It should be noticed that the field H inside a magnetized body is not in general the same as the field outside it, even when the body is placed in a uniform field, for the free poles on the outer surface of the body exert a "demagnetizing" effect, whose magni tude depends on the shape and dimensions of the specimen.
It may be shown, from energy considerations, that paramagnetics tend to move from weaker to stronger parts of the field, so that they are attracted by a magnetic pole, and if elongated and iso tropic (having the same properties in all directions) they tend to set themselves along the direction of the lines of force (axially) ; diamagnetics tend to move to weaker regions of the field, and to set themselves at right angles to the lines of force (equatorially) —though, in a perfectly uniform field, diamagnetics would set themselves axially, like paramagnetics. The behaviour of crystals may be more complicated owing to lack of isotropy.
The numerical value of the specific susceptibility of diamag netics is usually not greater than (for bismuth, one of the strongest diamagnetics, x = —1.4X For paramagnetics, at ordinary temperatures, the range of values is very wide, and no limits can be fixed, though for many x lies between o and 'cox Both for dia- and paramagnetics (not including under paramagnetics the ferromagnetics) the susceptibility is independ ent of the strength of such applied fields as are attainable, except at very low temperatures. Among paramagnetics, it is convenient to distinguish ferromagnetics (typified by iron), though the differ entiation cannot be made precise until the temperature behaviour is considered in detail. Typical ferromagnetics have a suscepti bility which varies with the field, and may attain a very high value.
(For electrolytic iron /.4 nuts =14,400 has been obtained, corre sponding to x=136; this is exceptional, but values of of several thousand are quite usual.) With ferromagnetics the mag netization, I, tends to a saturation value, which decreases with increasing temperature; in general magnetization is not uniquely determined by the field, but depends on the previous history of the specimen. If the field is gradually increased and diminished in a cyclic manner, the magnetization varies cyclically also, but the value in an increasing field is different from that in a de creasing one. There is a lagging of the magnetization behind the field, or hysteresis.
The behaviour of dia-, para- and ferromagnetics is represented in fig. 3, in which the specific intensity of magnetization (the magnetic moment per unit mass) is plotted against the field strength (a) for a soft iron, (b) for a paramagnetic salt CoSO4, (c) for a diamagnetic, antimony. In the iron curve it will be seen that the magnetization ap proaches a saturation value at A (usually specified by or that when the field is re duced to zero, at B, a certain magnetization remains (the rem anence, or retentivity, being specified by or Brem) ; and that to reduce the magnetization to zero (at C) a certain coercive field is required (the "coercivity," 11,, being specified by the value of the field in gauss to reduce the magnetization to zero). It should be noted that, on account of the different scales, the slope of (b) should be reduced by and that of (c) by to be compared with the (a) curve.