530 PROFESSOR KNOTT ON SOME RELATIONS BETWEEN 



should expect, takes place through a much diminished range. In one case, for instance, 

 the range of polarity in a nickel wire twisted through an angle of ±100° fell from 321 

 when a current of 2*5 amperes was flowing along it to 120 after six complete twistings 

 with the current off. The average polarity in the final cycle was slightly negative 

 although the current had been broken when the polarity was at its maximum positive 

 value. This persistence of the effect with residual magnetism is more marked in nickel 

 than in iron, as indeed are all the changes accompanying twisting in wires whether 

 magnetised circularly or longitudinally. The magnitude of the effect is, however, much 

 more striking in the experiments now under discussion than in those in which the 

 magnetising force is longitudinal. 



The explanation of this is probably to be looked for in the following general 

 considerations. 



27. In the first place, the whole phenomenon of hysteresis is an illustration of 

 Maxwell's general theory of viscosity in solid bodies, and depends upon the manner in 

 which molecular groups assume new configurations under the influence of varying forces. 

 Ewing's recent experiments on groups of little magnets form a very beautiful practical 

 corroboration of this view, which I believe was first stated distinctly by Barus. Now, 

 although we do not understand the mechanical connection between magnetic force and 

 electric currents, we see that they are so associated that electric currents never exist 

 without magnetic force, and that the distribution of the latter can (within certain limits) 

 be inferred from the former. It is usual, accordingly, to speak of the magnetic force 

 due to a current ; but such a view is probably incorrect, or at all events incomplete. 



The tendency of modern thought is to regard magnetic energy as transmitted in 

 some way through a medium, while the electric currents which in many cases 

 accompany this transmission are evidences of dissipation of electric (or magnetic) 

 energy. We see at once, then, that longitudinal magnetisation of a metal by means of 

 a current in a neighbouring conductor is a very different thing from circular magnetisa- 

 tion by a current through itself. In the one case, the dissipation of electric and 

 magnetic energy takes place (nearly) altogether outside of it ; in the other case it takes 

 place (nearly) whoDy within it. From the ordinary though incomplete point of view, 

 we may put it in this way. In cases of ordinary induced magnetisation, the magnetic 

 force must penetrate into the metal from the outside ; and many well-known phenomena 

 show that the magnetic metals have a screening magnetic influence. It is highly 

 probable that the thinnest of iron wires in a magnetising field is not magnetised 

 uniformly across its section. On the other hand, the magnetic action of a current along 

 an iron wire is quite different. Here any internal portion is, on the ordinary theory, 

 subject to the direct magnetic action of all current elements lying nearer to the axis 

 than itself. Under any twisting action, then, it is conceivable that the inner portions of 

 the wire will have a proportionately greater effect when the wire is under the domina- 

 tion of a current passing through it, than when it is being magnetised by a current 

 altogether outside of it. In our ignorance of the internal distribution in a magnetised 



