COMPARISON OF PLATINUM AND GAS THERMOMETERS. 93 



reservoir of " verre dur," was employed for the reduction of all the measurements 

 made with the gas thermometer with porcelain reservoir. 



XL. FIRST DETERMINATIONS WITH PORCELAIN GAB THERMOMETER. 



The mounting of the gas thermometer being completed, we proceeded to fill the 

 reservoir with nitrogen. The gas was prej>ared by the process previously described 

 and was thoroughly dried over phosphorus pentoxide. The reservoir was several 

 times pumped out and partially filled with the dry gas, it being heated meanwhile to 

 a temperature of about 250, and the final filling and adjustment of the pressure was 

 made at the same high temperature. 



We give in the following table the measurements of the initial pressure and 

 coefficient of expansion of the nitrogen, made immediately afterwards. 



The value here found for the coefficient of expansion of nitrogen is slightly higher 

 than that previously obtained with the glass reservoir thermometer (0'003 669 8 

 instead of 0-0036685). 



Before proceeding to the experiments at high temperatures, we thought it advisable 

 to heat the porcelain reservoir to the temperature of ebullition of sulphur, to see if 

 under the actual circumstances prolonged exposure to a high temperature would 

 produce a modification of the initial pressure. As is well known, certain bodies 

 retain traces of water or condensed gases up to very high temperatures, and, as the 

 reservoir had been washed with distilled water, there was some ground for appre- 



