( 153 ) 

 Tlie ('ompoiieiits are in equililu'iuin in bolh phases if 



SO lluil we uet (llic tei'nis with T loij T cancel each other): 



.^1 — q T-\ a .>r ^jrr lo,i ( i —x) = e\ ~ c\ T^ a' x'^ -\ RT lo(j (1 - .7;') 



r,-c, T^a {\—.ry-^RTlogx — e', — c\ T^a' {l—.v'y J^RTUxi.v'' 

 or witii 



l—x' 



RT lo<j — q,—Y, T-\- (« x'-a' .v"') 



1 — X 



x 



RTlog ~ = q-y, T+[«(l-.tf-«'(l-..7] 



iÜ 



If we pay attention to the circnnistance tliat for ,/'::=0, .r'=() (he 

 ([nantity T must be equal to 1\, and in the same way T=T., for 

 ,/'=!, ,r'=i (7\ and 1\ are the meltingtemperatures of llie j)uic 

 coni})oncnts), then we may wa-ite : 



— li — 1 



We lia\e therefore 



f <h l — x\ 



T[j^^R lor, ^-J = q^Jria x'-a' x!'^) 



or witli 



i+(g„,-,,.o ^ +,/'^ 



' i?r l-.i' ' Rl\ x' ^ ^ 



1 -| log 1 A lo() — 



7i 1— •^' ■ 72 ' •^" 



These are the two finidamental equations from which we may cal- 

 cuUite (he values of .u and T corresponding to each given value of 

 .1', and which represent a course of the meltingpoint-curve which is 

 perfectly continuous, at least theoretically. 



It is easy to see that in the case that no mixed crystals occur, x' 



is continuously equal to zero, and the equation is reduced to 



l^^x^ 



T = 7\ — 



' RT, 

 1 '-log{\—x) 



7i 

 an equation wliicii I have already deduced in a })revious paper. 

 Ihit in the present paper w^e will assume that the mixing-proportion 



