we put 
2 (f1)o = adn a 7 de +50 dnt =5 
Je ie de 
en de rn ayy) So ae de, dy, 
02100, vj dy, 00, 02404) 
and further 
d?v dv 
vn = (Goa) tnt +2 (Go) dr an + (Gr) te 
The quantities dv, dx, and dy, are determined by the properties 
of the plaitpoint as limiting values of v2—%, pai and vg. 
(ear by dex”, the 
Var 
cause why this quotient is indefinite is removed. I have already 
pointed out that (¢))) is always negative, and this sign is not 
reversed by division by de”. If the plaitpoint lies on the vapour 
sheet, then the denominator is also negative, as has been shown 
pag. 691. In that case we have therefore: 
Fa? Gr, 
If the plaitpoint lies on the liquid sheet, then we have: 
ERE 
Between the two branches of the p, 7 curve we might trace the 
p, I curve of the coincidence pressures. It represents the ordinary 
pressure curve for the saturated vapour pressure of a simple sub- 
stance. It docs not reach so far upwards as to meet the curve in 
which it is enclosed; but it ends in the critical point, if we thought 
the mixture to behave like a simple substance. Only in one case 
the three curves, which lie one above the other have an element 
in common, namely if the ternary system admits of two coexisting 
phases of equal composition. If this might still be the case at the 
critical temperature, then the transition from the lower branch of 
the p,Z curve to the higher one does not take place fluently, but 
both branches end in a point, where their tangents have the same 
direction. According to the law of the corresponding states we have 
in this case a= =f=7. For a binary system this occurred fre- 
quently in the experiments of KUENEN and Quint. 
If we divide numerator as well as denominator of —~— 
(Ta be continued.) 
