btained.

When metals in liquids are heated, they are more frequently rendered
positive than negative in the proportion of about 2.8 to 1.0; and
while the proportion in weak solutions was about 2.29 to 1.0, in
strong ones it was about 3.27 to 1.0, and this accords with their
thermo-electric behavior as metals alone. The thermo-electric order of
metals in liquids was, with nearly every solution, whether strong or
weak, widely different from the thermo-electric order of the same
metals alone. A conclusion previously arrived at was also confirmed,
viz., that the liquids in which the hot metal was thermo-electro-positive
in the largest proportion of cases were those containing highly
electro-positive bases, such as the alkali metals. The thermo-electric
effect of _gradually_ heating a metal in a liquid was sometimes
different from that of _suddenly_ heating it, and was occasionally
attended by a reversal of the current.

Degree of strength of liquid greatly affected the thermo-electric
order of metals. Increase of strength usually and considerably
increased the potential of metals thermo-electro-negative in liquids,
and somewhat increased that of those positive in liquids.

The electric potential of metals, thermo-electro-positive in weak
liquids, was usually about 3.87 times, and in strong ones 1.87 times,
as great as of those which were negative. The potential of the
strongest thermo-electric couple, viz., that of aluminum in weak
solution of sodic phosphate, was 0.66 volt for 100 deg. F. difference of
temperature, or about 100 times that of a bismuth and antimony couple.

Heating one of the metals, either the positive or negative, of a
voltaic couple, usually increased their electric difference, making
most metals more positive, and some more negative; while heating the
second one also usually neutralized to a large extent the effect of
heating the first one. The electrical effect of heating a voltaic
couple is nearly wholly composed of the united effects of heating each
of the two metals separately, but is not however exactly the same,
because while in the former case the metals are dissimilar, and are
heated to the same temperature, in the latter they are similar, but
heated to different temperatures. Also, when heating a voltaic pair,
the heat is applied to two metals, both of which are previously
electro-polar by contact with each other as well as by contact with
the liquid; but when heating one junction of a met