluid. This problem would 
be much simplified for investigators if eye-pieces pro- 
vided with a suitable Ehrlich diaphragm 1 were com- 
mercially procurable; 2 using this, together with definite 
tube-length and a definite objective, there could, to 
begin with, be marked out of the light-cone a disk of 
definite size, having a surface or area of 100 or 200/z2 
for instance. The operator has then only to take care 
that this disk is completely illuminated, and that the 
light-cone is correctly limited as to depth. By such an 
arrangement the errors in the determination of volume 
would be reduced to the errors in the determination of 
the depth, which are not important factors, because for 
the most part they arc smaller than the errors in the 
determination of the number of particles. 
If the investigators who use the ultraapparatus would 
take a little care to determine the mean number of 
particles in a volume of the fluid measured off as care- 
fully as possible, not only would the results of the 
different experimenters be mutually comparable, but 
there would also be obtained data for reckoning the 
which as a matter of fact is very disadvantageous, the length of the 
visible portion of the light-cone changes with the nature of the eyepiece 
used. 
1 Catalogue of C. Zeiss, Mikroskope, 1903, p. 97. 
2 Or a diaphragm with several different large openings so arranged 
that one can be easily substituted for the other. In fact, H. Siedentopf 
has recently introduced an eyepiece with checkerboard divisions, which 
is well adapted for the purpose of counting (Description of the Appa- 
ratus for Rendering Visible Ultramicroscopic Particles, Note 3, p. 97). 
To count submicrons approaching the limit of visibility, I should 
recommend, however, the eyepiece diaphragm referred to. 204 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
approximate size of the particles or their major limit, 
which information is of greatest importance in the 
further investigation of colloids. 
Furthermore, I consider it dangerous to dilute so far 
that only 3-4 particles are visible in the whole field, 
because then the dust particles incidently present in 
the water used for dilution are also counted in. It is 
more advantageous to dilute only to such an extent 
that about 1-6 particles arc present in a small, exactly- 
defined volume (of about 400-1000y3). (See remarks 
on the determination of size.) For dilution, of course, 
only such fluids must be chosen as produce no precipi- 
tative action on the colloidally dissolved substance. 
Investigations of Michaelis. From a preliminary 
communication of Leonor Michaelis,1 which has recently 
appeared, we quote the following. Michaelis subdivides 
dyestuff solutions according to their behavior in the 
ultraapparatus into: 
"(i) Dyes Totally Resolvable Optically. Their 
aqueous solutions even upon extreme dilution, exhibit 
ultramicroscopically numberless granules. Here belong: 
First, the aqueous solutions of many heavy molecular 
sulphoacid dyes, as indulin, violet-black, aniline blue; 
second, certain pseudo-solutions whose suspension-like 
nature is probable, even without the ultramicroscope, 
though it cannot be demonstrated; for example, (a) 
many dyes which, although their aqueous solutions cer- 
tainly do not belong to this class, exhibit the optical 
phenomenon of complete granular resolvability, if they 
are dissolved in aniline water even in a great degree of 
dilution (fuchsin); (6) a dilute solution of scarlet in 
alcohol, which is then diluted with five to six parts of 
1 Deutsche medizin. Wochenschrift, 1904, No. 42, and Virchows 
Archiv., 1905, Vol. CLXXIX, pp. 195-208. PUBLICATIONS OF OTHER SCIENTISTS 
205 
water. This procedure precipitates the dye, which of 
itself is not soluble in water, but it remains in ultra- 
microscopic suspension; (c) a dilute solution of fuchsin 
in hot, saturated NaCl solution, upon cooling becomes 
violet or blue to transmitted light, without yielding a 
precipitate. This solution can be totally resolved opti- 
cally. After twenty-four hours, the dye generally sep- 
arates out in flocks. 
" (3) Dyes Partially Resolvable. These also show 
granules in the ultramicroscope, but a certain concentra- 
tion must be reached before a large number is seen. 
The dye is contained in the solution in two phases, which 
bear a functional relation to each other; the first 
being a solution showing nothing in the ultramicroscope, 
the second being in a form optically resolvable. Here 
belong the aqueous solutions of fuchsin, methyl violet, 
and others. 
" (3) Dyes Completely Unresolvable, but fluorescent. 
They exhibit an optical inhomogeneity which is macro- 
scopically recoginzed as fluorescence. But with our 
present methods this inhomogeneity can by no means 
be resolved into granules even ultramicroscopically. 
Here belong the aqueous solutions of fluorescein (already 
examined by Siedentopf and Zsigmondy), eosin, toluidin 
blue, Nile blue, methylene blue." 
The fluorescence observed by Michaelis in methylene 
blue I can confirm from previous observation. 
From the second paper of Michaelis, already cited, I 
abstract the following remarks: 
Bavarian blue also belongs to the completely resolvable 
dyestuff solutions. Even upon extreme dilution, num- 
berless granules can be recognized in its solution. 
Although this dye consists of sulpho-acids, and it may 
be supposed to enter into solution, dissociated into ions 206 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
like an electrolyte, nevertheless its aqueous solutions 
act as hydrosols. 
In the second class, in addition to the solutions of 
fuchsin and methyl violet before referred to, belong 
the solutions of neutral red, Capri blue, and picric acid. 
Michaelis is of the opinion that the granules visible in 
them are not due to an impurity but to an integral 
constituent of the solution. He bases this opinion on 
the fact that particles microscopically visible are too 
profusely and too regularly distributed in the fluid to 
be regarded as an impurity.1 
1 This view is supported by observations made on solutions of gly- 
cogen, benzopurpurin, and others. The solutions of benzopurpurin, 
for instance, show with sunlight sometimes a light-cone resolvable 
into a large number of bluish-green particles, and at other times a 
light-cone of the same color, due to amicrons. A freshly prepared 
solution of this dye-stuff appears turbid because of large particles. In 
a dilute solution of fuchsin which had stood long enough I could 
see but few particles, and therefore considered the particles seen by 
other investigators to be impurities. Solutions of glycogen at the 
beginning show in the ultra-apparatus more ultramicrons than upon 
standing. (Raehlmann, Gatin-Gruzewska, and Biltz.) J. Lemanis- 
sier observed the same thing with haemoglobin solution. (L'etude des 
corps ultramicroscopiques, Paris, 1905, Jules Rousset). The solution 
of haemoglobin at first shows a large number of submicrons, but after 
forty-eight hours hardly a particle is left. According to Lemanissier 
this change is not due to a chemical alteration of the haemoglobin. 
These observations indicate that with many substances the course 
of solution is a double one, crystalloidal solution occurring simul- 
taneously with a colloidal solution, which for its part may consist 
either of submicrons or amicrons. From this it is quite evident that 
determinations of molecular weights with aqueous solutions of fuchsin, 
methyl violet, etc., made by Krafft (Ber. d. D. chem. Ges., 1889, 
Vol. XXXII, p. 1612), would lead to a higher molecular weight than 
determinations with solutions of the same substances in alcohol. 
With these substances the disappearance of ultramicrons may perhaps 
be due to a gradual transition from colloidal into crystalloidal solution. 
With other substance there seems to be a splitting up of submicrons 
into amicrons, according to the idea of Donnan (Chap. XXI). As may 
be seen, there are still numerous unsettled questions regarding the PUBLICATIONS OF OTHER SCIENTISTS 
207 
It is known that the acid dyes in general have Tout 
slight affinity for cell nuclei. An exception to this are 
those entirely resolvable optically. These dyes can be 
used as nucleus-stains, and in addition possess the 
property of staining evenly. According to Michaelis 
it is a rule that those dyes which in aqueous solution 
have the tendency of separating out in granular condi- 
tion, are the very ones which are adsorbed most readily 
and in least specific manner by all kinds of organic 
substrates. 
Michaelis further points out that in albumen solutions 
the granules do not form the whole mass of the albumen, 
but only a part of it; further, that a different number 
of particles are obtained according as water or physio- 
logical salt solution is used for dilution. The insolubility 
of globulin in pure water is very well known, as well as 
the ready partial coagulability of albumen solutions 
upon mechanical agitation. 
That the coagulated particles, before combining into 
larger ones, first appear as submicrons, is the more 
evident, because albumen, globulin, etc., are excellent 
protective colloids,1 which tend to prevent the formation 
of visible precipitates. These circumstances suggest care 
in judging facts ultramicroscopically observed, and, 
together with the observation, made five years ago, that 
outside of a few suspended dust particles and a light- 
cone due to amicrons, crystallized albumen shows noth- 
ing, they have led me to exclude albumen solutions 
from my sphere of work; for until a scientific basis for 
the method of examination was determined, complications 
aqueous solutions of many crystalloids, which can be decided by ultra- 
microscopy assisted by the application of physico-chemical methods. 
(Note added while in press). 
1 See Schulz and Zsigmondy, note 3, p. 82. 208 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
might easily arise. But this should by no means halt 
further research. The investigations of Raehlmann, 
Much, Romer, Siebert, and Michaelis have made evident 
the fact that submicrons are to be met with more fre- 
quently and in larger numbers than had formerly been 
thought, and they may play an important role in life 
processes which are involved in the mutual reaction 
between colloidal solutions and hydrogels. 
