Electromagnetic Stress. 267 



Fig. 3 represents the magnetization -curve and the de- 

 magnetizing force of the ellipsoid. 



Each observation was repeated several times, and the mean 

 value taken as the true value. The weights, however, never 

 differed by more than 1 or 2 per cent, for the same magne- 

 tizing current. 



It appears therefore that the present method of measure- 

 ment has great advantages over the ballistic method used 

 by Bosanquet and Threlfall, for Bosanquet obtained some- 

 times for the same induction weights differing by nearly 20 

 per cent. 



The apparatus as above described (or any apparatus of the 

 S. P. Thompson's " permeameter " * type) may therefore be 

 used for the accurate measurement of magnetic induction in 

 uniformly magnetized bars, the essential conditions being that 

 the contact surfaces are plane, that the upper bar can be 

 finely adjusted in position and very rigidly fixed, and that 

 the contact surfaces are as nearly as possible flush with 

 the ends of the magnetizing-coils, space being left for 

 the small guiding-ring on one of the bars. The lower bar 

 must also be guided, and for this it is sufficient that its 

 lower end pass freely through a ring properly adjusted in 

 position. 



A few experiments were made with the half-cores separated 

 by very thin sheets of silver. Two sheets were used, of thick- 

 nesses 1*5 and 8 hundredths millim.- respectively, and it was 

 found that the introduction of a sheet always caused a diminu- 

 tion of both magnetization and tractive force, greater with the 

 thicker sheet. No increase similar to that found by Wassmuthf 

 at low inductions was observed. 



In conclusion. I wish to express my obligations to the 

 late Prof. Kundt and Drs. du Bois and Rubens for the 

 interest they took in my work, and the help they gave me 

 in various ways. 



Berlin Phvs. Inst. d. Univ., 

 Jan. 1895. 



* Journ. Soc. Arts, Sept. 12, 1690, 

 t L. c. p. 336. 



