OP ARTS AND SCIENCES. 217 



Tables of the values of the Thomson effect show that nickel changes 

 twice between 17o° and 340°. There are two points of deflection in 

 the thermo-electric line. At these points, — 





■,(f,) = o. 



Let Tq be the temperature at which the Thomson effect equals 

 zero, and let The a. temperature above this. Let the circuit be so 

 heated that the temperature rises gradually from T^ — T, and then 

 falls abruptly to T^. Any heat effect at the plane must necessarily 

 take place at the higher temperature, if we suppose the Thomson 

 effect alone to exist. Hence frorn thermodynamic considerations the 

 absorption must take place here. In this case the direction of the 

 current is completely determined. 



Suppose, now, we pass a current from some outside source, through 

 the circuit, in such a direction as to absorb heat where the tempera- 

 ture falls gradually. There can be no evolution of heat at the plane, 

 as it can take place neither at T^ nor at T. If the whole energy of 

 the current is expended in heat we must have 



EI=RP- iL'dT. 

 : E=RI—icTdT. 



That is, the electromotive force of the outside current is diminished 

 by an electromotive force peculiar to the arrangement, and due to the 

 absorption of heat in the circuit. But if the Thomson effect is the 

 only effect in this circuit, the current produced by it does not obey the 

 fundamental principles of thermodynamics. We have an engine, 

 working between finite temperatures, in which there is no loss of heat. 

 Suppose, moreover, T < T^, and pass a current in such a direction 

 as to evolve heat where the temperature varies continuously. There 

 can be no absorption at TJ,, and 



E=RIA- la-dT. 



= RI-\- Co 



In this case the outside electromotive force is increased by a secon- 

 dary electromotive force. But unless some effect besides the Thomson 

 effect exists, the secondary current is produced alone by an evolution 

 of heat. The conclusion is absurd, and there must exist some other 

 effect at the plane of abrupt variation. If this effect exists for the 



