40 Prof. J. A. Ewing. On Effects of [May 25,.. 



method (as used by Rowland and others), I have examined at great 

 length the changes of magnetisation which occur in iron and steel 

 when the magnetising force is progressively increased, diminished, 

 again increased, reversed, and so on. The results show in the most 

 conclusive manner that all changes of magnetisation produced by 

 slow or fast, continuous or discontinuous, changes of the magnetising 

 force exhibit hysteresis. If we carry the metal through any cycle of 

 magnetisation, the curves giving the relation of I (the intensity of 

 magnetisation) to H (the magnetising force) form loops, and it does 

 not appear that the loops are different in any essential respect (except 

 size) when the action is confined to one sign from the loops given 

 when the sign of the magnetisation is reversed. 



The remarkable feature of the curves is, that when the magnetisa- 

 tion of iron is conducted in such a manner as to be uniform throughout 

 the piece experimented on, the initial change which occurs when we 

 pass from increase to decrease of the magnetising force, or vice versa, 

 is indefinitely small relatively to the initial change of the force. In 

 other words, say that we stop decreasing H and begin to increase it,. 

 dl 



then is at first zero. 



dtl 



The difference between the curves for increase and decrease of the 

 magnetic force is of a perfectly static character. If it is to be 

 explained by internal friction, the friction is analogous to that of 

 solids, and does not at all resemble the viscosity of liquids. The 

 phenomenon here described is independent of the quasi- viscous resist- 

 ance to changes of magnetisation which is due partly to the induction 

 of currents in neighbouring conductors, including the magnet itself, 

 and partly to the thermomagnetic properties of the metal discussed 

 by Sir W. Thomson ("Phil. Mag.," vol. v, 1878, pp. 24-25). The 

 influence of those causes disappears when the changes of magnetisa- 

 tion take place very slowly, or when a sufficient interval of time is 

 left after each change of magnetic force before a reading of the 

 magnetisation is taken. 



When any cyclic change of I is made to take place by varying H 

 cyclically, the area of the loop so formed, or fldH, is not only pro- 

 portional to, but actually the measure of the work done on the magnet, 

 per unit of volume, in performing the cycle. In cases where changes 

 of the magnetisation take place very slowly this is wholly spent on' 

 the magnet itself, and its equivalent is, no doubt, to be found in the 

 heating effect of the cycle. When, however, the changes of magne- 

 tisation take place at a finite rate, this area must of necessity be 

 greater, since the work done in performing the cycle is then greater 

 for two reasons ; first, because of the energy expended in inducing 

 currents in neighbouring conductors ; and, second, because of the 

 dissipation of energy involved in the heating and cooling effects 



