of Edinburgh, Session 1884-85. 
93 
vapour will not condense without the presence of a nucleus. This 
may he a solid body of finite size, a drop of water, or fine dust- 
particles. 
Both of these facts fit perfectly in to the hypothesis that the 
isothermal in question has an asymptote parallel to the axis of 
pressure ; the vapour requiring (in the absence of a nucleus) practi- 
cally infinite pressure to reduce it, without change of state or of 
temperature, to a certain finite volume ; while the liquid, also 
without change of state or temperature, may by sufficient hydro- 
static tension be made to expand almost to the same limit of 
volume. 
This limiting volume depends, of course, on the temperature of 
the isothermal; rising with it up to the critical point. 
The physical, not geometrical, discontinuity is of course to be 
attributed to the latent heat of vaporisation. The study of the 
adiabatics, as modified by this hypothesis, gives rise to some curious 
results. 
It is clear that the experimental realisation of the parts of the 
here suggested curve near to the asymptote, on either side, will be a 
matter of great difficulty for any substance. But valuable informa- 
tion may perhaps be obtained from the indications of a sensitive 
thermo-electric junction immersed in mercury at the top of a 
column which does not descend in a barometer tube of considerably 
more than 30 inches long, when the tube is suddenly placed at a 
large angle with the vertical ; or from those of a similar junction 
immersed in water, when it has , a concave surface of great curva- 
ture from which the atmospheric pressure is removed. 
Nothing of what is said above will necessarily apply when we 
have vapour and liquid in presence of one another, or when we 
consider a small portion of either in the immediate neighbourhood 
of another body. For then we are dealing w r ith a state of stress 
which cannot, like hydrostatic pressure or tension, be characterised 
(so far as we know) by a single number. The stress in these mole- 
cular films is probably one of tension in all directions parallel to 
the film, and of pressure in a direction perpendicular to it. Thus 
it is impossible to represent such a state properly on the ordinary 
indicator-diagram. This question is still further complicated by the 
possibility that the difference between the internal pressures, in a 
