This, at first sight, strange fact will be readily understood when we consider that since ice and solution are together in equilibrium with the same vapour, they must have the same vapour pressure. For suppose at any given temperature equilibrium to have been established in the system ice--solution--vapour, removal of the ice will not alter this equilibrium. Suppose, now, the ice and the solution placed under a bell-jar so that they have a common vapour, but are not themselves in contact; then, if they do not have the same vapour pressure, distillation must take place and the solution will become more dilute or more concentrated. Since, at the completion of this process, the ice and solution are now in equilibrium when they are not in contact, they must also be in equilibrium when they are in contact (p. 32). But if distillation has taken place the concentration of the solution must have altered, so that the ice will now be in equilibrium with a solution of a different concentration from before. But according to the Phase Rule ice cannot at one and the same temperature be in equilibrium with two solutions of different concentration, for the system ice--solution--vapour is univariant, and at any given temperature, therefore, not only the pressure but also the _concentration of the components in the solution must be constant_. Distillation could not, therefore, take place from the ice to the solution or _vice versâ_; that is to say, the solution and the ice must have the same vapour pressure--the sublimation pressure of ice. The reason of the coincidence is the non-volatility of the salt: had {129} the salt a measurable vapour pressure itself, the sublimation curve of ice and the curve for ice--solution--vapour would no longer fall together. The curve AO represents the pressures of the system ice--salt--vapour. This curve will also be coincident with the sublimation curve of ice, on account of the non-volatility of the salt. The equilibria of the fourth univariant system ice--salt--solution are represented by AE. Since this is a condensed system, the effect of a small change of temperature will be to cause a large change of pressure, as in the case of the fusion point of a pure substance. The direction of this curve will depend on whether there is an increase or diminution of volume on solidification; but the effect in any given case can be predicted with the help of the theorem of Le Chatelier.