931 
A+ice+L-+G and this is larger again than that of (/ce)— 
A+ BLG; this follows from the assumption that the line qq, 
intersects the line WB. [This also appears in the following way. 
We take in fig. 1 the 3 points r, s and ¢ in such a way, that 
T,= T,;= 7, and further 3 points r,, s, and ¢, (those are not 
drawn. in fig. 3), which represent the vapours belonging to 7, sand 
t. Then vs is the saturation-curve under its own vapour-pressure of 
B, 7, s, is the corresponding vapoursaturationcurve. From the change 
in pressure along this curve it follows P, > P;. When we also 
consider the other curves, then we find P, > P, > Eil. 
When we now consider the case that the vapour (gq, in (6) 
contains watervapour only, then equilibrium (6) passes into (5). Then 
in fig. 1 q, coincides with W, so that the singular equilibrium 
(M)= lee + G occurs. As A, B and L now become indifferent 
phases, (A), (B) and (LZ) become, therefore, singular curves, which 
consequently have to coincide. It appears from fig. 2 that this 
coincidence may take place only in such a way that the stable 
parts of (A) and (B) coincide and that (Z) coincides with the 
metastable parts of (A) and (5). Then we obtain fig. 3, in which 
the (M)-curve is therefore bidirectionable. 
The position of the curves in fig. 3 is in accordance with the 
rules, which we have deduced in the general considerations. As we 
are not able to transform the singular equilibrium (J/) = Lee + G 
into the invariant equilibrium (5), (J/) is, therefore, not transformable, 
so that (M) must be bidirectionable. 
When we take a reaction, in which occur the 3 indifferent phases 
A, B and L, eg. 
LZA+B+G consequenly A+b+G—L=0 
then it appears that the 3 indifferent phases have not the same 
sign. Hence it follows again that curve (M/) must be bidirectionable. 
As A and B have the same sign, the curves (A) and (5) have to 
coincide in the one direction -— and the curve (L) in the other 
direction with the (M)-curve. All this is in accordance with fig. 3, 
which we might have found reversally also from those data. 
We may deduce fig. 3 yet in another way, which we shall 
indicate briefly. We draw firstly in a P,7-diagram the curve 
(M)= Ice + G; this terminates in the triplepoint ¢ (fig. 3) of the 
pure water. The curves (A)=B+J/ee+L+G=(M)+6+L and 
(B)= A + lee + L+G=(M)+A+L gostarting from q towards 
higher 7’ and they have to coincide with the (M)-curve. 
