CONTEMPORARV ADVANCES IN PHYSICS 311 



Recently it became possible to make individual crystals so large that 

 one of them, or an aggregate of a few, is by itself large enough to 

 serve as a sample for magnetic testing. The I-vs.-H relation for an 

 individual crystal is very different from the relation for a mass of 

 tiny crystals of the same material. In fact, the vector I is usually 

 not parallel to the vector H. A magnetic field, applied to an ellipsoid 

 cut from a single crystal, magnetizes it askew unless the field, and an 

 axis of the ellipsoid, and an axis of the crystal happen to be all parallel 

 to one another. In the polycrystalline mass, these deviations between 

 direction of field and direction of magnetization must be averaged, 

 and cancel one another out; for otherwise, the universally made 

 assumption that / is parallel to // would not have been effectual. 

 No doubt it is fortunate that Nature, with a rare benevolence, simpli- 

 fied the data first presented to the students of magnetism by this 

 averaging and this cancellation. We cannot however conclude with 

 safety that an assemblage of small crystals will behave just like an 

 assemblage of equally many large ones. Evidently the size of the 

 crystal must influence its I-vs.-H relations, or else the boundaries 

 between adjacent crystals affect the magnetization, or there is some- 

 thing inherent which changes concurrently with the degree of crystal- 

 lization. At any rate, whenever the crystallization of a sample is 

 varied, the I-vs.-H curve is liable to feel it. 



Composition and strain, temperature and current, state of crystal- 

 lization — one must be prepared to find a new way of dependence of 

 / upon // for each combination and every gradation of these; and 

 yet the half has not been told. Whenever stress or heat are applied 

 to a magnetizable substance, they alter its I-vs.-H relation, twt 

 merely while they are being applied, but after they are withdrawn. After 

 such an experiment one may restore the original temperature and the 

 original stress or freedom from stress, but the material is no longer 

 quite the same. Vibrations and concussions, compressions and 

 tensions and twistings, bending and tapping and cold-rolling and 

 hammering, heating and cooling, annealing and quenching, the very 

 act of magnetization itself — each of these is liable not merely to affect 

 the I-vs.-H curve while it prevails, but to transform the substance 

 permanently into another and a distinct ferromagnetic substance, 

 with a system of magnetization-curves distinct from what the sample 

 showed beforehand. 



If we could see into the penetralia of a piece of iron, and discern 

 the conditions and the arrangements of its atoms, it is probable that 

 we should see that every such agency leaves behind it some definite 

 and enduring change; and then we should not wonder at (for example) 



I 



