66 PROCEEDINGS OF THE AMERICAN ACADEMY. 



original value by replacing the weaker of the two solutions witli that of 

 the connecting solution (of the same concentration). 



Cell No. 2, at 298°, remained constant for at least twenty minutes 

 after the 1% solution in the neighborhood of the electrode was replenished. 

 Of four cells measured, one decreased in value, and three increased, on 

 standing. 



This cell in the chinoline bath remained constant for several hours. 



Cells 3 and 4 were extremely constant for several hours at a time, and 

 the same solutions iu the same tubes could be lieated up several days in 

 succession, and would always give the same value. 



The inconstancy of the cells containing very dilute solutions would 

 seem to indicate that in these there is a slow reaction going on between 

 the solution and the electrode. The fact that the replenishing of the solu- 

 tion with that which had not been in contact with tlie electrode always 

 brought the cell back to the same value, shows that the reaction did not 

 take place throughout the solution as a whole, but only in the neighbor- 

 hood of the electrode. The more rapid variation at the higher tempera- 

 ture shows that, whatever the reaction may be, its speed increases with 

 the temperature. Obviously a slow dissolving of the silver electrode in 

 the fused salt would account for the observations, the weak solutions be- 

 ing much more affected by this possible irregularity than the strong ones; 

 but this hypothesis is not advanced as a certainty. 



The observed values for the cells No. 1 and No. 2 agree remarkably 

 well with the calculated. We must conclude that the osmotic theory 

 of electromotive forces can be extended to the case of fused salts, and 

 that even in a 10% solution the silver nitrate is almost wholly dissociated. 



In considering the observed potential to be the difference between the 

 two electrode potentials, we assume that the potential difference between 

 the two differently concentrated solutions is so small as to be negligible. 

 While the good agreement between the values found and calculated is 

 the strongest argument in favor of this supposition, it is to be inferred 

 also from what we know of aqueous solutions. Such potential differ- 

 ences depend on the difference in transference numbers of the two ions. 

 Nernst and Loeb,* and more especially W. Bein,t have shown that the 

 transference numbers all tend toward the value 0.5 as the temperature 

 increases. For AgNO^ especially, they are never far from 0.5, and ap- 

 proach very near it at the highest temperature measured. W. Beiu 



* Zeit. fur phys. Chemie, II. 962, 1888. 

 t Wied. Ann.,' XLVI. 69, 1892. 



