In. Nickel and Cobalt 161 



in 



the edge it would perhaps be two or three per cent. The 

 average of the real values of 35, therefore, at points along 

 the line running across the strip from one side connexion 

 to the other, is probably a rather small fraction of one per cent, 

 greater than the value obtained on the assumption that 35 is 

 equal to %. This error is, to be sure, not constant; but it is 

 nearly so up to 55= about 5000; and when it begins to change 

 rapidly, it grows smaller. 



Its influence upon the curve given further on must be very 

 small. We assume therefore, as stated above, that by deter- 

 mining the strength of the magnetic field by means of the coil 

 and galvanometer before the nickel is placed in the field, we 

 ascertain with sufficient accuracy the value of the magnetic 

 induction in the nickel strip itself when placed in the mag- 

 netic field. The advantage of determining this quantity is of 

 course very great ; for though we are probably unable to say 

 what is the exact physical nature of magnetic induction, we 

 do attach to the quantity represented by that term a very 

 definite and important mathematical significance. 



It was designed, therefore, to investigate the law of the 

 variation of the transverse effect with the variation of the 

 magnetic induction. Nickel was the best metal to experiment 

 upon, for the following reasons: the strip of this metal at 

 hand was very thin ; the transverse effect appears to be essen- 

 tially more powerful in nickel than in iron or cobalt ; the 

 magnetic permeability of nickel changes more rapidly than 

 that of iron or cobalt with high magnetizing-powers. 



As it was desired to determine simply what function of the 

 magnetization the transverse effect would prove to be, the pri- 

 mary current through the nickel strip has been kept approxi- 

 mately constant, the greatest variation from the mean being 

 probably not many per cent., as will be shown further on. 

 Within these limits it has been assumed that the transverse 

 effect may be considered a linear function of the direct current. 



It should be here stated that this latter relation has not yet 

 been proved to hold rigidly even in a non-magnetic conductor 

 like gold ; and the matter must some time be investigated, 

 though there seems to be no reason to think that the assump- 

 tion, as above limited, can prove to have involved any consider- 

 able error. 



The intensity of the magnetic field, and so the magnetic 

 induction in the nickel plate, has been varied from about 1600 

 to about 10,000 in absolute (cm. -grm.-sec.) measure. 



In the course of this investigation I have become indebted 

 to nearly every one connected with the Physical Laboratory 

 of the Johns Hopkins University, but particularly to Mr. S. 



