30(5 



Prof. W. Ramsay. 



[Mar. 



would be possible to determine osmotic pressures of various mixtims 

 by tbe freezing-point method, and so to construct isothermal curvt ^ 

 for such mixtures of solvents. And there can be no reasonable 

 doubt that, as the isobaric curves of liquid-gas and of solvent-solve 

 display so close an analogy, the isothermal curves would also closely 

 resemble each other. 



Granting then that this is the case, we may construct an imagii 

 isothermal curve on the model of the curve for alcohol published in 

 the ' Phil. Trans.' by Dr. Sydney Young and myself. Now, in one 

 series of papers on the liquid-gas relations, we showed that wit 

 constant volume pressure is a linear function of temperature ; am 

 we were thus able to calculate approximately the pressures and 

 volumes for any isothermal representing the continuous transition 

 from the gaseous to the liquid state (see 'Phil. Mag.,' 1887, vol. '2-i. 

 p. 435). It would be interesting to ascertain whether, if concentra- 

 tion be kept constant, osmotic pressure would also show itself to be a 

 linear function of temperature. But, this apart, it appears in the 

 highest degree probable that there should also exist, in theory, at 

 least, a continuous transition from solvent to solvent, the representa- 

 tion of which would be a continuous curve. In such a case, on 

 increasing the concentration of the solution by eliminating one 

 solvent, the other solvent should not separate visibly, but the two > 

 should remain mixed, until one solvent has been entirely removed. 



The accompanying diagram will make this clear. The sinuous curve 

 ABODE may represent either continuous change from <: 

 liquid along an isothermal on decrease of volume, or it may 



