Mr. J. Parker on Thermoelectric Phenomena. 357 



effect, make q return from M x to M without altering the 

 potential of either. The heat absorbed at the junction may- 

 be written 



—o'rq. 



(4) Again disconnect M and M 1? and then reduce their 

 temperatures to their original values. The increase of entropy 

 will be 



Lastly, reduce the potentials to their original values V, V 1? 

 which may be done without producing any thermal effect. 



If we can apply Carnot's principle to the cycle just de- 

 scribed, we get 



\t t + e) J f t dt J t+t t dt 



o* 1 df(t+0) 1<¥ =Q 



i. e. 



t t+e t+e d{t+e) t dt 



Hence - — - -J- is independent of t, and = &, say. 



Hence, by (6), dY _ , 



'all - kt ' 



If, now, we join two portions of the same metal, whose tem- 

 peratures t, t differ by a finite amount, and if V, V be the 

 potentials which they assume, then 



V-V =p.(^-tf) = A,say. ... (7) 



Also, if S . q be the heat absorbed at the junction when a 

 quantity of electricity q passes from t to t, 



fi=ik.(fi-t^+f(t)-/(t )=A+/(t)-f( t() ). . . (8) 



It follows immediately from (7) that if we take a piece of 

 homogeneous wire unequally heated, the difference of the 

 potentials of the two ends will be the same as if they were 

 actually in contact. The ends of the wire may therefore be 



