574 
PROFESSOR J. A. EWING- ON EXPERIMENTAL 
of 33 is very marked in this example : the steel has acquired a strong set towards the 
side of the first magnetisation. 
Values of the Coefficients p and k. 
§ 59. It follows from the form of the magnetisation curve, as has been pointed out 
by Stoletow and others, that the coefficient of magnetic susceptibility k, or P, and 
therefore also the permeability p, or rises, as magnetisation proceeds, from a com- 
s? 
paratively low value to a maximum, and then decreases continuously as <£> is further 
raised, at least within such limits of <£> as have ever been reached in actual experiments. 
The changes in k and p which occur during the process of normal magnetisation are 
best seen by Rowland’s method of plotting k in terms of 3, or p in terms of 33. 
Plate 62, fig. 29, gives two examples of this last mode of representation. One of 
the curves shown there refers to the annealed iron wire of § 21. In it the values of p 
(which are given numerically in § 21) pass from 128 for a very low value of 33 to a 
maximum of 2370 for <§ = 4’5, and down again to 705 for <£> = 22‘27. In the other, 
which refers to the annealed iron wire of § 22, the early values of p are not given, 
but the magnetisation has been pushed further by raising <£> to 89'8, with the result 
of reducing p to 183 and of giving the curve an upward inflection in its descending 
limb. That such an inflection occurs when the magnetisation is sufficiently increased 
has been already noticed and commented on by Fromme (Wied. Ann., xiii., p. 695). 
It makes the formula given by Rowland (Phil. Mag., xlviii., p. 339) fail to serve 
as an equation to the curve in this extreme portion. 
§ 60. Plate 62, fig. 30, gives two curves showing, in the same manner, the relation of 
k to 3 in the iron wire of § 31, first in its annealed state, and again after it had been 
hardened by stretching. They are drawn to the same scale, and* bring out well by 
contrast the great influence which permanent set has in reducing magnetic suscepti¬ 
bility. They are plotted from the observed values stated in § 31. Other curves of 
the same character may easily be drawn by aid of the data contained in previous 
paragraphs. 
§ 61. Curves of this kind suggest several interesting theoretical questions regarding 
the results which we should find if we were able to extend indefinitely the range of 
observations, both towards vanishingly low and towards indefinitely high values ol £>. 
It seems exceedingly probable, to judge from the portion of the curve of k and 3 
which we can actually determine, that, if produced backwards, the curve would cut 
the axis of k at a finite positive value of k* This, in other words, would imply that 
the susceptibility is not indefinitely small at the very beginning of magnetisation. 
To reconcile such a result with the idea of a static frictional resistance to the 
* Rowland has so produced his curves of /<. and 53 backwards, and has given values of /i corresponding 
to 33=0. These, however, are merely hypothetical. 
