1885.] Experimental Researches in Magnetism. 59 



(4.) The energy expended in producing cyclic changes of magneti- 

 sation. 



(5.) The ratio of residual to total induced magnetism. 



(6.) The changes of induced and residual magnetism caused by 

 changes of stress. 



(7.) The effects of constant stress on magnetic susceptibility and 

 retentiveness. 



(8.) The changes of magnetism caused by changes of temperature. 



(9.) The effect of temperature on magnetic susceptibility. 



The experiments were conducted on pieces of metal which gave as 

 near an approach to the condition of uniform magnetisation as is prac- 

 tically attainable. In some cases rings were used in a manner similar 

 to that employed by Stoletow and Rowland, except that in the present 

 case the process of magnetisation was effected in a series of sudden 

 steps, the total magnetisation being found by summing up the induc- 

 tive effect of the steps. In most cases, however, the author used 

 straight pieces of wire, whose length was 300 or 400 times their 

 diameter. These permitted the use of a direct magnetometric method 

 of observation, and consequently the magnetising force could be 

 applied and removed continuously and slowly, instead of by sudden 

 steps. Preliminary experiments had satisfied the author that wires of 

 this length gave as near an approach to uniform magnetisation as 

 could be practically obtained by using rings. 



Curves are given which show the behaviour of iron and steel in 

 various states of temper, when subjected to a first application of mag- 

 netising force, and also to subsequent cyclic changes of magnetising 

 force, such as complete or partial removal and re-application, or 

 reversal. The curves are drawn by plotting either |j, the intensity of 

 magnetisation, or ^, the magnetic induction, in relation to |p, the 

 magnetising force : the characteristics of these curves and their rela- 

 tion to the physical state of the piece under examination are pointed 

 out. Curves so drawn invariably exhibit the static lagging action to 

 which the author (in a former paper) gave the name " hysteresis," any 

 cyclic change of jp giving rise to a more or less nearly closed loop in 

 the curve. Attention was previously drawn to these loops by Warburg, 

 who also anticipated the author in pointing out their important phy- 

 sical meaning, namely, that the area of a loop, or — f^dfj, is the measure 

 of the energy expended (per unit of volume) in performing the cycle of 

 magnetisation which the loop describes. In the present paper numer- 

 ous absolute measurements of this energy are given, especially of the 

 energy which is thus dissipated in each reversal of the magnetism of 

 a piece of iron or steel. These show that while the dissipation of 

 energy by reversal of magnetism is very much smaller in soft iron 

 than in hard iron or steel, even in the latter its amount is very trifling, 

 so that the principal part of the heat which is produced in the cores of 



