1904-5.] Magnetic Quality in Molecular Assemblages . 1045 
impossible to tell the position of that point until the law connect- 
ing magnetisation with magnetising force is known. 
The special case in which the point lies on the quartic is of 
importance. For, since the internal force has then its variable 
part balanced by the external force, the medium is left magnetised 
under the constant force A (§ 9). The magnetisation is therefore 
constant in all directions. We then get, iu the previous notation, 
B.p/r = B/2r 4 = 6I/r 4 , if we put B = 25I where b is an absolute 
constant. This gives 
I = K/bJa Q + f3 Q + y Q . 
Now, if the medium were of such a nature that, in a given 
direction, I was proportional to H, this expression would also give 
the value of the induction under a force which was constant in all 
directions. But the curves of fig. 1 exclude this supposition. On 
the other hand, if we regard the curves B and T as normal, we see 
that the conditions when I is sufficiently large may be fairly well 
represented by the relation I — 1 0 = AH, where I 0 is an absolute 
constant and A varies from one curve to another. At least the 
representation is probably sufficiently close to indicate the general 
nature of the true relation over a range of force large enough to 
give equal magnetisation in all directions. If, therefore, there is 
any reason to modify the above expression for the force under 
constant magnetisation, so that the left-hand side is I - I 0 instead 
of I, we could identify A with l/bJa 6 + /3 6 + y Q and test the 
resulting values of the components of magnetisation under constant 
force in different directions. 
Now the assumptions made in § 2, as a basis for the present 
theory, cannot hold rigidly in an assemblage of molecules obeying 
the postulates of the kinetic theory. They can only represent an 
average condition. And in view of the extremely small range of 
molecular directional control (§ 19), it seems quite certain that, in 
actual materials and at ordinary temperatures, a considerable pro- 
portion of molecules are free to turn under the action of the external 
field without reference to crystalline direction. These molecules 
will therefore contribute to the magnetisation a term which is in- 
dependent of direction. Consequently a term I 0 must be added, giving 
I = I 0 + H/Va 6 + /3 6 +y 6 . 