Investigations of W. Biltz. A research carried out 
by W. Biltz in collaboration with W. Gahl, and reported 
by Biltz,1 illustrates how an insight may be obtained 
into solutions with amicroscopic particles by means of 
careful ultramicroscopic examination. First are given 
some precautions regarding the preparation and handling 
of distilled water to be used for solution and dilution. 
With both investigators ordinary distilled water yielded 
unsatisfactory results; for example, 1.6 particles in 
0.00004 mm.3 By combining Spring's precipitation 
method with careful distillation, they were able, how- 
ever, to obtain water almost optically clear. By care- 
ful filtration through a Pukall filter, there was also 
obtained water suitable for most uses. 
I can certainly confirm the fact that water can be 
rendered unsuitable for ultramicroscopic uses by stirring 
it with ground-glass stoppers or filtering it through felt. 
On the other hand, I have had no unsatisfactory experi- 
ence with rubber tubes, if they are cleaned by constant 
washing with water which must now and then be allowed 
to run through them.2 For closing flasks containing 
1 Gottinger Nachrichten, math-phys., Section 1904, No. 4. Con- 
tributed by W. Biltz. 
2 See p. 112. I have worked much by sunlight with fluids containing 
almost altogether amicroscopic particles, and was disturbed only a 
few times by the bad quality of the water. The reason why in my PUBLICATIONS OF OTHER SCIENTISTS 
209 
solutions to be ultramicroscopically examined, Biltz 
recommends cork stoppers covered with tin foil. 
The main subject of the investigation are two reac- 
tions: first, the decomposition of thiosulphuric acid; 
second, the formation of selenium from selenium dioxid 
and sulphurous acid, both reactions involving the sep- 
aration of a solid from a solution originally homogeneous. 
They next undertake to elucidate the phenomena of 
delay in the action of acids on thiosulphate investigated 
by Landolt, Foussereau, Hollemann, von Oettingen, and 
others. As we know, when thiosulphate solutions are 
acidified, it is some time before the fluid becomes turbid. 
Foussereau has shown that the decomposition of the 
acid begins the moment it is formed, and the retardation 
of the visible reaction is to be attributed to a delay in 
the sulphur separating out. 
It may be assumed that the sulphur is in crystalloidal 
solution, or that it is colloidally dissolved, in which case 
the increasing formation of sulphur must gradually 
change the colloidal solution into a distinct suspension. 
According to the former view, supported by Ostwald, 
it is a case of a supersaturation phenomenon, which 
within a short time is broken up; according to the latter 
view there is at first formed a microscopic imperceptible 
CQlloidal solution, whose particles in time grow large 
enough to be seen with the naked eye. In the former 
case the process must be continuous; in the latter, 
discontinuous. 
With the help of the ultramicroscope Biltz and Gahl 
were able to prove definitely the discontinuity of the 
process. For several minutes the microscopic picture 
investigations especially pure water was needed only in exceptional 
cases, is because in examining my fluids I generally used a volume 
fifty to one hundred times smaller than Biltz and Gahl. 210 
ULTRAMICROSCOPIC INVESTIGATIONS FROM THE 
remained unchanged, then suddenly appeared an increas- 
ing turbidity, with the formation of submicrons. 
The ultramicroscopic investigation of the second 
reaction referred to, illustrates how much difference 
there is between macroscopic and ultramicroscopic ob- 
servations. While in the case of the action of sul- 
phurous acid upon selenious acid a turbidity can be 
macroscopically recognized only after about thirty min- 
utes, upon ultramicroscopic examination, submicrons can 
be seen separating out in as soon as two minutes and 
twenty seconds, and this separation considerably increases 
within the next few seconds. Here too the discontinuity 
of the process indicates that there is at first formed a 
supersaturated crystalloid solution of selenium, whose 
supersaturation is broken down with the formation of 
submicrons. 
An interesting research 1 of Mme. Z. Gatin-Gruzewska 
and W. Biltz deals with the test of the observation of 
Raehlmann above referred to, that glycogen solutions 
are filled with ultramicroscopic particles. A particularly 
pure preparation was made by Z. Gatin-Gruzewska in 
Pfliiger's laboratory, and with the observance of all 
precautions was dissolved in distilled water, which had 
been purified according to Biltz' method; the solution 
was examined in the ultraapparatus. By the use of this 
preparation, Raehlmann's results could also be confirmed, 
for the 0.07 per cent solution of glycogen showed numer- 
ous extremely small white particles having an oscillatory 
motion. The 0.007 per cent solution also showed a dis- 
tinct light-cone as well as submicrons. With a con- 
centration of 1:300,000, contrary to Raehlmann's 
observation, no more than 1-2 particles were to be seen 
1 Archiv. fur die ges. Physiologie, 1904, Vol. CV, pp. 115-120. PUBLICATIONS OF OTHER SCIENTISTS 
211 
in the whole held of view; the light-cone was extremely 
faint but still recognizable. 
If, instead of water, alcohol in varying degrees of 
concentration was used for dilution, the particles increased 
proportionately with the increasing strength of the 
alcohol (the final concentration being 0.07 per cent of 
glycogen). 
These experiments prove that in addition to easily 
recognizable submicrons the aqueous solution of glycogen 
contains amicrons which, recognizable at first as a 
homogeneous light-cone, are clumped together into 
submicrons by the precipitative action of the alcohol. 
The separation of the dissolved glycogen is not discon- 
tinuous, but increases continually with the concentration 
of the precipitant. CHAPTER XXI 
ON THE FORMATION OF HYDROSOL AND HYDROGEL 
i. Sol Formation 
It is vain to attempt to convert a practically insoluble 
substance into a hydrosol by pulverization and levigation. 
By these means there are always obtained mere sus- 
pensions of the substance in question. In order to obtain 
a hydrosol something else is necessary. 
Cohesion very soon sets a fixed limit to mechanical 
subdivision. It would be of interest to determine this 
limit for different substances; it would vary from one 
substance to another, and in the case of friable, soft 
substances, should not be much lower than a diameter 
of | y. The difficulty of converting metals into fine 
powder is very well known; special technical appliances 
are necessary to do it. But the particles in such powders 
are still very coarse and cannot be compared with the 
hypomicrons in metal hydrosols. 
There are several ways to obtain hydrosols: (a) By 
the unaided subdivision of a water-soluble colloid in 
water; (&) by peptization, starting with a true hydrogel; 
(c) in some cases by electric atomization of metals 
(Breclig's process); and finally, (d) starting with crystal- 
loids. by the formation, by means of chemical reaction 
or change of temperature, of a substance within a fluid 
in which it has practically no crystalloid solubility; 
this at first leads to the finest subdivisions. 
212 SOL FORMATION 
213 
(a) Solution of Reversible Colloids. K start can be made 
with the solid form of a reversible hydrosol (solid hydro- 
sol or false hydrogel), which possesses the property of 
dissolving in water of itself. Donnan1 has advanced 
some interesting theoretical considerations regarding 
this process of subdivision. 
Applying the basic principles of capillary theory to 
the process of colloidal solution, Donnan shows that 
under certain assumptions the mutual reactions of the 
forces of cohesion and the forces of molecular attraction 
can result in an auto-subdivision of the substance, which 
ceases before molecular dimensions are reached.2 It 
should be carefully noted that such a process is possible 
without any particular solution-tension, that it takes 
place without the help of electrolytic dissociation and 
can be explained on the basis of the theory of capil- 
larity, and that it leads to fluids which of course possess 
no appreciable osmotic pressure. 
(6) Peptization and Pectization. Starting with true 
hydrogels,3 which already actually contain pre-formed 
an extremely fine subdivision, a hydrosol can no more 
be obtained by simple grinding up with water, than if 
we had started with an insoluble powder. In order to 
obtain a colloidal solution from hydrogels it is neces- 
sary to add a very minute quantity of a peptisizing 
1 Donnan, Zeitschr. f. phys. Chem., 1901, Vol. XXXVII, p. 735; 
also 1903, Vol. XLVI, p. 197. 
2 In passing it might be remarked that the process of subdivision 
can cease when the particles are of medium size, if the bulk of the 
reversible colloid consists of indivisible submicrons, which are pre' 
vented from combining into an inseparable whole by traces of pro- 
tective colloids, as is the case with Lea's colloidal silver and with 
colloidal indigo. 
3 A true hydrogel means a jelly which of itself is not soluble in pure 
water; a false hydrogel or "solid hydrosol" (see Lottermoser) is a 
water-soluble colloid in solid or semi-solid condition. 214 
FORMATION OF HYDROSOL AND HYDROGEL 
substance which need not have any actual chemical 
action on the hydrosol or its main constituents. The 
hydrogel then becomes fluid and can be diluted as desired. 
In this respect many substances show a difference between 
the true hydrogel1 and the water-free substance; the 
latter remain intact and under the same conditions yield 
no colloidal solution.2 Stannic acid may be mentioned 
as an example of this kind. By dilution and careful 
washing out, a hydrogel, sometimes acid-free, may be 
obtained from stannous chlorid. As van Bemmelen 
has shown, the hydrogel of stannic acid is not stannic 
acid hydrate, as had formerly been assumed, but an 
adsorption compound of SnO2 with H2O. A drop of 
ammonia can fluidify a large quantity of the gel,3 but 
upon being desiccated at ordinary temperature, the 
stannic acid completely loses the power of being 
fluidified. 
