S6 



SCIENCE. 



[N.S. Vol. XXI. Xo. 525. 



'^'d hoinogeneovis metal bar, unless the hot 

 .parts of this bar attracted a given quantity 

 kof electricity more or less strongly than the 

 cold parts. The two metals in which we 

 have stipulated that there shall be no 

 Thomson effect correspond in our thermo- 

 electric circuit to the horizontal pipes of 

 our imagined convection system ; and for. 

 the comparison which we are here making 

 it is well to go back to the usual disposition 

 of the thermo-electric diagram, in which 

 unequally heated metals having no Thom- 

 son effect are represented by horizontal 

 lines. 



ijet us now consider a case in which the 

 Thomson effect does play a part, such a 

 case as that illustrated by Figs. 1 and 3. 

 We can, apparently, account for the Thom- 

 son effect in any metal by assi;ming that 

 this metal has a greater attraction for elec- 

 tricity of one sign than for electricity of 

 the opposite sign, and that the difference 

 •of these attractions is a function of the 

 temperature of the meial. With this con- 

 ^lition the electricity of one sign at any 

 pnrt of a homogeneous but iinec[ually heat- 

 vi\ metal bar will be subject to a net attrae- 

 tii n, exerted by the metal, toward a place 

 ( f higher temperature or toward one of 

 .lower temperature, according as the attrac- 

 tion between the metal and this kind of 

 electricity increases or decreases with rise 

 of temperature of the metal ; and the other 

 kind of electricity will be subject to a dif- 

 ferent, greater or less, net attraction from 

 the metal ; so that a difference of potential 

 would be set up between the hot and cold 

 part of the bar, if the bar were left to 

 itself. 



If we take the view that the electromotive 

 forces which prevail are those at places 

 ■where heat is given out, we shall in Fig. 3 

 have the local electromotive force, due to 

 'the attraction between metal and elec- 

 tricity, opposite at every place to the elec- 

 tromotive force commonly supposed to re- 



side at that place; so that the unequally 

 heated metals and the hot junction will 

 still conspire again.st the cold junction; 

 but, as the direction of the current is 

 known by experiment to be that which is 

 indicated by the arrow points in Fig. 3, 

 we must in this case suppose that the cold 

 junction prevails over the opposing com- 

 bination. 



Let us now consider the magnitude of 

 the local electromotive forces. In any case 

 the net electromotive force of the whole 

 circuit is expressed, as we agreed at the 

 beginning, by the area CC'I'IC of Fig. 1 

 or Fig. 3. But knowledge of the net elec- 

 tromotive force of the circuit tells us little 

 or nothing of the magnitude of the indi- 

 vidual four electromotive forces of the cir- 

 cuit. Ordinary doctrine represents these 

 by the areas, already mentioned, under the 

 lines CC, CT, etc., in Fig. 3, down to the 

 line of absolute zero of temperature, but 

 as we now undertake to have the electro- 

 motive force at the cold junction prevail 

 over the other three, it is evident that we 

 must look for other areas on the thermo- 

 electric diagram to represent these local 

 forces. In this ease we find such areas 

 ahove the lines CC, CT, etc., in Fig. 3, or 

 in Fig. 4, which we will now use in place 



Fig. 



