1048 Proceedings of Royal Society of Edinburgh. [sess. 
force is always in the plane, the transverse field, parallel to the 
ternary axis, which arises because of the magnetisation, brings in, 
if unbalanced, a transverse component of magnetisation along that 
axis. This component alternates along the axis six times per 
complete revolution of the force, so that the arrangement could 
act as the core of a six-pole dynamo. 
The case of magnetisation always in the plane probably more 
nearly coincides with the conditions under which Weiss worked; 
for his disc of magnetite, though 20 mm. in diameter, was only 
0*3 mm. in thickness. Self-demagnetising force would therefore 
act strongly, so that the resultant force was not really in the 
plane ; the magnetisation would practically be in it, and he 
constant, as TFeiss found. 
18. Magnetisation and Field along Principal Axes. — From 
the expression (§ 16) connecting magnetisation and field, we can 
deduce relations connecting the magnetisations and fields in the 
directions of the axes of crystalline symmetry. Taking the 
suffixes T, B, Q to refer to the ternary, binary, and quaternary axes 
respectively, and taking account of I 0 , we get the result 
It + Iq = 2I b . 
Using the data given by fig. 1, the numbers in the second row 
below are calculated values of I B corresponding to values of H 
given in the first row. The observed values of I B are given in the 
third row. 
10 20 40 60 80 100 120 140 160 200 240 280 320 360 400 440 
107 167 245 286 311 329 340 350 358 368 376 383 389 394 401 405 
100 166 252 299 330 352 364 373 380 388 395 400 404 406 407 409 
In view of the marked crystalline defects which are indicated by 
Weiss’s diagrams, the differences between observed and calculated 
values do not seem to he excessive. Yet, apart from other facts, 
the very remarkable agreement between calculation and experi- 
ment, in the case of Wallerant’s formula (§ 5), would settle the 
question as between his formula and that given in § 16 in favour 
of his. On the other hand, as already pointed out (§ 14), the 
non-correspondence of the shape of the loops in figs. 5 and 6 to the 
shape of the observed loops, figs. 2 and 3, and the correspondence 
