280 MEASUREMENT OF HIGH TEMPERATURES. [bull. 541 
course excluded from the nature of the gases chosen. Passing fromi 
high to low temperature, it appeared to me that E. Wiedemann's results) 
are available for the interpretation of the discrepancy in question. Foil 
Wiedemann found that the exponent n=0.73 for air between 0° and 100°j 
actually changed to w=0.67 for temperatures between 100° and 185°.j 
But the latter value %=0.67 is practically indentical with the exponent] 
§ which my experiments prove to hold as far as 1,330°. Tbe conclusion,! 
therefore, is natural that below 200° air does not rigorously fulfill the 
conditions of a perfect gas. 
Looking for further data to substantiate this inference, it is well to re- 
mark that the frictional effect of decreasing temperature from the high 
to the low value is twofold in character; the purely kinetic friction is 
necessarily decreased. In proportion as temperature decreases, how- 
ever, the gas manifests cohesive friction in increasing degree. In other 
words, at low temperatures the phenomenon of residual affinity becomes 
of sufficient importance to lead to the formation of molecules, the parts 
of which cohere more or less loosely for a greater or shorter period of 1 
time. The occurrence of ephemeral molecular aggregates 1 at low tem- 
peratures seriously complicates the interpretation of the observed phe- 
nomena, and equations for the kinetics of imperfect gases have been in- 
vestigated only in a few instances. It is known, however, from the ex- 
periments of many observers, among whom v. Obermayer has examined ^ 
the greatest number of correlative results, that both the exponents n of 
the above formula^ as well as the coefficients of thermal expansion show 
a very marked tendency to increase in proportion as the gas loses the 
properties of a perfect gas and tends to become vapor. These known 
facts are almost conclusive evidence in favor of the view I have ex- 
pressed, and the- large .exponent n which applies for air at low temper- 
atures may be looked upon as a criterion of imperfect gaseity. 
All these inferences are materially substantiated by the data obtained 
for hydrogen. Pulufs low temperature exponent for hydrogen is 
M=0.69; Warburg's w=0.63 ; v. Obermayer's w=0.70. Hydrogen, 
therefore, being (cwt. par.) more nearly a perfect gas than air, shows the 
same value of n at all temperatures above zero as far as I have ob- 
served ; or, in other words, shows at zero the same law which in the 
case of air begins to apply at temperatures above 200°. Probably the 
strongest evidence in favor of the law I adduce is the fact that at high 
temperatures air and hydrogen behave identically, or that the same 
law 
appertains to each. This is proved conclusively by Tables 81 to 89, 
despite all thermal and incidental errors which the tables may. contain. 
For these errors, affecting both gases alike, will not interfere with the 
identity of behavior, should it obtain. My data show that it does. 
1 Cf. Nataiisou : Wied. Ann., vol 33, 1888, p. G83. 
(934) 
