(2) A = - a'i + b'i squared,

of the same form as the expression E:

E = - ai + bi squared,

of the heating of the thermometer electrode. Further, every cause which
affects the coefficients, a or b, also affects in the same way a' and
b': such causes being the greater or less dilution of the solution, the
nature of the salt, etc. It is, therefore, impossible not to be struck
by the direct relation of the thermic and mechanical phenomena of which
the negative electrode is the origin. The following is the explanation
which I offer: The thermometer indicates the mean temperature of the
liquid just outside it; this temperature is not necessarily that of the
metal which incloses it. The current, propagated almost exclusively by
the molecules of the decomposed salt, does not act directly to cause a
variation in the temperature of the dissolving molecules; these change
heat with the molecules of the electrolyte, which should be in general
hotter than those when a heating is noticed and colder when a cooling is
observed. Suppose it is found, in the first case, that the metal, at
the moment when it is deposited, is hotter than the liquid, and,
consequently, than the thermometer; it becomes colder immediately after
the deposit, and consequently contracts; the deposit is compressed.
The reverse is the case when the metal is colder than the liquid; the
deposit then dilates. If this hypothesis is correct, the excess, T,
of the temperature of the metal over the liquid which surrounds the
thermometer should be proportional to the contraction, A, represented
by the formula (2), and the neutral point, I', of the contraction
corresponds to the case where the temperature of the metal is precisely
equal to that of the liquid.