110 Proceedings of the Royal Society of Edinburgh. [Sess. 
and for that value of i the curve of fig. 3 shows that F,. should be 12'8, 
consequently [FJ becomes ^/o ^ which is almost exactly equal 
lur'^ lDX(3‘36p 
to the observed value of . On trying r = 3’35 or 3 37, substantial dis- 
crepancies are found in excess and defect; hence 336 cms. is accepted as 
the value of r for that spacing of the magnets. 
Similarly, when they were set nearer together r was found to be slightly 
greater, and when they were further apart it was slightly less. The follow- 
ing are representative figures, got by observing the current which produced 
rupture for various values of a, for the same pair of magnets, the angle 
being 90“ throughout : — 
Half distance a ..... . 
3‘80 cms. 
3-965 
4-28 
4-51 
Half length to centre of balls 
3-39 „ 
3-39 
3-39 
3-39 
Field producing rupture .... 
5T0 
2-20 
0-94 
0-62 
True half length of magnets, r . 
3'41 cms. 
3-36 
3-33 
3-32 
Value of t . 
1-115 
1-18 
1-29 
1-36 
Results of the same character were got when the observed breaking field for 
«=:180°, instead of for a==90“, was used as the criterion in determining r. 
16. The limiting stable defiection which preceded rupture was examined 
by attaching a light pointer to each magnet, which allowed the angle 0 to 
be read on a graduated arc. In this way values of d,, were measured which 
agreed with those calculated above. When the direction of the field was 
such that a was near 180° it was observed that as the field was gradually 
applied the adjacent poles at first deviated to opposite sides of the line 
joining the centres ; then with increasing H one of the two poles became 
dominant and pulled the other pole over to the same side as itself, and 
finally rupture took place in the manner described in § 10 and illustrated 
in fio\ 5. 
17. Experiments were also made on the rupture of long rows of 
magnets. The conditions that would apply in an indefinitely extended row 
were reproduced with a limited number of magnets by the device of using 
blocks to prevent the two terminal magnets from becoming more deflected 
than the others during the quasi-elastic stage which precedes rupture. 
Eight comparatively short magnets were used to form the row. Wooden 
blocks were set to limit the deflection of the end members ; the blocks were 
adjustable by hand, and it was easy to manipulate them, while the current 
was being slowl}^ increased, so that the deflection of the magnets at the 
ends of the row was kept, as nearly as the eye could judge, equal to that 
of the other mao-nets. The effect was that the other maonets behaved as 
O 
