[Reprinted from the American Chemical Journal, Vol. X, No. 5.] 
Contributions from the Laboratory of Cornell University. 
VIII.—THE SAFETY OF COMMERCIAL KEROSENE 
OILS. 
A series of experiments were made by the writers a year ago for 
the purpose of determining the maximum temperature reached by 
the oil in the reservoirs of the most powerful modern oil lamps. 
A brief outline of the results of these experiments was published 
in the Scientific American Supplement, July 16, 1887. The writers 
found that the oil in modern lamps often reaches a temperature of 
no0 to 1120 F., and strongly urged the use in these lamps of oil 
having a high flashing point, in order to avoid danger of explosion. 
The legal regulations in regard to the flashing point of illumi- 
nating oils vary greatly in different parts of the country, certain 
States having no statutes on the subject, others requiring that no 
oils shall be sold which do not reach a certain standard of safety. 
In New York State no oil is allowed to be sold which flashes below 
ioo° F. when tested in the manner prescribed by law, in the 
instrument known as the New York State Tester. Most of the 
prevailing legislation upon the subject is probably based upon the 
observations of Dr. C. F. Chandler, who sixteen years ago made 
a careful study of the question of the safety of burning oils.1 
Dr. Chandler found that the oil in the lamps generally in use at 
Spencer B. Newbury and W. P. Cutter. 
■Am. Chemist, Aug. 1872, 43. 2 
that time, while burning in rooms at ordinary temperature, rarely 
reached ioo° F., but nevertheless recommended the adoption of 
a legal standard of flashing point not lower than 120° F. In this 
connection he says,1 “ The point of ioo° F. does not seem high 
enough to secure immunity from danger, though it may be said 
that very few, if any, accidents occur with oil which does not flash 
below this temperature.” 
If the standard of ioo° was too low to afford protection with the 
old-fashioned single-wick burners, it must be far more inadequate 
at the present day when lamps are everywhere in use which raise 
the oil in the reservoirs to a point fully twenty degrees higher than 
that reached in the lamps which Dr. Chandler tested. In modern 
oil lamps with double wick or circular burner, the oil while the 
lamp is in use is often, probably generally, heated above its flash- 
ing point. It is evident, then, that the danger of explosion is con- 
stantly present while using these lamps with ordinary oils, unless, 
indeed, some other cause than temperature may contribute to the 
safety of oils which flash above ioo°. It is not proved, for example, 
that the vapor of such oils can, at any temperature, form violently 
explosive mixtures with air. So far as previous experiments have 
shown, kerosene explosions may be due to the presence in oils of 
low flashing point, of small quantities of very volatile or gaseous 
hydrocarbons; these constituents may be absent in oils of high 
flashing test, these latter may therefore not be capable of exploding. 
In short, no clear statement has yet been made of the relation of 
the flashing point of oils to their capability of giving violent explo- 
sions with air. It was in the hope of establishing such a relation, 
and thus determining the value of the flashing point indications as 
a guarantee of safety, that the experiments described in this paper 
were undertaken. 
The questions which presented themselves for study may be 
briefly stated as follows: 
1. Will all the hydrocarbons of which burning oils are composed, 
when diffused in air in proper proportions, yield distinctly explo- 
sive mixtures ? If so, at what temperature for each ? 
2. What are the most volatile constituents of ordinary kerosene, 
and how do the proportions of each of these lighter hydrocarbons 
affect the flashing point of the oil ? 
It is well known that it is difficult to produce explosive mixtures 
1Johnson’s Encyc., “Petroleum,’' 227. 3 
of gases at will; the experiment of exploding marsh gas with air 
requires very careful adjustment of proportions, or no explosion 
results. 
A series of hydrocarbons of nearly constant boiling point was 
obtained by careful purification and repeated fractional distillation of 
lighter petroleum products. The distillation was accomplished with 
the aid of a long Hempel tube. In testing the possibility of producing 
explosions with air with the several fractions, a strong glass cylinder 
30 cm. high and with a capacity of 300 cc. was employed. This 
was immersed nearly to its mouth in a water-bath, by means of 
which the experiments could be made at any desired temperature. 
