1882.] jRetentiveness in the Magnetisation of Iron and Steel. 43 



ing and removing 7 kilos, before the wire had been stretched at all 

 by any greater load. The lower part shows the effect of applying and 

 removing weights nearly equal in all to 16 kilos, after the wire had 

 been previously stretched by the^ame load. The initial magnetism for 

 zero stress had then fallen to 104, but during application it rose to 

 174 with 10 kilos., and again during removal it rose to 172 with 

 6^ kilos. 



The series of experiments from which this figure is taken shows 

 that the maximums of magnetisation during loading and unloading 

 (when conducted without shaking the wire) appear only after some 

 permanent set has been given, and that they gradually shift out to 

 the right as the amount of permanent set is progressively increased. 



A number of other iron wires tested in the same way agree with 

 this one in giving decrease of magnetism for " on," and increase for 

 ^'off" before stretching beyond their limits of elasticity, and after- 

 wards increase for "on " and decrease for " off." It appears that (at 

 least when the strain has occurred in the circumstances in which it 

 occurs here) this difference of behaviour forms an unfailing criterion 

 by which we may distinguish a piece which has received permanent 

 set from a piece in the annealed state. 



A careful examination of the initial parts of the curves which are 

 formed when after loading we change (without mechanical dis- 

 turbance) to unloading, or vice versa, has brought out the fact that, 



calling I the magnetisation and p the stress, — is always initially 



