and Electrical Resistance of Iron. 217 



the fact that, for pure annealed platinum wire, if j^oth part of 

 the increase of resistance which it experiences in passing 

 from 0° to 100° be called the increment for one " platinum 

 degree," so that the resistance of a platinum wire is a linear 

 function of its " platinum temperature,''' pt° ; then, if t° is 

 the actual temperature on the air-thermometer scale, the 

 difference t° — pt° can be expressed over wide ranges of 

 temperature by the form a + bt + ct 2 , which in this case must 

 be reducible to the form t°-pt° = 8{(t/100y-t/l00\. 



It was thought that, in the present investigation, an error 

 in the determination of the absolute values of the higher 

 temperatures of even several degrees was of comparatively 

 small moment where the recalescence point for different 

 specimens of iron is known to differ by 50° or more. The 

 resistance-temperature curve for the platinum thermometer 

 wire was not determined directly above 200°, but was extra- 

 polated by the help of Callendar's formula. The value of 8 

 for this wire was 2'10. 



The absolute values of the temperatures given must there- 

 fore be accepted with caution. They may be wrong by as much 

 as 5° in the neighbourhood of 800°, the error rising with the 

 square of the temperature. For the present purpose, how- 

 ever, it is the relative accuracy which is of importance, and 

 this is of a much higher order. 



To obtain reliable measurements of temperature, it was 

 necessary that the thermometer-wire should satisfy the fol- 

 lowing conditions : — (1) Its temperature must not differ 

 sensibly from that of the iron which it is intended to measure; 

 (2) its constants must not alter in any way during the heating 

 of the ring ; (3) its resistance must be measured with suf- 

 ficient precision. 



(1) The first condition was the most difficult to satisfy : 

 and it was only by actually burying the thermometric wire in 

 the core itself, and by having the temperature very nearly 

 constant before taking any observations, that the temperature 

 of the thermometer- wire and that of the iron core were brought 

 to a satisfactory coincidence. 



(2) With regard to the second condition : the resistance- 

 temperature curve of the thermometer-wire, besides being 

 taken previous to experiments, was also taken after those with 

 each of the specimens A and B ; in neither case could any 

 alteration either in the resistance-coefficient or in the value 

 of 8 be detected. (Alterations of these constants which had 

 taken place with former thermometer-wires were thought to 

 be due to the fact that the insulation in contact with them 

 had not been previously decarbonized (see p. 220).) 



