Wiedemann Effect in Ferromagnetic Substances. 657 



Tlie general feature of the twist for cobalt is similar to that 

 in nickel. In cast cobalt (fig. 15) the twist is increased first 

 slowly and then rapidly as the circular field is increased. 

 With the increase of the longitudinal field the twist reaches 

 a maximum and then gradually diminishes. If the field be 

 strong enough the twist occurs at first in the opposite direc- 

 tion and then in the ordinary. In annealed cobalt (fig. 16) 

 the twist is very small and the rate of increase is nearly con- 

 stant. Here also the twist obtained by first applying the 

 circular field and then the longitudinal is several times 

 greater than the twist when the order of application is 

 reversed. 



The effect of torque. — To study the effect of! torque, it is 

 convenient to keep the longitudinal field constant and to vary 

 the circular field ; for, though the application of the longi- 

 tudinal field is always accompanied by the twist due to the 

 change of rigidity, the passage of a longitudinal current does 

 not cause any appreciable twist ; hence, by varying the cir- 

 cular field, it is not necessary to apply the correction due to 

 the change of rigidity. The torque was given by means of 

 the suspended weight, as in the experiment on the change of 

 rigidity by magnetization. Keeping the longitudinal field 

 constant, we found that in all cases the effect of torque is to 

 diminish the twist by an amount which is nearly proportional 

 to the torque. Figs. 17, 18, 19, k 20 show the general 

 feature of the decrease of twist due to torque. In soft iron 

 and annealed cobalt the effect is very small, but in nickel and 

 cast cobalt it is considerable. 



In a paper* on the mutual relation between torsion and 

 magnetism Prof. Nagaoka and one of us have obtained from 

 Kirchhoff 's theory the result that for given longitudinal cur- 

 rent and field, the amount of twist is inversely proportional 

 to the square of the radius of the ferromagnetic wire. It is 

 interesting to notice that the comparison of the above results 

 in iron and nickel bars 1 cm. thick, obtained by varying the 

 longitudinal field, with the corresponding results in wires of 

 these metals about 1 mm. thick, shows the correctness of the 

 law of the inverse square of the radius. 



In conclusion, we wish to express our best thanks to 

 Prof. Nagaoka for useful suggestions in the course of the 

 present experiment. 



* Nagaoka and Honda. .Jour. Coll. Sci. xiii. p. 276; Pliil. Mag. ivj 



p. 68 (1902). 



