RESEARCHES IN MAGNETISM. 
623 
They show that the Yillari reversal occurs earlier in stretched iron the greater the 
stress is, but that even with a stress approaching the breaking strength of the 
material it does not come till 3 is more than 1000. For loads less than 12 kilos, it 
would, no doubt, occur at values of 3 greater than those reached in this experiment. 
§ 108. In soft annealed iron, however, the Yillari critical point occurs much 
earlier, as the experiment of § 91, fig. 44, serves to show. There the curves for 1, 2, 
4, and 6 kilos, all cross the normal or no-load curve, at values of 3 which (as in the 
case of stretched wire) are higher the smaller the load, but which are very much 
lower than the values of the critical point in stretched wire. The following are the 
values of 3 and at which the curves with load cross the normal curve in fig. 44 
(each kilo, of load gives a stress of 2T5 kilos, per sq. mm.):— 
Load. 
Y 
A 
kilos. 
1 
1220 
73 
2 
1040 
4-3 
4 
840 
3-4 
6 
690 
3’05 
Effects of Stress on Retentiveness. 
§ 109. In a previous part of this paper (§§ 37-47) I have described experiments 
by which the retentiveness of iron and steel was examined at various degrees of 
magnetisation, by applying a given magnetising force, reducing that to zero and 
observing the residual magnetism, then applying a stronger force, reducing that to 
zero, and so on. Those experiments showed that the ratio of residual to induced 
magnetism passes a maximum as the magnetisation is increased. 
It seemed desirable to extend the same method of experiment to the case of iron or 
steel under longitudinal stress. Accordingly corresponding observations have been 
made on pieces of iron (in the soft annealed and also in the stretched condition) and 
steel, in several different states of stress. 
(June 13, 1882.)—A piece of pianoforte sleel wire, 31 cm. long, after being annealed, 
was stretched about 1 cm. by a load of 22 kilos. The diameter after stretching was 
078 mm. Its retentiveness was then examined (by applying and removing progres¬ 
sively increased values of <£>) under four different states of load, namely, 0, 3, 10, and 
20 kilos. The process of demagnetising by reversals served to reduce the wire to a 
magnetically neutral state after magnetisation under each load, and before the next 
value of the load was applied. 
The table below shows the results of this experiment, by giving the observed values 
of the induced and residual magnetism, and their ratio, for various values of *£> :— 
4 l 2 
