438 Dr. P. Chappuis on Gas- Thermometry. 



resistance of a platinum wire traversing the whole length 

 of the tube and furnished with Callendar's compensating 

 leads. 



The observations, which are given in detail in the paper, 

 led Mr. Bedford to the following expression for the cubical 

 dilatation of Bayeux porcelain: — 



v t =v (l + 10275* + 3-24F)10- 9 . 



Treating the same observations by least squares, I found 

 the slightly different result 



v t =v (l + 9715- Qt + 4-43< 8 )10-». 



Or, if the temperatures thus found on Callendar's scale are 

 reduced to the new scale on which our experiments are 

 calculated : — 



^ = u (l-!-9781-4£ + 4-276£ 2 )10- 9 . 



The incontestable advantage of the method employed by 

 Mr. Bedford is that it allows the determination of the dila- 

 tation over the same temperature range as the thermometer 

 reservoir is subsequently to be employed ; but nevertheless 

 the fact cannot be ignored that to obtain very exact results 

 by a process of this kind must be a matter of extreme 

 difficulty. 



I would first remark that the marks made on the tube are 

 more than 8 millim. distant from the neutral plane; and under 

 these conditions the mode of support of the tube must neces- 

 sarily exercise a considerable influence on the results obtained. 

 And, further, apart altogether from the inevitable variations 

 in the distance between the supports, produced by the ex- 

 pansion of the medium in which they rest, we have the 

 possibility of sensible variations in the elastic constants of 

 porcelain, as yet undetermined at high temperatures. 



From the author's indications it would seem that during 

 the experiments the porcelain tube suffered a perceptible 

 change of form. Further, the homogeneity of tubes of this 

 kind is often somewhat doubtful, particularly when, like 

 Mr. Bedford's, they are glazed. 1 have the advantage of 

 possessing a specimen of Mr. Bedford's tube, obtained by the 

 kindness of Prof. Callendar. The hole through it is consi- 

 derably eccentric, so that opposite parts of the wall present 

 variations in thickness of from 2 to 4 millim. The glaze 

 having a coefficient of expansion probably differing slightly 

 from that of the mass, it seems quite possible that the tube 