The gel can again be precipitated from the hydrosol 
of stannic acid by small quantities of acid. Both pro- 
cesses, solution of the stannic acid by alkalis and pre- 
cipitation by acids, remind one so strongly of chemical 
reactions, that it seems as if this were a case of an ordinary 
formation of a salt, and that the hydrosol of stannic 
acid should be regarded as the solution of an easily 
soluble salt of the practically insoluble- stannic acid. 
A similar view has recently been adopted by Jordis.4 
Jordis regards all colloids as amphoteric substances 
which form salts by acting both as bases and acids, 
1 False hydrogels, such as albumen, dextrin, gum arabic, etc., dis- 
solve in water without the addition of foreign substances. 
2 Therefore in analytical chemistry directions are given in the deter- 
mination of silicic acid to evaporate to complete dryness. 
3 Zsigmondy, Liebig's Ann., 1898, Vol. CCCI, p. 370. 
4 Jordis, Zeitschr. f. Electrochem., Vol. X, p. 517, and . Ber. d. 
phys.-med. Soc., Erlangen, 1904. SOL FORMATION 
215 
that is, they can act like cations and anions. The same 
phenomena can be explained from another point of 
view which can be here only indicated k If the gel 
of stannic acid be treated with a less concentrated alkali, 
part of the alkali is by adsorption2 taken up by the 
hydrogel; a certain quantity of stannate, K2SnOs, may 
thus be formed. The solution of the gel with the for- 
mation of the sol takes place upon dilution with water. 
Upon dilution, the adsorbed alkali (or the stannate 
perhaps formed) undergoes electrolytic dissociation pre- 
cisely as if it were dissolved in water, the dissociation 
of course increasing with the dilution. In order to under- 
stand the process of colloidal solution as well as the 
behavior upon . electrolysis and toward reagents, we 
must assume in the case of this dissociation that the 
anion remains adsorbed by the amicrons or submicrons, 
of which the gel of stannic acid is composed. The 
adsorbed OH' or SnO3" anion gives the stannic acid 
1 I mention these observations only because I have made use of 
them to advantage for several years past, in order to review the reac- 
tions of the purple of Cassius and other colloids, and because their 
publication may perhaps be of use to others. In fact, quite similar 
assumptions have been made by Bredig (Anorg. Fermente, loc. cit., 
1901, p. 16) and by Swigel Posternak (Ann. Inst. Pasteur, 1901, Vol. 
XV, pp, 85, 169, 251, and 650), in order to explain peptisation and 
pectisation. 
1 Or absorption, see van Bemmelen (Chap. III). That adsorption 
or absorption is still somewhat "mysterious," as is the case with the 
action of gravity, chemical affinity, and other forces, cannot be doubted; 
but its presence is indicated by a multitude of phenomena which are 
neither to be explained by chemical affinity, nor to be referred to 
other kinds of energy, such as electricity, magnetism, etc. The con- 
densation of gases by charcoal, which was thoroughly investigated by 
Saussure in 1814 belongs to this class of actions. In Ostwald's Lehr- 
buch. d. allgem. Chem., 2d ed., I, p. 1084, is to be found a bibliog- 
raphy and critical discussion of this subject. See also E. du Bois- 
Reymond, Vorlesungen iiber die Physik des Stoffwechsels. 216 
FORMATION OF HYDROSOL AND HYDROGEL 
particles the negative charge which is made evident upon 
an electric migration test, and effects the separation 
of the amicrons from each other, as well as their distribu- 
tion in the fluid. The whole complex (particles of stannic 
acid with anion attached) acts upon electrolysis just 
like the complex anion of the alkali salt of a very weak 
acid; it separates out at the anode, but in contra- 
distinction to true complex ions, cannot pass through 
the parchment membrane. The reactions of colloidal 
stannic acid are likewise quite similar to those of the 
salt of a weak acid. While the behavior of colloidal 
stannic acid can perhaps be in part explained upon the 
assumption of the formation of a salt, according to 
Jordis' idea, such a view cannot hold with many other 
colloids, the purple of Cassius for instance. This acts 
just the same as stannic acid, and in order to explain 
its behavior upon the basis of salt formation, it would 
be necessary to return to the assumption of Berzelius 
that the purple is a chemical compound; but this assump- 
tion is untenable (see Chapter III, p. 67). Or recourse 
might be had to the equally erroneous assumption that 
the metallic gold particles themselves are the anions of 
a complex salt, or else constituents of such anions, just 
like platinum in platinochlorids, or iron in potassium 
ferrocyanid. This assumption contradicts all that we 
know regarding it: the metallic nature of the individual 
particles, their incapability of passing through parch- 
ment membranes upon electrolysis, and, finally, the 
fact that very large submicroscopic gold particles con- 
sisting of millions of atoms, upon electrolysis, act just 
the same as the smallest. 
In order to give a preliminary sketch as to how these 
relations may be imagined, a few illustrations are here 
given. SOL FORMATION 
217 
Designate an ultramicroscopic particle of stannic acid by 
and an ultramicroscopic particle of gold by 
In colloidal solution I believe them to be combined with 
hydroxyl or with other anions, for example: 
whereby they acquire their electric charge. In the 
purple of Cassius the gold always remains with the 
stannic acid, as long as the colloidal condition is main- 
tained. This combination (by "adhesion " [Gay-Lussac], 
"adsorption" or "affinity of constitution"1) can be 
expressed 2 by 
Those particles, too, are negatively charged just like 
those of stannic acid; their charge can be pictorially 
expressed in like manner, for example: 
1 This expression was proposed by W. Biltz. (Ber. d. Deutsch, 
chem. Ges., Vol. XXXVII, pp. 1112, 1904). 
2 Upon electrolysis the purple separates out at the anode as a red 
homogenous looking mass, no separation of gold and stannic acid 
being visible, which would be the case if both constituents were not 
united with each other (otherwise it may be assumed that the gold and 
stannic acid would separate the one from the other, according to the 
intensity of their respective charges and their frictional resistance 
in the fluid). In fact with an excess of SnO2 a separation sometimes 
takes place into red colored purple and colorless stannic acid. 218 
FORMATION OF HYDROSOL AND HYDROGEL 
Such signs may be used just the same as chemical 
formulae. In contradistinction to the very expressive 
chemical formulae, which incidentally express definite 
proportions by weight, the sign indicates that we 
are here dealing with a larger complex, with an amicro- 
scopic particle that may perhaps contain as few as 
several hundred molecules, or with a submicroscopic 
particle in which may be contained thousands, or even 
millions, of molecules.1 But this sign is by no means 
intended to give any idea of the shape of the amicron 
or of the way in which its space is filled up. 
Neither metastannic acid nor the purple obtained 
from it dissolve in hydrochloric acid. But solution cer- 
tainly takes place upon dilution. Just as in the case 
of peptization with alkalis, wo may assume that the gel 
of metastannic acid absorbs HC1; or else that some 
stannic chlorid or oxychlorid is formed and thereupon 
absorbed. Upon dilution, dissociation commences again, 
and the hydrogen ions (or perhaps the cation of a salt 
formed from stannic oxid and hydrochloric acid) which 
are then present in greater concentration, are adsorbed 
by the amicrons and give them a positive charge. 
With the aid of these signs can be expressed in graphic 
form the most important reactions of stannic acid and 
the purple of Cassius, as well as those of many other 
colloid reactions. This method is only indicated here; 
I must at the same time state, however, that my con- 
1 There is some indication that in this case numerous anions are 
united by adsorption with one submicroscopic particle, and they then 
resemble electrode rather than an ion, according to Billitzer's com- 
parison. In opposition to Billitzer, I consider it probable that each 
individual submicroscopic particle possesses a charge which equals 
or surpasses that of an ion. SOL FORMATION 
219 
ception is based on Hardy's law of isoelectric state, 
according to which the stability of irreversible hydrosols 
depends upon their electric charge,1 whereas if they 
lose their potential difference against the medium, they 
become unstable and coagulate. But Billitzer2 has 
shown that certain hydrosols are particularly stable 
when in isoelectric state, and upon this built up a new 
conception of coagulation, according to which the ions 
which produce the coagulation of colloids act as con- 
densation nuclei. Because of the consequences in- 
volved in Billitzer's explanation, I will later on take 
this up specifically, and also those points on which I 
differ with him. 
Whatever particular idea may be formed as to the 
process of peptization and coagulation, the assumption 
adopted by Bredig, and also by Billitzer and others, 
that the particles in the hydrogel are the same as those 
in the hydrosol and owe their electric charge to an 
adsorption of ions from the surrounding fluid or to a 
discharge of ions into it, is • certainly capable of more 
general application to the process in question than the 
assumption of the formation of chemical compounds, 
whose existence in some cases cannot be proved, and in 
other cases must be denied on the basis of the facts. 
In this section I must also mention that this very 
process of pectization and peptization, as well as the 
precipitation of reversible colloids by NH4C1, etc., show 
such manifold differences in phenomena with each indi- 
vidual colloid, that a complete explanation of these 
1 Other causes which increase stability are mentioned on p. 177. 
In fact, pure metal hydrosols form a group by themselves, which 
demands separate consideration. (See Chap. XXI, 2.) 
