MAGNETISM AND TWIST IN IRON AND NICKEL. 525 



the current is increased ; but in no case did I obtain a change of sign on working up to 



higher currents or down to lower for a given twist. For the cyclic twistings ± -, ± ir 



z 



±2 77, the lagging is positive ; for ±3 77, it becomes (in the first series of experiments) 

 negative even for small currents. It will be seen also that for this largest cyclic twisting 

 there is a maximum in the lag-area for a current of about 2\ amperes. Now it was very 

 natural to suppose that by working backward to lower twistings we should be able to hit 

 upon a twisting which should give a zero area, or one which should give a positive or 

 negative lagging according to the value of the current along the wire. The result of the 

 attempt to obtain such a critical twisting is shown in the last seven experiments of the 

 first half of Table VII. Gradually the cyclic range of twist was diminished from ± 2^ 7r 

 to ±1^ 77, and still the lagging persisted negative. The last experiment, however, gave 

 a positive lagging. It was quite clear, then, that we have here to do with a very pro- 

 nounced after-effect, so that, for a cyclic twisting of ± 2 77 the lagging was negative or 

 positive according as this twisting was or was not preceded by a twisting of ± 3 77. At 

 this stage a new specimen of the same wire was taken, a careful inspection having shown 

 that the first specimen had suffered a large permanent distortion near its middle point. 



With the new specimen, a twist of ± 2\ 77 was first tried, which for a current of 0'6 

 ampere give a small negative lagging. The smaller twist, ± 2 77, also gave the lagging 

 negative ; but with the twist ±1^ 77, the lagging became positive. On trying the some 

 what higher twist ±1| 77, I obtained the negative area again. The series of experiments 

 beginning with the current T42 for the twist ±lf 77, and finishing with the current 1*36 

 and the twist ± ^ 77, shows the change of sign in the lag-area, and the manner in which 

 this area rises to a positive maximum for the twist ±77, and falls off towards zero for 

 smaller twists. Similar conclusions are borne out by the succeeding series, which ends 

 with a large twist of ± 4 71-. 



24. Effect of reversing the current along a twisted ivire. — On 23rd April 1889, just 

 after completing the experiment on the first nickel wire for current 2 '5 9 and twist 

 ±180°, I tried the effect of changing the direction of the current along the wire. The 

 effect was to reverse the polarity from + 307 to — 300, that is, through almost exactly 

 the same range as had just been obtained by twisting the wire from —180° to +180°. 

 On carrying the operation of change of current slowly from its greatest positive value to 

 its greatest negative value, I found that until the current had passed through zero, and 

 reached a negative value of nearly one ampere, the change in the magnetic intensity was 

 very small. Thereafter, however, a rapid change of intensity set in as the current was 

 increased towards its negative limit. The features of the phenomenon are best shown 

 graphically. In figs. 5 and 6, PI. V., two examples are given, one for nickel and one 

 for iron. In the case of nickel (fig. 5) a current of 2' 7 amperes was reversed along the 

 wire, which had been twisted through an angle of +225° from its original position. 

 The dotted curve represents the same kind of curve obtained when the wire was 

 vigorously tapped before each reading. It will be seen at a glance that tapping very 



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



