the 1 her niody namic Scale of Temperature. 133 



We see, then, that the series % -~ must be chosen so as to 

 satisfy two conditions : — (i.) The Joule-Thomson effect must 

 be represented by S ^ ; (ii.) the value o£ (-pj mnst De 



represented by — S .\w> t for some single isothermal. 



Ii: we imagine that 2 — * has been chosen so as to fulfil both 

 these conditions, we shall have 



pv = 1 Rt— p% 



a, 



(n+l)t H 



By writing »= — in the small quantity »S -, ^— , we 



may obtain 



Application of the Theory to Nitrogen and to Hydrogen. 



The above theory, so far as it has been developed at pre- 

 sent, allows us to use any gas as the thermometric substance, 

 provided that the Joule-Thomson effect and the deviation from 

 Boyle's law are both small. In practice, however, some gases 

 are found unsuitable owing to their chemical properties, and 

 others have not had their thermal properties sufficiently well 

 examined. The only gases for which these objections do not 

 hold are nitrogen and hvdrogen. 



Both these gases were subjected to the porous- plug expe- 

 riment by Joule and Kelvin (Kelvin's Reprinted Papers, 

 vol. i. pp. 418-425), and their isothermal compressibility has 

 been studied by M. Amagat (Comples Rendu s, vol. xcix. 

 p. 1153; Ann. de Chlmie et de Physique, 5 ser. vol. xxii. 

 ]>. 353). 



It is found that we can reproduce all the experimental 

 results sufficiently well by putting 



v a n a l a 2 



and this formula also answers very well for the case of air. 

 This last gas, though not wholly suitable for thermometric 

 work owing to its tendency to attack mercury, seems worth 

 investigating, as its thermal properties have been studied more 