2 Billitzer, Sitzungsber. der. K. Akad. d. Wiss., Vienna; M. n. KI., 
CXI, Ila., Nov. 1902; Zeitschr. f. physik. Chem., 1903, Vol. XLV, p. 
107; ibid, 1905, Vol. LI, pp. 129-166. 220 
FORMATION OF HYDROSOL AND HYDROGEL 
processes may be safely left to the future. I need but 
recall the behavior of colloidal silicic acid, which Jordis 1 
is studying carefully; also that of globulin and albumen. 
(c) The conditions under which metal hydrosols are 
formed by electric atomization have been described in 
detail by Bredig.2 
(d) Production of colloidal solutions by the formation 
within a fluid, by chemical means or decrease in tempera- 
ture, of a substance possessing practically no crystalloid 
solubility. We may, in general, assume that most sub- 
stances are capable of forming crystals. It does not 
seem improbable that colloids, even if they form jellies, 
may perhaps consist of submicroscopic or amicroscopic 
little crystals. Such little crystals, with their water 
envelopes, must behave quite differently from ordinary 
microscopic crystals (sec further on). The greater the 
crystalloid solubility of a substance, the larger as a rule 
will be the crystals separating out from its supersaturated 
solution.3 Very insoluble substances like barium sulphate 
or silver chlorid yield precipitates whose crystal-like 
nature can hardly be recognized. Nevertheless their 
solubility is so great that stable colloidal solutions cannot 
be prepared from these substances. (See Chapter XIX, 
p. 193.) 
In order to obtain stable colloidal solutions by chem- 
ical interchange, it is necessary to begin with substances 
which as crystalloids are practically insoluble. Then 
i Jordis, loc. cit., and Zeitschr.anorg. Chem., Vol. XLIV, pp. 200-208. 
2 Anorg. Fermente, loc. cit. 
3 Compare Na2SO4 with CuSO4 and calcium oxalate. The rule stated 
serves only for temporary orientation, the size of the crystals depending 
upon numerous circumstances, important among which are tempera- 
ture, concentration of the supersaturated solution, degree of super- 
saturation, etc. But impurities and other influences exercise a con- 
siderable effect. SOL FORMATION 
221 
only will the little crystals finally formed remain so 
small and be able to grow so slightly that there are 
obtained subdivisions of the degree of homogeneity 
demanded by colloidal solutions; or else the presence of 
protective colloids is necessary, which powerfully retards 
the growth of the little crystals, just as it does the union 
of particles. 
It must not pass unnoticed that Nageli 1 had already 
advanced the idea that vegetable fibers are formed of 
micells, molecular complexes optically and structurally 
anisotropic, that is, of ultramicroscopic structures, whose 
chief properties are the same as those of tiny anisotropic 
little crystals. According to Nageli these micells are 
approximately uniformly oriented in fibers, thus pro- 
ducing double refraction in them. 
H. Ambronn2 has also frequently pointed out in 
several papers, that a series of phenomena with colloids 
(double refraction upon rotation of gelatin, gum solution, 
etc., or upon stretching gelatin, cherry gum, etc.; also 
dichroism of dyed fibers and doubly refracting gelatin) 
may be explained upon the assumption of ultramicro- 
scopic crystals or micells, according to Nageli's hypoth- 
esis. Thus fibers dyed with salts of gold show a wonderful 
red-greenish blue dichroism, the existence of which, 
according to H. Ambronn, is most easily explained by 
the oriented arrangement of the anisotropic little gold 
crystals. Many phenomena therefore indicate that sub- 
microscopic crystals may have something to do with 
colloids.3 
1 A brief and general statement of Nageli's views are to be found in 
Nageli and Schwendtener, Das Mikroskop, 2d ed., Leipzig, 1877. 
2 H. Ambronn, Ann. der Physik, 1888, Vol. XXXIV, p. 341; ibid, 
1889, Vol. XXXVIII, p. 160; Ber. d. D. Botan. Ges., 1889, Vol. VII, 
pp. 103-114; Ber. d. Kgl. Sachs. Ges. d. Wiss., Dec. 7, 1896. 
3Dichroism may also be due to differences in the spaces between the 222 
FORMATION OF HYDROSOL AND HYDROGEL 
As W. Biltz has also pointed out, the formation of 
supersaturated crystalloid solutions, even with practically 
insoluble substances, is more frequent than had been 
generally presupposed. It is therefore to be expected 
that upon the formation of a substance having practically 
no crystalloid solubility, there is at first formed a very 
strongly supersaturated crystalloid solution. Such solu- 
tions, however, are very unstable. In Chapter XVII and 
XVIII it has already been brought out how amicrons and 
submicrons can be formed in such supersaturated solu- 
tions.1 
Ostwald has determined the smallest quantity of solid 
matter which is just able to bring about the crystalliza- 
tion of the supersaturated solution of an easily soluble 
substance. He found that the limit was about 10-6 to 
10-9 mg.2 With substances difficulty soluble it should 
be still less, as brought out in Chapter XVII; and the 
smallest visible gold submicrons certainly have a mass 
millions of times smaller than the smallest little crystals 
extruded (thrown out) by supersaturations of easily 
soluble substances. But still smaller yet are the ami- 
crons which serve as centers of growth for them.3 (Chap- 
ter XVII.) 
particles, that is to an anisotropic lattice-like structure (Ambronn, 
Ber. d. Kgl. Sachs. Ges. d. Wiss., Vol. XLVIII, p. 622, 1896; F. Braun, 
Berl. Akad. Ber., 1904, p. 164; Drude's Ann., Vol. XVI; Kirchner 
and Zsigmondy, Drude's Ann., 1904, Vol. XV, p. 587). As Mr. Ambronn 
has stated to me that he has recently been able to prepare under 
certain very simple conditions, very thin needle-shaped microscopic 
crystals or even little plates united into druses, from gold and silver 
salts; and these showed just as strong a double refraction with con- 
commitant dichroism as is seen with fibers dyed with gold and silver 
salts. This rather strongly supports Ambronn's view, above given. 
1 See also Chap. XX, pp. 208-211. 
2 10-9 to 10~12 grams, Ostwald, Lehrb. d. allge. Chern., 2d ed., II, 
2, p. 784. 
3 The reason for this enormous difference in size is easily compre- SOL FORMATION 
223 
The conditions of colloid formation may be so chosen 
that only amicroscopic particles are formed (for instance, 
by extreme dilution of the reacting substances, or by 
the addition of protective colloids which interfere with 
growth), and a colloidal solution of great homogeneity may 
be thus obtained, or they may be so chosen that larger 
submicroscopic crystals are formed, as, for example, by 
the use of higher concentrations and by the proper choice 
of substances having greater speed of crystallization. 
Under certain circumstances such a case may terminate 
in a mechanical suspension. By such means can best 
be studied the influence of the degree of subdivision 
upon the properties of a substance; careful study along 
this line promises to bring to light much valuable infor- 
mation, towards which only an incomplete beginning 
has here been made. 
It might here be further stated that increasing the 
temperature of the reacting mixture (which as a rule 
increases the crystalloid solubility of slightly soluble 
substances, besides increasing the speed of crystallization, 
Chapter XVII) has the effect of causing the formation of 
hensible with the aid of Ostwald's classification. Ostwald distinguishes 
two classes of supersaturated solutions, labile, and metastable. In the 
former spontaneous crystallization occurs; in the latter it does not, 
but their supersaturation must be destroyed by a nucleus introduced 
from without. The limits determined by Ostwald refer to solutions in 
metastable condition. In the production of precipitates which are 
difficult to dissolve and of irreversible hydrosols, we are dealing with 
processes which belong to the labile group. 
But the limits given by Ostwald on p. 757, depend upon the degree 
of supersaturation. The greater the supersaturation in the metastable 
group, the smaller the crystals needed to resolve it. This very import- 
ant observation may be applied to the case in point. With increasing 
supersaturation of the solution of a certain substance, it seems to me 
that the limit referred to would be quickly carried over into the sphere 
of small dimensions. 224 
FORMATION OF HYDROSOL AND HYDROGEL 
larger particles in the case of irreversible hydrosols.1 
Thus upon the reduction of gold and silver solutions at 
boiling-point, larger particles are for the most part formed 
than at ordinary temperatures. It seems to me not 
improbable that the soluble metalumina and iron meta- 
oxid which Graham considered to be allotropic modifica- 
tions of colloidal alumina, are to be distinguished from 
the ordinary hydrosols of these substances chiefly by the 
size of the ultramicrons contained in them. The forma- 
tion of the former at higher temperatures and its be- 
havior, and the color and cloudy appearance of iron 
metaoxid, all point in this direction. Both modifications 
should yield interesting objects for ultramicroscopic 
examination. 
A considerably different influence of temperature may 
be observed with soluble starch, which is a reversible 
colloid. At boiling-point it appears to give a crystalloid 
solution with very high molecular'weight.2 A three per 
cent solution prepared at boiling-point shows in the ultra- 
apparatus even after cooling, a homogeneous polarized 
light-cone due to amicrons. If this solution is allowed 
to remain in an unheated room (at about 3-10° C.), it 
becomes opalescent after a few days, and in three or four 
weeks changes into a milk-white opaque fluid, which is 
quite thickly filled with brightly shining particles of 
50-100yy.3 
1 If the solubility of a substance practically insoluble be increased, 
then, according to the rule above-mentioned, conditions for the growth 
of the little crystals are more favorable, and finally larger structures 
can grow at the expense of the smaller ones. 
