( 689 ) 
We will call the direction of the line, for which de, de and dy 
are proportional to the cosines of the angles with the axes, the 
spinodal direction or direction of unstability. 
Applying equation (0) to the direction of unstability we may 
write it as follows: 
(a AE / En ae d?v dp 
ame wv - v 7.9 
5 Ca), de + aay), 2° + (ap), 
dy? da? 
The signification of the positive sign of the numerator, occurring 
in the first member, is as follows. If we imagine a surface p = constant 
to be constructed in the point in question, the curve situated in 
that surface and following the spinodal direction turns its convex 
side towards the plane »=0. For a binary system this signifies 
that the p-curve in a point of the liquid branch of the spinodal line 
turns its convex side towards the axis of « (see i.a. fig. 8 of our 
previous communication). From this follows as a special case for a 
plaitpoint, situated on the liquid sheet both of the coexistence sur- 
face and of the spinodal surface, that the curve for which p = constant, 
and which has moreover the ieee determined by the limiting com- 
position of the coexisting phases, is convex towards the side of 
the zy-plane. 
If for our investigation we had chosen a point on the vapour 
= 0 
sheet of the spinodal surface, = would have been positive, we should 
U 
have had to substitute the sign < for the sign > and we should 
have had to read “concave side” instead of “convex side” (see also 
fig. 8 of our previous communication). In some special cases the 
point on the spinodal surface may be chosen in such a way, that 
of 
dv 
tangent cylinder normal to the zy plane, touches that plane ; — in this 
ease the p-line, which follows the spinodal direction has a point of 
inflexion; — and in the second place in the point, in which the two 
sheets at the coexistence surface and also those of the spinodal 
f 3 OF 
do’ de dy 
equal to zero. Therefore we have also for that point: 
dv dv 
d —)\) dy? == 0. 
an 2), aa a sa) a ut eee), 
This signifies for a binary system, that the isobar passing through 
the point in which the plait splits up into separate parts has there 
a point of inflexion. 
= 0. This may be done: first in those points for which the 
surface coincide, and for which at the same time — 
