as a Source of small Constant Currents. 355 



altering the extent of the mercury surfaces, but we finally 

 convinced ourselves that the capillary assumptions made were 

 of such a nature as to require revision. This would in itself 

 have involved extensive research, and we did not feel that the 

 end was likely to prove in any way commensurate with the 

 labour of reaching it; consequently we are unable to say 

 whether the potential of the mercury surface does or does not 

 rise with increased density of hydrogen deposition. In any 

 case the small deviation of the value of A observed from its 

 value, as accounted for by the cell resistance, shows that very 

 little (if any) hydrogen really reaches the mercury surface. 

 On the other hand, it is easy to see several reasons why R" 

 might become smaller sufficiently to account for the effect 

 observed. It may be worth noting that the possibility of 

 collecting evidence as to the existence or nonexistence of 

 polarization when the change of A with flux of time is very 

 small becomes a question of indirect evidence. 



We did not habitually estimate changes of A of less 

 magnitude than 0' 000015 volt, and consequently what follows 

 refers only to changes of greater amount than this. By our 

 method the obvious test for polarization would be a measure- 

 ment of the resistance of the cell at the actual moment when 

 its terminal P.D. was obtained. This is experimentally im- 

 possible (as we consider) withXhe accuracy necessary, to say 

 nothing of the continual changes in the value of the current 

 taken from the cell which are to be expected during the 

 adjustment of the bridge. The tests which we apply can be 

 rendered evident most easily by means of a diagram. Con- 

 sider first the case of no polarization, i. e. no change of E or 

 R. If we plot values of A as ordinates when C is nearly 

 constant, and is assumed to be exactly so, and times reckoned 

 from the epoch of short-circuiting as abscissas, then the 

 resulting curve will be three sides of a rectangle if it is pro- 

 longed till the moment of reinsulating the plates of the cell. 

 This in fact corresponds simply to the case of A = CR. Now 

 let polarization set in. The curve giving the value of A will 

 rise or fall, and the amount of rise or fall will be a sort of 

 measure of the polarization. 



Again, let currents of di fie rent values be taken from the cell 

 and observe whether A = KO, where K is a constant (it is of 

 course assumed that R cannot be got with sufficient accuracy 

 by other means, and the determining of it in this way would 

 beg the question). We shall show that when certain con- 

 ditions are complied with, this is nearly but not quite true. 

 Consequently to the high degree of accuracy which has been 

 aimed at in our experiments, it is never true to say that the 



2 D2 



