diminutions and augmentations of about 12; that is to say, with 56 lbs. hung on, in addition to the constant 28 lbs., the magnetization of the wire would have been 53, and with only the 28 lbs. it would have been 65.

The M (−), on, off, and B which actually followed, added 154 to the magnetization, and so brought it up to about 804, or 52 more than it had at the beginning of Series XXXI. The subsequent M (-) and B probably produced little, if any, further change in the residual magnetization; and the "on," "off," "on," which followed, confirmed the previous result of augmentation 12 and diminution 12, alternately by off, and on, after something considerable had been shaken out permanently by the first "on." The "off,” "on," "off" after M (-) with current still flowing gave about 10 instead of the 12 found previously when the current was not flowing. These results agree in kind and amount with what was to be expected from Series XIV. and XV., considering that "on" and "off" in those series was 84 lbs. on and off (with nothing, or only a very slight steadying weight kept always on), whereas now the "on" and "off" means change from 28 lbs. to 84 lbs. and back.

192. Interpretation of Series XXXI. continued. Taking now the effects of the Ms and Bs, and (§ 191) subtracting 20 from each number, we see that the effect on the magnetization of the wire producible by repeating M (-) and B over and over again without other disturbance, would be about 16 each way with 84 lbs. on, and probably between 5 and 9 with 28 lbs. on. (This agrees in kind with the conclusion deducible from the comparison between the previous Series II. and VI.; but the absolute magnitudes of the results seem smaller, probably because of less battery-power in those series than in Series XXXI.) The whole effects of the M (+) after M (-) and B, and of the M (-) after M (+) and B, were still, as in Series XXX., each equal to 180 or 181, with only the 28 lbs. on, and therefore (subtracting 20 for the induction from coil to coil) we had 160 or 161, say 1601, for the sum of the demagnetization and reverse magnetization produced by a M following a B and a previous M of the opposite direction. Now came, in the course of a few minutes, a most startling discovery. With the 84 lbs. on, the sum of magnetization and demagnetization produced by the M after B from previous opposite M was 130, or less by 30 with those than with the 28 lbs. Thus we see that

while the magnetic effect of stopping the current is greater, the effect of subsequently instituting the current in the reverse direction is less, with the heavy than with the light weight; and less by about three times as great a difference.

193. Consider, lastly, the effect of a sudden reversal of the current. This may be regarded as the sum of the effects of stopping it, and starting it in the reverse direction; and therefore may be expected to be less with the 84 lbs. than with the 28 lbs. by about of the difference between the effects of starting the reverse current with the heavy and with the light weights hung on the wire. This inference is verified in the concluding thirteen results of the series before removal of the steel wire. Thus (§ 190) subtracting 40 for the induction of coil on coil in the reversal, we find 140, 139, 141, 136, 140, 139, 139 for magnetic effects of reversals with 84 lbs. hung on the wire, of which the mean is 1392; and 159, 158, 159, 160, 160, 160, for the magnetic effects of reversals with 28 lbs. on, of which the mean is 159.3.

194. After the conclusion of these experiments on the steel wire, I made many experiments of the same kind on soft-iron wires of various qualities substituted for it in the same apparatus, and I have obtained results of the same kind, as to the effects of hanging on and taking off weights, while the magnetizing current is kept flowing. I have also obtained some very remarkable and perplexing results by putting weights on and off with the current not flowing. In one of the iron wires the effect found was opposite to that in steel; that is to say, putting on weight augmented and taking off weight diminished the residual magnetism ; in another the same effect as in steel was found, that is, putting on diminished and taking off augmented the residual magnetism. Neither of these was as soft as some of the other wires tried, and the one ("bright soft iron wire," Johnson's) that agreed with steel was remarked on at the time as much harder than another that had been previously experimented on ("black soft iron wire," Johnson's). This latter seemed utterly destitute of retentive power under the influence of putting the weights on and off. Like all the others it always experienced a diminution of magnetism by weights on and increase by weights off, when the magnetizing current was flowing. But when the current was stopped large effects (larger than those when the current was

flowing) were produced by putting on and taking off the weights; and these effects were always of the same kind, whichever had been the direction of the current.

