398 Professor J. A. Ewing [May 22, 



successive applications and removals of the load (see Fig. 13), and 

 it is only when the process of loading has been many times repeated 

 that the magnetic change brought about by loading is just opposite 

 to the magnetic change brought about by unloading. 



Whenever, indeed, we are observing the effects of an alteration of 

 physical condition on the magnetism of iron, we have to distinguish 

 between the primitive effect, which is often very great and is not 

 reversible, and the ultimate effect, which is seen only after the 

 molecular structure has become somewhat settled through many 

 repetitions of the process. Experiments on the effects of temperature, 

 of strain, and so forth, have long ago shown this distinction to bo 

 exceedingly important : the molecular theory makes it perfectly 

 intelligible. 



Further, the theory makes plain another curious result of experi- 

 ment. When we have loaded and unloaded the iron wire many times 

 over, so that the effect is no longer complicated by the primitive 

 action I have just described, we still find that the magnetic changes 

 which occur while the load is being put on are not simply undone, 

 step by step, while the load is being taken off. Let the whole load 

 be divided into several parts, and you will see that the magnetism has 

 two different values, in going up and in coming down, tor one and 

 the same intermediate value of the load. The changes of magnetism 

 lag behind the changes of load : in other words, there is hysteresis in 

 the relation of the magnetism to the load (Fig. 14). This is because 

 some of the molecular groups are every time being broken up during 

 the loading, and re-established during the unloading, and that, as we 

 saw already, involves hysteresis. Consequently, too, each loading 

 and unloading requires the expenditure of a small quantity of energy, 

 which goes to heat the metal. 



Moreover, a remarkably interesting conclusion follows. This 

 hysteresis, and consequent dissipation of energy, will also happen 

 though there be no magnetisation of the piece as a whole: it depends 

 on the fact that the molecules are magnets. Accordingly, we should 

 expect to find — and experiment confirms this* — that if the wire is 

 loaded and unloaded, even when no magnetic field acts and there is 

 no magnetism, its physical qualities which are changed by the load will 

 change in a manner involving hysteresis. In particular, the length 

 must be less for the same load during loading than during unloading 

 so that work may be wasted in every cycle of loads. There can be no 

 such thing as perfect elasticity in a magnetisable metal, unless, indeed 

 the range of the strain is so very narrow that none of the molecules 

 tumble through unstable states. This may have something to do 

 with the fact, well known to engineers, that numerous repetitions of 

 a straining action, so slight as to be safe enough in itself, have a 

 dangerous effect on the structure of iron or steel. 



* See Phil. Trans. 1885, p. GU. 



