546 PROCEEDINGS OF THE AMERICAN ACADEMY. 



If in the testing of the (copper)-(german-silver) couples a part of 

 the german-silver wire lying in the place of steepest temperature gra- 

 dient, just outside the entrance to one of the heating vessels, happens to 

 have been severely stretched, it may give rise to such a discrepancy of 

 behavior as that noted in (3) above. But, as such discrepancy was 

 found in couples the german-silver of which had been carefully guarded 

 from any serious stretching, it seemed necessary to look further for a 

 satisfactory explanation. 



7. The application of the sharp edge of a lump of ice to a straight 

 piece of german-silver wire exposed to the air of the room, this applica- 

 tion being many centimeters distant from either end of the wire, 

 developed a perceptible local electromotive force, which frequently 

 changed its sign as the point of contact was moved, 1 cm. at a time, 

 along the wire. The cooling produced by the ice was plainly percep- 

 tible at a distance of 1.5 cm. along the wire from the point of contact, 

 though the cooling at 1 cm. from the contact point is perhaps not more 

 than one third as great as at this point. We may regard the ice test as 

 showing that the mean thermo-electric quality of the wire for a distance 

 of perhaps 1 cm. in one direction from the point of contact is percep- 

 tibly different from that of the corresponding piece lying on the opposite 

 side of the point of contact. 



Annealing was found to increase the range of thermo-electric hetero- 

 geneity of the wire and to change the local distribution of this 

 heterogeneity. 



Stretching annealed wire to a permanent extension of 10 per cent 

 made very little change in the range or distribution of the thermo- 

 electric heterogeneity. Reannealing a wire which had been thus 

 stretched changed again the distribution of its peculiarities and increased 

 still further the range of these peculiarities. In a wire so treated, con- 

 tact at one point with an edge of ice developed an electromotive force 

 about 0.2 per cent as great as that which would have been developed 

 by touching the ice to a junction of copper with german-silver. 



These tests show that there may be in straight, recently annealed 

 german-silver wire local differences of thermo-electric quality which may 

 cause serious error if a copper and german-silver couple is used to 

 measure a small difference between two high temperatures, the inter- 

 mediate part of the german-silver wire being exposed to air at the ordi- 

 nary temperature of a room. To avoid error from this source the 

 practice of reversing the temperature interval of the junctions should be 

 followed whenever this is practicable. 