2 Because of the size of molecules their general properties cling to 
colloidal solutions. Lobry de Bruyn figures the linear dimension of 
the starch molecule to be 5 «y. (Rec. des. Trav. Chern, des Pays-Bas., 
1900, Vol. XIX, p. 253). 
3 After half a month the hydrosol changed to a thin jelly; ultrami- SOL FORMATION 
225 
By boiling up, clear solutions can be immediately 
obtained again. The brightly shining appearance of the 
particles in the ultraapparatus already indicated that 
they were not actual hydrogel particles, but were granules 
of starch heavily compacted.1 An explanation of the 
process might perhaps be that at boiling-point an unsatu- 
rated crystalloid solution is formed, which becomes much 
supersaturated at ordinary temperatures; 2 contrary to 
the usual behavior of crystalloid solutions, the formation 
of centers of crystallization as well as the growth of little 
crystals take place with extraordinary slowness. At 
the same time the amicrons apparently unite to a jelly 
whose presence is indicated by various circumstances. 
The development of submicrons in starch solutions at 
3-10° C. recalls vividly the development of gold particles 
in ruby glass (Chapter XVII), or the process in the pro- 
duction of colloidal sulphur or selenium (Chapter XX). 
In all these cases a supersaturated crystalloid solution 
can first be assumed, in which grow the spontaneously 
formed amicrons and submicrons. Whereas in the last 
three cases growth takes place speedily at the proper 
temperature, partially because of the smallness of the 
crystalloid molecule, its very much more complete diffu- 
sion and probably because of a great specific rapidity 
of growth, that of the starch particles takes place with 
extreme slowness. It may also be assumed that the 
starch solution itself contains constituents which oppose 
growth. With soluble starch, reversion to the original 
molecular state of subdivision takes place at about 
croscopically it could be seen in the diluted fluid that 2-5 brightly 
shining particles were united together at resolvable distances. 
1 Or else granules made up of smaller but solidly filled submicrons. 
2 Soluble starch swells up in water at ordinary temperatures, but 
dissolves only to a very small extent. 226 
FORMATION OF HYDROSOL AND HYDROGEL 
100° C., and with ruby glass at white heat. It is not 
improbable that diluted sulphur or selenium solutions 
can be temporarily changed again into crystalloid solu- 
tion by heating in a sealed tube. 
One remarkable point of difference is that with starch 
both submicrons and amicrons are considerably larger 
than those of gold or sulphur, etc., providing that starch 
molecules have a linear dimension of 5 /z/z, as Lobry de 
Bruyn has assumed. 
Several cases have been considered in which the growth 
of the individual particles was explained by assuming 
that they grow like crystals in a crystalloid solution. 
Essentially different conditions may obtain, however, 
if the enlargement of the particles is consequent upon 
the union of the tiniest particles, as is the case in coagu- 
lation. This process is responsible for many irreversible 
changes of the condition of colloids. It may, according 
to the nature of the subdivided substance, lead some- 
times to a very intimate, sometimes to a very super- 
ficial union, as will be seen later on when we discuss 
the hydrogels. 
How the process of the particles uniting can affect 
color, has been stated in Chapter XII. 
We have described but two kinds of change of con- 
dition in hydrosols: the first concerns the growth of 
small hypothetical little crystals, and results in solid 
ultramicrons, filled with the matter of the subdivided 
substance; the other concerns union of the particles of 
amicrons and submicrons themselves. Both must lead 
to hydrosols with coarse particles which are different as 
to quality. Thus the growth of gold particles does not 
necessarily lead to a color change, whereas the union of 
particles always does, if the original subdivision was 
red colored. (See Chapters XII and XVII.) A FEW REMARKS CONCERNING GEL FORMATION 
227 
This does not by any means exhaust the changes in 
condition of hydrosols; their further discussion at present 
would not, however, be profitable. 
2. A Few Remarks Concerning Gel Formation 
We will now consider the final result of the union of 
particles, as they occur in coagulated masses. From this 
at once arises the necessity to divide the irreversible 
hydrosols, which alone are considered in this section, 
into two varieties, according to the nature and behavior 
of the coagulum. 
Two Classes of Irreversible Hydrosols. To the first 
class, which may, perhaps, be appropriately called com- 
pletely irreversible hydrosols, belong colloidal metals 
which are free from foreign colloids; to the second class 
belong most colloidal sulphids and many colloidal oxids, 
which may be called imperfectly irreversible hydro- 
sols. 
In Chapter XIX the difference between both classes 
has already been pointed out. This difference does not 
as yet appear to be sufficiently appreciated, or it is incor- 
rectly understood, for irreversible hydrosols of the first 
class have improperly been considered as suspensions, 
from which they differ not only in degree of subdivision, 
but also in the nature of their irreversible change of 
condition.1 (Chapters I and II.) 
1 There are also mechanical suspensions which undergo irreversible 
change of condition upon desiccation; but these are not levigated 
powders of insoluble substances, but levigated powders of hydrogels. 
The changes of condition of these suspensions, however-and therein 
lies the essential difference-are not changes of the relation of the 
levigated hydrogel particles to each other, but changes in the hydrogel 
itself. They are of exactly the same nature as those which under the 
same conditions gel-lumps undergo under water. 228 
FORMATION OF HYDROSOL AND HYDROGEL 
Difference Between the Two Classes. The most im- 
portant points of difference between these two classes 
of irreversible hydrosols are the following: 
The irreversible hydrosols of the first class undergo 
upon coagulation a much more complete separation from 
the surrounding water; the union of the ultramicrons 
is much closer; the coagulation is usually accompanied 
by a change in color. No true jelly is formed, but rather 
a spongy powder or a metal sponge. All attempts to 
peptisize the precipitate or change it directly into a 
hydrosol, are in vain. To obtain a colloidal solution, 
a salt of the metal must first be formed, and the metal 
again reduced from this; or recourse must be had to 
Bredig's method of atomization under water with the 
electric arc. 
The irreversible hydrosols of the second class can 
easily be prepared a hundred to a thousand times more 
concentrated than the completely irreversible; if, how- 
ever, the individual particles are brought near enough to 
each other (for example, in 1-10 per cent solution), coagu- 
lation spontaneously occurs. In dilute solution they 
are precipitated by traces of electrolytes, just like those 
of the first group; but here they show a great diversity 
of behavior; specific influences govern, and the pre- 
cipitate has all the properties of a hydrogel. When 
placed upon a filter the precipitate usually appears trans- 
lucent and jelly-like, just like the hydrogel formed by 
spontaneous coagulation. The water is so firmly held 
by it that its complete removal in a desiccator is almost 
impossible. These precipitates have often been considered 
to be hydrates, but van Bemmelen has shown that this 
view is untenable. 
While the precipitate is still moist, it is easy, as con- 
trasted with irreversible hydrosols of the first group, A FEW REMARKS CONCERNING GEL FORMATION 
229 
to convert it into a hydrosol again by electrolytes of 
certain kinds. When air-dried, however, it can as a 
rule no longer be peptisized. The resemblance and the 
difference between both classes can at once be seen. 
With metals there occurs a much closer union of ultra- 
microns under water, and upon coagulation there is 
immediately reached the end-point which, with imper- 
perfectly irreversible hydrosols, is reached only upon 
drying. 
A clear idea of the differences between the two classes 
may be formed by assuming that the water is held to 
the surface of the amicrons or submicrons of the second 
class, with much greater force than is the case with 
metals.1 In hydrogels of the second class the individual 
particles are probably separated at first only by an 
aqueous envelope, but they nevertheless attract each 
other just the same as do the metal particles, although 
with proportionately less force because of the greater 
distance. This attraction, in the absence of any pre- 
ventative, results in the separation of the hydrosol from 
the bulk of the water. As soon, however, as the parti- 
cles receive an electric charge (peptization by electro- 
lytes, Chapter XXI, 1 b), they can be separated from 
each other again because they are not as yet so closely 
united as the metal particles. Upon further dehydration 
(which sometimes occurs under water) the hydrogel 
gradually solidifies; because of the decreasing distance 
between the particles upon desiccation, the action of the 
1 Quite a rough comparison may serve for an illustration: while a 
powerful magnet covers itself with a thick coating of iron filings, a 
weak magnet can hold but few. It requires some force to remove 
them from the powerful magnet, but they can be easily rubbed off 
the weaker. The powerful magnet would correspond to amicrons of 
the second kind, the weak magnet to those of the first kind, but of 
course only in their behavior with water. 230 
FORMATION OF HYDROSOL AND HYDROGEL 
forces of cohesion speedily increases. They not only 
visibly affect the solidity of the gel, but also militate 
against the separation of the individual particles by 
peptization. This does not mean that the process pro- 
ceeds uniformly throughout the whole mass of the gel; 
more frequently it goes on at different speeds in different 
places. If, for example, an attempt is made to peptisize 
a slightly dried hydrogel, there is sometimes obtained a 
turbid hydrosol with numerous submicroscopic particles. 
Van Bemmelen,1 Butschli and Quincke have published 
notable work regarding the structure of hydrogels and 
their behavior upon desiccation. 
