318 H. XAGAOKA 



consequently, there exists a twist for which tlie luaximum slope of the 

 percentage curve attains a greatest vahie. Tlic rate at which the resi- 

 dual niagneti.'<in falls oft' in higher fields is diminished as the twist 

 is increased. 



The next series of experiments related to the comhined effect of 

 torsional and longitudinal stresses. For this jjurpose, a new wire prop- 

 erly prejjarod was loaded with 1.14 kgs. The normal eur\e (Fig. fi) 

 show a slight decrease of retentiveness ; and the curve obtained for 

 the twist of 0°.5 per cm. does not show those curious character- 

 istics of tlu' twisted nickel. The retentiveness, however, is greatly 

 increased, and at its maximum, the residual magnetism is .85 of the 

 induced. lUit when the twist is increased to 1°.5, all the curious 

 properties of twisted nickel already mentioned become apparent and 

 confirm the results already ol)tained for the unloaded wire. In this 

 case the nuiximum ratio of the residual to the induced magnetism 

 attains the enormous value of .985. The greatest maximum slope in 

 the ratio ciu'ves again occurs for the value of twist for which the 

 " Wendepunkt " occurs sooner. This is f^r a twist of nearly 3° per cm. 

 As may be seen from the experiments on the relation of 3 find !q , 

 this is the twist which gives the maximum differential susceptibility. 

 Hence it would appear that the twist of about 3° per cm. has a certain 

 critical significance in the relations of twist to magnetization for the 

 specimens of nickel wire used. 



When the load is increased to 3.14 kgs., the maximum ratio of 

 residual to temporary magnetism diminishes to .97, but the general 

 characteristics still remain the same. 



The experiments made on the wire which was loaded with a 

 weio'ht of 8.14 ko-s. also brings out clearly the effect of twist on the 

 retentiveness of nickel. The residua] magnetism for no twist is indeed 

 very small with surh a large load ; but merely twisting the wire 



