HALL. — ELECTRIC CONDUCTION AND THERMOELECTRIC ACTION. 79 



(2) If the thermal effusion law, ih(RiTi)* = n^T*)*, holds: — 

 In this case we use equation (6') and can find values of (/ -J- /3) 

 corresponding to chosen values of q' . 



TABLE II. 



Cases in which Equation (6') Holds. 



It may be that an equation intermediate between (6) and (6') 

 would come nearer to the truth than either (6) or (6')- For it seems 

 probable that thermal effusion of the electrons has some effect in the 

 interatomic spaces, though less effect than it would have in the case 

 of a thin partition pierced by a narrow hole. I shall, however, for the 

 present go on with equation (6), with occasional references to equation 

 6') and its consequences. 



The Thomson Effect: — Taking equation (6) as the condition for 

 electrical equilibrium in a detached wire having a temperature gradi- 

 ent, I shall now find the Thomson effect coefficient, a, in this metal. 

 This coefficient I shall so define that crdT will mean the mechanical 

 equivalent of the heat absorbed during the passage of the electro- 

 magnetic unit quantity of electricity, electrons, if e is the numerical 



value of the electron charge, from a part of the metal at temperature 

 T to apart at temperature T + dT, under the conditions of electrical 

 equilibrium. That is, I am dealing with a reversible process and 

 ignore for the present the consideration of resistance-heat, which, 

 according to custom, we may assume to be as small as we please to 

 make it. My only innovation here is the change of sign of a, — for 

 example, from + to — in copper and from — to + in iron, because the 

 current of electricity is now thought of as a movement of the negative 

 electrons. In my equations e is to be taken as a positive quantity. 



