576 
PROFESSOR J. A. EWING ON EXPERIMENTAL 
even if the substance has remained diamagnetic during the action of the field. This 
view is supported by the experiments of Lodge (‘Nature,’ 1886, March 25, p. 484; 
also April 1, p. 512), which have shown residual paramagnetic polarity in copper 
and other diamagnetics after exposure to a strong field. According to this view the 
curve of 3 and <§ for substances in general is (in its earlier portion) of the form 
sketched. If the region ab extends so far as to include fields of considerable 
intensity, we call the substance diamagnetic. In iron, b comes so near the origin that 
we are always dealing with the further portion of the curve.] 
§ 63. The extension of the curves of k and 3, or of /x and S3, at the upper end is 
subject to a like uncertainty. As Rowland has remarked in the introduction to 
his second paper, we do not know whether 3 or 93, or either of them, attains a 
maximum ; and Maxwell* has given a reason (similar to the above) for supposing 
that, after passing a maximum, 3 becomes negative. For, during the application of 
we may suppose that the induction of Amperian currents goes on in the molecules, 
giving a component of magnetisation opposite in sign to that which is produced by the 
turning of the molecules. The latter reaches a limit, the former increases indefinitely 
as <§ is increased, and therefore ultimately exceeds the latter. Hence, iron may be 
diamagnetic under a very strong magnetising field. 
In that case we should find that as S$ is raised, 3> after passing a positive maximum, 
decreases to zero, becomes negative, and then goes on always increasing negativelv. 
The curve of k and 3> if produced by increasing «§, would bend back, towards the 
origin, in its descending limb, pass through the origin and over to the negative side, 
k never exceeding a small negative value as 3 is increased negatively without limit. 
§ 64. On the other hand, even if iron be ultimately diamagnetic under strong 
magnetising forces, we have no reason to expect that 93 will pass a maximum. 
r*. . d^S 
Since 93 = 47 t3 + >§j 93 cannot pass a maximum with increase of unless -A attains a 
negative value greater than 4 b. This is a greater coefficient of magnetisation than 
any diamagnetic substance is known to possess, and, moreover, is too great to be 
consistent with Weber’s theory of diamagnetism, according to which /x cannot be 
negative. 
These considerations, if they serve no other useful purpose, show the futility of 
drawing conclusions as to the initial and ultimate values of the magnetic suscepti¬ 
bility of iron in indefinitely low and indefinitely high fields, from observations 
made, as all observations must be made, in fields of finite magnitude. For this 
reason it appears that an empirical formula, such as Rowland applies to the curves 
of /x and 93, must be misleading when pushed beyond the range of actual experi¬ 
ment. If we do not know whether 3 or 93, or either of them, attains a maximum, 
it is a truism to say that there is no use in assigning numerical values to that 
maximum. The results obtained by extending a formula past the limits of experience 
* Maxwell, ‘ Treatise on Electricity and Magnetism,’ ii., §§ 843, 844. 