The amount of each hydrocarbon which could be completely burned 
by this volume of air was calculated, and the corresponding quantity 
of each fraction weighed out in a thin glass bulb about £ cm. in 
diameter. When a test was to be made, a little mercury was placed 
in the cylinder, the little bulb of hydrocarbon introduced, and the 
cylinder closed by a well-fitting stopper. By shaking, the bulb 
was then broken and its contents diffused throughout the air in the 
cylinder. The explosiveness of the mixture was tested by cauti- 
ously removing the stopper and applying a small flame to the 
mouth of the jar. The experiments were made with varying 
qualities of each hydrocarbon, and at different temperatures. It 
was found that the most violent explosions were obtained by the 
use of nearly the amount of the hydrocarbon which could be com- 
pletely burned to water and carbon dioxide by the volume of 
air in the cylinder. It was shown at once, moreover, to the 
surprise of the experimenters, that sharp explosions could be 
obtained at ordinary temperatures with hydrocarbons of compara- 
tively high boiling point, as for example with heptane, which boils 
at about the same point as water. The following table gives 
a summary of the results obtained: 
Hydrocarbon. 
Formula 
B. P. 
Source. 
B. P. of 
Fraction. 
Temperature at 
which explodes 
with air. 
Butane . . . 
C4 IT] 0 
i° c. 
C2H5I and Sodium. 
1° C. 
Ordinary. 
Pentane (Iso) . 
c5h12 
3°° 
Petroleum Ether. 
3°° 
U 
Hexane . . . 
C (j H14 
68° 
U (6 
67-68° 
<c 
Heptane . . . 
c7h16 
98.5° 
“Abietene ” (Pinus 
Sabiniana.) 
O 
LO 
00 
. ON 
u 
Octane . . , 
CgHi e 
124° 
Illuminating Oil. 
<< (< 
125-130° 
u 
Nonane . . . 
Cg H2o 
149-5° 
148-150° 
Between 150° 
and 212° F. 
Decane . . . 
CI0H„ 
1610 (?) 
160-165° 
Between 150° 
and 212° F. 4 
Nonane and decane could not be made to explode with air at 
temperatures under 150° F., but by heating the bath to 2120 F. 
distinct explosions were obtained. As these hydrocarbons seemed 
to mark the limit of the constituents of a safe oil, experiments 
were made under conditions similar to those which prevail in a 
burning lamp. A cylindrical copper vessel of about 300 cc. 
capacity was employed, provided with a loosely fitting cover 
through which a hole about inch in diameter had been bored. 
This was charged with a few drops of the material to be tested, 
and gradually heated in a water-bath, a small flame being applied 
to the orifice from time to time, as in the ordinary method of test- 
ing oils. The point of explosion was shown by a distinct puff 
which threw off the cover of the vessel. The results were more 
definite than those obtained by the use of the glass cylinder, and 
show that the legal limit as to flashing point lies between these 
two hydrocarbons. 
Nonane exploded sharply at 790 F. 
Decane “ “ “ 104° F. 
A sample of illuminating oil which showed a flashing point of 
1 ii° F. by the New York State tester gave, when tested as above, 
a distinct explosion at that temperature, m° F. 
The above experiments lead to the following conclusions : 
1. All the paraffines up to decane, and probably also higher 
members of the series, may at suitable temperatures form explosive 
mixtures with air. 
2. Oils which flash at a point considerably above the legal 
requirement may, under proper circumstances, give violent explo- 
sions at their flashing temperature. 
3. An oil consisting of pure decane, C10H22, would be accounted 
a safe oil by the legal flashing point test of New York State, while 
one consisting of nonane would be below the standard. 
In order to determine what are the lowest hydrocarbons present 
in ordinary burning oil, a sample of commercial “water-white” 
oil of excellent quality, showing a flashing point of 108° F., 
was carefully distilled, at first in a vacuum, and the lighter por- 
tions repeatedly fractionated by means of a Hempel tube. Two 
kilograms of oil yielded 30 grams of a distillate boiling at no°- 
130° C., which on further rectification proved to be chiefly octane; 
and 100 grams, boiling at I30°-i6o° C., which appeared to consist 
principally of nonane and decane. Only traces of hydrocarbons 5 
of lower boiling point than octane were present. The addition of 
5 per cent, of heptane to a sample of this oil lowered the flashing 
point to 970. A specimen of “ mineral sperm ” oil which flashed 
above 2120 F. was treated with different proportions by weight 
of pentane, hexane, and heptane, and the reduction of the flashing 
point noted. 
Mixtures of Mineral Sperm Oil with Light Hydrocarbons. 
Mineral sperm oil 95 per cent., pentane 5 per cent., flashes 1 io° F. 
