( 141 ) 



tempemtnres a liif»licr value must bc chosen in oi'dei- to establish 

 agreement. For a higher vabie of ƒ yields the same result as a not 



rji 



higher vabic of/' in ƒ — ^, from which a smaller ({uantity is sublraeted. 



It might appear that the dependency of /> on 7' is strongly increased 

 by the difference between the values of z for different temperatures. 

 The following i-elation however always holds good if h is indepen- 

 dent of T: 



a 

 I' dp Vj 



^~dT~ ~~~RT 



and therefore (see p. 127) 



a 



T dp h^ v^ — h^ 



pdT RT b, 



or 



Tdp _27Tth,j v,-b, 

 '^dT~ 8 Tb, b. 



In the supposition made here, this is equal to 



Tdp 27 Tk z z 



p dT 4 T l-^z l-^z 



which expression does not vary much with z, if z remains small. 



T dp 

 Yet we find the value of -, at low temperatures for most sub- 



p dl 

 Stances to be somewhat higher than is indicated by this fonuula. 

 We should in fact have found a higher value if we had assumed 

 /a;>2/>„. If therefore we had only to deal with the formula for the 

 vapour tension, then it w^ould be rational lo in\estigate the conse- 



1 1 



(picuces of the suppositions: ;^ = 2 --^ or y/ = 2 -^ . Other experi- 

 mental quantities however follow less perfectly the foi-mula chosen 

 for h, if we give n these values. Therefore I will conline myself 

 to the investigation of the consequences of the equation chosen for 

 h with 11 = 2. 



I thiidv the following theoretical oliscM-vatioii to be of some inqtor- 

 tance, even if we disregard the (juestiou wiiether we have established 

 a ])erfect, numerically accurate agreement with the expei'iinents, by 

 assuming the cpiantity l> only to be ^•ariable, and even this varia- 

 bility to be independent of 7'. The pressures in two coexisting phases 

 which lie at a great distance from the critical conditions satisfy, if 



