HALL. — THERMO-ELECTRIC HETEROGENEITY IN ALLOYS. 571 



The values obtained before the policy of heating the middle of ger- 

 man-silver wires was adopted are at every stage somewhat greater, and 

 on the average about 1 per cent greater, than those given by the later 

 experiments. The reason for this is not apparent. It seems best to 

 take the mean value for each stage. The numbers thus obtained indi- 

 cate a nearly uniform increase of " thermo-electric height, " electromotive 

 force for 1° difference of temperature of the junctions, for the interval 

 from 100° to 218°. Perfectly uniform increase would require the 

 values found for the 140° to 180° region to be raised about 1 in 400. 



If we take 2194 X 10 -8 as the thermo-electric height for 120°, the 

 middle of the a interval, and 2562 X 10~ 8 as the thermo-electric height 

 for 199°, the middle of the y region, we get 4.7 X 10 -8 as the mean in- 

 crease per degree from 120° to 199°. This rate of change applied 

 downward from 120° would give 2100 X 10 -8 as the thermo-electric 

 height for 100°. 



At the end of the paper 4 referred to at the beginning of this article 

 a summary is given of the results of calibration experiments on sets of 

 thermo-electric couples other than those here studied but made from 

 the same materials in the same way. From this summary we get 



thermo-electric height at 89°.4 = 2063 X 10~ 8 

 " « « « 17 o 4 _ 1713 u 



From these values we get 4.9 X 10 -8 as the rate of increase per 

 degree from 17°.4 to 89°. 4. This rate continued gives 2115 X 10 -8 as 

 the thermo-electric height at 100°. This differs about 0.7 per cent from 

 -the value found above. The value for 199° calculated from the data 

 obtained in the region below 100° would be 2600 X 10 -8 , only 1.5 per 

 cent greater than the value found in this paper. 



Taking 2100 X 10 -8 as the thermo-electric height at 100° and taking 

 the rate of increase per degree rise of temperature as 4.8 X 10 -8 , we 

 should get values which would not at any point between 0° and 

 218° differ as much as 1 per cent from the most probable value which 

 we have for that point. 



It has been stated earlier in this paper that in the (2) set of couples 

 the maximum difference of mean electromotive force among the various 

 couples was about 0.3 per cent of the mean electromotive force. The 

 corresponding difference was greater in the (3) set of couples, being, on 



4 These Proceedings, May, 1905. 



