January 20, 19G.5.] 



SCIENCE. 



87 



of Fig. 3. Thus the area Cbb'C'C, ter- 

 minated above by the temperature line T^, 

 characteristic of metal 1, may represent the 

 thermo-eleetromotive force directed from 

 C to C in the unequally heated (1). 



Similarly the area I'd'dir, terminated 

 above by the temperature line T^, charac- 

 teristic of metal 2, may represent the 

 thermo-eleetromotive force directed from 7 

 to /' in the unequally heated (2).* The area 

 C"6'e/d7'C", terminated above by the broken 

 line h'efd', depending on both and T,, 

 may represent the thermo-eleetromotive 

 force directed from 7' to C" at the hot junc- 

 tion. Finally the area Cbb'efd'dIC, term- 

 inated above the broken line bb'efd'd, de- 

 pending on both and T.,, may represent 

 the thermo-eleetromotive force directed 

 from 7 to C at the cold junction. This last, 

 larger than the sum of the others, which 

 oppose it, would be the prevailing electro- 

 motive force. The net electromotive force 

 of the circuit would be, as m Fig. 3, repre- 

 sented by the area CC'l'IC, and the cur- 

 rent would run, as before, clockwise with 

 respect to the boundary of this area. 



We have apparently succeeded in ac- 

 counting for the circulation of the elec- 

 tricity by means of differential attractions 

 conditioned by differences of temperatiire 

 and in showing that the local electromotive 

 forces of the thermo-electric circuit may be 

 opposite in direction to those which are 

 commonly supposed to exist. But we have 

 as yet given no conclusive reason why heat 

 should go in at one part and out at the 

 other, and we have not yet made any at- 

 tempt to show how heat is used up in the 

 circuit. Our explanation, so far as it has 



* 7*1 is apparently the temperature at which the 

 differential attraction of for the two kinds of 

 electricity becomes zero. A like explanation holds 

 for T... [The sloping lines might curve so as to 

 strike the lines and T.,, respectively, at any 

 angle.] 



now gone, utilizes difference of tempera- 

 ture but does not utilize heat. 



If we return to the consideration of our 

 analogical convection system, we see that, 

 if we were to put in heat at any point p 

 only and take out heat at the point p' only, 

 these two points being on the same level, 

 there would be no continued circulation, 

 as we should presently have the fluid at a 

 uniform temperature all the way over from 



Fig. 5. 



p to p' and at a uniform, though different, 

 temperature all the way under from p' to 

 p. To maintain circulation we must have 

 the point p, at which heat enters, at a lower 

 level, and therefore at a higher pressure, 

 than the level and the pressure of the point 

 p', at which heat comes out. The work 

 and the absorption of heat at expansion 

 under high pressure would be greater than 

 the return work and the emission of heat 

 at the lower pressure, and the difference 

 between the inflow and the outflow of heat 

 would be utilized in maintaining circula- 

 tion against some resistance. 



Do we naturally find anything in our 

 thermo-electric circuit corresponding to 

 this heat differential? 



We have already assumed that the elec- 

 tricity within each metal acts like an ex- 

 pansible fluid, and it is natural to assume 

 that the rise of temperature which causes 

 the expansion of the electricity absorbs 

 heat. That is, we naturally assume next 

 that there is a real thermal capacity of 



