194 Dr. W. C. M C C. Lewis on the 



van der Waals' is nearer the truth and would yield more 

 accurate values of K provided an accurate value of a were 

 obtained. 



The simplest assumption we can make as regards the 

 density changes in the surface layer is that there is a gradual 

 decrease as we pass from the liquid to the vapour phase. 

 This is the usually accepted view (of. for example, 

 G. Bakker, Annalen der Physik, xvii. p. 4:1 5, 1905). As 

 regards liquid /liquid or liquid /solid interfaces, it is not 

 known what may be the density distribution. It is not 

 impossible that in such cases the density of one of the phases 

 does pass through a maximum at the surface. Considering 

 the case of the liquid/vapour interface, since the density 

 decreases gradually it is extremely likely that K also decreases 

 uniformly as we pass from liquid to vapour. 



As regards the connexion between the true value of K (in 

 the bulk of the liquid) and the latent heat of vaporization, we 

 may proceed as follows. K denotes the internal work done 

 per unit volume of the liquid against the molecular attracting 

 forces when the molecules are supposed to be drawn apart 

 until they are outside the range of their mutual attractions. 

 In the process of vaporization, the internal energy change 

 per unit volume of liquid, which is identical with the internal 

 latent heat per unit volume, is not (necessarily) equal to the 

 internal work done, which is measured by K. The equality 

 would only hold good if K were independent of temperature,, 

 which is not the case. The true form of the Dupre rela- 

 tionship may possibly take the following shape, 



K _L; =T /BK\ 



v \ d< A 



where L z = internal latent heat of vaporization per gram, 



or 



k=k Dup , + t(^) ; 



In this it is assumed that K and L; bear a relation to one 

 another analogous to that between external work and heat 

 change in other cases. 



Since it is known that K decreases as temperature rises, 

 Krjupre can only represent at best an upper limit to the value 

 of the internal pressure. 



To obtain reliable values of K, it will be seen, requires 

 some further assumption with respect to K in relation to T\. 



