( 431] ) 



8 / (/' , 1 \ , \ 



3 y3{d—d') "^ 3— dj 



s r d i \ 



8 \3{d~d') '' 3 — d'J 



01' l)\ cXtlditioii : 



d-i-d' 1 1 Ö~idi-d') 



Ml — 7. . + --,-; = 



3{d-d') ■ 3—d: 3 — d' (3-rf)(3 — (/') 

 If relation (i") is taken into consideration, this becomes: 

 d 3-d'\ 8 d-d' 



As we shall see iji what follows, we find back the two equations 

 (1) and (2) in about the same form for no'U-con.stant b ; only some 

 numerical values ai-e slightly different then. 



Division of (1) by (2) makes in disappear, and gives the following 



relation between the two coe.cistlny densities : 



d 'd' 

 log loq 



1, i^) 



d d' d^d' 



3—d 3-d! 



which enables us by a[)proximation to calcidate for any arbitrarily 

 assumed value of d the corresponding value of the vapour density 



a. If we viz. put ^ = .1 and ; =-4', the calculatioji from the 



3 — d. 3 — (/ 



perfectly symmetrical equation 



log A — log A' 6 



(3-) 



A— A' d^d' 



is vei'y sinq)le indeed, and so it is not necessary for the solution of 

 this problem to introduce goniometric or other auxiliary variables, 

 as Planck and others did. 



When the density of the vapour [)hase d' can be neglected by the 

 side of that of the liquid phase d, which is the case for values of 

 iti <^ 0,4, it follows immediately from this, that 



K = 0) m = -^rf(3-ö?),^) (4) 



while it follows from (3) for d' , that 



3 / ( X 32 A 



1) Reversely d = — ( 1 J- I / 1 — — ^ \ follows trom this, which passes 

 2\ ^ V 21 J 



8 



into (Z = 3 T^vn lor very small values of m, 



9 ^ 



