THERMO-ELECTRIC DIAGRAM FROM -200° C. TO 100° C. 



761 



Tait-line has two asymptotes, ECF at the temperature 170*7° C. and CD at the value of 

 dE/dt = — 544*02. As in the case of copper, 170*7° C. being a readily accessible tempera- 

 ture, it would have been interesting had the experiments been carried so high. 



From the A-curve, by the method already described, a series of values of dE/dt 

 were measured. In these measurements a correction had to be made on the value of t 

 corresponding to any observed value of dE/dt, because the slope of the axis was so great. 

 Having observed the value of tan for any dE/dt, two equal deviates FL, GrM (fig. 11), 

 were found, and their corresponding temperatures, OB = ^, and OA = t 2 , were read off 

 the curve. If the axis of the curve had been vertical, ON, the temperature of the point 



of contact of the tangent, LTVI, would have been the mean of OA and OB. But as this 

 was not the case, the mean of OA and OB was OC, and it became necessary to determine 

 the correction CN. Let $ be the value of the deviate FL, or GM, or DK ; then in the 

 triangle DKP, we get 



KP = DK 



sin a) 



cos (oj -- oy 



whence 



CN = KPcos0 = 



and the value of t at P is given by 



cot o) + tan 6 



8 



tJi + h , 



2 27201 + tan 6 



■ (47) 



The values adopted for $ were, conveniently, 2, 4, 6, and 8 mm. If t be the correction 

 for S = 8 mm. , we get 



/ I- : 



+ r, 



for8 = 8mm.,t = t ± + t2 



„ =6 „ ^ 



t[ + t' 2 



+ ir, 



t l + t 2 , 

 „ = 4 >» t= — o~ +t T > 



= 2 



t- 2 +*t, 



always supposing that it had been possible to measure these temperatures accurately, 



which, it need hardly be added, was not the case. Nevertheless, some test of the 



correctness of such observations would be got if the various values of t obtained at any 



TRANS. ROY. SOC. EDIN., VOL. XLVTI. PART IV. (NO. 25). 112 



