of Electro-Magnetism by Heat. 55 
an increase in either of these dimensions was not attended with 
an equivalent increase of power. A cylindrical rod of bismuth, 
9 inches long by 3 diameter was rather more powerfal than another 
of 25 inches by 2; but, between this, and a thin plate 43 inches 
long by 1 broad and } thick, the difference was scarcely perceptible. 
By increasing the surfaces in contact, or rather, as it afterwards 
appeared, by adding to the conducting surface of the connecting 
wire, there was a slight addition to the effect produced. The 
cylinder ef bismuth of 25 inches by 2, having a plate of copper 
of the same diameter soldered upon it, with four connecting wires, 
(Table III.) was equally, if not more powerful, than the longer 
cylinder, and on the whole, this seems to be the best form ; yet 
the gain of power, by increasing either the diameters of the bars or 
of the surfaces in contact, is not such as to promise any advantages 
by the use of large metallic bars, analogous to those obtained by 
the employment of large plates in the galvanic apparatus. When 
two metallic rods in connexion, were heated at the same time, there 
was some accumulation of power; which appeared to be greater 
when they were in sequence than when the wires from both the 
heated ends were placed in one cup of the galvanoscope, and the 
wires from the cooler ends in the other. (Fig. 8 and9). A bar AB 
which gave a deviation of 16°, when placed in sequence, as Fig. 8, 
with another CD, whose deviation was 21°, gave a deviation of 
25°, but when connected, as in Fig. 9, gave only 23°. 
As in this experiment there was some, though not a consi- 
derable increase of power, a battery was formed of eight plates, 
four of antimony, and four of bismuth, placed alternately, and 
connected in sequence, (Fig. 10.). This (Table IV.) at the mean 
temperatures of 175° and 100°, gave 17° and 73° of deviation ; 
a single plate of bismuth, at the same temperatures, gave the 
deviations 73° and 23°, and one of antimony gave 67° and 2°; the 
effect therefore of the eight plates is but little more than double 
