Thermoelectricity of Molten Metals. 503 
of particular interest to repeat Magnus’s experiment with 
this fluid alloy. Ifa portion of the amalgam was plunged 
cold into the rest heated to about 300°, no thermo-current was 
produced the electromotive force of which exceeded 0°5 micro- 
volts, which was the lowest I was in a position to measure. 
The curve for this element shows a remarkable peculiarity. 
At 180° it is convex towards the axis of abscissee, but between 
180° and 210° it becomes concave, and then again convex. 
In this respect it is qualitatively exactly like the curve which 
platinum-iron elements give, only that in the latter the pecu- 
liarity is more decided. With a Pte element the curve 
rises steeply to about 360° and somewhat concave towards 
the axis of abscissee, then bends rapidly towards the horizontal 
direction, having a maximum at 420°, then falls from this very 
little (almost horizontal) to a minimum at 520°, rises then 
slowly, and then from about 630° ascends again to 1000° as 
steeply as from 0° to 36°. 
The element consisting of molten potassium and mercury 
shows an exactly similar behaviour, the curve rising with 
oscillations. Consistent results are, however, only obtained 
with this latter element when vapours of constant boiling- 
point are employed as source of heat. Using the oil-bath or 
the air-bath, the values obtained at the same temperature 
often vary considerably. 
We shall probably only be able to explain these phenomena . 
by the assumption that molecular transformations take place 
also in liquid bodies (as is known to be the case, for example, 
with sulphur); and these changes appear, at least as far as their 
velocity is concerned, to depend on the rate of heating. The 
question then suggests itself whether molten cadmium, which 
in the form of vapour is monatomic, combined with a thermo- 
element with the similarly constituted mercury, would show 
linear dependence of electromotive force upon temperature. 
The interval of temperature available seemed to be too 
small to determine this point with certainty; and I considered 
it beside the mark to employ cadmium amalgam, since we 
could no longer assume the existence of isolated atoms. 
8. Taking all the results together, we may come to the 
conclusion that in thermoelectricity we are still further 
from an insight into the true nature of things than most phy- 
sicists have supposed. We must assume the existence of 
electromotive forces even in the interior of molten metals of 
unequal temperature ; and, moreover, these follow no simple 
laws. It would not be difficult to find formule for the elec- 
tromotive forces supposed to exist in the interior of metals 
which should satisfy the observations; but it seems to me 
