

308 Mr. E. II. Hall on the new Action of 



that, so far as observation has gone, the relative direction of 

 the transverse current is always the same for any particular 

 metal. This uniformity in so many cases could hardly be 

 accidental. 



This matter of direction is evidently one of fundamental im- 

 portance. The direction was found to be the same for silver 

 as for gold, these being the two metals first examined. Pro- 

 fessor Rowland, however, predicted that the direction would 

 be reversed in iron ; and experiment verified prediction. Pro- 

 fessor Rowland's comments upon the significance of this dis- 

 covery are already before the public*. It is a seemingly 

 awkward fact that in nickel, next to iron and cobalt the most 

 strongly magnetic substance, the direction of the transverse 

 Ncurrent is the same as in gold. This fact will be discussed 

 further on. The conductors which have up to this date been 

 subjected to experiment are gold, silver, iron, tin, nickel, and 

 platinum. The direction is the same in all except iron. 



The extreme irregularity in the results obtained in the early 

 part of this course of experiments was due to various causes, 

 only one of which is worth mentioning here. This source of 

 error was the shape of the magnet-poles, which, being intended 

 for the study of the magnetic rotation of polarized light, were 

 perforated axially by a hole several millims. in diameter. With 

 these poles the magnetic force was found to vary many per 

 cent, in different parts of the field. These poles were subse- 

 quently replaced by solid ones ; and a sufficiently uniform field 

 was thus secured. It will, however, be noticed that even after 

 this change the results obtained on the same day and with the 

 same plate often vary by several per cent. Probably quite a 

 part of this irregularity was due to the faulty manner in which 

 the tangent-galvanometer, which measured the strength of the 

 primary current through the strip, was introduced. This source 

 of error can probably be avoided in future measurements. 

 Again, it is to be remembered that the strength of the trans- 

 verse current was determined by a delicate Thomson galvano- 

 meter, an instrument far more sensitive than accurate. In 

 using comparatively thick strips of metal there is especial 

 liability to error from this source, as a low-resistance galva- 

 nometer must then be employed, which may easily change in 

 sensitiveness several per cent, within an hour. 



Much of the disagreement to be observed in the results ob- 

 tained with different plates of the same metal is no doubt to 

 be explained by the difficulty of determining, with any thing 

 like accuracy, the thickness of the various strips employed. I 

 have tried to determine approximately the thickness of the thin- 

 * Amer, Journ. Math. vol. ii. p. 35-5. 



