( 688 ) 
Q Q 
2 af (6, dlog A, 2aT (0, dlog A, 
= A EE ig et ee ene 
da: 33 de 
O ed 6 
Je Mig 
At the same time the intensities of the partial currents are given by 
_ 9,5, oO, 
= : 0 
i= 7 (E,—E) = , #= = (EE) =. 
These values, which are equal with opposite signs, will in general 
vary along the circuit, so that, even in this simple case, we cannot 
avoid the complications I have pointed out in $ 17. Nor can the 
difficulty be easily overcome. Indeed, we can hardly admit that the 
state of two pieces of different metal, in contact with each other 
and kept at a uniform temperature is not truly stationary. If, in 
order to escape this hypothesis, we have recourse to the considera- 
tions I presented at the end of § 17, we must suppose the neutral 
electricity to be continually built up in some parts of the system 
and to be decomposed in other parts. The first phenomenon will be 
accompanied by a production and the second by a consumption of 
heat. That these effects should take place in a system whose state 
is stationary and in which there are no differences of temperature, 
is however in contradiction with the second law of thermodynamics. 
The only way out of the difficulty, if we do not wish to confine 
ourselves to one kind of free electrons, seems to be the assumption 
that there is no accumulation of neutral electricity at all, 1. e. that 
7, and 7, are simultaneously 0. This would require that E, = E,, 
2 
or in virtue of (53) 
] dV, 2 al d log Ae : 1 dV, 2 a ak d log A, 
: Nee ee - 58 
En da 3 2 da (Op da 5 GC da: ( ) 
Since e, = —e,, we might further conelude that 
2, 1 log (A, A.) L(V. V. 
z on ged core: a) zi ae ra Vv) — 0. 
3 da da 
which means that 
3 ye re Id Al 
log tt) nn ar A) == th (IE) 
ought to have the same value in all parts of the circuit. We should 
therefore have to regard this expression as a funetion of the tempe- 
rature, independent of the nature of the metal *). 
If we suppose the contact of two metals to have no influence on 
the number of free electrons in their interior, we must understand 
by A, and A, in the above equation quantities characteristic for each 
1) Cf, Drupe, Annalen der Physik, 1 (1900), p. 591. 
