645 
1907-8.] Magnetic Quality of Molecular Magnets. 
permeabilities along these axes are in descending order of magnitude. 
The reverse occurred in the closest packed arrangement. 
6. At the surface, supposed to be a plane perpendicular to a quaternary 
axis, evaluation of the sums from v = 0 to */=10 gives, for the longitudinal 
components of the internal force along the quaternary, binary, and ternary 
axes respectively, the proportionate values 306, —19, —127, the same units 
as in § 4 being used. The transverse components also are larger by about 
one-third than they are in the interior, and their directions are unaltered. 
Thus the quaternary axes are lines of more stable magnetisation at the 
surface than they are in the interior. 
10 
Evaluation of the terms M/4yo B .2(3Ni/ 2 y) _ ® — Np - ^), which give the longi- 
0 
tudinal effect due to the so-called “ surface ” magnetisation, shows that it 
is equal to 2T I, where I = M/p 3 is the intensity of magnetisation. It 
must be remembered that this is merely the value of the force, due to 
these terms, at a definite set of points in the surface plane. It increases 
the stability of the surface magnets. 
In the closest packed arrangement the corresponding force is 2*84 I. 
In the former paper the values of the terms 3S.r 2 (2 \p)~% and 2.N(2p) - * were 
given as 19*75 and 14*54 respectively, the summation being taken from 
v— — 10 to v= 10. When the summation is from — oo to +oo the 
sums should be identical, so their approximate equality was regarded as 
indicating that the limited summation was sufficiently accurate to give 
a correct general description of the facts. The correspondence is actually 
much closer, for, in Table I. of that paper, the value N = 2 should be N = 1, 
and consequently the numbers in the first rows of Table V. and of the second 
half of Table IV., under the headings v = 2, i/ = 4, . . , i/ = 10, require to be 
halved. Also the decimal point is misplaced in the second rows of these 
tables under the heading v = 0. The numbers should read 0T25 and 0*5 
respectively. These errors do not essentially affect any condition dealt 
with in the paper ; but when they are corrected, the above sums are found 
to be 14*69 and 14*73 respectively. This gives ample verification of the 
sufficiency of the partial summation. In the most open cubic arrangement 
now dealt with the sums are 32*70 and 32*74 respectively. 
7. From Weiss’s experimental observations on magnetite, it results 
that the internal magnetic structure of the substance possesses, like its 
crystalline structure, cubic symmetry. It appears now that the internal 
magnetic arrangement cannot be that of most open cubic packing, though 
it may be that of least open homogeneous cubic packing. Exhaustion of 
the possibilities must discriminate the real arrangement. In this connec- 
