The conditions for the formation of these two compounds, by passing ammonia over silver chloride, to which reference has already been made, will be readily understood from the above tables. In the case of the triammonia mono-chloride, the dissociation pressure becomes equal to atmospheric pressure at a temperature of about 20°; above this temperature, therefore, it cannot be formed by the action of ammonia at atmospheric pressure on silver chloride. The triammonia dichloride can, however, be formed, for its dissociation pressure at this temperature amounts to only 9 cm., and becomes equal to the atmospheric pressure only at a temperature of about 68°; and this temperature, therefore, constitutes the limit above which no combination can take place between silver chloride and ammonia under atmospheric pressure. Attention may be here drawn to the fact, to which reference will also be made later, that _two_ solid phases are necessary in order that the dissociation pressure at a given temperature shall be definite; _and for the exact definition of this pressure it is necessary to know, not merely what is the substance undergoing dissociation, but also what is the solid product of dissociation formed_. For the definition of the equilibrium, the latter is as important as the former. We shall presently find proof of this in the case {85} of an analogous class of phenomena, viz. the dissociation of salt hydrates. Salts with Water of Crystallization.--In the case of the dehydration of crystalline salts containing water of crystallization, we meet with phenomena which are in all respects similar to those just studied. A salt hydrate on being heated dissociates into a lower hydrate (or anhydrous salt) and water vapour. Since we are dealing with two components--salt and water[154]--in three phases, viz. hydrate _a_, hydrate _b_ (or anhydrous salt), and vapour, the system is univariant, and to each temperature there will correspond a certain, definite vapour pressure (the dissociation pressure), which will be independent of the relative or absolute amounts of the phases, _i.e._ of the amount of hydrate which has already undergone dissociation or dehydration. [Illustration: FIG. 19.] The constancy of the dissociation pressure had been proved experimentally by several investigators[155] a number of years before the theoretical basis for its necessity had been given. In the case of salts capable of forming more than one hydrate, we should obtain a series of dissociation curves (_pt_-curves), as in the case of the different hydrates of copper sulphate. In Fig. 19 there are represented diagrammatically the vapour-pressure curves of the following univariant systems of copper sulphate and water:-- Curve OA: CuSO_{4},5H_{2}O <--> CuSO_{4},3H_{2}O + 2H_{2}O. Curve OB: CuSO_{4},3H_{2}O <--> CuSO_{4},H_{2}O + 2H_{2}O. Curve OC: CuSO_{4},H_{2}O <--> CuSO_{4} + H_{2}O.