1050 Proceedings of Royal Society of Edinburgh. [sess. 
withdrawal of the field would be followed by alignment along the 
ternary axes, so that the residual magnetisation would be only one- 
third of the original magnetisation even if there were no molecular 
motion. The effect of molecular motions is to make the residual 
magnetisation less than the original magnetisation along a ternary 
axis, for these motions are controlled by smaller magnetic force 
after the external field is removed. The tendency is similar in the 
other cases ; but the fact that the internal field is stronger along the 
ternary axes, to which the molecules set from the binary or 
quaternary axes, works oppositely. It is probable that the large 
value of the reduction factors, 1/3 and 2/3 respectively, will 
ensure that the magnitudes of the residual effects are in decreasing 
order from the ternary to the quaternary axes. Because of the 
entire withdrawal of the magnetisation I 0 , which takes place 
under the action of an external field on groups of molecules 
whose constituents are randomly conditioned because of heat 
motions, the actual residual magnetisations will be uniformly less 
than the above considerations indicate. 
Weiss’s curves T' and B', fig. 1, verify these conclusions, but 
the curve Q' is in conflict with them. These curves of the residual 
effects indicate ver}^ strongly the abnormal nature of magnetisation 
parallel to a quaternary axis. For the binary effect is not inter- 
mediate between the ternary and quaternary effects as it is in 
the other curves ; on the contrary, the usual conditions are so 
entirely departed from that Q' lies on the remote side of T\ 
Now the existence of flaws perpendicular to the quaternary 
axes might naturally be expected to lessen, not to raise, the 
residual magnetisation along these axes; for large demagnetising 
forces might be expected to exist because of the effective intro- 
duction of boundaries. As a matter of fact, the present theory leads 
to the result that the effect of such flaws may really be to raise the 
residual magnetisation along a quaternary axis. To prove this we 
have first to determine the stable directions of magnetisation, 
under internal force, in the neighbourhood of a boundary which 
we assume to be a plane perpendicular to a quaternary axis. In 
the evaluation of the constants involved in the expressions for the 
parallel and transverse components of internal force we have now 
to consider the action of molecules lying in, or on one side of, the 
