1054 Proceedings of Royal Society of Edinburgh. [sess. 
From the above expression we can obtain also an estimate of 
the value of p. The ratio M/H is the molecular susceptibility. 
Consistent determinations of the value of that quantity, obtained 
from observations on solutions of different salts of iron, have been 
made by Liebknecht and Wills (Du Bois, Rapports, Oongres 
International de Physique, tome ii., 1900). The mean value is 
0*014, the mass of a hydrogen atom being taken as unity. To 
obtain the value of this mass independently of any estimate of the 
number of molecules contained in unit volume, we may postulate 
that the atomic charge is identical with the ionic charge. From 
the magnitude of this quantity, together with the value of the 
electro-chemical equivalent of hydrogen, we find that the mass of 
the hydrogen atom is 1*2(10) -24 grammes. Multiplying 0*042 by 
this number, we obtain 5(1 0) -26 as the absolute value of the 
molecular susceptibility in a molecule containing three atoms 
of iron. 
If, now, disregarding the fact that the susceptibility of magnetite 
is not constant, we identify the number just obtained with the 
ratio M/H, we find p = 3(10) -9 , a result not greatly different from 
the values usually given for molecular distance. 
If, on the other hand, we adopt the usual estimate of molecular 
distance, we obtain a value for the molecular susceptibility of 
magnetite which is about twenty times greater than that found 
above from observations on dissolved salts of iron. This is by 
no means an impossible result. 
21. Conclusions . — The following conclusions may, I think, be 
drawn from the foregoing results. 
(1) The theory of molecular magnetism, when applied to 
magnetic crystals of the cubic class, leads to results which are not 
in discordance with any observed results, but which, on the 
contrary, are in general agreement with observation. 
(2) Wallerant’s formula is correct in mathematical form, but 
should be interpreted with reference to internal force instead of 
with reference to magnetisation. 
(3) It is possible that parallel investigations of the properties of 
other assemblages having cubic symmetry may lead to a discrimina- 
tion of actual molecular arrangement. 
