MAGNETISM AND TWIST IN IRON AND NICKEL. 521 



nounced negative lagging. In other cases the positive lagging is diminished, but still 

 remains positive ; while, in still a third group, the negative lagging is greatly increased. 

 In all, however, the effect of the tapping is the same. Now the very operation of 

 twisting a wire must be somewhat of the character of jarring or tapping ; and the 

 greater the twist the greater the accompanying jarring. This consideration seems to 

 explain the earlier occurrence of the negative lagging with the larger twist, and the 

 disappearance of the positive lagging altogether for high twistings. Similarly, if we 

 assume that stronger currents are associated with more pronounced molecular dis- 

 turbances, we may look upon the change from positive to negative lagging, when the 

 current is high enough, as an illustration of the same principle. But the recognition of 

 this principle can carry us no distance in explaining the existence of positive lagging 

 under the conditions favourable to it. 



21. After-effect of the Second Order. — It is worthy of notice that the change of sign 

 in the lagging occurs for descending as well as for ascending values of current. This is 

 shown in Table VI. for the wire of diameter +3 mm. and for the twist ±90°. The 

 change of sign does not, however, occur for exactly the same combination of twist and 

 current in the ascending and descending series. For descending values of current, there 

 is a persistence of the negative lagging that characterises the higher values, so that the 

 change of sign occurs for a value of current lower than that for which the change of sign 

 occurs in the ascending series. Thus the effect of a cyclic twisting with a given current 

 depends upon the previous history of the wire. This is a true after-effect, which we 

 must regard as of the second order, since already in the individual cycle itself an after- 

 effect exists, as evidenced by what we have throughout called the lagging. 



A similar phenomenon shows itself when, with steady current, we subject the wire to 

 increasing and decreasing ranges of twist. An example of this is given in the following 

 table, which shows the lagging for a series of cyclic twistings from ±135° down to ±5° 

 and up again to ±90°. The twist numbers occupy the first column, and the other 

 columns contain the lag numbers, their arrangement sufficiently indicating to what cyclic 

 range of twisting they belong. Along the top row are entered the corresponding values 

 of the current, which fell off slightly as the experiment proceeded ; and along the bottom 

 row the complete ranges of intensity are given. It should be mentioned that each 

 column of lag numbers corresponds to a cycle which was observed after the permanent 

 state for that cycle had been established by repeated to-and-fro twistings. 



[Table 



