182 
IOWA ACADEMY OP SCIENCE 
It is not my desire to discuss here the methods by Avhich the imria- 
tions in the other conditions A\mre eliminated. Suffice it to say that it 
was possible to make tests of illuminating poAver that AAmre actually 
comparative. 
For the sake of accuracy it should be stated that ''illuminating 
poAA^er’’ as used in this paper refers to the candle-poAA^er-hours-per-gram, 
the measurement's of candle-power being made in one direction only, 
that direction being perpendicular to the plane of the flams. It Avas 
found that the illuminating poAver of the samples as received did not 
seem to have any definite connection Avith any usual tests such as resi- 
due, density, and viscosity. Of course no relation could be anticipated 
as to the fire and flash tests, and none vmre found. It is impossible to 
conceive of any relation betAveen the illuminating power and density 
and viscosity, yet the results of these experiments do shoAA^ a definite 
relation, though not a simple one. 
The illuminating power of samples obtained by the distillatioil of a 
good grade of kerosene oil Avas measured, and it Avas found that these 
distillates shoAved a definite relation between illuminating power and 
density. The former gradually increases with the latter, the actual 
change in illuminating poAver being 3 per cent between the densities 
0.78 and 0.83. There Avas also found a definite relation between vis- 
cosity and density. The fact that such a relation should exist in .the 
distillates of one sample of oil and yet not occur in the various sam- 
ples of oil taken in the market, is most significant. It Avas this fact 
that led finally to a method of comparison of the various samples of 
oil Avhich shoAved a definite resuit. 
The method of procedure in this comparison Aviil be expiained Avith- 
out giving the train of reasoning invoived. A curve Avas plotted shoAv- 
ing the relation betAveen viscosity and density in the case of the distil- 
lates of the one sample of oil. This curve Avas assumed to represent 
the ideal conditions. The viscosities of the tested samples were then 
compared Avith the viscosities as shown by the curve for oils of the 
given densities. The ratio betAveen the so-called "ideal” viscosity and 
the actual viscosity was thus obtained. (Where viscosity is used, ref- 
erence is made to the coefficient of viscosity.) A curve Avas now plotted 
Avith illuminating power as ordinates and this ratio as abscissae. The 
result, astonishing as it may seem, is that Avith 28 out of 30 samples 
tried, the curve showed a definite relation. That is to say, if this curve 
can be assumed to represent the oils correctly, the illuminating power 
of an oil should be measured quite closely by measuring its viscosity 
and itJ density. This result suggests the possibility of testing an ilium- 
