1904.] 



Thermoelectric Power and Magnetisation. 



413 



" On the Changes of Thermoelectric Power produced by Magneti- 

 sation, and their Eelation to Magnetic Strains." By Shelford 

 Bid well, M.A., Sc.D. 5 F.E.S. Beceivecl April 11, — Eead 

 April 28, 1904. 



Summary. 



It is well known that magnetisation generally produces a change both 

 in the thermoelectric quality of a magnetisable metal, and also in its 

 linear dimensions. In an article published in October, 1902,* I directed 

 attention to the remarkable qualitative correspondence which appeared 

 in several cases to exist between the two classes of phenomena, and 

 the experiments which form the subject of the present paper were 

 undertaken with the view of investigating their apparent relation. 

 Some of the results were of an unexpected character. The accepted 

 statements regarding certain thermoelectric effects to which in this 

 connection special importance was attached, turn out to be at least not 

 generally true, and, as far as my own observations go, altogether 

 erroneous. Several experimenters appear to have been misled by 

 regarding as unmagnetisecl a piece of metal which either retained some 

 permanent magnetism or was continuous with the piece subjected to 

 magnetisation. What they in fact observed was not the thermoelectric 

 power of magnetised with respect to unmagnetised metal, but that of 

 more strongly - magnetised with respect to less strongly magnetised; 

 and the effects may 'be, as will appear, directly opposite in the two cases. 

 Probably also mistakes have arisen from the assumption that for a 

 lengthened period the electromotive forces in the circuit underwent no 

 changes other than those clue to magnetisation. It is hardly possible 

 to keep the temperatures throughout so nearly constant as to avoid 

 gradual changes of electromotive force considerably greater than those 

 which it is desired to measure. In my own work care M precautions 

 were taken against both these sources of error ; before every observa- 

 tion the metal was demagnetised by reversals, and the galvanometer 

 reading for no magnetisation was noted. 



Although, as mentioned, some of the chief grounds which formed 

 the basis of my conjecture as to a possible relation between the thermo- 

 electric and the strain effects have disappeared, strong evidence is 

 nevertheless forthcoming, that for iron and nickel there is such a 

 relation, and that not merely qualitative but quantitative. As regards 

 one important detail there is an unexplained inconsistency in the 

 behaviour of the two metals, but the coincidences observed in both 

 cases are too numerous and too varied to be the result of accident. 



Iron. — For iron, when allowance is made for the purely mechanical 

 * ' Encj. Brit.,' yol. 30, p. 449, article " Magnetism." 



VOL. LXXIII. 2 G 



