454? Dr. Draper on the Electro-motive Power of Heat, 
We are therefore led to the general conclusion, that in 
these six different systems of metals, the developments of electri- 
city do not increase proportionally with the temperatures, hut in 
some %mth greater rapidity, and in others with less. 
The results here given, I have corroborated in a variety of 
vrays and with a variety of wires. A pair consisting of copper 
and platina, gave for the temperature of tin when in the act 
of congealing 452° Fahr. instead of 442° Fahr. the point usually 
taken. For the melting point of lead, it gave 942|° Fahr., 
instead of 612° Fahr. The melting points of tin, lead, zinc, and 
occasionally of antimony and bismuth, were in this manner 
employed, for they allow time for the working of the torsion 
balance, and with the exception of bismuth, their temperature 
appears to be steady all the while they are in a granular con- 
dition, before they finally solidify. The action of these me- 
tals on the thermo-electric pair is easily prevented by dipping 
it into a cream of pipe-clay. 
A pair of copper and platina gave for a dull red heat 1416° 
Fahr., and for a bright red 2103° Fahr. 
A pair of palladium and platina gave for a dull red 1850° 
Fahr., and for a bright red 2923° Fahr. 
Some of the combinations into which iron enters as an ele- 
ment, give rise to remarkable results; thus if we project the 
curve given by a system of copper and iron, we shall find it 
resembling fig. 3, where the maximum ordinate h occurs at a 
temperature of about 650° Fahr. ; the point c appears to be 
given between 700 and 800 degrees ; by a dull red heat ; e 
is very nearly the point at which an alloy of equal parts of 
brass and silver melts, for if the pair be soldered with this 
substance, it fuses when the needles have returned almost 
exactly to the zero point. With harder solders or with 
wdres simply twisted, the curve may be traced on the opposite 
side of the axis towards f, its ordinate increasing with regu- 
larity. At 60° Fahr., taking the length of the ordinate cor- 
responding to a temperature of 212° Fahr. as unity, the length 
of the maximum ordinate at h, is 1*85 very nearly. 
A system of silver and iron gives also a similar curve, the 
point h occurring at a temperature rather higher than the 
analogous one for the preceding system, but still below the 
boiling point of mercury. 
Now all these things serve to show, that we cannot deter- 
mine with accuracy unknown temperatures by the aid of 
thermo-electric currents, on the supposition that the incre- 
ments of the quantities of electricity are proportional to the 
