262 Mr. S. Bidwell's Experiments illustrating 



movement of the spot of light occurred. The electromagnet 

 was then excited, the key was depressed, and the spot of light 

 was deflected. The magnet-current was reversed, the key 

 was again depressed, and there was again a deflection of the 

 spot of light. And the directions of the deflections were in 

 both cases opposite to those which occurred before the slits 

 were cut in the plate. The amount of deflection thus pro- 

 duced was about 85 scale-divisions *. 



Now it is quite clear that, if HalFs effect were due to the 

 direct action of the magnet upon the current, it ought to 

 occur in each of the two halves of the divided plate. The 

 transverse current in the lower half w r ould (supposing its 

 direction to be upwards in the figure) pass from the point 

 B to the metallic bridge at the centre, and the transverse 

 current in the upper half would pass from the centre to 

 the point A. Upon no conceivable hypothesis could the 

 mere division of the plate have the effect of actually reversing 

 any direct action which the magnet might exert upon the 

 current. 



The strain theory, however, explains the matter perfectly. 

 Supposing a south pole to be beneath the plate, and the cur- 

 rent through it to be flowing from left to right, then, before 

 the cuts were made, the point A would be in a stretched dis- 

 trict, and the point B in a compressed district (see fig. 3) ; 

 and the resulting Peltier effects would cause the transverse 

 current to flow in the direction A Gr B, fig. 5. But in con- 

 sequence of the longitudinal cuts, the distribution of stress is 

 altered. Instead of the whole plate, each half of it may now 

 be mapped out into six areas which are differently strained, 

 and the point A will now be in a compressed district, and B 

 in a stretched district. The Peltier effects will therefore be 

 reversed, and the transverse current will flow in the direction 

 BGA. The existence of the central bridge does not prevent 

 the new strain, because there is no shearing at points midway 



* The experiment as described is an exceedingly troublesome one, and 

 no attempt was made to exhibit it at the meeting. The difficulty arises 

 from the fact that, in order to show the effect with certainty/ a very 

 perfect balance must be obtained, and this alone generally occupies a 

 great deal of time. But even when two exactly equipotential points 

 have been found, it often happens that the mere reversal of the polarity 

 of the magnet disturbs the adjustment of the galvanometer-contacts to 

 such an extent that the effect cannot be observed : more time must 

 therefore be consumed in readjusting the contacts before a successful 

 experiment can be made. Believing this inconvenience to be due to 

 small displacements resulting from the attractive action of the magnet 

 upon the iron, I have since tried the experiment with gold-foil. This 

 turns out to be a much more suitable metal for the purpose, rendering it 

 possible to observe the effect with ease and certainty. 



