398 HENEY A. EOWLAXD 



is 1-0033. Regnault's formula gives 1-005; but going directly to his 

 experiments, we get about 1-004, the other quantity being for 110. 



The agreement is very satisfactory, though one would expect my 

 small apparatus to lose more of the heat of the boiling water than 

 Regnault's. Indeed, for high temperatures my apparatus is much 

 inferior to Regnault's, and so I have not attempted any further experi- 

 ments at high temperatures. 



My only object was to confirm by this method the results deduced 

 from the experiments on the mechanical equivalent; and this I have 

 done, for the experiments nearly all show that the specific heat of water 

 decreases to about 30, after which it increases. But the mechanical 

 equivalent experiments give by far the most accurate solution of the 

 problem; and, indeed, give it with an accuracy hitherto unattempted in 

 experiments of this nature. 



But whether water increases or decreases in specific heat from to 

 30 depends upon the determination of the reduction to the air ther- 

 mometer. According to the mercurial thermometers Nos. 6163, 6165 and 

 6166, treating them only as mercurial thermometers, the specific heat of 

 water up to 30 is nearly constant, ~bui by the air thermometer, or ~by the 

 Kew standard or Fastre, it decreases. 



Full and complete tables of comparison are published, and from them 

 any one can satisfy himself of the facts in the case. 



I am myself satisfied that I have obtained a very near approximation 

 to absolute temperatures, and accept them as the standard. And by 

 this standard the specific heat of water undoubtedly decreases from 

 to about 30. 



To show that I have not arrived at this result rashly, I may mention 

 that I fought against a conclusion so much at variance with my precon- 

 ceived notions, but was forced at last to accept it, after studying it for 

 more than a year, and making frequent comparisons of thermometers, 

 and examinations of all other sources of error. 



However remarkable this fact may be, being the first instance of the 

 decrease of the specific heat with rise of temperature, it is no more 

 remarkable than the contraction of water to 4. Indeed, in both cases 

 the water hardly seems to have recovered from freezing. The specific 

 heat of melting ice is infinite. Why is it necessary that the specific 

 heat should instantly fall, and then recover as the temperature rises? 

 Is it not more natural to suppose that it continues to fall even after the 

 ice is melted, and then to rise again as the specific heat approaches infin- 



