(458) 
when the pressure decreases. The isobar traced in fig. 2 remained 
entirely within the limits of e=0 and «=1, and was therefore 
a continuous curve without interruption. If the ae is decreased, 
the point D' will move to the left, and will have reached the ide 
of our diagram at a certain value of p, so that for that pressure 
ep’ =0. Then the pressure must have the value of the maximum 
pressure of the isothermal of the first component. This pressure 
is of course much greater than the pressure for A or for A’. This 
isobar cuts the connodal curve somewhere between A and B or 
between A' and B'. Nor has the isobar any interruption then. The 
modification of the ¢-curve consists then in the following two points 
Ist the crest has become broader, 2rd the crest has become higher, 
so that the point D' has reached the O¢-axis. The point which it 
has in coramon with this axis must then lie higher than the ney, 
where the liquid branch in the ¢-axis begins, as appears from fig. 1 
This case is represented in fig. 4. If the pressure descended still 
lower, the pressure curve would lie partly outside the ovv diagram, 
and we should get the case represented in fig. 5. Then we should 
have to add a part on the left side of the ¢-axis, in order 
to be able to consider the ¢-curve as one coherent curve. ') 
oe A 
OE oe 
Fig. 4, Fig. 5. 
If the pressure should have come below that of A and 4A’, then the 
vapour branch has become the lowest branch all over its breadth 
— and the remaining part of the erest has entirely separated from 
the vapour branch. Then there is no longer question of tracing a 
double tangent and so there are no longer coexisting phases. 
A few points are still to be discussed, viz. the signification of the 
double point of the ¢-curve, and the displacement of the points of 
!) See for fig. 4 and fig. 5 the note on page 456, 
