123 
1921-22.] On Models of Ferromagnetic Induction. 
symmetry, the least number of orbits constituting it (paired as in fig. 19) 
would be twelve. These might be made up of three groups of two pairs 
each, arranged as in fig. 21, in mutually perpendicular planes. But a. group- 
ing apparently more appropriate to an atom which is to take its place in 
a centred-cube space-lattice, would be one in which three pairs of orbits 
are symmetrically placed (with their major axes inclined at 60°) in each 
of the four octahedral planes — that is to say, in planes perpendicular to the 
trigonal axes. 
30. Reverting to the model of § 25, it is interesting to notice that 
the pivoted octet which there constitutes the Weber element may be 
deprived of magnetic moment (without other change) by reversing, end for 
end, four of its eight spokes, with the result that all the projecting poles 
have the same name, the central steel 
boss then forming a common pole of 
the opposite name. When this is 
done, the pivoted element, having 
as a whole no magnetic moment, 
is not caused to turn by applying 
an external field. It sets itself in a 
stable position with reference to the 
eight fixed poles of the shell, namely, 
with its spokes pointing towards 
them (just as in fig. 15, PI. II) if its 
own poles have the opposite name, or 
in the position of fig. 22 (PL II) if its 
own poles have the same name. It 
is obvious that in either case a substance made up of such atoms has lost 
its capacity for magnetisation, though each atom is still composed of the 
same fixed parts, and of a moving part comprising the same magnetic con- 
stituents, similarly grouped except that, instead of exposing four north poles 
and four south poles, they now expose eight poles of the same name. The 
geometry of the atom is otherwise unchanged. This transformation, which 
is readily produced in the model by making a simple rearrangement in the 
magnetised spokes of the central octet, is suggestive of what occurs when 
iron is heated to the critical temperature at which it loses magnetic quality. 
As is well known, when iron passes from the magnetisable to the non- 
magnetisable state there is an arrest in the process of heating — a point at 
which energy is absorbed — and there is a corresponding arrest when the 
magnetisable state is recovered in the process of cooling. But as Westgren 
has shown. X-ray analysis reveals no change at this point in the crystal 