195. I have not yet been able to explain these effects by terrestrial magnetic force*, nor to even guess any other possible cause; and have in fact, since the 23rd of December last, been exceedingly perplexed by seeming anomalies which the various soft-iron wires tried have presented, commencing that day with a reversal of the electro-magnetic effect of the "off" and "on" which the first of the wires experimented upon showed, after the magnetizing current had been made for the first time and broken, and a weight of 14 lbs. put on and off several times had first shaken out nearly one-third of the residual magnetism, and then given alternate augmentation by "on" and diminution by "off" of the magnetism that remained. A weight of 28 lbs. was then hung on, and it stretched the wire permanently by about 8 per cent. of its length. Then immediately I found reverse effects by putting off and on and off the 28 lbs., and on and off smaller weights—the "on" giving diminution and the "off" augmentation of what would have been the residual magnetism, if residual magnetism there was, from the first magnetization by the current. This quality remained until an hour or two later, when the current was once more made in the same direction as the first, and broken again. Then 14 lbs. put on actually gave augmentation of the residual magnetism (instead of shaking out a considerable quantity as the first "on" after the first "B" had done), and the "offs," "ons," and "offs" following gave alternate diminution (by the "offs") and augmentation (by the "ons"); that is to say, the effects were the same in kind as the effects which had been observed before the stretching by the 28 lbs.; but they were nearly three times as great in amount as they had been then, with the same weight of 14 lbs. on and off. Thenceforth the same piece of wire (experimented on several more days up till Jan. 12, 1875) and other pieces of similar wire tried after it, with the current made sometimes in one direction and sometimes in the other, and broken,

* [Note added January 1877.-Nearly six months later I ascertained that these startlingly great effects actually were due to the vertical component of the earth's force, though this was only about of the magnetizing force of the currents used.]

1

300

and different amounts of weight up to 28 lbs. put on and off, always showed increase of the residual magnetism by the "on and diminution by the "off." Even the first "on" after the stoppage of the current gave always an increase of the magnetism; but when the weight was as much as 28 lbs. the "shaking out tendency was remarkably shown in the increase by the first "on being much less than the diminution by the first “off.”

"

196. The soft-iron wire experimented upon, on the 23rd of December gave, with 28 lbs. hanging on it, smaller effects of successive "makes," in one direction, and "breaks" of the current; a greater effect when it was made in the reverse direction; and a smaller sum of these two effects (that is to say, a smaller effect) on the reversal of the current than when it was un-pulled.

197. The investigation is being continued with special arrangements to discover the explanation of the seeming anomalies described above, and with the further object of determining in absolute measure the amounts of all the ascertained effects, at different temperatures up to 100° Cent. It is needless to give in the mean time any minute details of the experiments already made on the soft-iron wires by which the results now described were obtained.

PRELIMINARY NOTICE TO PART VII.

[From the Proceedings of the Royal Society, June 10, 1875.] EFFECTS OF STRESS ON INDUCTIVE MAGNETISM IN SOFT IRON.

197. At the last ordinary meeting of the Royal Society (May 27, 1875), after fully describing experiments (Part VI. above) by which I had found certain remarkable effects of stress on inductive and retained magnetism in steel and soft iron, I briefly referred to seeming anomalies presented by soft iron which had much perplexed me since the 23rd of December, 1874. Differences presented by the different specimens of soft-iron wire which I tried complicated the question very much; but one of them, the softest of all, a wire specially made by Messrs Richard Johnson and Nephew, of Manchester, for this investigation, through the kindness of Mr William H. Johnson, gave a result standing clearly out from 23

T. II.

the general confusion, and pointing the way to further experiments, by which, within the fortnight which has intervened since my former communication, I have arrived at a complete explanation of all that had formerly seemed anomalous. These experiments have been performed in the Physical Laboratory of the University of Glasgow by Mr Andrew Gray and Mr Thomas Gray, according to instructions which, in my absence, I have sent them from day to day by post and telegraph.

197,'. The guiding result (described near the end of my former paper, and referred to in the last paragraph but one of the Abstract [II., Appendix below] in Proceedings of the Royal Society for May 27, 1875) was, that the softest wire, tried with weights on and off repeatedly, after it had been magnetized in either direction by making the current, in the positive or negative direction, and stopping it, gave effects on the ballistic galvanometer which proved a shaking out of residual magnetism by the first two or three ons and offs, and a gradual settlement into a condition in which the effect of " on was an augmentation, and the effect of "off" a diminution, of the inductive magnetization due to the vertical component of the earth's magnetizing force. When a fresh piece of the same wire was put into the apparatus and tested with weights on and off it gave this same effect. If the wire had been turned upper end down and, tried again in the course of any of the experiments, still the same effect would have been shown. It seemed perfectly clear that in these experiments there was no other efficient dipolar quality of the apparatus by which the positive throw of the ballistic galvanometer could be given by putting on the weight, and the negative throw by taking it off, than the vertical component of the earth's magnetic force.

3

197,'. Yet I did not consider that I had explained the result by the terrestrial influence, because, for all the specimens of steel and soft iron, the effect of weights on had been uniformly to diminish, and of weights off to augment the magnetism when the magnetizing current was kept flowing. And I was, moreover, perplexed by the magnitude of the result-the effect of weights on and off shown by the very soft iron wire, under only the feeble magnetizing influence of the earth, being many times (from three times to nine or ten times) as great as the effects which the same weights on and off produced in the same wires when