1911-12.] The Molecular Theory of Magnetism in Solids. 241 
permeability along the diagonals in the face FAB. This may occur to an 
extent practically sufficient to annul that normal excess of permeability 
along the diagonals over that along the edges, which is due to the molecular 
structure, and is exhibited in the value of 7 L / 2 . Hence we have a possible 
explanation of the rectilinearity of the curve, at H = 31'5, in the plane of 
the axes b and c, fig. 6. At higher fields, magneto-striction may tend to 
efface the fissures and allow the structural effect 7 L' 2 to become dominant, 
as the diagram shows. 
Similarly, equal foliation parallel to the two octahedral planes at the 
corners A or H, C or F, preserves symmetry in the plane of the axes b and 
c ; while, if it be not so marked as the preceding foliation, it will not 
obliterate its effect in the plane of the axes a and b. 
Foliation parallel to any of the other octahedral planes intersects the 
plane at the corner A in lines which are perpendicular to one or other of 
the projections of the axes a, b, c on the plane. It thus tends to make the 
permeability in the directions of these projections small, and so gives rise to 
a cos 69 term in the magnetisation having its minima in the directions of 
the projections of the quaternary axes. This is exactly the nature of the 
effect which would be due to the internal field in an octahedral plane in 
a non-foliated molecular arrangement with twelve nearest neighbours. 
And, if the foliation perpendicular to the projection of the c axis be more 
effective than that perpendicular to the other two axes, a cos 20 term, with 
its minimum in the direction of the c axis, which the analysis of the curves 
in fig. 7 shows to exist at the higher fields, results. The only argument 
against this explanation of the existence of the cos 20 term in the value of 
the magnetisation in an octahedral plane is that, if the foliation in planes 
perpendicular to the diagonal DG be more effective than that in planes 
perpendicular to the diagonal BE, the cos 20 term in the plane ab would 
have its minimum in the direction of the diagonal AG, while fig. 6 exhibits 
the reverse effect. We shall see immediately How that adverse result may 
be overbalanced. 
Octahedral foliation cannot give rise to a cos 4 0 term in an octahedral 
plane. For the production of this effect by foliation, we have to postulate 
foliation parallel to dodecahedral planes. 
Foliation parallel to the dodecahedral planes which pass perpendicularly 
through the diagonals of the face ABCD might be sufficient to practically 
annul the structural effect 7 L 9 in that face at low fields, while maerneto- 
striction would diminish the foliation effect at higher fields and allow the 
structural effect to preponderate, as fig. 6 shows. We have already seen 
that the same effect may be produced by octahedral foliation. If 
VOL. xxxii. 16 
