liquid during the act of uniting together. Corrosion therefore is an
effect of molecular motion, and is one of the modes by which that
motion is converted into and produces electric current.

In accordance with this theory, if we take a thermo-electric pair
consisting of a non-corrodible metal and an electrolyte (the two being
already electro-polar by mutual contact), and heat one of their points
of contact, the molecular motions of the heated end of each substance
at the junction are altered; and as thermo-electric energy in such
combinations usually increases by rise of temperature, the metal and
liquid, each singly, usually becomes more electro polar. In such a
case the unequally heated metal behaves to some extent like two
metals, and the unequally heated liquid like two liquids, and so the
thermo-electric pair is like a feeble chemico-electric one of two
metals in two liquids, but without corrosion of either metal. If the
metal and liquid are each, when alone, thermo-electro-positive, and if,
when in contact, the metal increases in positive condition faster than
the liquid by being heated, the latter appears thermo-electro-negative,
but if less rapidly than the liquid, the metal appears
thermo-electro-negative.

As also the proportion of cases is small in which metals that are
positive in the ordinary thermo-electric series of metals only become
negative in the metal and liquid ones (viz., only 73 out of 286 in
weak solutions, and 48 out of the same number in strong ones), we may
conclude that the metals, more frequently than the liquids, have the
greatest thermo-electric influence, and also that the relative
largeness of the number of instances of thermo-electro-positive metals
in the series of metals and liquids, as in the series of metals only,
is partly a consequence of the circumstance that rise of temperature