64 Proceedings of the Pioyal Society 
relation between curvature and pressure as in vessels open to the 
air. The permanence of this equilibrium implies physical equi¬ 
librium between the liquid and the vapour in contact with it at all 
parts of its surface. But the pressure of the vapour at different 
levels differs according to hydrostatic law. Hence the pressure of 
saturated vapour in contact with a liquid differs according to the 
curvature of the bounding surface, being less when the liquid is 
concave, and greater when it is convex. And detached portions of 
the liquid in separate vessels all enclosed in one containing vessel, 
cannot remain permanently with their free surfaces in any other 
relative positions than those they would occupy if there were hydro¬ 
static communication of pressure between the portions of liquid 
in the several vessels. There must be evaporation from those 
surfaces which are too high, and condensation into the liquid at 
those surfaces which are too low—a process which goes on until 
hydrostatic equilibrium, as if with free communication of pressure 
from vessel to vessel, is attained. Thus, for example, if there are 
two large open vessels of water, one considerably above the other 
in level, and if the temperature of the surrounding matter is kept 
rigorously constant, the liquid in the higher vessel will gradually 
evaporate until it is all gone and condensed into the lower vessel. 
Or if, as illustrated by the annexed diagram, a capillary tube, with 
a small quantity of liquid occupying it from its bottom up to a 
certain level, be placed in the neighbourhood of a quantity of the 
same liquid with a wide free surface, vapour will gradually become 
condensed into the liquid in the capillary tube until the level of 
the liquid in it is the same as it would be were the lower end of 
the tube in hydrostatic communication with the large mass of 
liquid. Whether air be present above the free surface of the 
liquid in the several vessels or not, the condition of ultimate 
equilibrium is the same; but the processes of evaporation and 
condensation through which equilibrium is approached will be 
very much retarded by the presence of air. The experiments of 
Graham, and the kinetic theory of Clausius and Maxwell, scarcely 
yet afford us sufficient data for estimating the rapidity with which 
the vapour proceeding from one of the liquids will diffuse itself 
through the air and reach the surface of another liquid at a lower 
level. With air at anything approaching to ordinary atmospheric 
