RICHARDS. — CALOMEL ELECTRODE, 15 



for we cannot assume either P or p to be constants with this form of 

 electrode. Dividing by d T, we obtain an expression for the temperature 

 coeflicient of a single electrode, as follows : — 



___ _- w _ + __ y^^rp ^^^y 



Since the general tendency of the actual results is logarithmic, my 

 observations show that the last quantity must be so small as to be with- 

 out any very great influence. 



In order to compute the maximum influence of this last quantity, let 

 us assume that P is constant, hence that dP =. ; and the quantity con- 

 sidered becomes 



But according to the law of Dalton, which applies approximately to ions, 



T:dT = p:dp, or — x "^ == 1. 



dT p 



Hence in this case the last term becomes = —0.000086, which is 



only about a tenth of — itself. 

 ^ dT 



For the present, then, let us neglect this last factor, assuming that on 



d P dp 

 the average — = _:L , and our formula becomes 

 P p 



^= ^In- = 0.000198 log - , 

 dT e^ p p 



in which log stands for the common Briofcrs's losarithm. 



In this case, since we are considering a negative ion, and P > /?, it is 

 obvious that the potential of the mercury with respect to the solution 

 must increase with the temperature, an inference which corresponds with 

 the facts. The current may be supposed to be carried by chlorine ions 

 moving from the hot electrode, where they are given oflT, to the cold, 

 where tliey combine with mercury. 



We have now at our disposal a formula of the sort required ; the 



table above furnishes values for — ^ , and p may easily be calculated 



dT ■^ 



from the relation a = the dissociation factor = -^ , which changes only 



