Prof. H. L. Callendar on Platinum Thermometry. 205 



resistance of one of their wires had already nearly reached 

 6R° at a temperature of about l(U0 n C, whereas the maximum 

 calculated resistance in the case of one of my wires (with a coeffi- 

 cient c= '00340) was only (r576R°. He omits to notice that 

 the result depends on the coefficients of the wire. 



The wire used by Holborn and Wien had a fundamental 

 coefficient c = '00380, and the highest value of the resistance 

 actually observed was not 6R° as suggested, but B/R° — 5'53, 

 at a temperature £ = 1610° C, deduced from thermo-j unctions 

 at each end. If we assume d=l'70 as a probable value of 

 the difference-coefficient for their wire, the difference-formula 

 (2) would give, 



at £=1610°, D = 414°, ^=1196°, whence R/R° = 5'54. 



It w T ould be absurd to attach much weight to so rough an 

 observation, but it will be seen that, so far as it goes, the 

 result is consistent with the usual formula, and does not bear 

 out Mr. Dickson's contention. A more important defect in 

 arguments (1) and (2) lies in the fact that maximum and 

 minimum values of the resistance are known to occur in 

 the case of manganin and bismuth within the experimental 

 range, and that such cases can be at least approximately 

 represented by a formula of the type (3), but cannot be 

 represented by a formula of the type (6). 



As shown by Table I. above, the formula proposed by 

 Dickson agrees fairly well with formula (3), in the special 

 case of platinum, through a considerable range. Rut the 

 case of platinum is exceptional. If we attempt to apply a 

 formula of Dickson's type to the case of other metals, we are 

 met by practical difficulties of a serious character, and are 

 driven to conclude that the claim that it is 6l more represen- 

 tative of the connexion between temperature and resistance 

 than any formula hitherto proposed/' cannot be maintained. 



observations are somewhat inconsistent, and lead to values of d which 

 are rather large and variable, ranging from 3:7 to 2-0. These variations 

 are probably due to errors of observation or reduction. This is shown 

 by the work of Mr. Tory (B.A. Report, 1897), who made a direct com- 

 parison between the Pt — PtRh thermocouple and the platinum-thermo- 

 meter by a much more accurate method than that of Holborn and Wien. 

 He found the parabolic difference-formula for the platinum thermometer 

 to be in very fair agreement between 100° and 800° C. with the previous 

 series of observations of Holborn and Wien on this thermocouple (Wied. 

 Ann. 1892), and there can be little doubt that the discrepancies shown 

 by their later tests were due chiefly to the many obvious defects of the 

 method. For a more detailed criticism of these observations, the reader 

 should refer to a letter by Griffiths in ' Nature,' Feb. 27th, 1896. It is 

 sufficient to state here that the conclusions which these observers drew 

 from their experiments are not justified by the observations themselves. 



