724 PROFESSOR THOMSON ON THE ELECTRO-DYNAMIC QUALITIES OF METALS. 
with it, a portion of the slip next the other end which was left uncoiled, was 
carried out from the outer part of the spiral, and cut to such a length as to let the 
two ends be brought close together. Copper wires, to lead to the galvanometer 
electrodes, were soldered to these ends, and the junctions of dissimilar metals thus 
formed were arranged with the usual precautions (§ 92) to ensure equality of tem- 
perature and electrical insulation. Contrary poles of two steel bars, each about 3 
feet long and of rectangular section, 4 inches by i inch, were placed pressing on each 
side of the spiral, as shown by the dark shading pjg., 32 . 
in the diagram, but insulated from it of course. 
Four rectangular pieces of thick plate glass, two 
of them very hot (perhaps about 300° Cent.) and 
two cold, were applied, touching the coil on each 
side, and symmetrically arranged on the two 
sides of the steel magnets. 
The galvanometer showed a current in the direction indicated by the arrow-heads. 
The pieces of hot and cold plate glass were interchanged, and the current became 
reversed. The magnets were removed, and their effects became scarcely perceptible, 
or altogether ceased. On repeated trials a current was found always in the direc- 
tion, from parts of the coil between the magnets towards parts touched by the hot 
glasses. The experiment was repeated with a powerful electro-magnet, and gave 
the same result, but not with the same ease, because of difficulties in applying the 
heaters, &c. 
137 . The very strong tendency iron has to assume longitudinal rather than trans- 
verse magnetization, when of any form extended in one direction more than in others, 
was partially done away with by the mutual influence of the different turns of the 
spiral used in the experiment which has been described ; and the symmetrical 
arrangement of the heaters was such as to nearly exclude all thermo-electric action, 
except what is due to the thermo-electric difference between that part of the coil 
touched on each side by the steel magnets, and the part diametrically opposite. Any 
thermo-electric effect there may have been from longitudinal magnetization in the 
parts of the iron ribbon on each side of the steel magnets, must, so far as I could 
judge, have been contrary to the effect observed. The result obtained, therefore, 
demonstrates an electro-motive force urging a current from transversely magnetized 
parts of the iron conductor, through hot parts, to comparatively unmagnetized parts. 
Hence a transversely magnetized iron conductor deviates from unmagnetized iron 
towards bismuth, or in the reverse direction to that of the deviation discovered in 
wire longitudinally magnetized, in the first experiment on the thermo-electric effects 
of magnetism. It may be concluded, d fortiori, that in uniformly magnetized iron, 
directions transverse to the lines of magnetization differ thermo-electrically from direc- 
tions along the lines of magnetization and differ in such a way, that if we could get 
