332 C. Barns — Status of High Temperature Research. 



Akt. XLVII. — The Present Status of High Temperature 

 Research • by Cael Baeus. With Plate X. 



1. Preliminary. — Some time ago* I adduced reasons for 

 supposing that the electromotive force, d, of a thermocouple, 

 can be expressed by an equation of the form 



^P + Qd P 1 + Q x , . 



e + e = 10 + 10 , (1) 



where P, Q, P\ Q\ are constants, 6 denotes temperature, and 

 where e is zero, when 6 = O . I wish now to test this equation 

 by the aid of data formerly foundf in direct comparisons of 

 the iridioplatinum thermocouple^: with the porcelain air ther- 

 mometer, and therewithal to exhibit the degree of accordance 

 of known results in the region of high temperature. 



Inasmuch as e must pass either through a maximum or a 

 minimum, Q and Q l must have different signs ; for m , the 

 temperature of this point, is 



log (-Q/Q 1 ) -(**-?) 



Q l -Q 



when the sign connecting the two terms in (1) is positive. 



2. First extreme case. — The simplest form which the above 

 equation takes is the catenary, and it is interesting to note in 

 what respect this curve falls short of reproducing the experi- 

 mental results. Taking therefore the equation y = k cos hyp 

 x / k, it is necessary to shift the vertex to the position (— a, 

 — Jc) in order that the curve may pass through some fiducial 

 point near the origin ; or in other words, that the electromotive 

 force may be zero when the hot junction has the stated tem- 

 perature of the cold junction. The equation then becomes 

 e 4- h 4- k 1 = k cos hyp (0 4- a) / k, where e is the electromotive 

 force for the temperature 6 of the hot junction, k the constant 

 of the catenary, and where (— a, — k 1 ) are the coordinates of 

 the vertex or the position of the thermoelectric minimum. In 

 the absence of suitable tables it is then possible to find the 

 constants involved by actually using a chain in connection with 

 a carefully constructed and mounted chart of results. For if 

 e + k 4- k 1 = y, then the length of the arc from the vertex to 

 the end of the ordinate e is s = Vy* — h 2 ; and therefore since 

 a and &' are obtained by direct measurement, k can be computed 

 from the measured length s of the arc or chain in question. 



Constants so obtained are of course approximate and they 

 must be improved by successive trials. Doing this I found for 



* This Journal, xlvii, p. 366, 1894. •)• See § 4. 



X One metal containing about 20% of iridium, the other being soft platinum. 



