1212 
is stable, but it may be negative when the equilibrium is unstable. 
Consequently 7' is a maximum when the equilibrium is stable, but 
it may be a minimum when the equilibrium is unstable. 
When we summarize the previous considerations, then we find 
the following: 
In an equilibrium of components in n phases under constant 
P the temperature is maximum or minimum when a phase-reaction 
can occur between the phases. 
When one of the phases is a liquid and when the n—1 other 
phases are solids with invariable composition, then 7’ is a maxi- 
mum when the equilibrium is stable (or metastable); Z’ can be a 
minimum when the equilibrium is unstable. 
We may apply those general considerations to special cases; with 
this we assume that the equilibrium is stable (or metastable). 
In the binary equilibrium # = L, + F,... L, represents the 
liquid, saturated with solid F,. In a Z-concentration-diagram L, 
follows, therefore, the saturation-curve under its own vapour pres- 
sure. Consequently this curve must have its maximum-temperature 
in the point, in which Z, has the same composition as F,; this is, 
therefore, in the melting-point of /,. 
In the ternary equilibrium H= L,+ F,+ F,...LZ, is a liquid, 
saturated with /#’, + F,. In the concentration-diagram 1, follows, 
therefore, the saturationcurve of /’, + /, under its own vapour-pres- 
sure. 7’ changes along this curve from point to point. It will be 
necessary that 7’ is a maximum in the point of intersection of this 
curve with the line F, F,. 
Similar considerations are true for systems with 4 and more 
components. 
In a following communication we shall refer to unstable conditions. 
Equilibria of n components in n phases under constant pressure 
and at a temperature which differs little from the maximum- or 
minimum temperature. 
As between the mn phases of an equilibrium Zr a phase-reaction 
may occur, (13) may be satisfied. The ratios between Az, Ay,... 
Av, Ay,... are then defined by (9) and (10). When we imagine 
Aw,... Ay, Ay,... to be expressed in Az, and this to be substi- 
tuted in (18) then it appears that we may write for >(2d?Z) a form 
like A Aw,’. Herein A has a definite positive or negative value, 
Then follows from (15): 
