1398 
of the invariant point; when we wish, however, to convert X into 
a bivariant equilibrium, then from (10) two phases have to disappear. 
It is evident that we may take for this two arbitrary phases; for 
this we choose 4 and u in such a way that the reactioncoefticients 
of these phases become zero in (10). But herefrom it does not yet 
result that the remaining five phases form a bivariant equilibrium ; 
for this it is viz. necessary that the coefficients of the remaining 
phases are all positive. 
In order to examine now whether the complex may be con- 
verted into the bivariant equilibrium : 
(OV) = Pre SRT 
we equate in (LO) the coefficients of the phases Q and WV to zero; 
we find: ‘ 
consequently A= —1 and «= 6. Equation (10) passes into: 
X—16P+4R—78S4+8T—10U 
so that the complex Y cannot pass into the bivariant equilibrium 
(QV). 
When we act in the same way for the other bivariant equilibria 
(PU), (SV), (FPU) and (UV) into which the metastable equilibrium 
(R) may pass, then we find that the complex A can only be con- 
verted into the equilibrium: 
(UN = POP Ree Seep 
Then we find: 
X=4P+12Q414R+13S72T 
by which also is defined the ratio of the phases which oceur in 
this equilibrium. 
In this communication we have deduced several general properties 
and we have elucidated them with the P,7-diagrams. We should 
have been able also, however, to deduce with the aid of those pro- 
perties a method to deduce the P,7-diagramtypes from the concen- 
tration-diagramtypes. As those types have been deduced already in 
different ways, I have not followed that way here. 
Leiden, Anorg. Chem. Lab. (To be continued). 
