RESEARCHES IN MAGNETISM. 
G13 
when (after applying and removing load) we take a curve of magnetisation. We find 
the magnetic susceptibility now changed from what it was. The difference of 
condition is merely this, that in the former case, after removing a previously applied 
load the wire was passed through the process of demagnetising by reversals before a 
new curve was taken, while in the latter case the wire was demagnetised before the 
load was removed, and the load then removed before the curve was taken. So slight 
a difference of procedure as this might fairly be expected to have no sensible influence 
on the results of the test, but it was only after recognising that this difference and 
others like it had an extraordinarily great influence that I was able to obtain the 
mutually consistent results stated in the above paragraphs, by adhering to a strictly 
uniform mode in the taking of every curve. I found that after the wire had 
been demagnetised completely, the application and removal of any load, though 
producing no immediately visible effect, affected most materially the subsequent 
behaviour of the wire. Indeed the curve of magnetisation under any given load 
depends not only on the load actually present, but on any changes of load which have 
been permitted to take place since the preceding demagnetisation. For example, if 
a curve was to be taken with (say) a load of 3 kilos, on the wire, and if, after complete 
removal of all visible magnetisation, the load were accidentally raised to 4 kilos, and 
1 kilo, then removed, the resulting curve was very sensibly different from what it 
would have been if the weight had simply been raised to and kept at 3 kilos., and this, 
too, in spite of the fact that the wire had been frequently subjected to more than 
four times that amount of stress, and was therefore in a mechanically stable state. 
§ 97. This obscure but very interesting phenomenon of the residual effects of 
previously applied stress formed the subject of a large number of experiments, made 
for the most part on the ware of § 94, which had been previously stretched by a load 
of 18*5 kilos. In describing these experiments, the interest of which lies only in the 
comparative and not in the absolute values of the induced magnetism, it is needless 
to reduce the observations to absolute measure, and I shall generally give merely the 
galvanometer and magnetometer readings, instead of <§ and 3- 
(May 12, 1882.) After many applications of a load of 18*5 kilos, the wire was 
demagnetised by reversals without load, and then on applying magnetising force the 
readings in Column I. below were taken. 
Next, after removal of the load, the wire was again demagnetised, then loaded 
to 18*5 kilos., then unloaded, and the observations of Column II. wei*e taken. 
Lastly, the wire was again demagnetised, loaded to 18*5 kilos., unloaded, tapped, and 
the observations of Column III. taken. In all three cases there was no load on the 
wire during the process of magnetisation. 
MDCCCLXXXV. 
4 K 
