107 
1921-22.] On Models of Ferromagnetic Induction. 
fields which produce instability in pairs and rows of permanent steel 
magnets have been experimentally measured, under various conditions as 
to length, proximity of poles, and direction of the deflecting field. All 
the conclusions stated above for rows and pairs of magnets have in this 
way been verified. The experiments were made, with the permission of 
Professor Bar Ida, in the Physical Laboratory of the University, and I have 
to thank Dr C. G. Knott for the use of his room there and for the loan of 
a large pair of Helmholtz coils, with a diameter of 49 cms., set 241- cms. 
apart, which gave a nearly uniform field of sufficient extent.^' Dr G. A. 
Carse, lecturer in the same department, kindly volunteered to help me 
in these experiments, and I am much indebted to him for his skilful 
co-operation in this part of the work. 
The magnets used were of the ball-ended type originally introduced by 
J. Robison, Professor of Natural Philosophy in the University of Edin- 
burgh from 1774 to 1805, and reintroduced in modern magnetic work by 
Dr G. F. C. Searle.f Their behaviour, as Searle has pointed out, approxi- 
mates closely to that of an ideal pair of magnetic poles situated at, or very 
near to, the centres of the balls. In the experiments now to be described 
I found that the position as well as the strength of the poles was but 
little altered by varying the conditions to which the magnets were sub- 
jected. When the pivots were so placed as to bring the magnets very near 
together, the poles shifted a little out; when the pivots were drawn further 
apart, the poles shifted a little in. These changes of position, on the part 
of the poles, were to be expected; they were in all cases sliglit, and their 
amount was readily inferred from observations of the field which was 
required to produce instability. Tests by rupture of a pair of pivoted 
magnets in which the ratio ajr or i is small afford a delicate method of 
finding the true value of r and so determining the position of the poles. 
An example is given in § 15 below. 
13. The magnets were rods of steel wire 0’21 cm. in diameter, with 
lengths ranging up to 10 cms. Their ends were screwed into quarter-inch 
bicycle balls (diameter 0'62 cm.). The rods and balls were softened to 
have the screws cut, were rehardened before magnetising, and were 
magnetised by being placed between the conical poles of a powerful electro- 
magnet. They were supported on needle-point centres, as in figs. 6 and 6a. 
Most of the magnets were straight, as in fig. 6a, which is the better form 
for exact work, and had gravitational stability given them by a tubular 
brass weight shown in section at A, which could be slipped sideways along 
* A description of these coils will be found in Proc. vol. xiii, 1885, p. 523. 
f Searle, Proc. Camh. Phil. Soc., vol. xii, 1902, p. 27. 
