G62 PROCEEDINGS OF THE AMERICAN ACADEMY. 



temperatures of the two solutions, he had reduced the thermometric error 

 to a low value. 



The method of Pfaundler has since been somewhat modified and used 

 effectively by Magie, in whose hands it has attained very considerable 

 accuracy.* 



The method of Andrews, of Thomsen, of Marignac, and to a less 

 extent the electrical method of Pfaundler and Magie. involve a more 

 or less fortuitous correction for the heat exchange witli the surround- 

 ing. All of them further involve the thermometric difficulty of accu- 

 rately measuring a rapidly rising or falling temperature. Both these 

 errors seem at first sight unavoidable in specific heat determinations. 

 In 1887 Hesehus f attempted to avoid these difficulties by making the 

 final temperature that of the surrounding air. This he accomplished in 

 the following very simple fashion. After the hot body, whose specific 

 heat was to be determined, had been brought into the calorimeter, 

 he added cold water, little by little, at such a rate that the temperature 

 in the calorimeter remained constant. Knowing the initial tempera- 

 tures of the hot body, and the cold water, and the temperature in the 

 calorimeter, and knowing the weight of the hot body and of the water 

 added, he could evidently calculate the specific heat. Since there was no 

 temperature change in the calorimeter, thei'e was no heat exchange to 

 correct for, and since the temperature of the hot body, the cold water, 

 and the water in the calorimeter were stationary, the temperature meas- 

 urements could be made under the most favorable conditions. Hesehus 

 determined the specific heat of brass in this way, and Waterman, $ after 

 improving the technique of the method, made extended determinations 

 of the specific heats of metals. His results showed excellent concord- 

 ance (0.05 per cent) and established the efficacy of the method. 



The method of Hesehus and Waterman is a great improvement over 

 that of Andrews, because the final medium temperature at least is 

 measured accurately. This medium temperature in specific heat deter- 

 minations is of more importance than either the higher or the lower 

 temperature, since an error in it is magnified in the calculation. Never- 

 theless, the cooling of the warm body just before the mixing, and the 

 warming in transit of the cold water gradually added, bring in here also 

 the old difficulty of the dissipation of heat in a new form. 



* W. F. Magie, Physical Review, 9, 65 (1890) ; 13, 91 (1001) ; 14, 193 (1902) ; 

 17, 105 (1903). 



t Jour, de la Soc. Phys. Cheni. Russ., Nov., 1887 ; Jour, de Physique, 7, 489 (1888). 

 t Pliys. Review, 4, 161 (189G). 



