460 Dr. Draper on the lEilectro-motive Power of Heat. 
Hence there cannot be any doubt, that the quantities of 
electricity evolved by compound batteries, at the same tem- 
perature, are directly proportional to the number of the pairs. 
With some general remarks, arising from the foregoing 
subjects, I shall conclude this communication. 
1. It is of importance to remember, that thermo-electric 
currents traverse metallic masses only on account of differences 
of temperature existing at different points. 
2. When a current of electricity, flowing from the poles of 
a battery, is made to traverse a metallic sheet, the whole of 
it does not pass in a straight line from one pole to the other, 
but diffuses itself through the metal, diverging from one point 
and converging to the other. The greater part of the cur- 
rent is found, however, to take the shortest route. 
3. Combining therefore the foregoing observations (1.2), 
we perceive, that there are certain forms of construction 
which will give to thermo-electric arrangements peculiar ad- 
vantages. For example, the surfaces united by soldering must 
not be too massive. Let a, fig. 5, be a bar of antimony, and b 
a bar of bismuth; let them be soldered together along the line 
c c?, and at the point d let the temperature be raised ; a current 
is immediately excited; but this does not pass around the bars 
«, b, in as much as it finds a shorter and readier channel 
through the metals, between c and c?, circulating therefore as 
indicated by the arrows. Nor will the whole current pass 
round the bars, until the temperature of the soldered surface 
has become uniform. 
An obvious improvement in such a combination is shown 
in fig. 6, which consists of the former arrangement, cut out 
along the dotted lines: here the whole current so soon as it 
exists is forced to pass along the bars. And because the 
mass of metal has been diminished at the line of junction, 
such a pair will change its temperature very quickly. 
One of the very best forms for a thermo-electric couple is 
given in fig. 7, where « is a semicylindrical bar of antimony, 
b one of bismuth, united together by the opposite corners of a 
lozenge-shaped piece of copper c. From its exposing so much 
surface, the copper becomes hot and cold with the greatest 
promptitude, and from its good conducting power it may be 
made very thin without injury to the current. With a pair 
of bars fths of an inch thick, and a circular copper plate c, 
having both surfaces blackened, I have repeated the greater 
part of those experiments which M. Melloni made with his 
multiplier. 
4. The currents which circulate in a steel magnet are to 
