100 K. HONDA. 



in the most significant case. I believe that the principal causes 

 of discrepancy above ennmerated are sufficient to account for 

 the difference between the calculated and the experimental 

 numbers. 



19. Thus qualitatively the above result and the experiment 

 are in complete agreement with each other, although there are some 

 discrepancies in quantitative details ; there are, however, probable 

 causes to account for the discrepancies. According to Knott, the 

 change of length in cobalt by longitudinal magnetization is very 

 little affected by the presence of a circular field, but the above 

 consideration leads to a result which contradicts his anticipation. 

 Hence a single experiment on this point for cobalt will deci- 

 dedly establish the correctness of the one explanation against 

 that of the other. 



In conclusion, I wish to express my best thanks to Prof. 

 H. Nagaoka, and also to Prof. A. Tanakadate for useful advice 

 and kind guidance. 



