436 THE SCIENTIFIC PAPERS OF 



It is well known that the measurement of electrical currents and 

 resistance is susceptible of very great accuracy compared with all 

 thermal measurements : hence my endeavour has been to estimate 

 thermal effects entirely by electrical methods. In the Bakerian 

 Lecture for 1871, which I had the honour of delivering before the 

 Royal Society ("Proc. Roy. Soc.," vol. 19, p. 443), I showed that 

 the resistance of a platinum wire can be expressed as a linear 

 function of its temperature by an empirical formula, the constants 

 of which must be determined for each individual wire ; hence 

 conversely, if resistance of a wire previously calibrated is measured, 

 its temperature can be deduced. From theoretical considerations 



I showed that - = T* + /3T + y might be expected to represent the 



?9 



relation between the resistance and absolute temperature. This 

 formula agreed closely with my own experimental results for 

 platinum, copper, silver, iron, and aluminium wires ("Journal of 

 the Society of Telegraph Engineers and Electricians," vol. i. p. 

 123, and vol. iii. p. 297),* and has since been verified by Professor 

 A. Weinhold in the case of platinum from 100 to 1000 C. 

 (" Annalen der Physik und Cheinie," 1873, p. 225). 



The apparatus (Fig. 3, Plate 23) which I propose for determin- 

 ing the dependence of radiation on temperature consists of a 

 platinum or other wire, 0'76 millim. in diameter, suspended 

 between two binding screws, marked (A) and (B) on the diagram, 

 carried on two suitable wooden stands. The binding screws are 

 connected through an electro-dynamometer (D), for the purpose 

 of measuring the current, to a secondary battery, the number of 

 cells in which can be varied. A high resistance galvanometer (G) 

 is also inserted between the binding screws as a shunt to the 

 platinum wire. 



The electro-dynamometer is of the ordinary form, in which the 

 current passes through a fixed coil, and a movable coil consisting 

 of a single twist, hung by a torsion spring in a vertical plane at 

 right angles to the plane of the fixed coil. The couple due to the 

 current is balanced by the torsion of the spring, hence the angle 

 of torsion is proportional to the square of the current. The 

 current through the high resistance galvanometer being a measure 

 of the difference of potential between the extremities of the plati- 

 * See ante, p. 148. 



