122 Mr. H. Nagaoka on the Efects of Torsion and 



by sucli twisting and untwisting a curve of the form given in 

 PI. II. fig. 5, since the behaviour of nickel with regard to the 

 effect of stress in magnetization seems to be just opposite to 

 that of iron (see Sir W. Thomson's figures for iron, Philoso- 

 phical Transactions, 1879). But in this experiment the 

 curve of magnetization seems to have no resemblance at all to 

 any figured by Sir W. Thomson ; and then there is the very 

 curious fact that the magnetization of nickel in a steady field 

 can be made to change sign by twisting. It first occurred to 

 me that this very extraordinary result must be due to some 

 defect in the arrangement, but careful examination discovered 

 no flaw. The question naturally suggests itself. Was this phe- 

 nomenon a function of the load as well as of the twist ? Hence, 

 as a next step, the weight was increased by '5 kilog., and the 

 experiment was performed in the usual manner. 



The result is shown in fig. 2. Here the effect is quite 

 similar to the former, the differences being only differences 

 of detail. The march of magnetization with positive twisting 

 is sensibly the same as in the former case ; but during 

 negative twisting the opposite magnetization has increased to 

 more than ten times its amount in the first experiment. 

 However, the rate of recovery has very much diminished, 

 and even after a twist of 180° the magnetization is far from 

 reaching the former value. 



The same experiment was then performed with the load 

 increased to 5*l4 kilogs. The result is shown in fig. 8. 

 There again we find the opposite magnetization still more 

 increased. Indeed, the two kinds of magnetization do not 

 differ much in intensity at the two extreme twists, although 

 the initial or positive magnetization is somev/hat predominant. 

 In the two 'former experiments there was a distinct tendency 

 towards recovery of positive magnetization as the wire was 

 twisted more and more in the negative direction. Here, 

 however, no such tendency shows itself except in the diminished 

 rate of growth of negative magnetization. 



Still increasing the load, we see that the curve (fig. 4) 

 becomes nearly symmetrical with respect both to the line of 

 zero magnetization, and the line of zero twisting. In this 

 case we find further that the range of the change of magneti- 

 zation has considerably diminished. The slight excess of the 

 initial magnetization over the other still shows itself, a fact 

 which is probably to be referred to the direction of the mag- 

 netizing force. 



The general conclusion from these experiments is that in a 

 weak field of "34 units, the increase of load makes the manner 

 of change of magnetization in nickel under the influence of 