Formation of Jellies. According to Butschli2 and 
van Bemmelen,3 gel formation is to be considered a 
segregation process, whereby there occurs a separation 
into two fluids, one of which subsequently solidifies. 
According to Quincke4 there exists at the bounding 
surfaces of what he calls the oleaginous fluid A, a surface 
tension against the solution B which is richer in water. 
Considering the segregation of hydrosols, in accord- 
ance with the idea above explained, as conditioned by 
the mutual attractions of amicrons or submicrons (sur- 
rounded by aqueous envelopes), it is at once evident 
firstly that the precipitate will act like a fluid (even if 
the individual particles are solid), and secondly, this 
1 See note, Chap. Ill, p. 73. 
2 O. Butschli, Untersuchungen uber mikroskopische Schaume und 
Protoplasma, Leipzig, 1892; Untersuchungen uber Strukturen, Leip- 
zig, 1898; Untersuchungen uber die Mikrostructur kiinstlicher und 
natiirlicher Kieselsauregallerten, Heidelberg, 1900, p. 343, uber der 
Bau quellbarer Korper, Gottingen, 1896. 
3 van Bemmelen, Z. anorg. Chem., 1898, Vol. XVIII, pp. 14-38. 
4 Quincke, Drude's Ann., 1902, Vol. VII, pp. 57, 631, 701; ebenda, 
1903, Vol. IX, pp. 1, 796, 969, ebenda, 1903, Vol. X, pp. 478, 673; 
ebenda, 1903, Vol. XI, pp. 54 and 449. A FEW REMARKS CONCERNING GEL FORMATION 
231 
fluid will exert an osmotic pressure against the surround- 
ing medium. The fluid properties may be ascribed 
to the slight friction of the aqueous envelopes against each 
other, the surface tension to the attraction which the 
amicrons exercise upon each other through these envelopes. 
Such attractive forces tend to contract the surface of 
the precipitate, just as does ordinary surface tension 
between crystalloid fluids. The fact that hydrogels 
solidify under water or in the open air, can be considered 
as consequent upon the co-operation of the attractive 
forces. As before mentioned, segregation and dehydra- 
tion do not proceed uniformly throughout the whole 
mass, but with the formation of a large number of specks 
which are much coarser than those of the original colloidal 
solution. Biitschli's microscopic work in particular has 
been of service in casting light upon this subject. 
Butschli examined a large number of hydrogels and 
almost everywhere found a fine-webbed microstructure, 
which is usually distinctly visible. By measuring, for 
example, the diameter d of the little cavities in different 
silica jellies and by calculating the thickness of their 
walls m, Butschli found: 
TABLE X 
With 
d 
m' 
Tabaschir 
1 A5ji 
1 .00/2 
1.50/t 
0.152-0.187/1 
0.27/1 
0.30/1 
van Bemmelen's silica gel 
Biitschli's silica gel 
In a carefully prepared solution of silica I could find 
ultramicroscopically, in harmony with the macroscopic 
1 The figures in column marked m are to be considered the major 
limits of the thickness of the walls. See Butschli, Unters. uber die 
Mikrostrutur kilnst. und nat. Kieselsauregallerten, p. 313. 232 
FORMATION OF HYDROSOL AND HYDROGEL 
observation of Picton and Linder, only the slightest 
suggestion of a light-cone; the silica therefore exists in 
solution in the form of amicrons, of which size nothing 
can be accurately stated; it is a substance particularly 
unsuitable for exhibiting its individual particles. But 
if we assume that its particles are smaller than 0.005 n 
or O.Oly, an assumption which very probably expresses 
actual conditions, wc ean then form an idea of variations 
in the inhomogeneity of colloidal solutions on the one hand 
and of hydrogels on the other. In any event their 
amicrons, being invisible upon intense side illumination, 
are very much smaller than the walls or cavities in the 
silica gels. 
To determine the extent of the effect produced by 
surface tension upon the formation of the gel skeleton, 
Quincke conducted a series of interesting researches. 
"Under the influence of surface tension thin-flowing, 
oil-like lamellae form spheres, bubbles, spherical foam- 
walls, and under some conditions spiral surfaces. Solid 
thin lamellae curl up together into a hollow cylinder or 
hollow sphere. Lamellae of very sticky, oily fluid lie 
between them, acting like thin solid lamellae and curling 
up into a hollow cylinder or hollow sphere, which then 
gradually subdivides into spherical bubbles like a fluid 
tube, or else forms swellings and constrictions." (Drude's 
Ann., Vol. XI, p. 1035 [8]). 
Quincke furthermore could in many cases demonstrate 
the existence of invisible layers of fluid consisting of 
bubbles and foam walls; he gives eight methods by 
which this proof can be established. (Drude's Ann., 
Vol. VII, p. 638.) 
According to Quincke, surface tension plays a part in 
the formation of membranes of precipitates, metal- 
foliage, and stellate crystals, in modern photogra- A FEW REMARKS CONCERNING GEL FORMATION 
233 
phy, and in the double refraction of jellies and 
colloids. 
This is merely mentioned in passing, just as is the 
important research of van Bemmelen1 regarding the 
physico-chemical behavior of hydrogels, for this book 
deals chiefly with irreversible colloidal solutions. 
In connection with the theory of hydrogels one fact 
appears to me to be of significance, that from hydrosols 
of the first class, which of themselves form no true jellies, 
there can be formed by the addition of small quantities 
of other colloids, a hydrogel with its characteristic prop- 
erties (transparency, ability to swell up and [in the case 
of irreversible hydrosols] to be peptisizecl). (See "purple 
of Cassius," etc., Chapter III, also Chapter XII). With 
a hydrogel, there may therefore be incorporated particles 
foreign to its nature in such a way that the foreign con- 
stituents sometimes constitute the greater part, without 
materially changing the character of the gel. 
In Chapters XVIII and XXI an attempt was macle 
to assemble uniformly several important phenomena 
concerning irreversible hydrosols, together with the 
known views published by other investigators, mainly 
upon the assumption that the ultramicrons in hydrosols 
consist of the same substance as those in the hydrogels 
obtained from them.2 This assumption is by no means 
new, for Graham, van Bemmelen, Hardy, Bredig, Billitzer, 
and others assume essentially the same thing. And it 
1 Loc. cit. 
2 The fact supported by Jordis in particular, that irreversible 
hydrosols have not as a whole the same chemical constitution as the 
hydrogels obtained from them, is not affected by these statements, 
but only explained from another point of view. I also agree with 
Jordis that chemical reactions in cases where they are to be expected 
must by all means be allowed for. 234 
FORMATION OF HYDROSOL AND HYDROGEL 
is also supported by the fact that such submicrons as 
are concerned in gel formation, as, for example, gold in 
the purple of Cassius, do not change their properties if 
they are changed into a hydrogel and then back again 
into a hydrosol. In the course of time other examples 
will doubtless be found. 
The second assumption made necessary to explain the 
peptization and pectization of irreversible colloids, namely, 
that the ultramicrons of colloids can, without being 
chemically bound, unite with ions which give them an 
electric charge, naturally follows from the application 
of the view supported by Nernst in particular-that ions 
are an independent kind of molecules-to the results of 
Hardy's investigations. This is by no means a new 
proposition, for Bredig has already referred the electric 
charge of colloid particles to the differences of the specific 
particle coefficients of the ions present; and Billitzer 
has directly maintained and established by experiments 
that the electric charge of the individual particles-is due 
to the acquisition or loss of ions. 
The third assumption necessary to explain plainly and 
satisfactorily the course of colloidal solution, the coagu- 
lation of metals, the existence of surface tension at the 
boundary between gel and surrounding fluid, and espe- 
cially numerous colloid reactions, viz., the assumption 
of attractive forces between the ultramicrons themselves 
and between them and the molecules of the surrounding 
medium, is nothing new either, for the capillary theory 
of La Place assumes molecular attractive forces between 
similar and dissimilar substances and thus is able to 
explain the great variety of capillary phenomena. Similar 
attractive forces were assumed in colloids even by Gay- 
Lussac, by Graham (to explain pectization), by Nageli, 
and by Barus and Schneider. Donnan has pointed out A FEW REMARKS CONCERNING GEL FORMATION 
235 
the applicability of the fundamental ideas of the capillary 
theory to what happens during colloidal solution, and 
paved the way to show that a great number of phenomena 
can be explained by a few special assumptions regarding 
the forces in question, without the addition of any new 
hypotheses. That these forces will vary from case to 
case, can be seen from the fact that the cohesion of 
different substances varies greatly in value (van der 
Waals, Die Kontinuitat des gasformigen und fliissigen 
Zustandes, 2d Edition, Vol. I, p. 175), and that the 
surface tension at the boundary between two media is 
dependent upon it. 
From the assumption made by Donnan, of attractive 
forces between colloid and surrounding medium, to that 
of attractive forces between colloid particles of different 
kinds (which in addition to electric charge are connected 
with colloid reactions), is only a short step. W. Biltz 
also arrived at a similar conclusion and expressed it in 
the same way by proposing the term "affinity of condi- 
tion" (Zustandsaffinitat) for the basic causes of absorp- 
tion processes. 
