the Primary Laws of Elastic Fluids. 281 
in which A represents the deviation, 
¢ the temperature on Centigrade scale, 
0 the observed cooling effect, 
ep a, : 
n the ratio = (which is 3°78, taking the mean of the 
experiments from the Mécanique Céleste). 
and 
§ 5. In the demonstration of this process, it is assumed as an 
axiom that air, in passing from a higher to a lower state of den- 
sity without performing work, does not gain or lose heat upon the 
whole. (This is called Mayer’s hypothesis by Prof. Thomson.) 
Thus suppose S W, fig. 2, to represent a cylinder impervious to 
heat, and that it is divided Ki 
b tition Kk. Onone ie 2 
yy a partiti : L 
side of this (S) let there be 
air, and on the other (W)a 
perfect vacuum. If we now 
imagine kk to be instanta- 
neously withdrawn, the air 
in § rushes violently into W. 
At first, great motion in the . 
direction SW is generated, and a corresponding lowering of 
temperature: then the motion is reconverted into heat. After 
the action has subsided, the resulting temperature of the expanded 
air is assumed to be the same as before, and this whether or not 
there was a vacuum originally in W. Mr. Joule has put this to 
the test of experiment (see Phil. Mag. May 1845, vol. xxvi. p.369), 
and found no sensible deviation from it. (Note A, Appendix.) 
If the resulting temperature were lower than before, then heat 
has disappeared without performing any apparent work; but 
there may be conversion of heat into work not apparent to the 
senses, into work concealed in the molecules of air, or in their 
physical habitudes with the higher agents of force; and vice 
versd, the resulting temperature may be higher ; heat appearing, 
without apparently the equivalent work being converted, the 
separation of the molecules being the antecedent of the phzno- 
menon of absorption or evolution, as the case may be. To deter- 
mine this is a fair subject for experiment; and if the deviation 
from the law of Mariotte was known by direct observation to 
the degree of accuracy required, we should, by comparing it with 
the amount assigned to the deviation by the experiments with 
plugs, have the means of testing the point in question. If the 
results differed sensibly, we then would have direct proof that air, 
on being compressed or dilated, did not alter its temperature in 
correspondence with the mechanical equivalent of the apparent 
work. Assuming, therefore, that there is no recondite evolution 
