1904.] Thermoelectric Power produced by Magnetisation. 425 



may be found by dividing the change of microvolts by 85. In all 

 the experiments the temperature of the cold water differed little 

 from 14°, the mean temperature being therefore 57°. 



Table I. 



Amperes. 



Galvanometer 

 deflections. 



Mean. 



H — amperes 

 x 92. 



[ 



Microvolts 

 deflections x 0'283. 



0-35 



28, 28, 28, 29 





28 -3 



32 



8-0 



0-8 



31, 33, 34, 32, 



32 



32-2 



74 



9-1 



1 -o 



34, 33, 33, 33, 



34 



33 3 



92 



9-4 



1 -6 



31, 32, 31, 31, 



30 



31 -Qp 



147 



8-8 



2-6 



29, 29, 29 





29 -0 



249 



8'2 



4-0 



24, 24, 25 





24 3 



368 



6-9 



5-5 



20, 19, 20 





19 7 



506 



56 



6-8 



19, 19, 19 





19-0 



626 



5-4 



8 -4 



16, 17, 16 





16 -3 



773 



4-6 



11 -1 



15, i5 





15 



1020 



42 



Curve (B) was given by a piece of commercial iron wire, which was 

 hard, springy, and easily broken. Changes of thermoelectric force, 

 like changes of length, are greater for such a specimen than for one 

 which is purer and softer. 



Curve (C) shows the results for a piece of steel knitting needle in 

 the state in which it was bought. The experiment was repeated with 

 the wire in a glass hard state, but the curve was almost unchanged. 



The dotted curve (a), copied from a former paper, * shows for 

 comparison the changes of length exhibited by an iron wire in fields 

 up to 1500. The main point of difference is that whereas the curve 

 for change of length crosses the axis of H, indicating a reversal of the 

 sign of the phenomenon in moderate fields, the other does not. Two 

 other thermoelectric experiments, made with different pieces of iron, 

 gave results of a. like character. Even with the high field of 1600, 

 which greatly exceeds any before used in similar work, there is no 

 indication that a reversal of sign would ever be reached. 



The curves in fig. 4 are intended to show the effect of " correction " 

 for mechanical stress. The pieces of wire used in the three experi- 

 ments concerned were all cut for the same hank, and imperfectly 

 annealed by heating in a Bunsen flame. Curve (D) is the thermo- 

 electric curve for the sample, and (b) the curve of elongation and 

 retraction.! Curve (c) shows the relation of lifting power to field, | 

 the ordinates being referred to the scale of grammes weight on the 



* « Phil. Trans.,' vol. 179, p. 228, fig. 6, 1888. 

 f Loc. ext., p. 224, fig. 5. 



% ' Boy. Soc. Proc.,' vol. 40, p. 491, fig. 1, 1886. 



