G54 Messrs. K. Honda and S. Shimizu on the 



gradually less as the field is increased. The diminution is 

 approximately proportional to the applied tension. 



To test the effect of heavy loading, thin wires about £ mm. 

 thick were examined. Even by a tension at which contrac- 

 tion occurs by magnetization, the direction o£ twist in nickel 

 steels is not reversed, though the amount of the maximum 

 twist is reduced to about ^ or i its value corresponding to no 

 tension, as ^een from figs. 5 & 6. 



Whichever theory we adopt, whether Maxwell's or 

 Kirchhoff 's, the direction of twist is principally determined 

 by the sign of the quantity 3\— a, where X and a are respec- 

 tively the length- and the volume-change of the ferromag- 

 netics. When there is no tension acting on the wire, the 

 sign of BX.— a must be positive, because the direction of twist 

 in the alloy is the same as that of iron. By applying a heavy 

 load the magnetization is accompanied by contraction, so that 

 X is negative. Hence, in order that ?>\ — a should be posi- 

 tive, a must necessarily be negative under a heavy load ; 

 that is, the change of volume by magnetization must 

 change its sign from positive to negative as the load is 

 increased. 



Twist by varying the circular field. — In figs. 7 & 8 we 

 notice that under a constant longitudinal field the angle of 

 twist at first increases at a constant rate, but later at a 

 gradually diminishing rate. As the longitudinal field is in- 

 creased the curves approximate to right lines, a result which 

 is to be expected from Kirchhoff 's theory of magnetostriction. 

 For, according to the theory, if the circular field is small 

 compared with the longitudinal field, the amount of twist for 

 a given longitudinal field is proportional to the longitudinal 

 current. The amount of twist is greater in 45 per cent, 

 nickel steel than in 35 per cent, nickel steel. 



By figs. 7 & 8 we can obtain the twist under a given 

 longitudinal current by gradually increasing the longitudinal 

 field ; the result so obtained, if it is compared with figs. ] & 

 2, shows that the twist produced by the interaction of the 

 circular and longitudinal fields is independent of the order of 

 applying them. 



The apparatus for studying the Wiedemann effect in ferro- 

 magnetic bars was that used in the experiment on the change 

 of rigidity by magnetization ; the longitudinal current was 

 led to the bar by means of mercury contact without causing 

 sensible resistance. The ferromagnetic bar was soldered at 

 both ends to brass bars of thicker diameter, as in the former 

 experiment just referred to. It was fixed by means of the 



