594 
PROFESSOR J. A. EWING ON EXPERIMENTAL 
When the wire is annealed in the magnetising field, the amount of magnetisation 
which it possesses when it has become cool is an uncertain quantity, and depends 
much on the rapidity of the cooling, being greater for slow than for fast cooling. As 
soon as loading begins the magnetism rises rapidly, especially when it has chanced 
(perhaps from the cooling having happened somewhat quickly) to have been initially 
low. In fact, the annealed wire is, magnetically, in an extremely unstable state, and 
responds with great sensitiveness to the first application of load, every step of which 
(up to a certain limit) causes a rise of magnetism. This rise is not lost on the with¬ 
drawal of the stress; the application and removal of load leaves the wire permanently 
more magnetic. Indeed, the first application and removal of a small load produces 
an augmentation of magnetism not unlike that which would be produced by tapping. 
This continues during several successive loadings, and it is only when the load has 
been applied and removed many times that the changes of magnetism become 
sensibly cyclic. 
§ 78. Taking now the particular case of the annealed iron wire already dealt with, 
the character of the effect of load when it becomes cyclic depends on the limit to 
which loading has been carried. If the total load has been only 1 kilo, (corresponding 
to an intensity of stress of say 2'1 kilos, per sq. mm.), then its application gives 
increase and its removal gives decrease of magnetism, when the cyclic state is reached. 
But if the load has been 3 kilos., or anything greater (still not exceeding the limit 
of elasticity), the effect when it becomes cyclic is that on gives fall and off gives rise 
of magnetism. This corresponds to the effects shown in fig. 36 for the cycles of load 
from 0-3-0 up to 0-9-0. 
Now it would seem, from the fact that when 1 kilo, only is applied and removed on 
gives rise and off gives fall of magnetism, that there is essentially a maximum of 
magnetisation in this annealed wire corresponding to a very small load, and this view 
is confirmed by curves taken with vibration, for in these a maximum appears at or 
near 1 kilo, of load. But this maximum does not appear in cyclic curves given by 
applying and removing higher values of the load. Its absence is, however, satisfac¬ 
torily explainable as one of the effects of hysteresis. In fact, it still exists, but 
hysteresis shifts it to the left of the origin, so that successive applications and 
removals of pull lie on one side of it. If we could trace the effects of push as well as 
pull, the presence of this maximum would be immediately detected. The diagram 
below will make my meaning more clear. 
Starting from A (a point reached, say, by tapping the annealed wire, when the load 
is zero), and beginning to load, we find the magnetism change in the manner repre¬ 
sented by A B C. Then on removal of the load, the curve is C D E, E being the 
value of magnetism corresponding to no load. The off curve is still rising at this 
point, and my explanation of this rise, and of the apparent absence of a maximum 
point, is that the maximum point is shifted to the left by hysteresis so much as to 
