437 



made at the close of the experiment is obviously less than the 

 truth, for two reasons : 1st. Because it was not taken until m' 

 minutes after the gas was cut off. 2nd. Because, during the 

 m minutes that the gas was being absorbed, the copper cylin- 

 der was hotter than the surrounding air, and, therefore, con- 

 stantly losing heat. After a good deal of reflection on the 

 subject, it finally occurred to me to adopt the following method 

 of calculating the necessary corrections. 



Having introduced into the copper cylinder, furnished with 

 its thermometer, 4.66 cubic inches of water, very nearly the 

 quantity used in all the experiments, and raised the whole to 

 93.4° (the air being 53.7°), the temperatures were noted at 

 intervals of a minute, until the thermometer indicated ^^.b°. 

 To those I then applied the expression for the velocity of 

 cooling, deducible from the Newtonian law, viz. : 



T(h.l.A — h.l.r) 

 v = ~ , 



A being the excess of temperature of the cylinder over the air 

 at any instant, and t the excess after t minutes, and thus 

 obtained the velocities of cooling corresponding to the succes- 

 sive values of t separated by intervals of a minute. These 

 being reduced to a tabular form, furnish, by mere inspection, 

 the means of applying the first of the two corrections already 

 indicated, or of ascertaining the temperature which the ther- 

 mometer would show, had it been read at the instant the gas 

 was cut off, or m' minutes previous to the actual time of 

 observation. 



" But the rise of temperature actually produced is less than 

 that which we are in search of, in consequence of the cooling 

 power exercised by the joint influence of radiation and atmos- 

 pheric contact during the time m. The method I have adopted 

 of determining the effect of refrigeration, and which, as far as 

 I am aware, has not been previously used, I shall now explain. 



" The velocity of cooling at any instant while the gas is 

 2p 2 