The above explanation is of course not intended as 
a theory of colloids, for the study of colloids will become 
a great and extensive science, in the development of 
which many must assist; it is only when the numerous 
facts, developed from basic physico-chemical experi- 
mental research, have been systematically arranged, that 
the theory of colloids will be raised from the stage of a 
generalization of special instances to the rank of an 
exact science. RECAPITULATION 
The starting point of this book is a fundamental ques- 
tion of colloidal chemistry, viz., the question whether 
or not the polarized dispersion of light recognizable in 
hydrosols by the Tyndall test, is one of the essential 
properties of hydrosols. 
By means of the method developed by Siedentopf and 
myself, of rendering visible ultramicroscopic particles, 
this question could in some cases be decided. In the 
case of the hydrosols examined, the diffuse dispersion 
of light has proved to be an essential characteristic; it 
is caused by the same material particles which give to 
those fluids their otherwise remarkable properties. 
The newly developed method for making visible and 
measuring ultramicroscopic particles, made it possible to 
obtain a closer insight into the processes and relations in 
colloidal solutions, hitherto inaccessible to direct observa- 
tion. 
Thus by a series of experiments, the size of the parti- 
cles in colloidal metal solutions was determined. It 
was shown that in coarser hydrosols the largest particles 
have a mass corresponding to a linear dimension of about 
60-80 nn (Chapters IX and XIX), but that carefully 
prepared solutions show an inhomogeneity which is 
scarcely discernible and hardly to be distinguished from 
that of many crystalloid solutions; further, that numerous 
intermediaries exist between these extreme cases (Chapter 
IX). 
It was shown that the smaller metal particles in hydro- 
236 RECAPITULATION 
237 
sols possess an extremely active oscillatory and translatory 
motion, which in many respects differs from the typical 
Brownian movement (Chapter X). 
The statement macle in conjunction with II. Sieden- 
topf, that in the case of metal subdivisions (in ruby 
glasses) there is no connection recognizable between the 
size of the particles and their color, was confirmed by the 
examination of colloidal gold solutions (Chapter XI). 
In collaboration with F. Kirchner a closer insight 
could be had into the course of events responsible for 
the color changes of metal hydrosols (Chapter XII). 
To obtain a preliminary insight into the question so 
important to bacteriologists, as to the relative size of 
the pores in the clay and other filters used, some experi- 
ments were conducted toward this encl (Chapter XIV). 
My own experiments and especially the experiments of 
other investigators (Chapter XX) have shown that the 
individual particles in many reversible colloids and dye- 
stuff solutions are also capable of direct observation. 
The possibility of thus forming an idea as to the size of 
these particles, renders it desirable to have in mind the 
size of the particles when classifying hydrosols. An 
attempt in this direction is made in Chapter II. In 
Chapter XV the sizes of the particles in some hydrosols, 
experimentally determined, are compared with those in 
ordinary suspensions. 
Since this work makes frequent reference to earlier 
publications concerning irreversible hydrosols, it seemed 
desirable to give a brief resume of the papers on this sub- 
ject in an historical section, especially as abstracts of 
them do not always give the authors' views with sufficient 
accuracy. It seemed desirable to make readily accessible 
to everyone interested therein the contents of the basic 
work of Graham on colloids. In the historical portion 238 
RECAPITULATION 
his publications on this subject are reviewed at length 
and in part quoted verbatim. 
While actively engaged with the ultramicroscopic 
examination of fluids, the author learned many facts of 
general interest, and certain precautions which should 
be observed; these are mentioned in Chapter VIL 
Numerous observations made upon ruby glass have been 
most simply explained by applying the results of Tam- 
mann's work upon spontaneous crystallization of super- 
cooled melts, to what happens in the formation of ruby 
glass. This made it probable that the formation of red 
ruby glass is due to the growth of amicroscopic crystals 
in a crystalloid solution of metallic gold. 
It seemed desirable to apply the preliminary insight 
thus obtained into actions only partially susceptible of 
direct observation, to what occurs during the formation 
of hydrosols and hydrogels, and to test upon what 
premises the assumption of ultramicroscopic crystals 
could be reconciled with the facts observed in connection 
with colloids. These and some other questions concern- 
ing the theory of colloids, are briefly discussed in Chapters 
XVIII and XXL 
The numerous methods of chemistry, of physical 
chemistry, and of physics, to which has recently been 
added ultramicroscopy, will in proper combination make 
it possible to settle certain questions the answers to which 
are of importance for the further development of the 
study of colloids. Even if the last-named method can- 
not fulfil the exaggerated hopes sometimes based upon 
it, present results have shown it is an important supple- 
ment to other methods, and permits an insight into rela- 
tions which, until now, were not accessible either to 
direct microscopic observation, or to ordinary, physico- 
chemical methods. EXPLANATION OF THE COLORED PLATES 
The gold solutions are arranged in order of the increasing size of 
their particles; at A (Plate I) is illustrated their appearance in inci- 
dent light, at C (Plate II), their appearance in transmitted light. 
B (Plate I) shows the corresponding ultramicroscopic pictures, D 
(Plate II) the macroscopic appearance of the light-cones. The num- 
bers of the fluids are marked upon the cork stoppers. 
In Plate I, under the bottles are given the linear dimensions of the 
gold particles, as determined according to the method described and 
conditions given in Drude's Annalen, Vol. X, pp. 16-30. For estimat- 
ing approximately the size of the particles in amicroscopic hydrosols, 
there was used a new method, the principle of which is given in the 
Zeitschrift fur physikalische Chemie, Vol. LVI, p. 68, and in the 
Zeitschrift fur Elektrochemie, 1906, p. 631. 
The solutions shown each contain the same quantity of gold, i.e., 
0.005%. The ultramicroscopic images as well as the macroscopic 
light-cones were observed and drawn after diluting each fluid ten 
times. (In examining the ultramicrons of fluids, 0 to 4, the isolated, 
brighter, yellow particles should be disregarded. Because of dust 
particles they form an incidental impurity of the fluid, difficult to 
avoid; it must also be stated that it is impossible to reproduce, true 
to nature, the brilliant light emitted by the gold particles in Fluid 5a' 
and especially in Fluid 6. The gradations of brilliancy are naturally 
considerably greater than appears from the illustrations. The red 
colors of the fluids 0 to 4 are more vivid than in the illustration. The 
light-cone of Fluid 1 is distinctly green; that of Fluid 0 has no recog- 
nizable color.) 
At the right is a suspension that settles (No. 6), with particles of 
100-150 yy, a turbid, faintly colored fluid whose ultramicroscopic 
image consists of few very intensely luminous submicrons. 
With decreasing size of the particles the brilliancy of the submicrons 
considerably decreases; from about 50 yy and under, the brilliancy of 
the light-cone also decreases. 
The almost-clear fluid marked No. 3, contains gold particles 20-25 yy 
in diameter, all quite visible and mostly green. As is the case with EXPLANATION OF THE COLORED PLATES 
the subsequent numbers, 0 and 1, it no longer settles, and furthermore 
exhibits the properties of a colloidal solution, especially the irrever- 
sible changes of condition characteristic of such solutions. 
With further subdivision, the brilliancy of the ultramicrons speedily 
diminishes, and we soon reach the amicroscopic field, where can be 
seen the light-cone but no individual particles. The fluid marked 0 
(with particles of about 2-3 na, not 0.2-3 as erroneously given on the 
plate), is an example of this class. It appears completely clear both 
in incident and in transmitted light, and its Faraday-Tyndall light- 
cone (see D, Plate II) is fainter than those of the other fluids. 
Sometimes there can be obtained still finer subdivisions, that show 
in the ultraapparatus only individual dust particles, between which 
the fluid appears optically clear. Fluids of this kind contain particles 
still smaller than those previously described; they resemble the latter 
completely in superficial appearance and behavior, and their exist- 
ence is a proof that gold itself can be subdivided until it is optically 
homogeneous. Subdivisions of (told in order of increasing size of particles (R. Zsigmond if). 
A. Appearance in incident daylight. B. Ultramicroscopic images of the subdivisions. 
COLORED PLATE I Subdivisions of Gold in order of in creasing size of particles (R. Zsigmondy f 
C. Appearance in transmitted light. D. Macroscopic appearance of the corresponding 
Faraday-Tyndall light-cone. 
