Intelligence and Miscellaneous Articles. 227 



and the results I have now obtained agree in general with those 

 of Mr. JNagaoka. Some discrepancy between our observations may 

 be explained as due to a divergence in the conditions of experi- 

 ment *. The method I used consisted in the application of inter- 

 ference-fringes and sensitive lever. Its sensitiveness was nearly 

 the same as that in the method of Mr. Nagaoka: one division 

 of the microscope scale corresponding to a change of length 

 = llxlO~ 7 cm. Influence of temperature effects was compen- 

 sated by a peculiar bimetallic suspension of the sensitive lever. 

 The results I have obtained are given in the following Table, 



where H represents magnetic force (C.G-.S. units), y elongation 

 per unit of length of my iron wire t, and I its magnetization. 



[(C.G.S.). 



yXlO 9 . 



I(C.G.£ 



674 



-137-1 



1495 



50-5 



-107-0 



1495 



337 



- 62-2 



1448 



253 



- 28-6 



1395 



16-9 



- 1-7 



1283 



13-5 



4- 13-6 



1206 



8-4 



4- 24-3 



1142 



51 



+ 295 



1065 



34 



+ 29-5 



1012 



1-7 



4- 266 



965 







4-217 



920 



- 1-7 



4- 139 



683 



- 34 



4- 3-8 



312 



- 5-1 



4- 7-8 



- 418 



- 67 



4- 12-2 



- 559 



-10-1 



4- 15 1 



- 900 



-13-5 



4- 11-3 



-1060 



-169 



4- 4-1 



-1107 



-253 



- 21-4 



-1219 



-337 



- 45- 1 



-1272 



-50-5 



- 88-8 



-1378 



-67-4 



-124-7 



-1454 



Theoretical views permit us to suppose the following correlation 



11 



between -=- and corresponding values of H and I — 



y=aI 2 +6IH + cH 2 (1) 



My friend Mr. Weinberg has kindly calculated the coefficients 

 of this formula by the method of least squares. These are as 

 follows : — 



a=4-4834 x 10" 14 , b= -365-023 x 10" 14 , c=3015-31 x 10" 14 . 



* The iron bars of Mr. Nagaoka were comparatively thick and short, 

 the ratio of their length to their diameter was not greater than 70, whereas 

 my wire had for this ratio about 494. This wire was moreover stretched. 



t Mean of several sets of observations. 



