( $94 ) 
‘Aya d 
form a closed curve, which lies on the left of (3) = 0 and which 
‚0 
aw Ar 
must pass through the point in which (=) = 0 has its minimum 
Ren 
d’ d 
volume. In fig. 25, in which ($2) = 0 lies on the left of (2) = 
xv v 
v av 
2, 
. . ) 
the case has been drawn, in which (Ee) == 0 no longer forms a 
av 
q 
loop-line. The closed curve, the loop, has then got detached from 
the other part. This latter part, which, however, has not been 
represented in fig. 25, then forms a continuous curve in the region 
d d 
in whieh (£ < 0, always following the line “P \ — 0 at a certain 
da v 3 de vt 
variable distance. The loop then passes through the maximum volume 
d? 
and the minimum volume of (=) =("'), and probably closes on 
enw he 
the lefthand side of this curve. In fig. 44 I have drawn the course 
dv d?v 
of the two curves =() and ae = ( for the case that 
Pp q 
dx? & 
dp dp ay 
E\ =0 intersects both *) = 0 and mie while in fig. 25 
dz), dv ] » dx? 
av 
the double point of (5) =( has disappeared, and so a closed 
av q ’ 
portion of this curve has detached itself from the other part, and 
d 
lies entirely on the lefthand side of (7) 0); 
hs 
9 
dv dv | 
If we now ask where (S) == 0 and (Ge) = 0 intersect, we see 
/ q 
bevy io 
that this can only take place more or less in the neighbourhood of 
d 
()=6 on the left of the point in which this curve has mini- 
a/v 
mum volume. And this being also the place where minimum value 
of Tr occurs, we may expect the splitting up of the spinodal line 
for mixtures which have minimum value of 7%. 
A first possibility of intersection of the two curves we have below 
5 q? 
1) In fig. 25, however, this particularity has been overlooked, and ae = 0: 
a? 
has been drawn erroneously on the right of (5 4) =d. 
j Kij 
