356 ^Ir. J. Ro<e-Inne< on the Attainment of 



This last equation^ after some manipulation, may be written 



In this equation we may use the numerical data furnished 

 by M. Chappuis in his paper, and we thus obtain the follow- 

 in o- numbers : — 



Hydrogen 273'04 



Nitrogen 273-13 



The last three pairs o£ estimates haye been obtained with- 

 out employing the Joule-Thomson measurements directly. 

 We are obliged to rely on their experimental results in order 

 to arrive at the formula used, but none of their numerical 

 data enter into the final calculations. As the result the 

 difference between the estimates formed from hydrogen and 

 from nitrogen is in all cases considerably diminished. AVe 

 may therefore infer that errors in the accepted numbers for 

 the Joule-Thomson effect were really responsible for a large 

 portion of the original difference. 



But though the difference between the estimates derived 

 from hydrogen and from nitrogen has been diminished in the 

 later calculations, it has been by no means completely removed. 

 AVe note also that the difference, though varying in amount, 

 remains persistently of one sign. If it were a matter of 

 one pair of estimates only the ditference between them might 

 very well be attributed to some accidental disturbing agent 

 interfering with that particular set of experiments : but it 

 seems hardly likely that a difierence persistently in one 

 direction can be the result of mere accident. At any rate it 

 seems worth while to take into consideration an alternative 

 hypothesis, viz. : — that there is some permanent physical cause 

 at work, independently of the Joule-Thomson experiments, 

 which serves either to raise slightly the estimate of the 

 freezing-point derived from nitrogen, or to lower slightly 

 the estimate derived from hydrogen. The amount of error 

 introduced into the uncorrected estimate of the freezing- 

 point in either case is supposed to be comparable with 

 0°-08 C. 



We are not able at present to assert that any known physical 

 cause is certainly the source of the error here spoken of ; 

 but there are indications which lead us to suppose that its 

 introduction may be due to the influence of the walls of the 

 containing vessel. Thf^ ordinary mathematical investigation 

 concerning the Joule-Thomson effect leads to the conclusion 

 that the thermodynamic scale agrees very closely with the 



