BANCROFT. — TERNARY MIXTURES. 367 



the rough statements of the laws for the solvent. The corresponding 

 expressions for the solute have not yet been worked out. The distinc- 

 tion between solvent and solute is very clear in solid solutions of 

 metals in metals. Starting from either of two pure metals a depres- 

 sion of the freezing point is noted when the other is added, the two 

 curves thus formed meeting at the melting point of the eutectic alloy. 

 Here there can be no question that along one curve the first metal is 

 solvent, while on the other it plays the role of solute. In the case of 

 two partially miscible liquids there is also no difficulty in determining 

 which is solvent and which solute. When ether and water are shaken 

 tosfether, the upper layer contains water as dissolved substance, the 

 lower ether. With completely miscible liquids having a maximum (or 

 minimum) vapor pressure at some concentration, such as propylalco- 

 hol and water (formic acid and water), it is probable that the change 

 of solvent occurs at the concentration corresponding to the maximum 

 (or minimum) vapor pressure. With such things as ethylalcohol and 

 water, which are infinitely miscible and which show no maximum or 

 minimum vapor pressure, it is impossible at present to say at what con- 

 centration alcohol ceases to be the solvent and water assumes that 

 duty. As soon as we have worked out the relation between the con- 

 centrations in the solution and in the va[)or, I feel certain that we 

 shall find that it requires two curves to express the relation, and not 

 one. The intersection of these curves will be the point where the 

 solvent changes. I look upon my own results with ternary mixtures as 

 very significant in this respect, the change from one curve to another 

 coming at the point where the precipitate or the solvent changed. It 

 is interesting to note that at the point, for instance, where an excess of 

 one of the partially miscible liquids first has no effect, the solubility 

 curve of the dissolved substance has a " break." The possibility of 

 such a case has always been denied except by the upholders of the 

 " hydrate theory." 



The effect of temperature on the various equilibria will form the 

 subject of a special paper, and I shall reserve for it the discussion of 

 changes of temperature coefficient at the intersections of two curves, 

 one or two very striking instances of which I have come upon inci- 

 dentally in my work so far. I hope also to be able to present a 

 paper on equilibrium in two liquid layers, a subject which is of especial 

 interest because the theoretical treatment based on the experimental 

 work in this paper gives results which are not in accordance with the 

 assumptions on which Nernst bases his Distribution Law. Besides, 

 there is the application of the Mass Law to the case where one or more 



