MAGNETIC EFFECTS OF TWIST. 175 



kept steadily flowing. Obviously the wire is, to begin with, circularly 

 magnetised ; and, if it is isotropic as regards magnetic susceptibility, 

 it should show no longitudinal polarity. It is almost impossible, 

 however, to obtain a wire of magnetic metal free from such polarity. 

 All the usual precautions were taken. The wires were roasted and 

 allowed to cool as they lay perpendicular to the magnetic meridian ; 

 and the wire which was being studied lay horizontally in the same 

 direction : opposite the one end of the wire a magnetometer of the 

 usual mirror device was set ; and the other end of the wire was 

 fixed to the twisting apparatus. In order to avoid jars and jolts, 

 the twist was not applied directly by the hand, but was effected 

 by means of a gearing of toothed wheels. For every complete 

 rotation of the handle worked by the hand, the axle which form- 

 ed the continuation of the wire moved through a few decrees. At the 

 same time the resistance of the gearing was considerable, so that it 

 was an easy matter to keep the hand working at a fairly uniform speed. 

 In this way the twisting of the wire was effected very gently and 

 gradually. 



We know already from Wiedemann's experiments that, when 

 such a circularly magnetised wire is twisted right-handedly with 

 reference to the current flowing along it, it becomes longitudinally 

 magnetised, the magnetic intensity being co-directional with the current 

 in the case of iron wire, but anti-directional in the case of nickel wire. 

 After the wire has been twisted and untwisted repeatedly through a 

 considerable range — 4° per centimetre in Wiedemann's experiments — 

 a corresponding cyclic variation is impressed upon the longitudinal 

 intensity. From a large positive value at or near the one limit of 

 tw T ist, the magnetic intensity changes to a nearly equal negative value 

 at or near the other limit of twist. The average apparent polarity of 

 the wire, as indicated by the deflections of a magnetometer placed oppo- 



