1414 
its length above the vanishing temperature, which would immediately 
cause considerable potential differences. In this connection we thought 
particularly of Joure heat. PELTIER 
heat, which we had noticed before 
(Comm. N°. 124c) but which for the 
present we attributed to impurities 
in the mercury in the legs, and 
assumed to be present only in the 
neighbourhood of the transition 
from solid to liquid, I took to be 
as far as possible excluded by the 
fact that the whole current system 
was of pure solid mercury at the 
very low temperatures. Now this 
belief may be untrue, because 
owing either to tension caused by 
a difference of expansion to that 
of glass which it seems can be 
fairly great as the mercury sticks 
to the glass, or through the contact 
between crystals of different kinds 
or sizes, even in the purest mer- 
cury. considerable thermopowers 
may possibly appear. But then 
they have their seat, as shown by 
the previous experiments, chiefly 
in places above the temperature of liquid air and Petter heat in 
these places need not be feared. To avoid disturbances of the sort 
to which we referred the experiment was repeated with resistances 
of such a kind that the conduction of any kind of heat from a 
part of the apparatus where there was higher temperature was 
made very difficult. The accompanying figure, which should be 
compared with figs. 1 and 2 on the Plate in the previous Comm. 
N°. 124e (VI of this series) shows the form chosen. The mercury 
threads which lead the current to and from the apparatus, run 
first through the liquid helium downwards, before they come 
out into the widened parts of the resistance. The potential 
wires do the same’). Close to the surface of the liquid the leading 
1) Corresponding parts are indicated by the same letters, modified parts by the 
addition of an accent A small additional improvement was further introduced into 
the contacts at the upper end, the four leading tubes were simply left open 
