of the Electromotive Forces in a Voltaic Cell, 59 



of a volt. But if, as we have proved, the coefficient of the 

 Peltier effect is not sl measure of the contact-potential difference, 

 but of the product of the absolute temperature into the rate of 

 variation with temperature of the contact-potential difference, 

 it follows that the smallness of the Peltier effect does not 

 require that the contact-potential difference must be also 

 small. 



We have endeavoured to experiment on this rate of variation 

 of the contact-potential difference with temperature ; and in 

 the discussion on Dr. Lodge's paper before the Society of 

 Telegraph Engineers and Electricians, we gave a number of 

 hitherto unpublished results which we had obtained on this sub- 

 ject. The apparatus we employed was the same as that which 

 Dr. Lodge described as being used by us in our former expe- 

 riments on contact-potential difference, and with it we mea- 

 sured the contact-potential difference of hot and cold mercury 

 and of hot mercury and cold iron, hot mercury and cold 

 copper, also for hot mercury and cold platinum, in each case 

 for various differences of temperature. The results obtained 

 were large (amounting in the case of mercury and iron to as 

 much as 3 volts for about 16° C. difference of temperature), 

 and not small, as might have been expected from the preceding 

 equation and the smallness of the coefficient of the Peltier 

 effect. We explained, however, that the very large potential- 

 differences we obtained were probably due to invisible films of 

 oxide on the surface of the warm mercury, which probably 

 existed in spite of the fact that every effort was made to 

 remove this oxide by distilling the mercury between every 

 two sets of experiments ; and we showed that in such cases 

 the summation law does not hold true — that is to say, the 

 E.M.F. of a combination is not equal to the algebraical sum 

 of the various potential-differences at the contacts, measured 

 inductively, and which we have always found to be true in 

 other cases when the experiments were conducted with as 

 much care as was devoted to those made with hot mercury. 

 For example, we found that 



Copper at 16° C., with clean mercury at 16° C. ; had a 

 contact-difference of O308 volt ; 



Mercury at 16° C, with mercury at 26° C, had a 

 contact-difference of 0*75 volt ; 



Mercury at 26° C, with copper at 16° C, had a contact- 

 difference of —1*5 volt. 



Therefore, if these numbers represented the true contact- 

 potential differences, a thermopile consisting of two pieces of 

 cold copper at 16°C. ; dipping into two vessels of mercury, 



