MAGNETISATION ON THE THERMOELECTRIC QUALITY OP IRON. 
379 
load as to appear as a mere inflection in the “ on ” curve, though still present in the 
“ off” curve; while, as part of the same action, the outer or ascending limb of the 
curve becomes the conspicuous portion and occupies nearly the whole figure. 
§ 45. At the conclusion of the above experiments with iron, the effects of stress on 
the thermoelectric quality of certain other metals were examined in the same way, 
with the view of testing for hysteresis in changes of thermoelectric quality caused by 
applying and removing longitudinal pull. Wires of silver, copper, lead, magnesium, 
and german silver were tried, but in no case did the results of loading and unloading 
afford any evidence of hysteresis. In curves showing the relation of thermoelectric 
E.M.F. to load, the “ on ” and “ off” curves did not differ by more than the limits of 
experimental error. With all the wires named, loading produced in the part loaded a 
change towards 'positive* and the change remained nearly proportional to the stress up 
to the largest value to which the load was increased. In silver and magnesium the 
E.M.F. rose to about 4 microvolts when the load was increased to as large a value as 
it was judged safe to apply without risk of breaking the wire ; in the other metals the 
range of thermoelectric change was much smaller. 
Another metal examined was aluminium, which, unlike the others, changed to 
negative when loaded. The change was very small, amounting to only half a micro¬ 
volt under the greatest load. A tin wire showed no measurable change of thermo¬ 
electric quality, either positive or negative, when loaded up to its breaking point. 
So far there is no evidence that the peculiar behaviour of iron, the study of which 
has been the object of this paper, is not peculiar to that metal. It will be interesting 
to see whether similar characteristics are exhibited by the other strongly paramagnetic 
metals, nickel and cobalt. 
§ 46. The changes of thermoelectric quality through which a piece of iron passes 
when subjected to varying amounts of stress are so complex that it seems impracti¬ 
cable, until our knowledge of molecular structure is greatly extended, to attempt any 
mechanical hypothesis by way of explanation. With respect to one very curious 
feature, however, to which attention has been drawn in §§ 13, and 39-41, it may be 
worth while to point out an analogy which suggests a connexion betwnen the 
structure of iron and that of a granular medium. Professor Osborne Reynolds has 
pointed outf that when a substance composed of granules in contact is subjected to 
* Cf. Thomson, Phil. Trans., vol. 146 (1856), p. 729, § 144. 
t Phil. Mag., ser. 5, vol. 20 (1885), p. 469, “ On the Dilafancy of Media composed of Rigid 
Particles in Contact.” 
3 C 2 
