43(5 Dr. P. Chappuis on Gas- Thermometry. 



that of the constant-volume thermometer. From experiments 

 made at the Bureau International in 1888 *, it would appear 

 that the divergences of the former are about double the latter. 

 Reduction to the normal scale would raise the result of 

 Messrs. Callendar and Griffiths at least a tenth of a degree. 



(2) The coefficient of expansion of air at constant pressure, 

 determined by Messrs. Callendar and Griffiths and used by 

 them in the determination of their temperatures, is 



a = O003 6749. 



This coefficient is very sensibly greater than that obtained 

 for air under similar conditions of pressure by Reguault, who 

 found 



«=0-003 6700 ; 



or the result given by my own more recent experiments, 



a = 0-003 6708. 



If in the calculation of Callendar and Griffiths's experiments 

 they had taken the latter value, 



«=0-003 6708, 



their value for the boiling-point of sulphur would become 

 445 a instead of 444°-3. 



(3) A possible cause of part of the divergence may also be 

 sought in one of the elements entering into one of the cor- 

 rections involved in the determination of temperatures with 

 the gas-thermometer, namely, the dilatation of the material 

 of the reservoir, discussed at some length by Prof. Callendar 

 in a recent article in this magazine, vol. xlviii. p. 542. 



It is customary to express this dilatation by a function of 

 the form 



»* = t'o(l + «* + # a ), 



in which a and /3 are constants. 



The determination of the constant a is of no great im- 

 porance in the measurement of temperatures by the gas- 

 thermometer. 



The value of a. only affects the value obtained for the co- 

 efficient of the dilatation of the gas, It is, however, other- 

 wise with the coefficient /3, the influence of which increases 

 with increase of temperature. (See in this connexion, 

 Callendar, loc. cit. p. 544.) 



Of the methods actually employed for the determination of 

 coefficients of expansion, the most precise is undoubtedly 



* Proces-verbanx ties seances de 1888, p. 98 et se-jr. 



