286 MORRIS LOEB 



tivity are a function of the total number of molecules of the 

 electrolyte between the electrodes; according to Arrhenius, 

 the coefficient of activity, i.e., the proportion of molecules 

 actually engaged in conduction, not only varies for different 

 compounds, but also increases for any one compound with 

 its dilution, and approaches unity for extreme dilutions. It is 

 only in this limiting case that Kohlrausch's equation becomes 

 absolutely true; this is the view that Kohlrausch himself 

 had expressed with regard to his law, whose accuracy, he said, 

 could only be tested by experiments with highly dilute solu- 

 tions. Ostwald's examination of the conducting power of 

 highly attenuated solutions of a very large number of electro- 

 lytes l has afforded very valuable support to these recent 

 views. 



Since, however, the quantities u and v are so closely con- 

 nected [117] with the rate of transference of the ions, a study 

 of the latter in dilute solutions must afford powerful means 

 of testing the truth of Kohlrausch's law with Arrhenius's 

 emendations. This was the chief motive of our work, and we 

 must now, therefore, proceed to compare the values which we 

 have given in section 5 with the conducting power of the same 

 salts. Accordingly, the necessary measurements of conduc- 

 tion were made by one of us (W. N.), by Kohlrausch's method 

 which with various valuable modifications has been described 

 by Professor Ostwald. 2 The values which we present are 

 scaled upon the conducting power of mercury as unity, and 

 the calculations are based upon Kohlrausch's determination 

 of the molecular conductivity of argentic nitrate. 3 



The following table gives the conducting power of the dif- 



1 Zeitschr. 1, 61 and 97. * Ibid. 2, 563. 



3 As Kohlrausch worked at 18, while the present determinations were made at 

 25, the coefficients of temperature were determined by special experiments for va- 

 rious dilutions of AgNOa. For the details of these results we must refer to the Ger- 

 man version of our paper. 



