753 



However, notwithstanding the constancy of the quotient a:\ d^ 

 is remarkable (disregarding tlie value at Th, on an average 13,84), 

 all this is nevertheless very little convincing. Really, from 



dPx', 



dt 



v-vj \dt),'^'' T{v,~v,)j Y + v^J '^' ' 



holding when a and h are assumed to be independent of 7', hence 

 (/p,, 1 r • 1 f a 1 f a 



a a a {v^ — v^) 

 And in this I --dv=: nr is only valid when a 



-J^ 



is assumed to be independent of v. And only then we have: 



when for the pressure of coexistence p■^^ we simply write again p. 



And as van der Waals, just as we in our table, used for the 



calculation of a a formula for which the' constancy of a with respect 



to V was assumed ^), it follows in my opinion by no means from 



the fact that after all the thus found values of a appear to be 



proportional with \'\l^ (or with another function of c/,) that the real 



values of a are proportional with it. For this we should have to 



fa 

 carry out the integration I - dv on the supposition of an assumed 



J f' 



dependence of a with respect to v. And at any rate — at least for 

 higher temperatures — the vapour density d^ would have to occur 

 in the result'), not only d^. From the fact, established by van der 

 Waals and by me, follows only that a can be dependent on v. 



But it is, of course, just as well possible that - — v^ being a func- 

 tion of T along the boundary line of the region of saturation — 

 the quantity a is not directly dependent on v, but that it is a pure 

 function of the temperature, and only indirectly dependent on v^. 



If however a is dependent on T, the above formula must be 



^) He puts namely the internal latent heat proportional with aidi — d^). 



2) The calculation of the critical quantities would also have to undergo a far 

 reaching modification. Much of what has been pretty well established now, would 

 then have become entirely uncertain again. 



