tlie TIu'TDwdijnamic Scale of Temperature, 357 



bvdrooen or nitrooen scale ; but tliis invostioation assumes 

 that the gas in the thermometric vessel possesses the same 

 state as in the Joule-Thomson experiments. Thus in obtaining 

 the fundamental differential equation we have to consider the 

 work done by the streaming gas on both sides of the porous 

 plug ; now such an expression as pv, the work done by the 

 issuing gas against the atmosphere, clearly refers to the 

 pressure and volume of the gas actually in use, and not to 

 the pressure and volume of the same gas contained in a sepa- 

 rate vessel, should these last two chance to be different from 

 the former. In a palladium vessel the volume of unit mass 

 of hydrogen might be — probably would be — something very 

 different from the volume of unit mass of hydrogen at the 

 same temperature and pressure in the stream of the Joule- 

 Thomson experiments. We should therefore be travelling- 

 en ti rely beyond our data if we were to assume on the strength 

 of the Joule-Thomson experiments that the readings of a 

 constant-pressure hydrogen thermometer were approximately 

 those of the thermodynamic scale when the containing vessel 

 was made of palladium. 



For thermometric purposes hydrogen is usually inclosed 

 in a glass or platinum vessel ; we may feel sure that in such 

 a vessel the effect of the walls on the density of the gas 

 would be nothing like as big as in the case of a palladium 

 vessel, but we cannot be certain at present that there is no 

 effect at all. The most trustw^orthy experiments bearing on 

 this point are some measurements made by M. Chappuis- 

 That physicist examined the increase of pressure at constant 

 volume of hydrogen in a vessel of iridio-platinum, and also 

 in a vessel of hard glass. He found that when the pressure 

 at the freezing-point is 100 cms. of mercury, the increase 

 with the former vessel was 36*6254 cms. of mercury, and with 

 the latter vessel the increase was 36*6217. {L'echelle thermo- 

 metrique norniale et les echelles pratiques, p. 3.) The diffe- 

 rence between these two numbers is small, but it is worth 

 taking into account. If we assume that the walls of the 

 containing vessel exert an influence which has the same sign 

 for both vessels, but which is greater for one vessel than for 

 the other, then the true value of the increase of the pressure 

 will be outside the interval between the observed values, 

 differing from either of them by an amount comparable with 

 the said interval. Thus if we imagine that the influence of 

 the platinum vessel upon the gas is the more marked, we 

 might readily suppose the true value of the increase of the 

 pressure to be as low as 36*6150 without straining the evidence. 

 The reciprocal of 0*00366150 is 273*11, while the reciprocal 



