1911-12.] The Molecular Theory of Magnetism in Solids. 243 
tion transverse to it. In each case there is evidence of a cos 20 term with 
a weak cos 60 term superposed upon it. The former corresponds to a 
rhombohedral structure, L' 0 and T' 0 , either real or induced by foliation. 
From §§ 11, 14, 16, we see that no term of the forms cos 20 and cos 4 0 can 
appear in the case of a truly hexagonal structure, while a term of the 
form cos 60 is apparent. 
Weiss found it possible to obtain by fracture a crystal in which the 
cos 60 effect was almost absent, and he regarded the substance as constituted 
crystallographically of three conjoined rhombohedral crystals which thus 
simulated hexagonal form — a view adopted also by some crystallographers. 
This is a thoroughly plausible view, quite in consonance with the molecular 
theory. But another possible view, so far as magnetic quality is concerned , 
is that the cos 60 term is due to a real hexagonal structure ; in which case 
the cos 20 term must be induced by foliation. Indeed, the cos 60 effect may, 
in part at least, be due to foliation also. 
Some evidence of variability in the constitution of the rhombohedral 
effect is given by Weiss’s observations. In the right-hand part of fig. 10, 
copied from his paper, curves of the component of magnetisation parallel 
to the external field are given, four curves being shown at different values 
of the field. 
When we put y = 0, L 0 takes the form A sin 2 d-f-B cos 2 d (eq. 10), and 
(eqs. 18, 23) I/ 2 and I/ 4 each take the form C + D sin 2 0 + E sin 2 2d : so the 
internal field up to this order, in the rhombohedral arrangement, is of the 
form 
C + D sin 2 # + E sin 2 # . 
The four full-line curves in the left-hand side of fig. 10 have the equations 
l — Od sin 2 0 — J sin 2 2d, 1 — 02 sin 2 0 — J sin 2 2$, 1 — 0*5 sin 2 0 — J sin 2 2d, and 
1— 0-8sin 2 d — J sin 2 2d, the numerical values having been chosen merely so 
