﻿in the Moisture condensed on Glass Surfaces. 339 



positive terminal while it shrinks away from the other. The 

 neaping up of the water would cause an increase in resistance 

 owing to the departure from uniform cross-section of the 

 conducting film. 



If this current be now reversed, water falls on one side and 

 rises at the other. This would cause a diminution in resist- 

 ance in the first instance, but subsequently an increase. Thus 

 on reversal the current would, as observed, commence by 

 rising, while finally falling in amount. 



II. An electrolytic action may occur in which sufficient 

 oxygen is released to produce a sensible covering of oxide, at 

 least in the case of tin. 



In this way a very considerable increase in the resistance 

 may be brought about. On reversal of the current the 

 hydrogen evolved would reduce the oxide, and thus diminish 

 the resistance, and consequently the current would increase 

 in amount. A difficulty in this second hypothesis is the 

 rapidity under some circumstances with which the reduction 

 must be effected ; for with even a dead-beat galvanometer the 

 current is often several times the direct value on reversal. 



Rate of Change of Current ivith Time. 



The time-rate of change of the current was experimentally 

 determined and compared with that deduced from the second 

 hypothesis. 



Assuming the change of resistance as due to the formation 

 of electrolytic products, we may reasonably take 



dR/dt = a.dX/dt, 



where X = amount of material formed deleterious to the passage 

 of the current ; but 



dXldt = bxC, C = E/R; 

 so dB/^ = m.E/R, 



R 2 -R 2 = m.E.£; 



or 



that is f7:L— _L\- 



\C 2 cW 



mt. 



This curve has been calculated and plotted. From Table L 

 and fig. 2 it will be seen that the curve fits the experimentally 

 determined points with a fair accuracy. This result obviously 

 points strongly to the second hypothesis. 



Z 2 



