MAGNETISM AND TWIST IN IRON AND NICKEL. 529 



that external magnetic action can lead to no certain knowledge of internal magnetic 

 distribution. 



26. Although it seems to have escaped distinct notice by other experimenters, there 

 is something very striking in the magnitude of the magnetic changes produced by twist- 

 ing. Thus, in the three thinnest nickel wires experimented on, a cycle twisting of ± 90° 

 in association with a current of 1 ampere along the wire produced a range of intensity 

 of fully 400. We may suppose half this quantity, or 2O0, to be the intensity due to a 

 single applied twist of 90° in a wire 76 cm. long, — that is, a twist of 1°*2 per centimetre 

 length. For a twist of 2° '4 per centimetre length with the same current, the resulting 

 longitudinal intensity becomes 270. The similar quantities for iron wire (see Table VII., 

 sect. 22) are 190 and 310. 



Now the magnetic field at the circumference of a wire 0'86 or 0*87 mm. in diameter, 

 when that wire has a current of 1 ampere passing along it, is barely 5 units. In a 

 longitudinal field of this value, soft, well-annealed iron wire might be magnetised to an 

 intensity of nearly 1000. If we may suppose this to be something like the value of the 

 circular magnetisation induced at the circumference of the wire by the current that 

 passes along it, we may perhaps have no great difficulty in explaining the production of 

 a longitudinal intensity of 190 or 310 in terms of the average rotation of the molecules 

 under the influence of the twist of 1°'2 or 2° "4 per centimetre length. It is conceivable 

 that such twist should produce an average rotation of molecules through angles whose 

 sines are 0"19 or 0*31 respectively. 



But with nickel it is very different. Ordinary unstrained nickel wire, in a 

 longitudinal field of 5, acquires an intensity of at most 50. Twisted nickel wire, 

 however, as shown by Mr Nagaoka, may in the same field acquire an intensity of 200 or 

 300, so long as the twist does not exceed 3° per centimetre length. In these cases the 

 magnetisation curve rises very abruptly between the fields 3 and 5, almost immediately 

 attaining a value approximating to practical saturation. Let us assume that twist has a 

 similar effect on the susceptibility of the wire to circularly magnetising forces, and that 

 the intensity induced by the sustained current has the value 300 at the circumference. 

 The average value across a section of the wire will of course be less ; and yet by simply 

 twisting the wire to and fro we can obtain a variation of longitudinal intensity of as 

 much as 540 units. Further, it must be borne in mind that this longitudinal intensity 

 is largely residual — falling off very slightly, if at all, when the current is reduced to 

 zero. But, again, it is of an essentially different character from the residual longitudinal 

 intensity produced in the ordinary way by removal of a longitudinal magnetic field, for 

 (sect. 24) it is reversed if the current is reversed along the wire. This pronounced 

 polarity, then, in a twisted nickel wire must be derived directly from the circular 

 magnetisation sustained by the current along the wire, assisted to a certain extent by the 

 helical aeolotropy produced by the twisting. 



If the current is broken, a succession of to-and-fro twistings continues to be accom- 

 panied by a corresponding cyclic change in the magnetic polarity, but this change, as we 



