523 
The question might rise whether in fig. 3(V) curve (4,, may 
also coincide with the last curve of bundle (D). From the series of 
signs (10) it should follow that this cannot be the case; for uw, and 
42 cannot be equal to one another. However, we have to bear 
in mind, that not only the series of signs (10) causes the fig. 3 (V), 
but that also each other series of signs, which may be deduced 
from (10), gives the same figure 3(V). It may e.g. arise also from 
the series of signs: 
R B S C ft D A 
EN DE ed apes halen” eee Nn pl GED) 
which is the same as the series of signs (10). Now it appears from 
(11) that the first curve of bundle (A) and the last curve of bundle 
(D) can coincide. 
In general, therefore, in a P,7-diagram every two curves, of 
which the stable part of the one curve is situated next the metastable 
part of the one curve, may coincide in opposite direction. [Further 
we shall refer to an exception]. 
In fig. fa (V) may, therefore, coincide in opposite direction : curve 
(A) with (Z) or with (D), curve (B) with (£) or with (#), etc; in 
fig. 1d(V) curve (R) with (U) or with (V), curve (7), however, 
only with (S), curve (P) only with (U) ete. 
In fig. 2e(V) we find two curves and a fivecurvical bundle; when 
in this figure we let coincide the two curves in opposite direction, 
then all curves are situated on the same side of the singular curves 
or in an angle of 180°; it is evident that this is not possible in a 
P,T-diagram. k 
The rule, mentioned above, that every two curves, of which the 
stable part of the one curve is situated next the metastable part of 
the other curve, may coincide in opposite direction, is, therefore, no 
more valid, when a P,T-diagram consists of two curves and one 
bundle. The two curves of such a diagram viz. cannot coincide in 
opposite direction. 
We may show the above also in the following way: We take 
for reaction (1) the series of signs: 
B, Fis Jig sis Ent? (12) 
Sl + 
Consequently the P,7-diagram consists of the curves (/,) and (/,) 
and of a bundle with the curves (/;), (/,)...(/i42). Now we may 
show that #, and #, cannot be indifferent phases. It follows viz. 
from (12) that all coefficients in (1) are positive, except a,. When 
we replace a, by —a, then must be true: 
a,—-da,t+a,ta,+...+&42=0. . . « 13) 
