88 



SCIENCE. 



LX. S. Vol. XXI. Xo. 525. 



electricity, or of the corpuscles moving 

 with it, which would come to the same 

 thing. Moreover, we can hardly avoid 

 supposing that the attraction which we 

 have assumed to exist between metal and 

 electricity holds the electricity within the 

 metal in a state of pressure ; and accord- 

 ingly we must recognize in the thermal 

 capacity of the electricity a part accom- 

 plished against this pressure in the expan- 

 sion which accompanies rise of tempera- 

 ture. 



Returning, with these additional ideas, 

 to the examination of a thermo-electric cir- 

 cuit showing no Thomson effect, we find 

 that we must in such a case suppose that 

 in each metal the heat absorbed by the 

 current of electricity, positive or negative, 

 which is flowing from cold to warm within 

 that metal is balanced by the heat given 

 out by the current of opposite sign, nega- 

 tive or positive, which is flowing in equal 

 strength from hot to cold within the same 

 metal. 



But at the junctions the case is different. 

 At the junction which is the prevailing 

 one, across which each kind of electricity 

 flows from the metal by which its kind is 

 attracted less to the metal by which its 

 kind is attracted more, that is, from n 

 place where the pressure caused by the 

 attraction is less to a place where the pres- 

 sure caused by attraction is more, each kind 

 of electricity will, without change of tem- 

 perature, suffer contraction of volume in 

 the transition, and evolution of heat will 

 result. On the other hand, at the other 

 junction, where each kind of electricity 

 moves, without change of temperatiire, 

 from a place of high attractive pressure, to 

 a place of low attractive pressure, each 

 kind will expand in the transit, and ab- 

 sorption of heat will accompany this ex- 

 pansion. 



Thermodynamic considerations show us 

 that in such a case as that which wc are 



considering, in which there is no Thomson 

 effect, heat must be taken in at the hot 

 junction and heat must be given out at 

 the cold junction. Hence our theory, with 

 its later assumptions, assumptions suggest- 

 ed, as others have been, by reflection on 

 the manner and reason of the working of 

 an ordinary convection cycle, has led us 

 clearly to the conclusion that the cold junc- 

 tion should be, in the case considered, the 

 prevailing junction. But thermodynamic 

 considerations go further. They require 

 that the amount of heat, Q' , taken in at the 

 hot junction at temperature T', must bear 

 to the heat, Q, given out at the cold junc- 

 tion at temperature T, such a relation that 



T' T' 



Can we without a straining extension of 

 our assumptions meet this condition ? Ap- 

 parently we can do so by supposing that 

 electricity in its state of compression with- 

 in each metal obeys the law of a perfect 

 gas. At the hot junction we have the posi- 

 tive electricity going, at constant tempera- 

 ture T', from the attractive pressure p to 

 the attractive pressure p — dp, with con- 

 sequent expansion, work of expansion, W, 

 and absorption of heat equivalent to this 

 amount of work. At the cold junction we 

 have the positive electricity going, at con- 

 stant temperature T, from the attractive 

 pressure p — dp to the attractive pressure 

 p, with consequent compression, work of 

 compression, W, and evolution of heat 

 equivalent to this amount of work. From 

 the gas law, pv — KT. we have, when T 

 Ls constant, 



pan = — nip = an. 



P 



This gives us, since p and dp are the same 

 at the hot jimction as at the cold junction, 



W : W : : pdv' : pdv : : T' : T. 



And so 



Q' : Q : : T' : T. 



