344 BELL SYSTEM TECHNICAL JOURNAL 



and thereupon, those which belong to the chains originally inclined 

 at more than 90° to the field are more than halfway turned around, 

 and when the field is nullified they will realign themselves with their 

 first associates, but every one will be reversed. Originally the net 

 magnetization of the assemblage of chains was nil, for half neutralized 

 the other half; now it is considerable, for half have been inverted. 

 Its ratio to the total magnetization of all the chains when parallel is, 

 in fact, one half. This consequently would be an adequate model 

 for a substance of which the remanence is one half of the saturation- 

 intensity. 



Other values than one half for the ratio of remanence to saturation 

 can be derived from Ewing's picture by choosing a suitable arrange- 

 ment for the elementary magnets. Suppose, for another and a final 

 example, that they are arranged in a cubic lattice, so that each has 

 the choice (as it were) of orienting itself along any one of the directions 

 parallel to the cube-edges. Chains of magnets may then form them- 

 selves along any one of six possible directions (counting the two 

 opposite senses of any line parallel to a cube-edge as two distinct 

 directions). In a demagnetized crystal, one may imagine that the 

 elementary magnets in the lattice fall into groups or "complexes," 

 within each of which all the chains are parallel, while from one complex 

 to the next they change over from one to another of the six specified 

 possible directions. In a demagnetized piece of metal composed of 

 many small crystals oriented quite at random, there will be chains of 

 magnets pointing in all directions. To such a piece of metal let a 

 field be applied, increased to so great an amount that it saturates 

 the material, and reduced gradually to zero. Whatever the direction 

 of the field, it will be inclined at 45° or less to one or more of the six 

 possible directions for the magnet-chains in every crystal. As the 

 field is varied in the manner which I have described, the magnets in 

 each crystal will be wheeled into parallelism, and subsequently will 

 relapse into chains pointed in that direction (or those directions) 

 which makes the least angle with the field. The ratio of remanence 

 to saturation for a polycrystalline sample, resulting from this model, 

 should then be 0.893. 



By adjusting the disposable constants, Ewing's model maybe made 

 to predict not only the general shape of the I-vs.-H curve, but the 

 values of fieldstrength and magnetization at which the first segment 

 of the curve should pass into the second. Apparently no very pleasing 

 agreements between experiment and theory have yet been attained in 

 this way. Nevertheless I will show how the attempt is made; by 

 doing so, I can at least bring out the influence of the various dispos- 

 able constants upon the result. 