COLORED PLATE II INDEX 
PAGE 
Absorption 73, 215 
Absorption and adsorption. 73 
Absorption compounds 68, 69 
Adsorption 73} 215 
2Escorcein, examination of 197 
Albumen-Grahams' work on 44 
Albumen, protective action of 81, 82, 87 
Albumoses, action of, on colloidal gold 82, 83 
Alcogels 50 
Alumina, soluble. . 38 
Amicrons, definition of 110 
Aniline dyes, ultramicroscopic examination of 204 
Antitoxins and toxins, mutual reactions of 73 
Antitoxins, gold figures of 84 
Argentous oxid 70 
Argentum Crede 191 
Aurous sulphite 70 
Barium sulphate suspension 193 
Benzopurpurin 206 
Bredig's hydrosols . 188 
Brownian motion 137 
Caesium subchlorid 69 
Capillary theory as applied to colloidal solution 213 
Caramel 43 
Carey Lea's colloidal silver . ... ; 190 
Carmine 194 
Casein, protective action of 81, 87 
Chamberland filter, experiments with 153 
Chemical compounds, simulation of, by colloidal mixtures 67 
Chromium oxid, soluble 41 
Cigarette smoke, ultramicroscopic examination of 196 
239 240 
INDEX 
Classification of colloids: PA0E 
Graham's 22 
Hardy's. 19, 23 
Muller's 19 
Zsigmondy's 25 
Cohesion, influence of, on optical constants 146 
Colloid compounds 68, 69 74 
graphic representation of 217 
simulation of chemical compounds of 70, 74 
Colloid mixtures 68 
Colloidal antimony sulphid 57 
Colloidal arsenic sulphid 57-62 
Colloidal condition 44 
Colloidal copper sulphid 57 
Colloidal ferric oxid 191 
Colloidal gold 74 
Colloidal gold and alumina 151 
Colloidal gold and gelatin 150 
Colloidal gold, coagulation of 147 
Colloidal gold, coagulation of by electrolytes 78 
Colloidal gold, color changes of 144 
Colloidal gold, color of particles of 141 
Colloidal gold, color-producing particles in 161 
Colloidal gold (Bredig's) examination of 188 
Colloidal gold, finest subdivisions of 182 
Colloidal gold, limits of size of particles in 160, 173 
Colloidal gold, motion of particles of 134 
Colloidal gold, preparation of 124 
Colloidal gold, relative size of particles in 157 
Colloidal gold, ultramicroscopic examination of 129 
Colloidal mercury 191 
Colloidal metals 63 
Colloidal metastannic acid 54 
Colloidal mixtures 67, 68 
Colloidal molybdic acid 56 
Colloidal palladium 189 
Colloidal platinum 189 
Colloidal selenium 210 
Colloidal silicic acid 35-47 
Colloidal silver 59 
Colloidal silver, examination of 189, 190 
Colloidal silver, heat developed by the coagulation of 15 
Colloidal silver iodid. 192 
Colloidal solutions, characteristics of 11 INDEX 
241 
PAGE 
Colloidal solutions, definition of 23 
Colloidal solutions, distinguished from crystalloidal solutions 11 
Colloidal solutions, distinguished from suspensions 12 
Colloidal solutions, nature of 89 
Colloidal solutions, ultramicroscopic examination of 113 
Colloidal stannic acid 54 
Colloidal sulphids 57 
Colloidal titanic acid 54 
Colloidal tungstic acid 55 
Colloidal ultramarine 58 
Colloids, Jordis' views on 126, 214 
Colloids, preparation of by dialysis (Graham) 34 
Colloids, protective action of 185 
Color lakes, colloidal nature of 74 
Crystallization, prevention of, by protective colloids 186 
Devitrification 167 
Dextrin, protective action of 81, 87 
Diastase, appearance of action of, on glycogen 198 
Dichroism in fibers 221 
Distilled water, ultramicroscopic examination of 112 
Dyeing, colloidal theory of 74 
Dyes, examination of 198, 204 
Energy of colloidal solutions 90, 91 
Faraday-Tyndall method 3 
Faraday-Tyndall test, sensitiveness of 96 
Faraday's researches on colloidal gold 75 
Ferric arsenite 73 
Filtration experiments 153 
Fluorescein, examination of 196 
Fluorescent dyes, examination of 196 
Gamboge solution 194 
Gelatin, relative protective action of 80, 81 
Gelatin solution, ultramicroscopic examination of 195 
Gels, formation of 230 
Globulin, digestion of 200 
Globulin, protective action of 82 
Glue, protective action of 81, 87 
Glycogen, action of diastase on. 158 
Glycogen, disappearance of particles in 206 242 
INDEX 
PAGE 
Glycogen, examination of 210 
Gold, crystaloid solution of 162 
Gold figure 80 
Gold figure, determination of 80 
Gold ruby glass 165 
Graham's classification of colloids 22 
Graham's original nomenclature 30 
Graham's original paper on diffusion 31 
Graham's original paper on silicic acid 47 
Gum arabic, protective action of 81, 87 
Gum tragacanth, protective action of 80, 81 
Haemoglobin 206 
Hardy's classification of colloids ;. 23 
Heterogeneity of solutions 3 
Heterogeneity of crystalloid solutions, demonstration of 6 
Homogeneity of solutions 3 
Horse sera, examination of 201 
Hydrogel, definition of 24 
Hydrogels, formation of 212 
Hydrogels, remarks on 227 
Hydrogels, structure of 116 
Hydrosols, formation of 212 
Hydrosols, limits of size of particles in 160 
Hydrosols, nature of 89 
Hypomicrons, definition of 110 
lodin suspension 193 
Irish moss, protective action of 81 
Irreversible hydrosols, definition of 25 
Irreversible colloids, definition of . 25 
Irreversible hydrosols, effect of heat on the formation of 223 
Irreversible hydrosols, two classes of 227 
Isinglass, protective action of 81, 87 
Isoelectric state, Hardy's law of 219 
Jellies, formation of 230 
Maassen filter, experiments with 153 
Matter, colloidal condition of 44 
Mechanical subdivision, limitations of 212 
Metal fog 70, 163 
Metalumina, soluble 40 INDEX 
243 
PAGE 
Micelis, Nageli's views on 221 
Microscopic resolvability 101, 109 
Microscopic visibility 101 
Milk, digestion of 201 
Molten salts, solubility of metals in 1, 102 
Mutual precipitation of colloids 72 
Optical constants, influence of cohesion on 146 
Osmotic method as applied to colloids 90 
Osmotic method, uncertainty of 90 
Pectization 213 
Peptization 213 
Peptones, action of, on colloidal gold 82, 83, 85 
Polarization of light by small particles 115 
Potassium subchlorid 69 
Precipitation of colloids by each other 72 
Protective action 25, 148 
Protective action of colloids: 
relative values of 79 
variation of 80 
Protective action on suspensions 86 
Protective colloids 77, 78, 79 
Protective colloids, action of 185 
Pukall filters, experiments with 153 
Purple of Cassius 65, 76 
Red phosphorus, suspension of 86 
Reversible hydrosols, definition of 25 
Reversible colloids, definition of 25 
Reversible colloids, solution of 213 
Rice starch suspension 195 
Rubidium subchlorid 69 
Ruby glass 165 
Ruby glass, gold particles in 165 
Ruby glass, spontaneous crystallization in 169 
Ruby glass, technology of 171 
Selenium 209 
Sera, examination of 201 
Sera, gold figures of 84 
Silica, ultramicroscopic appearance of 231 
Silicic acid, soluble 35 244 
INDEX 
PAGE 
Silicic acid, Graham's research on 47 
Silver chlorid suspension J94 
Silver purple 70 
Silver suboxid 74 
Smoke, examination of 196 
Sodium subchlorid 74 
"Sol-formers" 126 
Soluble starch, effect of heat on solutions of 224 
Soluble starch, examination of 195 
Soluble starch, size of molecule of 88 
Solution, definition of 8, 21 
Solutions, heterogeneity of 3 
Solutions, homogeneity of 3 
Spontaneous crystallization » 167 
Stalagmometer, tests with 83, 84 
Stannic acid, pectization and peptization of 214 
Stannic acids 71 
Starch paste, ultramicroscopic examination of 196 
Subdivision, definition of 9 
Submicrons, definition of 110 
Subchlorids 69 
Sulphagels 51 
Sulphur, separation of from thiosulphates 209 
Supercooled solutions 167 
Supersaturated solutions, starting crystallization in 222 
Surface tension, effect of, on gel formation 232 
Suspension, definition of 9, 10, 22 
Suspensions, contrasted with colloidal solution. . . 8-17 
Suspensions and colloidal solutions, intermediate forms between. . 16 
Suspensions, flocculation of 13-15 
Suspensions, ultramicroscopic examination of 193 
Tetraiodofluorescein, examination of 197 
Tin sesquioxid 71 
Toxins and antitoxins, mutual actions of 73 
Tragacanth, protective action of 80, 81 
Ultramarine suspension 58 
Ultramarine, color of 146 
Ultramicrons, color of 117 
Ultramicrons, determination of size of 117 
Ultramicrons, limit of visibility of 122 
Ultramicrons, definition of 110 INDEX 
245 
PAGE 
Ultramicrons in gold hydrosols, relative size of 157 
Ultramicrons, motion of 134 
Ultramicroscope, description of 103 
Ultramicroscope, development of 95 
Ultramicroscope, manipulation of 106, 111 
Ultramicroscopic examination of fluids Ill 
Ultramicroscopy, development of . 95 
Ultramicroscopy, questions for 92 
Ultra-value, determination of 202 
Ultra-value of solutions 199 
Urine, ultramicroscopic examination of 199, 200 
Vapor pressure of metals 163 
Vegetable gum 42 
Water for ultramicroscopic uses 208 
Water, protective action of 175 
Water, purification of 78 
Weight, influence of, on suspended particles 20 
Wheys, electrolysis of 202 
Wheys, experiments with 201 1ft 
"y yl I 
Ml • 
' " L^TBmi' 
h^^j. h ^^MWalllg5^ 
Apparatus for Ultra-Microscopy and 
Dark Ground Illumination 
An illustrated and descriptive catalog is issued by Carl 
Zeiss, Jena, copies of which, as well as the apparatus can be 
obtained from us. 
Rausch &Jomb ©ptiral (£0. 
NEW YORK />V<Z>XX CHICAGO 
WASHINGTON SAN FRANCISC0 
LONDON FRANKFORT 
ROCHESTER, N. Y.