1417 
it had been shown that the kind of lead had little influence on 
the phenomena, so as to be free from the troublesome thermocurrents 
in the potential wires, when these were of mercury, from the resistance 
which was immersed in helium to where the ordinary temperature 
began, and all four were replaced in order to be free in the choice 
of the pair of threads which were to be used as potential wires or 
as current leads. The thermopowers were now only about 10 
microvolts, 
The experiments of Jan. 1912 were made with two mercury 
threads, one with a resistance of about 50 2, the other of about 
130 &. These resistances were joined up in a cireuit with a milliam- 
meter, which could be shunted, and to each of them one of the 
coils of a differential galvanometer was connected as a shunt. By 
using only one coil at a time the resistance of each of the mercury 
threads could be measured separately; by connecting the two coils 
in the opposite direction the change in the ratio of them with the 
temperature could be investigated as long as the difference was small. 
The ratio 
Wvo = 128,4 — 9 55 
W,, J/r=2900 50,4 
became, through cooling to the boiling point of helium 
Wir = epee oe. 
oe 40,25 0,024 
The ratio changed, as had been found before, and as could be 
readily explained by a slightly different manner of freezing of the 
mercury in the two tubes. 
On changing the current strength at 4.25 K. we found 
Current in Amp. Ws, Wo 
0,006 0,0545 0,0251 
0,010 0,0250° 
0,016 0,0249 
0,030 0,0549 0,0260 *). 
Up to currents of 0.03 amp. therefore it is confirmed that there 
is no reason to assume a deviation from Onm's law above the 
vanishing point. | 
On lowering the temperature from the boiling point to where the 
1) As regards the deviation at 0,03 amp. of Wi, we may perhaps conclude 
from the comparison of the ratio of the resistances at 7’= 290° K and 7’ = 4°.25 K. 
in the two resistances, that there is a thinner place in the thread Wz, by which a 
greater heating takes place locally at temperatures above the vanishing point, than 
would be expected from the average section. 
