(591 ) 
variability of Nj and N7 in so small an interval, we may write 
for (35) 
The value of the first factor on the righthand side may be taken 
from what, in § 9, we have deduced from the electrochemical 
aequivalent of hydrogen’). We found for 7’= 291 
Te 
B LOP 
é 
so that 
Nir 4 
log —" — 0,00011 Fe. 
Nr 
In the case of bismuth and antimony, #9 amounts to 12000, 
corresponding to 
Nir Ni 
(Qe — a, 
Á Ni ’ Nz : 
I see no difficulty in admitting this ratio between the number of 
free electrons in two metals wide apart from each other in the 
thermo-electric series *). 
1) The numbers of that § contain an error which, however, has no influence 
on the agreement that should be established by them. The value of 3 p and that of 
7 
deduced from the measurements of JaeGer and DiessetHorst are not 38 and 
é 
47, but 
op == 38 x 10° 
and 
T 
a ATCO 
e 
N 
2) Let x be the mean value of log - " between the temperatures 7” and Tv. 
Nr P 
Then the equation (35) may be put in the form 
9 
Be 5 na(ir" — 7"). 
This may be expressed as follows: The work done by the electromotive force 
9 
in case one electron travels around the circuit is found if we multiply by 5 n 
the increase of the mean kinetic energy of a gaseous molecule, due to an elevation 
of temperature from 7” to 7", 
