1911-12.] The Molecular Theory of Magnetism in Solids. 239 
the strongest field, but maxima and minima are as nearly as possible 
interchanged. 
In the right-hand side of the same figure, the curves r = cos 2# — cos 4 0 
±2 cos 6# are represented. The correspondence in general form between 
these curves and Quittner’s curves (4) and (3), respectively, is very evi- 
dent. Now an expression cos 20-\-p cos 4# + g cos 6$ is proportional to 
sin 6 # + P sin 4 # + Q sin 2 #, where P and Q are functions of p and q ; and an 
expression L' 0 +_p'L / 2 + q'E\ — the variable parts of L' 0 , L' 2 , and L\ (§ 17) 
being proportional respectively to sin 2 #, sin 2 2#, and sin 2 #(3-4 sin 2 #) 2 — is, 
apart from a constant, proportional to sin 6 # + P' sin 4 # + Q' sin 2 #, where P' 
and Q' are functions of p' and q'. Hence the presumption is very strong 
that the variation of the magnetisation is due to three constituents of the 
form L' 0 , L' 2 , and L' 4 in the expression for the internal field. 
Therefore, if the crystalline structure is truly cubic, since the true terms 
L' 0 and L' 2 in an octahedral plate are not variables (§ 17), the actual terms 
I/ 0 and L' 2 , now postulated, can only be due to deformation or to false 
structure caused by flaws. The change in sign of the term involving cos 6#, 
which means a change in sign of the term L' 4 , could only, in a truly 
cubic arrangement, be due to a change from one to another cubic 
arrangement, e.g. a transformation, in high fields, from the cubic molecular 
arrangement with twelve nearest neighbours to that with six nearest 
neighbours ( Proc . R.S.E.. 1907). But, if this transformation occurred, 
maxima and minima, in a cubic face, should appear respectively in the 
