318 YOSHIJIRO KATO ; 



altogether looked upon as experimental errors, because they occur so 

 regularly in relation to time. If forced to give an explanation of it, 

 I would say that it is due to heterogeneity in the distribution of the 

 molecules in the iron mass, some groups of molecules being more 

 stable than others. Heterogeneity would seem to give hindrance to 

 the diffusion of magnetisation. Thus, when the iron is well annealed, 

 homogeneity will be obtained, and in this case magnetisation can diffuse 

 more easily, which accounts for its great susceptibility in low fields 

 and great rate of growth. On the other hand, when the homogeneity is 

 broken by stretching the iron beyond its limit of elasticity, the suscepti- 

 bility and the rate of growth are both reduced. In the latter case, and 

 probably also in steel, the magnetisation will be only superficial. 

 Grotrian is of opinion that an iron cylinder not very short in comparison 

 w 7 ith its diameter, exposed in a moderate magnetic field in its axial 

 direction, is magnetised very unequally, the magnetisation being 

 stronger at the circumference than in the interior. 1 Du Bois objects 

 to this view, as being quite contrary to usual ideas, and explains 

 the experimental results of Grotrian as being the effect of the 

 demagnetising force. 2 The influence of the demagnetising force would, 

 of course, lead to the same result, but from the later experiments 3 

 of Grotrian, it can be seen that this is not the only cause, and 

 that the distribution of magnetisation is really that stated by 

 Grotrian. This view has a certain likeness to the above explana- 

 tion of the time-lag. But, after all, there is no certain ground, 

 theoretical or experimental, for presenting the above explanation as 

 representing the true nature of the process of magnetisation. I mean 

 only this : that in the absence of any definite theory about the pheno- 



1 Wied. Ann., 50, p. 703. (1893). 



2 Wied. Ann., 51, p. 529. (1894). 



3 Wied. Ann., 52, p. 733. (1894). 



