214 Mr. R. H. M. Bosanquet on 



the two ends ; there are also parallel air-circuits which start 

 off all along the sides of the rod. These latter are under 

 smaller differences of potential than the divergences at the 

 end, and may, in the first instance, with rods not very long, 

 be neglected in comparison with the resistances at the ends. 



Now suppose the rod cut up into n pieces : we have the 

 same total magnetomotive forces, the same total steel resist- 

 ance, and the resistance of 2n ends. The magnetic induc- 

 tion in each piece is therefore altered in a ratio which depends 

 on the ratio of the steel-resistance of one of the little pieces 

 to the air-resistance of its two ends. 



It is not possible to cut up a hard steel bar without dis- 

 turbing the magnetism ; in fact it is hardly possible to cut 

 it up at all. I have therefore preferred to cut a soft steel 

 bar into short lengths, finish these as accurately as possible 

 so that they may be put together to form one long bar ; then 

 harden, glass-hard ; then grind the ends with emery and oil 

 till the pieces will pick each other up when firmly pressed 

 together with a trace of oil on the faces; then magnetize. 



The compound bar is then suspended in a cradle by means 

 of a bifilar suspension arranged with its equilibrium-plane at 

 right angles to the magnetic meridian. If it were true that 

 the moments of the separate portions were the same whether 

 joined up or not, the deflection should be the same in both 

 cases. But it is not so. The deflection when the bar is 

 joined up and pressed together is many times as great as that 

 obtained when the pieces are so dispersed about the tray 

 which carries them as to be fairly removed from each other's 

 influence. 



A rough preliminary arrangement showed the existence 

 of a difference, but did not lead to the detection of the small- 

 ness of the effect produced by the separated pieces. 



A bifilar suspension was then constructed. It has a pulley 

 for the wire to pass over ; adjustable slides with metre- 

 and inch-scales carrying the holes for the wire to pass 

 through ; and a circular seat, with a circle divided to de- 

 grees; this is fixed on a firm crossbeam at a good height. 

 The wires enter, through a hole in the cover, a cylin- 

 drical case, whose sides are made of narrow pieces of flat 

 glass. Within this case swings the cradle which carries the 

 magnet. It has three V-shaped troughs. The pieces can 

 either be wedged together in the middle trough, or be placed 

 at considerable distances in all three. The cradle is 16 inches 

 long. It is suspended over a circle divided to degrees and having 

 nearly 16 inches diameter. Pointers are attached to the ends 

 of the cradle, which play in front of the circle. There is not 

 more than ^ of an inch to spare between points and circle on 



