Eqidpotential Lines of an Electric Current 288 



direct current through the cross or by reversing the facing 

 direction of the plate between the poles. 



On Dec. 31, 1887, observing the dissymmetry, which I did 

 not remember reading of, to be very great in a field of about 

 19000, I cut off the direct current through the cross, and leav- 

 ing the transverse circuit closed, tried the effect of the magnet 

 upon the cross in this condition. When the magnet current 

 ran in one direction, it had little, if any, effect upon the. gal- 

 vanometer in the transverse circuit ; but when it ran in the 

 other direction, it produced a deflection more than one-half 

 as large as the dissymmetry which had been observed and in 

 the ri^ht direction to account for it. When the transverse 

 circuit was broken, the effect almost entirely disappeared, 

 showing that the direct action which the electro-magnet ex- 

 erted upon the galvanometer was very small. The poles of 

 the magnet were about 3 mm apart in this case. The plate bear- 

 ing the bismuth was not pinched between them, but it became 

 warm enough to soften the cement to a certain degree. On 

 Jan. 4th, with the poles about 7 mm apart, the magnet had a 

 slight effect upon the transverse circuit, the direct circuit being 

 broken, the change of deflection produced by reversal being in 

 the same direction as on Dec. 31. Whether this small effect 

 was unsymmetrical was not observed. This was after magnet 

 and cross had been in use. At the beginning of operations on 

 this day, while the poles were about 4 mm apart, there was no 

 certain effect of this sort. I have no record of further observa- 

 tions upon this point, which certainly demands investigation. 

 The deflections observed may possibly have been merely acci- 

 dental, but I do not think they were so. In the June experi- 

 ments the poles were always about T mm apart and wads of 

 cotton were placed between the plate and the poles on either 

 side. 



Ettingshausen and Nernst state that the R. P. of u bismuth 

 sinks almost to J. .... , when M increases from 1000 to 

 16000," and that in a very strong field the transverse effect 

 actually diminishes when the field is made stronger. The P. 

 P. of the specimen which I have used diminishes greatly as 

 M increases, but in no case under my observation has an in- 

 crease of M failed to produce an increase of the transverse 

 current, due allowance being made for casual discrepancies. 

 In the following table the column Tr. gives the relative 

 values of the transverse effects obtained with the respective 

 values of M. The results of Dec. 31, 1887 and Jan. 4, 1888 

 have been reduced, as nearly as may be, to the scale of the 

 June results by making all agree when M is about 10500. 

 The column following Tr. shows roughly the dissymmetry 

 which has been described. The column headed P gives the 



