Jan. 1 6, 1890] 



NA TURE 



251 



retains a large part of this property after the current has 

 ceased. We may push the experiment a stage further. 

 Suppose that the current in the primary is raised to a 

 great value, and is then slowly diminished to a smaller 

 value, and that the ring is opened and the secondary 

 coil withdrawn. With most substances we find that 

 the galvanometer deflection is precisely the same as if 

 the current had been simply raised to its final value. It 

 is not so with iron : the galvanometer deflection depends 

 not alone upon the current at the moment of withdrawal, 

 but on the currentto which the ring has been previously sub- 

 jected. We may then draw another curve (Fig. 2) represent- 

 ing the galvanometer deflections produced when the current 

 has been raised to a high value and has been subsequently 

 reduced to a value indicated by the abscissa. This curve 

 may be properly called a descending curve. In the case 

 of ordinary bodies this curve is a straight line coincident 

 with the straight line of the ascending curve, but for iron 

 is a curve such as is represented in the drawing. You 

 observe that this curve descends to nothing like zero when 

 the current is reduced to zero ; and that when the current 

 is not only diminished to zero, but is reversed, the galvano- 

 meter deflection only becomes zero when the reversed 

 current has a substantial value. This property possessed 

 by magnetic bodies of retaining that which is impressed 



upon them by the primary current has been called by 

 Prof. Ewing " hysteresis," or, as similar properties have 

 been observed in quite other connections, " magnetic 

 hysteresis." The name is a good one, and has been 

 adopted. Broadly speaking, the induction as measured 

 by the galvanometer deflection is independent of the time 

 during which the successive currents have acted, and 

 depends only upon their magnitude and order of succes- 

 sion. Some recent experiments of Prof. Ewihg, however, 

 seem to show a well-marked time effect. There are 

 curious features in these experiments which require more 

 elucidation. 



It has been pointed out by Warburg, and subsequently 

 by Ewing, that the area of curve 2 is a measure of the 

 quantity of energy expended in changing the magnetism 

 of the mass of iron from that produced by the current 

 in one direction to that produced by the current in the 

 opposite direction and back again. The energy expended 

 with varying amplitude of magnetizing forces has been 

 determined for iron, and also for large magnetizing forces 

 for a considerable variety of samples of steel. Different 

 sorts of iron and steel differ from each other very greatly 

 in this respect. For example, the energy lost in a com- 

 plete cycle of reversals in a sample of Whitworth's mild 

 steel was about 10,000 ergs per cubic centimetre ; in oil- 



FlG. 2. 



hardened hard steel it was near 100,000 ; and in tungsten 

 steel it was near 200,000 — a range of variation of 20 to i. 

 It is, of course, of the greatest possible importance to 

 keep this quantity low in the case of armatures of dynamos, 

 and in that of the cores of transformers. If the armature 

 of a dynamo machine be made of good iron, the loss from 

 hysteresis may easily be less than i per cent ; if, how- 

 ever, to take an extreme case, it were made of tungsten 

 steel, it would readily amount to 20 per cent. In 

 the case of transformers and alternate-current dynamo 

 machines, where the number of reversals per second is 

 great, the loss of power by hysteresis of the iron, and the 

 consequent heating, become very important. The loss of 

 power by hysteresis increases more rapidly than does the 

 induction. Hence it is not well in such machines to 

 work the iron to anything like the same intensity of in- 

 duction as is desirable in ordinary continuous current 

 machines. The quantity O A, when measured in proper 

 units, as already explained — that is to say, the reversed 

 magnetic force, which just suffices to reduce the induction 

 as measured by the kick on the galvanometer to nothing 

 after the material has been submitted to a very great 

 magnetizing force— is called the " coercive force," giving 

 a definite meaning to a term which has long been used in 

 a somewhat indefinite sense. The quantity is really the 

 important one in judging the magnetism of short per- 



manent magnets. The residual magnetism, o B, is then 

 practically of no interest at all ; the magnetic moment 

 depends almost entirely upon the coercive force. The 

 range of magnitude is somewhat greater than in the case 

 of the energy dissipated in a complete reversal. For 

 very soft iron the coercive force is r6 C.G.S. units ; for 

 tungsten steel, the most suitable material for magnets, it is 

 51 in the same units. A very good guess may be made of 

 the amount of coercive force in a sample of iron or steel 

 by the form of the ascending curve, determined as I de- 

 scribed at first. This is readily seen by inspection of 

 f '^- 3> which shows the curves in the cases of wrought 

 iron, and steel containing 0*9 per cent, of carbon. With 

 the wrought iron a rapid ascent of the ascending curve is 

 made, when the magnetizing force is small and the 

 coercive force is small ; in the case of the hard steel the 

 ascent of the curve is made with a larger magnetizing 

 current, and the coercive force is large. There is one 

 curious feature shown in the curve for hard steel which 

 may, so far as I know, be observed in all magnetizable 

 substances : the ascending curve twice cuts the descend- 

 ing curve, as at M and N. This peculiarity was, so far as 

 I know, first observed by Prof. G. Wiedemann. 



I have already called emphatic attention to the fact 

 that magnetic substances are enormously magnetic, and 

 that non-magnetic substances are hardly at all magnetic : 



