ins .A. TAXAKADATE AND H. XAGAOKA. 



Thi:^ result is greater tlian the value used in reducing the 

 observations of 1.S87. Tlie induction coefficient then used was 3.0, 

 and was deduced from, the observations made in 1883 on the magnet 

 from Avhich the present magnet was prepared ]:)y cutting off its end. 



AVe tested the result by the ballistic galvanometer method. For 

 this purpose a coil, into which the magnet fitted almost exactly, was 

 wound, and its ends Avere led to a low resistance ballistic galvanometer 

 whicli was 10 metres distant from the coil. The coil lay on a horizon- 

 tal plane with its axis parallel to the magnetic meridian, and it was 

 adjusted to this positi(Mi by means of a wooden block. When the coil 

 was quickly turned round through half a revolution in the liorizontal 

 plane the ballistic galvanometer gave deflections proportional to the 

 number of lines of force through the solenoid due to the horizontal 

 coiu])onent of the earth's field. Now the bar magnet Avas put into the 

 solenoid, and the experiment repeated; the magnet was introduced 

 again with its axis reversed, so as to eliminate its direct effect on the 

 galvanometer. The following is the result of this experiment : — 



Moan of 10 readings of galvanometer t-hi-ow without magnet 9.1 

 „ „ „ „ „ „ -with magnet 125.7 

 Volume of magnet 4.5 c. cm. 



The ratio of the volume included by the solenoid to that of the 

 magnet is 1.44; whence /^ = ' x ^x 1.44 =^ 0.62, which is greater 



than the value obtained by the other method. The difference is such 

 as might be expected from the difference in Û\c methods, as the first 

 partly depends upon the distribution of induced magnetism in the 

 magnet, and the second on the integral effect of induction on the sub- 

 stance. In applying the correction due to induced magnetism in the 

 reduction of our o])servations, we have taken the value obtained by 

 the first experiment, as it refers more direct]}^ to the point in 

 question. 



