38 PROCEEDINGS OF THE AMERICAN ACADEMY. 



v' = — q' -T- (2.515 X 38.1 X 305) = - 715 X lO^ 10 . 



The value of this coefficient found from the data belonging to the stretch 

 m to h alone is 



v "=-q'i + (25.15 + 39.2 X 344) = - 793 X 1CT 10 . 



The difference between v' and v", about 10 per cent, is too large to be 

 readily accounted for by experimental errors. Taken with the evidence 

 found by Battelli, it may perhaps be regarded as raising a presumption in 

 favor of the opinion that v does increase, numerically, with rise of tem- 

 perature in iron.* On the other hand, the common practice of drawing 

 a straight line for iron, at low temperatures, on the thermo-electric dia- 

 gram is not based on pure imagination or on mere theory. For example, 

 experiments made on a thermo-electric couple, consisting of the iron 

 here studied combined with copper, indicated f that the " thermo-electric 

 height" of this couple diminished at a nearly uniform rate with rise of 

 temperature, which may seem at first sight to require either lines very 

 nearly straight for both iron and copper on the thermo-electric diagram, 

 or lines having very nearly equal changes of curvature with the region 

 of temperature used, 26.6° to 71.1°. Inspection of the ordinary diagram 

 will show, however, that in this range of temperature a considerable 

 relative change in the inclination of either line would make a much 

 smaller relative change in the thermo-electric height, which is represented 

 by the vertical distance between the two lines. Such change as there 

 was in the rate of change of this height with rise of temperature was in 

 the right direction to be accounted for by an increase iu the steepness 

 of the slope of either the iron line or the copper line. The values found 

 for the " thermo-electric height " of the copper-iron couple were : 



at 26.6° C. 1028 X 10" 8 volt 



" 41.3° 980 X 10" 8 volt 



" 54.5° 938 X lO" 8 volt 



" 71.1° 870 x 10- 8 volt 



* Tait, Trans. Roy. Soc. Edin., 27 (1872-76), 137, found that at high tempera- 

 tures the iron line has at least two curvatures, each sufficient to change the sign 

 of its inclination. Tait made no absolute determinations of the Thomson effect, 

 but based the inclination of the lines of his diagram on the assumption (supported 

 by the experiments of Le Roux) that this effect is null in lead. Battelli (Atti della 

 R. Accademia dei Lincei, serie 4, vol. 3, 1887, p. 212) found data which give 

 4.3 X 10 ~ 10 as the value of v in lead. 



t These Proceedings, 36, 130, August, 1900. 



