500 Scale-Value of the late Dr. Joule's Thermometers. 



obtained directly by experiment, as they are given to the 

 thousandth part of a division. Joule probably only gave the 

 mean between a certain number of successive observations 

 ranging over several divisions of his scale. He would in this 

 way eliminate the errors of division, and the regularity in the 

 difference between his and Rowland's thermometer shows that 

 some such process must have been adopted. There are certain 

 corrections also no doubt applied by Joule, such as that due to 

 the emergent stem, about which it would be necessary to have 

 further information, before any definite conclusions can be 

 drawn. 



The important question as to a possible difference in the 

 air-thermometers of Rowland and Cbappuis can only be set at 

 rest by a direct comparison of one of Rowland's thermometers 

 with one compared at the Bureau International des Poids et 

 Mesures. 



But as regards the main point of the present investigation, 

 this question does not arise. We are only concerned with 

 Joule's thermometer, and the comparison between it and the 

 Paris standard. 



The relation between the intervals obtained by combining 

 the two series of comparison was found to be, in terms of the 

 Tonnelot nitrogen and hydrogen scales, 



t. = t T (l- -00027) 



= ^(l+*0024) 



= * H (l + *0028). 



Joule's final value for the equivalent of heat therefore 

 reduces as follows : — 



Joule's value for a temperature 61°*69 F. (16°*5 C). 772*65 



On the scale of the French hard-glass thermometer. 772*44 

 Gn the scale of the nitrogen thermometer of the 



Bureau International des Poids et Mesures . . 774*51 

 On the scale of the hydrogen thermometer of the 



Bureau International des Poids et Mesures . . 774*81 



Rowland applies a small correction to Joule's value of the 

 heat-capacity of his calorimeter. This would raise the equi- 

 valent by *2. Taking account of this, and considering that 

 Joule's thermometer was never intended to measure tem- 

 peratures nearer than one part in a thousand, and is not 

 graduated sufficiently well to allow the decimal place to be 

 determined with any certainty, we may state it as the result 

 of this investigation that 



Joule's equivalent of heat resulting from his own investigations 



