CONTEMPORARY ADVANCES IN PHYSICS 349 



ones. This notion is endangered by the discovery that a single 

 crystal of iron displays only a slight degree of hysteresis, much less 

 than a polycrystalline mass — a discovery which likewise weakens the 

 force of calculations of remanence based upon the assumption of a 

 cubic lattice, such as I gave earlier. In fact, it seems quite probable 

 that in the course of assimilating the newly-acquired data concerning 

 single crystals, all of the numerical agreements hitherto derived from 

 Ewing's and other theories of ferromagnetism may be swept away. 



Nevertheless the basis of Ewing's theory is likely to persist; for it 

 has two great advantages which are nearly independent of numerical 

 agreements. Hysteresis is derived from an atom-model in which 

 nothing of the nature of hysteresis is introduced by postulate; and 

 the general effect of mechanical and electrical jerkings, bumpings and 

 jogglings is explained in a way which seems most natural and plausible 

 to our mechanical instincts. As for the first point: to explain hyster- 

 esis by the mutual interactions of magnets which in themselves are 

 constant and do not possess it is so eminently satisfactory a solution 

 that any theory or model in which hysteresis is introduced ab initio 

 or derived from some extra assumption will start under a great handi- 

 cap. As for the second: to take one illustration, it is well known 

 that a demagnetized piece of iron exposed to a weak field, and endowed 

 thereby with the moderate magnetization corresponding to some point 

 or other on the first segment of the initial curve, becomes enormously 

 more intensely magnetized when it is jolted or jerked. Visualized by 

 Ewing's model, this seems the most natural thing in the world: the 

 elementary magnets which were on or close to the verge of capsizing 

 are pushed over that verge by the mechanical shock. Equally 

 natural seem the annulment of the residual magnetism of a piece of 

 iron, by mechanical shocks and jerks; the like effect of rapidly- 

 alternating magnetic fields; the tendency of a current along an iron 

 wire to favor magnetization of the wire; and the fact to which I 

 alluded earlier, that in a very strong rotating magnetic field a piece 

 of iron does not grow hot, and consequently there can be no hysteresis- 

 loss. This last-named feature may be visualized by supposing that 

 the chains, having been once completely broken up, do not form 

 again as the magnets are whirled round and round. It seems natural 

 also to expect that as the temperature is raised, the chains will be 

 broken up by thermal agitation, and the reversible first segment of 

 the initial curve will mount more sharply and continue longer. 

 In trying to deal with the effect of temperature, however, we soon 

 reach the limits to which Ewing's theory can be forced; and another 

 method of attacking the problem of ferromagnetism recommends 

 itself. 



