528 PROFESSOR KNOTT ON SOME RELATIONS BETWEEN 



This was established in the earlier experiments carried out by Mr Imagawa. These same 

 features, namely, the change of sign of the cyclic area and its high negative value for 

 large twistings, are also characteristic of the phenomena which accompany the cyclic 

 twisting of longitudinally magnetised wire. As mentioned above (sect. 13), we owe to 

 Mr Nagaoka the complete working out of this line of experiment, — complete, excepting 

 in one point. He did not investigate the effect of tapping, which we may, however, 

 infer to be very similar to its effect in the experiments on circular magnetisation. The 

 following general conclusions seem to be applicable to both lines of experiment. 



It is important to bear in mind the essential difference between experiments in which 

 mechanical straining is the cause of magnetic change, and those in which magnetising 

 force is the cause. Take, for example, the well-known case so frequently discussed, of 

 a wire subjected to a continuously varying longitudinal field, and, to fix our ideas, 

 consider more especially the behaviour of the wire in a diminishing field. Here there is 

 true hysteresis ; true magnetic lagging as the magnetic force is in part removed. But 

 when the wire is twisted through a large angle, the operation of untwisting the wire is 

 no mere removal of twist, but is really a superposition of an opposite twist. Now we 

 know that the induced magnetism due to a given field is apparently destroyed by a 

 reversed field of smaller value. It is this effect, as much as the effect of mere diminution 

 of the field to zero, that is to be compared to the effect of untwisting, especially if it is 

 untwisting from a large twist. If the twist, however, is small, — that is, not greatly 

 beyond the limits of torsional elasticity, — the untwisting will be aided by the elasticity of 

 the wire, so that it will have something of the character of a mere undoing. From this 

 point of view, then, small twistings and untwistings will be to a certain extent compar- 

 able to applying and removing magnetising force ; while large twistings and untwistings 

 are to be compared rather to applying first a given magnetising force, then removing it, 

 and then applying a reversed magnetising force. Thus for small cyclic twistings there 

 is true magnetic lagging, while for large cyclic twistings there is magnetic " priming." 

 The fact that the limits of torsional elasticity for iron are much higher than the same 

 for nickel fits in admirably with the result that the critical twist at which lagging 

 changes sign is much higher for iron than it is for nickel. 



The effect of tapping in always diminishing the positive lagging, reversing it indeed 

 under certain circumstances, and increasing the negative lagging, seems to indicate that 

 the tendency of cyclic twisting is to produce a negative lagging, modified however by 

 the property of magnetic retentiveness. In the case of a twisting cycle with steady 

 current, tapping may reverse the sign of the graph area (see sect. 19) ; but it can only 

 diminish the positive area in the case of a current cycle performed on a twisted wire (see 

 sect. 23, fig. 5, PI. V.). In the comparison of these two cases, we have the difference 

 between them strongly accentuated. The changes of molecular configuration which 

 result from these operations, although they are accompanied, broadly speaking, with 

 very similar total changes of magnetic intensity, can be in few other respects compar- 

 able. Here, in fact, wc have an experimental demonstration of the well-known theorem 



