( 134 ) 



equations. The \alue of p — the vahies of T and r being the 

 same for both curves — for the moditied isothermal is smaller than 

 that for the isothermal with constant a and h, and the difference is 

 greater according as the volume is smaller. According to the figure 



discussed — — the value of — being the same for both curves — 



k 



will therefore have a smaller value for the modified isothermal tlian 

 for the unmodified one. A value of a increasing with decreasing 

 value of V would have the same effect. But I have not discussed a 

 modification of this kind, at least not elaborately, because I had con- 

 cluded already' l)efore (see "Livre Jub. dédié a Lorkntz" p. 407) that 

 the value of the coefficient of compressibility in liquid state can only 



a 

 be explained by assuming a molecular pressure of the form — . The 



supposition of complex molecules in the liquid state would involve 



RT 



a modification of the kinetic pressure to <f{vT), where y (r, T) 



V — b 



must increase with decreasing value of r. Also this supposition would lead 



to a smaller xalue of — for the same value of — . This is namely 



pk Tj, 



certaiidy true, if tlie greater complexity has disappeared in the critical 

 state, and if therefore the values of 71- and pk n,re unmodified ; pro- 

 bably it will also he the case if still some complex molecules occur 

 even in tiie critical state. But whetiier this is so or not can only l)e 

 settled by a direct closer investigation, and for this case the property 

 of the drawn figure alone is not decisive. I have, however, alreadj' 

 shown abo\'e, that we cannot regard this circumstance as the 

 probable cause of the considerable difference between the real xalue 

 of the vapour pressure and tliat calculated from the equation 

 of state with constant a and h. So we have no choice but to 

 return to my original point of view of 30 years ago and to siqjpose 

 h to be xariable, so that the value of h decreases with decreasing 

 volume. It is clear that a variability of this kind causes the kinetic 



RT 



pressure to be smaller than we shonld fiiid it with constant h, 



V — b 



and the more so according as b is smaller. Moreover it is possible 

 in this way to account for the fact, that liqnid volumes occur smal- 

 ler than the value which b has for very large volumes and which 

 I shall henceforth denote by bg. Or I may more accurately say that 

 I do not return to that point of view, for properly speaking I have 

 never left it. As the law of the variabilitv was not known, I could 



