392 Cryoscopic Relations of Solutions of Cane-Sugw Sfc. 



results were too high. Thus the problem stands. In v the light 

 of this new work, I do not see a sufficient justification for the 

 conclusion that non-electrolytes, in fairly dilute solution, 



give lowerings which conform to the equation £ = 0'02^. A 



very satisfactory agreement for more concentrated solutions 

 has been shown by others and confirmed by myself*. That 

 such a discrepancy should exist for very dilute solutions does 

 not appear to me to be a priori impossible. Take the case of 

 dilute gases. The law of gaseous expansion with temperature 

 is fairly well established as an approximate expression of the 

 truth, yet Ramsay and Balyf have recently shown that it 

 does not hold when the gases become very dilute. The 

 coefficient of expansion for dilute gases is not a constant, but 

 is sometimes greater, and at times less, than the normal. To 

 illustrate this, I give a few of their results : — 



Oxygen. 



Pressure, 

 millim. 



Hydrogen. 



Pressure, 

 millim. 



347 



Coef. Exp. 



273-74 



0-096 ^r 



1_ 



297 



0-077 



300-5 



51 



1-4 



0-07 ~ 



1. 



Nitre 



gen. 



ef. Exp. 



Pressure, 

 millim. 



Coef. Exp. 



1 



260 



53 



1 



304 



1 



233 



0-8 



1 



331 



1 



240 



0-6 



1 . 1 

 355 377 



If dilute gases, then, do not obey the expansion law, to 

 which more concentrated gases conform, might it not be that 

 dilute solutions of non-electrolytes do not conform to the 

 law of the lowering of the freezing-point which holds for 

 more concentrated solutions of the same substances? It 

 seems to me that the analogy is at least suggestive, since, 

 thanks to van't HofF, we know that solutions resemble gases 

 in a striking manner. 



However this may be, I believe the question must be 

 settled by direct experiment, rather than by the introduction 

 of doubtful corrections, which, when applied to the experi- 

 mental results, give values which lie somewhere in the neigh- 

 bourhood of the calculated. 



The point raised by Wildermann J, in connexion with the 

 use of my thermometer, that when the ice separated from the 

 water, a layer formed around the bulb, seemed to be worth 

 testing again. I could not think that this was possible, since 

 in earlier work this point had received attention, yet experi- 



* '/At. f.phys. Chem. xii. p. 653; Phil. Mag. xxxvi. p. 494 (1893). 

 t Phil. Ma?-, xxxviii. p. 301 (1894). 

 \ Zeit. f. pfiys. Chan, xv. p. 358. 



