( 737 ) 



XXV. — Thermo-Electric Diagram from -200° C. to 100° 0., deduced from the 

 observations of Professors Dewar and Fleming. By J. D. Hamilton Dickson, 

 M.A., Peterhouse, Cambridge. 



(MS. received July 15, 1910. Read November 7, 1910. Issued separately March 17, 1911.) 



In 1895 Professors Dewar and Fleming made an extensive and careful series of 

 observations connecting temperature and the thermo-electromotive force of some twenty 

 metals compared with lead. The greatest care was taken in the preparation for, and in 

 the conduct of, the experiments, and about thirty readings, at temperatures ranging 

 over the 300 c between the boiling point of oxygen and the boiling point of water, were 

 recorded for each metal. A full description of the mode of experimenting, with the 

 results for each metal, was published in the Philosophical Magazine, July 1895.* In 

 their discussion of the results a further examination of them was promised, but has not 

 appeared. Professor Fleming, however, in a Friday Evening Discourse at the Royal 

 Institution in 1896, indicated that the lines of thermo-electric power did not appear as 

 straight lines although they were approximately so. 



In the summer of 1896, I attempted to determine the nature of the connection 

 between electromotive force and temperature for platinum. The selection of this metal 

 was made for two reasons — it was a piece of the same wire as that used by Callendar 

 in his researches on thermometry, and another portion of this same wire formed the 

 Standard Platinum Thermometer used in these observations by Dewar and Fleming. 

 The observations were plotted, taking electromotive force as ordinate to platinum- 

 temperature as abscissa ; a free-hand curve run through these points showed both sides 

 of the curve in considerable magnitude, thus affording a good opportunity of testing its 

 nature. A very great number of attempts, extending over some months, were then 

 made to connect this curve with an equation of the form y n = px, n and p being constants. 

 Since n = 2 would give a parabola, and as the curve appeared of parabolic form, the 

 values chosen for n were all in the neighbourhood of 2, and corresponding values were 

 chosen for^>. From each pair of associated values of n and p values of E.M.F. were 

 calculated for the temperatures at which observations were recorded, and the differences 

 between the observed and calculated values of E.M.F. were plotted as ordinates to 

 temperatures as abscissae. For short ranges in the middle of the temperature-interval 

 of 300 degrees, these differences could always be made small ; but in every case, for all 

 higher or lower temperatures, they became excessively and systematically large. It was 

 clear, therefore, that no such simple law as y n - px would satisfy platinum. A new and 

 enlarged " plot " of the observations was then made, and a free-hand curve drawn with 

 the greatest care among the representative points. In drawing the curve, as it was 



* Phil. Mag., Ser. V., vol. xl. pp. 95-119. 

 TRANS. ROY. SOC. EDIN., VOL. XLVII. PART IV. (NO. 25). 109 



