on the Thermoelectrical Properties of Bismuth. 291 



before magnetization and the spot of light was driven off the 

 scale. The light- could be again brought on the scale by 

 putting the adjusting-magnet low down ; but this of course 

 materially diminished the sensibility of the galvanometer, and 

 as the effect to be looked for was likely to be very small this 

 was not desirable. Accordingly the following plan was 

 adopted : — There were two sets of needles in the galvano- 

 meter, connected with each other and the mirror ; round one 

 set was wound a coil of about 6000 ohms resistance, and 

 round the other a coil of between 7 and 8 ohms resistance ; 

 the extremities of these coils were attached to separate ter- 

 minals, and the latter coil was employed to measure the 

 thermoelectrical effect of magnetization. The thermoelec- 

 trical current due to want of homogeneousness in the bismuth 

 was balanced by the current from a Daniell cell sent through 

 the other coil, which was shunted, and through a very large 

 resistance : by altering this resistance the spot of light could 

 easily be brought again to the middle of the scale. Some two 

 hours were allowed to elapse, the steam during the whole 

 of this time passing through the air-chamber, C, and the cold 

 water through the air-chamber, D, after which the spot of 

 light remained steady. The solenoid, 8, was now actuated 

 by a current from six Grove cells, and a deflection ensued 

 indicating a current from unmagnetized to magnetized bismuth 

 through the hot junction. The current through the solenoid 

 was then stopped, and the spot of light returned sensibly to its 

 old position. The observation was repeated ten times, and 

 then the current through the solenoid being reversed, ten 

 other observations were made, after which the current was 

 again reversed. The readings had to be corrected for a slight 

 direct action of the magnetizing solenoid on the needles of 

 the galvanometer. 



The deflection due to the thermoelectrical current between 

 magnetized and unmagnetized bismuth was very small ; but, 

 so far as could be made out, it was the same for both direc- 

 tions of the magnetizing current. The mean of the readings 

 gave a deflection of 3 - 5 divisions of the scale ; and, by inde- 

 pendent observation with a Daniell cell, it had been ascertained 

 that a deflection of 1 division of the scale would, under the 

 conditions of the experiment, indicate an E.M.F. of - 143 

 microvolt. Consequently the E.M.F. produced by tempe- 

 ratures of 100° C. and 16° C. at the two junctions of mag- 

 netized and unmagnetized bismuth would be 0*143 x 3*5 

 microvolts = ^ microvolt. The magnetizing force was 226 

 C.G.S. units ; so that the E.M.F. for unit magnetizing force 

 would be -0022 microvolt, or '22 C.G.S. units of E.M.F. If 



