192 Prof. H. L. Callendar on Platinum Thermometry, 

 Historical Summary. 



The earlier experiments on the variation of the electrical 

 resistance of metals with temperature were either too rough, 

 or too limited in range, to afford any satisfactory basis for a 

 formula. The conclusion of Lenz (1838), that the resistance 

 reached a maximum at a comparatively low temperature, 

 generally between 200° and 300° C, was derived from the 

 empirical formula, 



R°/R=l + at'+bt 2 , (L) 



in which R° and R stand for the resistances at 0° and t° C, 

 respectively. This conclusion resulted simply from the 

 accident that he expressed his results in terms of conductivity 

 instead of resistance, and could be disproved by the roughest 

 qualitative experiments at temperatures beyond the range 

 0° to 100° C, to which his observations were restricted. 

 Matthiessen (1862), in his laborious and extensive investiga- 

 tions, also unfortunately fell into the same method of 

 expression. His results have been very widely quoted and 

 adopted, but, owing to the extreme inadequacy of the formula, 

 the accuracy of his work is very seriously impaired even 

 within the limits of the experimental range to which it was 

 confined. The so-called Law of Glausius, that the resistance 

 of pure metals varied as the absolute temperature, was a 

 generalization founded on similarly incomplete data. The 

 experiments of Arndtsen (1858), by which it was suggested, 

 gave, for instance, the temperature-coefhcients '00394 for 

 copper, '00341 for silver and *00413 for iron, all of which 

 differ considerably from the required coefficient '003G65. 

 The observations, moreover, were not sufficiently exact to 

 show the deviation of the resistance-variation from lineality. 

 The experiments of Sir William Siemens (1870) did not 

 afford any evidence for the particular formula which he pro- 

 posed, at least in the case of iron. These formulas have been 

 already discussed in previous communications'^, but con- 

 sidering the extent to which they are still quoted, it may be 

 instructive to append the curves representing them, as a 

 graphic illustration of the danger of applying for purposes of 

 extrapolation formulas of an unsuitable type. The curves 

 labelled Morrisf and Benoit, which are of the same general 

 character but differ in steepness, may be taken as representing 

 approximately the resistance-variation of specimens of pure 

 and impure iron respectively. 



The first experiments which can be said to have afforded 

 any satisfactory basis for a general formula were those of 



* Callendar, Phil. Mag. July 1891 ; G. M. Clark, Electrician, Jan. 1897. 

 t Phil. Mag. Sept. 1897, p. 213. 



