286 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



generally known that his computations involve one step which modern 

 work has shown to be erroneous. He made four sets of experiments, 

 all at atmospheric pressure, and all covering about the same range of 

 superheat. In each experiment, first slightly superheated steam, and 

 later highly superheated steam at the same pressure, was condensed in 

 a water calorimeter. The heat released per gram of steam in the first 

 process was then subtracted from that released per gram of steam in 

 the second process and the difference divided by the difference in 

 superheat to give C P . The results which he deduced from his experi- 

 ments will be found in the third column of Table IV. below. 



The error which he made was in the determination of the quantity 



TABLE IV. 



A Recomputation of Regnault's Values of C p . 



of water in his calorimeter. This he accomplished, not by weighing, 

 but by a volumetric measurement in a sheet iron tank filled each time 

 to a scratch on the glass tube that formed its neck. Regnault knew 

 that the coefficients of expansion of the water and of the tank were 

 such that the tank would hold fewer grams of water at the room 

 temperatures at which he worked than at 0°, the temperature at 

 which he had calibrated the tank. But he supposed that he also knew 

 the specific heat of water to be an increasing function of the tempera- 

 ture at room temperatures as well as above 100° where he had care- 

 fully studied it. He therefore neglected both temperature changes, 

 thinking that, they neutralized each other, and used at all room 

 temperatures the weight that would have filled the tank at 0°, and 

 the specific heat 1. 



