518 Intelligence and Miscellaneous Articles. 



large pattern) astaticized as completely as possible, and whose coils 

 were of greatest sensitiveness for thermoelectric currents. The 

 deflections were read with a telescope on a scale a metre in length 

 divided in millimetres, and arranged at a distance of about two 

 metres from the mirror of the galvanometer. 



The current due to the difference between the electromotive force 

 of the bismuth-copper couple and the compensating-couple was 

 so weak as to produce a deflection of a few centimetres only in 

 the scale of the telescope ; but either of the batteries would have 

 sent the spot of light off the scale. 



The galvanometer was at a distance of about twenty metres from 

 the electromagnet. 



By means of two keys, one in the thermoelectric circuit A and 

 one in the magnetizing current B, arranged near the telescope, 

 either of the two circuits could be opened or closed. 



Opening first the circuit A, I investigated the influence of the 

 electromagnet on the galvanometer, which, notwithstanding the 

 distance of the two apparatus, was sufficiently apparent, and then 

 tried the position at which the least action was exerted. It was 

 thus found that the deflection of the needle when the electromagnet 

 was excited was only two millim. It remained constant for the 

 entire duration of the experiment. This was made as follows. 



The galvanometer being at zero, the circuit A was first closed ; 

 then the circuit B was opened, and the deflection of the galvano- 

 meter read off (first reading). Then A was opened to avoid 

 induction-currents, which, notwithstanding the most minute pre- 

 cautions, were sufficiently apparent ; then B was closed, and imme- 

 diately afterwards A, to make the second reading ; then A being 

 opened, and afterwards B, A was again closed, and the third read- 

 ing made. Lastly, it was ascertained if the needle returned to zero 

 of the scale, those experiments being rejected in which this was 

 not the case. The difference between the second reading and 

 the mean of the first and third, which generally were but little 

 different, corrected for the above-mentioned influence of the elec- 

 tromagnet on the galvanometer, measured the intensity of the 

 phenomenon I investigated. Many measurements were lost owing 

 to the action on the galvanometer of carriages passing in the 

 neighbouring street, an influence which made itself more felt when 

 the galvanometer was arranged in a room of the laboratory, than 

 when it was on the first floor in the courtyard. 



With a magnetizing current of twelve Bunsen's elements, freshly 

 filled, the following result was obtained : — The thermoelectric force 

 of bismuth with respect to copper was considerably iveakened by mag- 

 netism ; the difference between the second reading and the mean of 

 the first and third, which, as mentioned above, measured the inten- 

 sity of the said diminution, amounted ultimately to 45 milhm. of 

 the scale. 



In order to be certain that the phenomenon observed was solely 

 due to the influence in question, and was not caused by any dis- 

 turbing influences, I made the following experiment. 



Adjusting both the solderings of the bismuth-copper couple to 



