MAGNETISM AND TWIST IN IRON AND NICKEL. 533 



inner layer of molecules. In short, if we imagine the existence of an inner and outer 

 layer of oppositely polarised groupings, we see at once that they will be separated by a 

 neutral layer, and will have a mutually sustaining action. 



Let us suppose such a distribution of magnetic polarity to be self-annulling as 

 regards external action, and consider the probable effect of a twist applied to the wire. 

 The wire is, of course, most strained towards its surface ; and where the greatest strain 

 exists, there presumably will the magnetic changes set in most quickly. The outer layer 

 of polarised molecules will be more sensitive than the inner layer, and the magnetic 

 balance will be destroyed. 



The effect of the first twist in the experiment given above is almost startling in its 

 magnitude. A left-handed twist of barely 1° per centimetre length suffices to bring into 

 view an average negative polarity a quarter greater than the original positive polarity 

 which the current annulled. Now we know from Wiedemann's experiments that the 

 first effect of twist is to increase the magnetic moment of a magnetised nickel wire. In 

 the present case the effect of the twist is clearly to increase the magnetic moment of 

 what we have supposed to be the negatively polarised outer layer. Thus the hypothesis 

 fits in admirably with known facts of experiment. 



On untwisting the wire and applying a right-handed twist of the same amount, we 

 get the negative polarity to increase to nearly four times the original compensated positive 

 polarity. This imposed polarity persists in spite of the removal of the positively 

 magnetising force, and in spite of tappings and twistings. 



By suitable alternations of currents combined with tapping the wire is brought a 

 second time to a neutral condition. The current required for this purpose is still 

 negative, but only one-fourth of the current that was required during the first 

 experiment. 



A left-handed twist of the same amount as before imposes on the wire a positive 

 polarity ; but it was noticed, as the twist was being applied, that the deflection rose first 

 to 650 and then fell, passed through zero, and finally settled at 535. Here is evidence 

 apparently of the sub-surface layers of positively polarised groupings coming more 

 strongly into play as the twist grows. The very first effect, however, is that due to the 

 negatively polarised surface layer under the influence of the magnetising current ; while 

 over all there is a preponderance of the negative polarity. 



In the third experiment the magnetic distribution in the wire must be very compli- 

 cated indeed. There is an accumulation of the after-effects of a succession of different 

 magnetising forces and of many to-and-fro twistings. The wire is in a practically 

 neutral condition, although it is resting in an uncompensated horizontal field ; and yet it 

 has at no time been subjected to an extra magnetising force greater than that originally 

 applied to compensate the field. 



A very minute twisting of less than 7 minutes of arc on either side of its normal 

 position produces a very distinct rise and fall in the magnetic moment. 



Other experiments of a similar character were tried, with various modifications 



VOL. XXXVI. PART II. (NO. 18). 4 K 



