Prof. H. L. Callendar on Platinum Thermometry. 209 



surprising when we consider the very large difference in the 

 fundamental coefficient c, which is approximately '00390 in 

 the case of the purest obtainable wire. The remarkable fact 

 is that, as stated in my original paper (see above, p. 195), the 

 values of pt for such different specimens of wire should show 

 so close an agreement through so wide a range. The differ- 

 ence in the fundamental coefficients in this extreme case is about 

 20 per cent.; but the values oipt for the two wires differ by only 

 4° at the S.B.P., and this difference, instead of increasing in 

 proportion to the square of the temperature, remains of the 

 same order, or nearly so, at the freezing-points of silver and 

 gold. Thus the wire c = '00320 gave p£ = 830° at the 

 Ag. F.P., but I shortly afterwards obtained with a specimen 

 of very pure wire (c = '003897), the value pt = 835° for the 

 same point. Messrs. Reycoek and Neville, using the same 

 pure wire, have confirmed this value. They also find for the 

 F.P. of gold, with different instruments, constructed of the 

 same wire, the average value p£ = 905 o, 8. I did not test this 

 point with the pure wire, but the value found by Messrs. 

 Heycock and Neville may be compared with the value 

 ^ = 902°'3 (Phil. Mag., Feb. 1892), which I found at the 

 Mint with one of the old instruments. 



From these and other comparisons of the platinum scales of 

 different wires, it appears likely that the deviation of the 

 impure wire from the parabolic curve is generally of this 

 nature. As shown by the comparison curves in my original 

 paper, the deviation follows approximately the parabolic law 

 up to 400°, beyond that point the curves tend to become 

 parallel, and at higher temperatures they often show a 

 tendency to approach each other again. The application of 

 the S.B.P. method of reduction to impure wires at high 

 temperatures will therefore give results which are too high, 

 because the value of d is calculated from the S.B.P., where 

 the difference between the wires is nearly a maximum. Thus, 

 taking the values of d from the S.B.P. for the two specimens 

 of wire above quoted, we find, calculating the values of t for 

 the Ag.F.P., and Au.F.P. from the data, 



Impure wire, c = '00320, d= 1*751 ; 



Ag. F.P., £ = 98P6 ; Au. F.P., £ = 1092'0. 

 Pure wire, c = '00390, d = l'520 ; 



Ag. F.P., £ = 960'7 : Au. F.P., £=1060'7. 



The results for the impure wire obtained by the S.B.P. method 

 of reduction are not so high as those found by Bams with a 

 Pt-Ptlr thermo-element, which he compared with an air- 

 thermometer up to 1050°. There can be little doubt. 



