( 825 ) 
v dv V,—v v,—v 
a — (=) = —*_*—_ is then positive. So the denominator 
L,— 2, de JpT %y—2, &,—2, 
dp 
of has always the same sign, and does not pass through zero. 
123 
v nt dp Hh 
But now that ——— is positive, (2) has also become positive. 
Us TU, Av, 81 
The figs. 41¢ and 41/ do not call for further elucidation. 
Whether the experiment will succeed in showing this double retrograde 
condensation, the investigation will have to prove. The fact that it 
only undoubtedly exists for such small values of « which do not 
only lie below 2, but must moreover be smaller than the value of 
x of the plaitpoint of the equilibrium 2,1, and the circumstance that 
for such small differences in the value of « the condensed part is 
exceedingly slight, will certainly impede the observation greatly. 
Moreover we shall have to find cases in which the occurrence of 
the solid state is no impediment. We might begin to try and show 
that after the termination of the first retrograde condensation of the 
equilibrium 3,1 another condensation makes its appearance with 
further raised pressure. The very appearance of this new conden- 
sation after the first is finished, might be considered as a not unim- 
portant addition to our knowledge of the complex phenomena of not 
perfectly miscible binary mixtures. And it must be called surprising 
that if the possibility of splitting up into two phases differing so much 
in concentration as the states 1 and 3 is past, at still higher pressure 
the possibility of splitting up reappears for phases differing as little 
as the states 1 and 2. A value of n= + and eas might be 
serviceable. 
To attain these results we might also have made use of the 
properties of the p,7 sections of the surface of saturation; and it is 
even by the consistent application of the rules for the change in 
direction when such a p,7-section is intersected by the three-phase- 
sheet that I have arrived at these results. The complication only 
occurring in sections on the left of Q,, we shall confine ourselves 
to the discussion of these sections in what follows. Thus at a value 
of « somewhat smaller than that of Q, the upper branch will be 
cut by the three-phase-curve, and that at a temperature somewhat 
higher than that of the plaitpoint of this section. In such a point 
0 : 
(5e) is either small, or perhaps already negative. But whatever 
may be the value of this quantity, the upper branch will suddenly 
change its direction at the temperature of the intersection with the 
