the Thermal Unit. 439 



be conducted by means of Bunsen's calorimeter, and thus 

 no advance would be made towards the solution of the 

 difficulty in which we are placed by the existing uncertainty 

 regarding the heat-capacity of water at different temperatures. 

 The value of the " mean calorie " in terms of a " thermal 

 unit at t° *" has been indirectly investigated by Dieterici. 

 His determination * of the value of J has, I think, received 

 far less attention than so important a communication deserves. 

 His value is 4*2436 X 10 7 f in terms of the " mean calorie" t ; 

 and he points out that this result is in fair agreement with 

 Rowland's, if we make the following assumptions (ibid. 

 p. 422) :— 



(1) That the capacity for heat of water changes between 



0° and 5° in the same linear manner as given by 

 Rowland's observations from 5° to 30°. 



(2) That the capacity from 30° to 100° changes linearly 

 with the temperature. 



If we assume number (1) and hence deduce Rowland's 

 value of J at o, 5, then (using Die tenor's corrected value) : — 



Mean calorie Dieterici's value of J 4*233 _ - #nn9 



Thermal unit at 0°*5 " Rowland's value at 0°*5 " 4*224 " = y 



as against 1*005, the value of the same ratio according to 

 Regnault. 



If we apply the same method to determine the ratio 



Mean calorie . 4233 ., ~..~ . . 



Thermal unit at IF ™ get 4l89 =1 '° 10 ' aS agalDSt 



t Q = 1*013, the value obtained from the experimental 



numbers of Regnault and Rowland. The discrepancies are 

 thus comparatively small. 



It appears to me, however, that if we make the assumptions 

 mentioned by Dieterici (supra, 1 and 2) we are " begging 

 the whole question." As Rowland remarks §, "The specific 

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

 specific heat should instantly fall and then recover ? . . . We 

 have physical evidence that the water has not recovered its 

 natural condition until above 4° C." 



In a recent communication to the Royal Society || I gave 



* Wied. Ann. xxxiii. 

 t Tom. cit. p. 418. 



t A necessary correction, of which I give particulars later, reduces this 

 to 4-233 XlO 7 . 



§ Proc. American Acad., vol. xv. p. 131. 



|| Phil. Trans, clxxxvi. (A) 1895, pp. 316-323. 



