DIMENSIONS OF RINGS AND RODS OP IRON AND OTHER METALS. 229 



the crosses, and the dots are seen to lie evenly and regularly a little below it. A 

 discrepancy of this kind might, under the circumstances, have been foreseen. It is 

 beyond doubt due to the fact which I have frequently noticed and already remarked 

 upon in this paper, that the degree of retraction which nickel undergoes when 

 magnetised is materially affected by differences of temperature,* a fact which may 

 be explained by the well-known influence of heat in diminishing the magnetic suscep- 

 tibility of that metal. Now the temperature of Mr. PRBECE'S engine-room on the 

 afternoon of May 25, when the secondary battery experiments were carried out, was 

 most uncomfortably high, whereas the air pf my laboratory when the other experiments 

 were made, three days later, was bitterly cold. Unfortunately no thermometer 

 readings were taken, but the difference of the temperatures on the two occasions 

 could not have been less than 10 C. Such a difference is, I think, sufficient to 

 account for the discrepancy of the two series of results. 



The results of the experiments referred to in this addendum may be stated as 

 follows. 



Save as to mere details, which may be expected to vary more or less with different 

 specimens of the metals used, according to their purity and physical condition, the 

 results lastly obtained are in agreement with the former ones, so far as these go. 



With regard to details, the new specimen of cobalt reaches its maximum retraction 

 at an earlier stage than the old one, namely, at about 300 units instead of 400. Both, 

 however, agree in not yielding the smallest indication of either retraction or elongation 

 in weak fields. 



The retraction exhibited by the new nickel rod is enormous, being for equal mag- 

 netising forces more than twice as great as that of the old one, and ultimately 

 amounting to nearly l/40000th part of its length. I believe this to be accounted for 

 by its superior purity. 



The new iron curve differs from that in fig. 4 in showing a somewhat smaller 

 maximum elongation and a greater amount of retraction, but the two are of just the 

 same general character. 



The latest results are so far merely confirmatory of the earlier ones. But they go 

 further, and afford information concerning the behaviour of the metals in far stronger 

 tields than were before obtained. 



They show clearly that the retractions of iron and of nickel reach a limit in fields 

 of a certain intensity, about 1000 for iron and 800 for nickel in the specimens 

 examined, the retraction in stronger fields being neither greater nor less. My conjec- 

 ture that a minimum length might possibly be passed is therefore not supported. 



On the other hand, it appears that the length of the cobalt rod, after passing through 

 a minimum, regains its original value in a field the strength of which is for this 

 particular rod about 750 units, and then rapidly increases, the increments of length 



I find that Professor BARBKTT had observed this effect of heat. ' Nature, 1 vol. 26, 1882, p. 586. 

 June 13, 1888. 



