120 
Proceedings of the Koyal Society of Edinburgh. [Sess. 
of repelling poles, in the manner already suggested in § 28. In either 
case the Weber element becomes stable in whichever of the possible 
positions it may assume ; in the first case, because it is a member of a row 
of other Weber elements which have turned into the same direction ; in 
the second case, because the symmetry of the opposed internal forces is 
disturbed by the Weber element itself, and the disturbance travels round 
with it as it turns from one to another of the possible positions of stability. 
Of the two hypotheses, the second appears to consort better with 
the idea that the atom is an elastic structure, in which the shell at least is 
liable to have its form altered by such forces as come into play when the 
piece is strained. It may be added that it is easy to imagine, and to 
realise in a model, a Weber element which will have its projecting poles 
of one name stronger than those of the other name, and this offers another 
(but less probable) means of accounting for the stability. 
27. The effects of stress on magnetic quality are too intricate * to be 
deduced in detail from the behaviour of so simple a model, which, at the 
best, cannot claim to do more than represent crudely the mechanism of 
magnetisation. We should need a fuller knowledge of the distribution of 
the electrons and of the forces which determine their positions, before 
speculating about such features as the Villari reversal in iron, or about 
those differences which have been observed in the magnetic effects of stress 
in iron, cobalt, and nickel. j* But a notable characteristic shared by all 
ferromagnetic metals is that a simple stress — a pull or a push-— produces 
magnetic seolotropy. This is well reproduced in the model. A pull, for 
example, applied along one line will increase the clearance for those “ fixed ” 
poles which lie more or less in that line, and reduce it for others, with the 
result that a piece composed of model atoms will show different qualities 
as to magnetic susceptibility and retentiveness in the longitudinal and 
transverse directions. The main magnetic effect seen in iron, under fairly 
strong fields, and more conspicuously in nickel, is that a simple push gives 
the metal greater susceptibility and very much greater retentiveness in the 
direction of the push, which is the kind of change the model would lead 
one to expect. 
28. The model also throws light on the influence of impurity in a ferro- 
magnetic metal. When a foreign atom occupies one of the places of the 
* For a summary of effects of stress see chapter ix of the author’s book on Magnetic 
Induction in Iron and other Metals. 
t Hull (Phys. Rev.., May 1921) finds that the space-lattice of nickel is the face-centred 
cube. This gives each atom twelve nearest neighbours. If the shell electrons assume 
a corresponding grouping, the nickel model should have twelve “fixed” magnets, set in 
lines inclined to one another at 60°, 120°, and 180°. 
