1905.] On the Platinum and Normal Scales of Temperature. 529 

 If the coil has a resistance R at the platinum temperature T p , 



R = R (l+a'T' ;; ). 



Callendar and Griffiths* have shown that between and 444-55, 

 the boiling point of sulphur on the constant pressure nitrogen scale* 

 the difference, A, between the platinum and the gas scales can be 

 calculated by means of a parabolic formula, 



VTOO /100' 



where T is the temperature (Centigrade), on the gas scale, and 9 is a 

 constant. Callendar has suggested that this principle should form the 

 basis of a practical scale of temperature, and Messrs. Johnson and 

 Matthey have taken much trouble to prepare pure platinum for the 

 purpose. For this material a is about '00388 or '00390, and the 

 value of 3 about 1 '5. 



Previous Investigations. Marker and Chappuisf have compared a set of 

 platinum thermometers, for which the value of a was about 0'003865, 

 with the constant volume nitrogen thermometer, and have found that 

 the difference between the two scales of temperature can be calculated 

 by means of Callendar's formula, taking for 8 values between 1'54 and 

 1*55. The comparison was carried out over a range of temperature 

 between - 23 C. and 455 C., and the errors at the higher tempera- 

 tures did not exceed 0'2. 



More recently, Harker} has extended this investigation, comparing 

 the platinum thermometers with a constant volume nitrogen thermometer 

 standardised at 0, 100, and 444'55, the latter being Callendar's value 

 for the boiling point of sulphur on the constant pressure air scale. The 

 observations, therefore, really refer to the constant pressure air scale. 

 In this research he employed a resistance thermometer described in the 

 British Association Report for 1899 (p. 243), for which the value of 

 a is 0-00389. Taking 1'51 for the value of 8, he found that the 

 temperatures calculated from observations of the two thermometers 

 showed a maximum difference which only exceeded the probable error of 

 measurement over the higher part of the range. 



At the end of his paper Harker gives a table of the corresponding 

 values of Tp and A, calculated on the basis 8=1'50, between -200 

 and 1,100. As it is well-known that the formula cannot be applied 

 to the calculation of the difference between the two scales at low 

 temperatures, this appears to be unnecessary and misleading. 



Several independent investigators have compared platinum and gas 

 thermometers at low temperatures, but for the reason that they did not 



* ' Phil. Trans.,' A, 1891, p. 119. 



f ' Phil. Trans.,' A, 1900, vol. 194, p. 37. 



J ' Phil. Trans.,' A, 1904, vol. 203, p. 343. 



