1426 
horizontal axis in fig. 7 is probably drawn too sharp; at 3°.96 K. 
the potential difference was < 0.03.10 6 V. 
The whole gives one the impression that the lower temperature 
of the bath at greater strength of current is required (a comparison 
of 0.004 and 0.04 amp. shows that an almost constant shift of 
temperature would change the potential differences per unit of current 
in the one case into those of the other) to cool the part.of the 
thread that has an ordinary resistance strongly enough to prevent it 
imparting its temperature to the part which is below the vanishing 
point, and to prevent the temperature in the latter part from being raised 
above the vanishing point by the greater local development of heat. 
With the same thread in the manner of table II the results of 
table VII were found, in which experiments are included with a 
second thread with a section of about 0.012 mm’. 
It appears that in the thread Wy}, to which the experiments just 
quoted refer, local heating takes place more easily at the same 
current density than in W,,, (see $ 5). The fact that the latter thread 
gives off heat more readily also explains why in JV,,, a greater 
current density checks the disappearance of the resistance less than 
in the case of IW, (June 1912). 
As regards the threshold value of the current density for different 
temperatures with the same thread, it would seem from Table VII 
and Table V roughly speaking to change linearly with the temperature, 
TA BLE VII 
Potential differences at the extremities of mercury 
threads carrying current 
| Current density in | Potential difference 
ie | amp. per mm? | in microvolts 
ie ed 
i Ce Wij 
3220 KK 129 | | 05 
| 141 | | very large 
Saget heir | 
412 | 3,8 
429 | EZ 
| 431 | very large 
| | DAA 
