438 
Proceedings of Royal Society of Edinburgh. [sess. 
certain angle. Point b.c. for 25'35° C. (mixture a , fig. 1), as 
determined without special accuracy, will be seen to fall entirely 
outside the real border-curve, as determined by interpolation. 
The explanation of the shape of the isothermals inside the 
border-curve is simple : the liquid which is formed near b.c. is 
butane with some ethane dissolved in it. As the quantity of 
butane is relatively small, the condensation of butane changes the 
composition of the mixture rapidly, and, accordingly, the condensa- 
tion pressure also. But the further the process is continued the 
smaller the changes in composition, and therefore also in the 
pressures, become. With carbonic acid and a little air it is just 
the reverse : the liquid condensing in the beginning is C0 2 with a 
trace of air : but the composition of the mixture above the liquid 
hardly changes at all in the beginning : it is only in the end that 
the mixture becomes appreciably richer in air, and shows a slight 
rise in pressure. 
Andrews’ mixture contained so little air that the critical pheno- 
mena, even when tested with a view to detecting an influence of 
the admixture, would probably not have shown any abnormalities. 
The mixtures of ethane and butane, a, and in a greater degree b, 
contained sufficient quantities to allow the critical phenomena 
typical of mixtures to be completely followed. 
The peculiar features of the pv diagram for mixtures near the 
critical point were described by me in an article in the April 
number of Science Progress , 1897. 
The isothermal for the critical temperature (as for pure sub- 
stances) touches the border-curve, but for mixtures not at the 
top M, but at a point to the right of M, C in the diagrams 1 and 2. 
dp 
For the critical isothermal ^ <0 in all its points. For the iso- 
thermals at low temperature, a theoretical part with the double 
waveshape, as proposed by James Thomson and Van der Waals, 
must be assumed, and the transition from these isothermals to 
those without an unstable part takes place inside the border-curve 
for an isothermal of lower temperature than tC. That temperature 
is the critical temperature of the mixture as it would be, if it 
remained homogeneous and did not separate into phases of different 
composition. 