“ “ “ go “ “ “ 10 “ “ “ ord. temp. 
“ “ “ 95 “ “ hexane 5 “ “ “ no° F. 
“ “ “ go “ “ “ 10 “ “ “ ord. temp. 
“ “ “ 95 “ “ heptane 5 “ “ “ 1450 F. 
“ “ “ go “ “ “ 10 “ “ “ iio° F. 
“ “ “ 83 “ “ “ 15 “ “ <f 96° F. 
“ “ “ 80 “ “ “ 20 “ “ “ ord. temp. 
It is evident, therefore, that the low flashing point and danger 
of oils may be due not to the presence of small quantities of very 
volatile products, but rather to the large proportion which they 
contain of constituents of moderately high boiling point. 
The burning test which determines the temperature at which a 
sample of oil may be set on fire and will burn in an open vessel, 
has been generally conceded to be unreliable as an indication of 
safety. Nevertheless, the laws of New York State prohibit 
the sale of oil which burns below 300° F. An oil may show 
a very high burning test and yet be of low flashing point and 
therefore extremely unsafe. This is especially the case with oils 
which consist of a mixture of very heavy and very light hydro- 
carbons. The more nearly homogeneous an oil is in composition 
the closer the flashing and burning points will approach each 
other. For example, the fractions corresponding to nonane and 
decane used in the experiments described above take fire at a tem- 
perature but little above their flashing points. 
Nonane flashes 790 F., burns no° F. 
Decane “ 104° “ “ 136° “ 
The flashing point of an oil is also, as is well known, dependent 
not only on the proportion of light constituents present, but upon 
the character of the oil as a whole. An unsafe oil may be brought 
up to legal standard by adding to it a sufficient quantity of heavy 
lubricating oil, but such a heterogeneous mixture will be of poor 
quality for burning purposes. It is doubtless this which has led 
manufacturers to make the oft-repeated claim that a high flashing 6 
point is inconsistent with good burning qualities. If refiners 
would be content to make a little less oil, leaving out some of the 
benzine and some of the lubricating oil as well, there would be no 
difficulty in producing illuminating oils of high flashing point and 
excellent illuminating power. The well known Astral oil, which 
flashes at 1250 F., is a standing protest against the claim of supe- 
rior burning qualities for low-grade oils. This is simply the heart 
of the burning oil freed from both lighter and heavier deleterious 
ingredients. Tests made by the authors have shown that the 
above mentioned brand burns admirably in modern lamps, and by 
photometric tests gives results fully equal to any other oil in the 
market. 
We are at present engaged in a series of experiments to deter- 
mine the relative value as illuminating agents of the different 
hydrocarbons composing kerosene. The results of these experi- 
ments will form the material of another communication. 
Only a very small amount of volatile material requires to be 
removed from an oil of ordinary character in order to raise its 
flashing point materially, as the following experiment shows. 
One kilogram of an oil flashing at 108° F. was placed in a flask, 
and a gentle current of air at ordinary temperature aspirated 
through it during 24 hours. At the end of that time the oil was 
found to have lost 28 grams, or 2.8 per cent, in weight, and showed 
a flashing point of 1160 F. Treatment of ordinary oils by a some- 
what similar process would doubtless be effective in making them 
materially safer, and would probably add but a fraction of a cent 
per gallon to their cost. 
It would be too much to expect of refiners that they should 
furnish oils of much higher flash test than they are required to do 
by law. The passage of a statute raising the legal standard to 
1200 would, however, cause no serious hardship to manufacturers, 
and would effect a saving of life and property which would far 
outweigh the very slightly increased cost of burning oil. It has 
been urged that with such a law manufacturers would simply add 
heavier products to their oils and thus bring them up to the legal 
standard, to the detriment of their burning qualities. This would, 
however, be in opposition to their own interest, since oils of high 
flash and excellent quality are already in the market and are 
extensively used. 
The experiments described above show conclusively that an oil 
heated above its flashing point is dangerous, whether that point be 7 
high or low. There can, therefore, be no doubt that in using kero- 
sene of ordinary quality in the powerfully heating lamps which are 
now common, we are dependent for our safety upon the accurate 
construction of the burners, which render it difficult for the explo- 
sive vapors in the lamp to become ignited. An oil which flashes 
below 120° F. is not safe to use in these lamps, and with oils of 
ordinary grade the conditions of explosion are constantly present.