132 H. Nagaoka on Hysteresis attending Change of 



I also tried an experiment with a system of levers ; but it 

 did not work smoothly, so that the readings were capricious 

 and could not be trusted. 



These faults were, to a great extent, removed by the 

 apparatus which I describe below. The horizontal and 

 vertical projections of the apparatus are represented in fig. 1 

 and fig. 2 (Plate I.) respectively. The essential part consists 

 of a stout brass bar 53 centim. long, 1 centim. broad, and 

 1*1 centim. high. It is provided with three levelling-screws 

 (Ji, h> h)' A carefully polished Y-groove is cut along the 

 bar. A small rectangular brass pillar (p) is erected at one 

 corner of the bar. A small vertical Y-groove is cut on it, 

 and on this two points of the lever rest. The lever with 

 a mirror attached is shown in fig. 3, both from the front and 

 from behind. It is a small rectangular piece of brass with 

 three steel points (p^^g,^), of which two (fii, p 2 ) res ^ on 

 the Y-groove in p. The, other point (_p 3 ) comes in contact 

 with a small plane glass plate, which is fixed to the end of the 

 movable brass rod. The point of contact is in the prolonged 

 axis of the wire whose change of length is to be determined. 

 The distance of the line p x p 2 from p 3 is 1*125 millim. Pre- 

 liminary testing showed that the relative positions of these 

 three fine steel points were not directly affected by the 

 magnetizing forces. The plate has three holes (h 1} A 2 , h s ). 

 To the holes h 2 A3 is attached a thin brass wire, which is 

 pulled at its middle by means of a small spiral spring (s 2 ) 

 of hard brass wire. Another spring (si), similarly made, is 

 attached to the other hole. These springs can be adjusted by 

 means of slide arrangements, n± and n 2) attached to the sides 

 of the bar. The circular mirror, m, attached to the lever was 

 obtained from Hartmann and Braun. 



The greatest difficulty in the measurement of change of 

 length by magnetization arises from the temperature-changes 

 produced by a current passing into the magnetizing coil. 

 On this account most experimenters have passed the current 

 only for a very short time, and observed the change before 

 the temperature produced any effect. The consequence 

 is that the changes are traced only by jumps. I found 

 that the temperature-effect could be greatly compensated for 

 by applying the principle of the gridiron-pendulum. This 

 end was achieved by using zinc rods of different lengths such 

 that, in any combination, the total expansion due to small 

 changes of temperature in particular lengths of zinc and iron 

 (or nickel) was equal to that in a particular length of brass. 



