Proof of Arrhenius's Generalization. 135 



In these Tables are given under a[ the degrees of dis- 

 sociation as deduced from the freezing-point, when the value 

 of Van't HofFs constant is taken to be 1*87 ; a 2 gives the 

 degrees of dissociation as determined by electric conductivity 

 by a careful interpolation. The result is that the agreement 

 between the values a[ and a 2 is excellent. I think, therefore, 

 that the latent heat of ice is more correctly estimated at 80 

 than at 79 cal. It appears that the degrees of dissociation 

 deduced from the freezing-point under a\ are somewhat 

 smaller than those deduced from electric conductivity. Now 

 the values of a, a' v and a 2 have been determined at various 

 temperatures (from about 0° and 18° or 25°), and in view of the 

 investigations of Kohlrausch and Ostwald (on acetic acid, &c), 

 we have good ground for presuming that the degree of disso- 

 ciation deduced from electric conductivity would be somewhat 

 smaller for the same concentration at lower temperatures. I 

 also think — being, till lately, singular in this opinion — that we 

 may assume that in the solutions there are always present, in 

 greater or lesser number, still higher undissociated and disso- 

 ciated molecules which cause greater or lesser deviations from 

 the strict agreement of the degrees of dissociation as deduced 

 from the freezing-point and from electric conductivity. But 

 the most probable reason is, that the absolute value of the 

 graduated scale of the 1 ^qq° thermometer is, between 0*49 and 

 0*36, 1-2 per cent, too small. From various causes 1 regard 

 this as very probable. My further investigations will decide 

 this question. In every case we ought, no doubt, in our 

 calculations of the degrees of dissociation to use in all our 

 given concentrations, or in concentrations measured by the 

 scale 0-49 to 0*36, the value 1-84 or 1*85 instead of the 

 theoretical 1*87, since we have practically found in the case 

 of non-electrolytes under similar conditions in the more 

 dilute solutions the value 1*85 or 1 # 84. If we do so, the 

 law of Arrhenius finds in the observations given above the 

 most wonderful, unexpected confirmation. See in the table 

 above, a Y and a 2 . In any case, look at the above results 

 how we may, there is no doubt that the law of Arrhenius 

 has found remarkable confirmation in the cases of the non- 

 conductors cane-sugar, urea, and alcohol, and of the conductors 

 potassium chloride, sulphuric acid, dichloracetic acid, tri- 

 chloracetic acid, orthonitrobenzoic acid. 



