( 427 ) 



only occur if is siliiatc<l liotwooii V', '"^'kI x- l'^'"»'' /9 =:r ^/.^, .r = ; 

 for tJ =1 :c \vc tiiul on the otlicr liaii<l ,r :=r ().8(>o. A poiiil of iiillcclioii 

 further than ,r = ().iS()5 can only occin' willi MCfiativc xaliics for O 

 {0 — — CC till = 0, when ,/• z= ().8(i5 till .r = 1). lint there is //<y 

 point of inllection if <^ \\, that is to say, if 



or in gram-cals. 



'?>4 2V 

 In our case therefore, where 7'„ = 505 — wlien r/ ^ 2()()() gi-ani-cals. 

 This last conclusion will however be nuulitled, when we ap[)ly 

 the necessary correction to the a])proximate forninla (3). But the 

 fact of the possible occurreiice of a point of iiijlecf/oii, may nUvudy he 

 completely exj)lained by the simple formula (3), and this by the course 

 of the function hn/ (1 — .f). 



II. We now proceed to write dow^i a more stringent relation 

 than (3). 



Assuming an eipiation of condition of the van dkk Waals's kind, 

 the value of jj^ (the molecular potential of the com[)onent //J becomes 

 as follows: 



^,= - k, T{logT-\) - RT{log{V-h)-l) + ({e,), - T {^X ) + 



Hn.)?j 



:^n, .RT 2 



-1 ^;^ ^'i — -jMi <'u + ''. <'i2 + •••) + ^^ Ï' % "i . . . ■ (4) 



For b has been written : 



b = n^ b^ 4- n^ //, -f . . ., 



whilst for a the cpiadratic relation 



a =ii^'a,, + 2«, ;/,0j, -f . . . 

 has l)een taken. 



Now, /o7 ( f" — b) can be supposed to be independent (»f ,/•, whilst 

 the expression 



::£u^.R7' 2 RT 2 



iji regard to ./■ will become not of the ordei- ,r, bul of ./■". Let us, to prove 

 this, rather start from a nu)re general exj)ression for the total poten- 

 tial ^ (in our case we have only to deal with two single components 

 III and ;/.J, namely 



)i, + «, 



-f iï 2' ( «1 loif — -' (- «, lo(f — ^— j . 



