96 K. HONDA. 



lu the case of nickel, the change of volume is negligibly small 

 compared with that of length ; hence we may put with tolerable 

 accuracy f(H) + 2F{H)=^0. With steel and soft iron, the 

 change of volume is not very small comj)ared with the change of 

 length. But if t does not exceed 50 C. G. S. units, the effect of 

 volume change on the change of length by combined action of 

 I and t is negligibly small, for in these strong fields at which the 

 change of volume is pronounced, the ratio f/H'- in the above 

 expression becomes very small. Hence even for these metals, 

 we may neglect the change of volume, provided the circular 



> 7- 



field is not very large, and the expression for --^ becomes, 

 in all cases, 



L IP 



Since the material is supposed to be isotropic, /{H) is the same 

 as the ordinary change of length by longitudinal magnetization. 

 Thus the change of length by longitudinal magnetization with 

 a constant circular field can be calculated from the change of 

 length by longitudinal magnetization aloue. The same expres- 

 sion can also be used for the calculation of the change of 

 length due to circular magnetization with a constant longitu- 

 dinal field. 



In order to compare the above result with that of the 

 expeiiment, it is obviously necessary to subtract from -^^ the ex- 

 pression F{i) for the change of length by longitudinal magne- 

 tization with a constant circular field t, aud/(7) for the reciprocal 

 case. 



Assuming for the expression /(//) a suitable empirical for- 

 mula for iron, steel or nickel, a simple analytical discussion of 



