72 
STR WILLIAM THOMSON ON THE 
226. One very interesting feature common to all the diagrams, and presented even to 
some degree by the exceptional ones 9, 10, 12, and 14, shows that the effect of twisting 
the wire first in one direction and then in the other, and leaving it free from torsive 
force, was in every case to leave it with less magnetization than it had at the beginning. 
227. These exceptional diagrams and later continuation of the operations through a 
second positive torsion and a second negative torsion, represented in the latter halves 
of the curves of Diagrams XXIV. to XXIX., show what would be the general character 
of the effect of continued periodic applications of positive and negative torsion, through 
equal angles on the two sides of zero. 
In every case there is a lagging of quality, showing a residue of effect from pre¬ 
viously acting causes. Thus, beginning with a wire which has been reduced to a 
normal condition by having had the weight off and on, and having been left to itself 
for twenty-four hours, we found on the 24th of August that the magnetization fell from 
a normal value of 685 down to 550 as the result of twisting it to +260,° then to 
— 260, and then to zero of torsion. The second application of positive and negative 
torsion reduced the magnetization further to 534. It is curious to find that not 
merely does torsion diminish the magnetization temporarily, but that it leaves so large 
a permanent diminution. Whether this permanence is absolute in respect to time or 
not is an interesting question to be solved by leaving a wire which has been thus dealt 
with absolutely quiescent from day to day, month to month, year to year, century to 
century. It seems, however, that but slight mechanical disturbance suffices to shake 
out the diminution of magnetization left at the end of each of these experiments. 
228. The general lagging of effect is shown by the fact that in every ascending 
branch the curve is lower than in the immediately previous descending branch; and 
the dotted latter halves of the long curves of the Diagrams XXIV. to XXIX. show, 
by the intersections of their convex portions near the zero, that if the experiment was 
continued long enough, the history of the variation of magnetization would in every 
case be represented by a curve like that in the annexed sketch. 
