Practical Thermometric Standard. 521 



errors of measurement with the gas-thermometer, and the 

 possible accuracy of measurement with a platinum-thermo- 

 meter, especially at high temperatures, is of a much higher 

 order than with the gas-thermometer. Measurements directly 

 referred to the British Association scale would therefore be of 

 greater permanent value, because they could be subsequently 

 corrected when the relation between the scales had been more 

 accurately determined.) 



" (3) That the value of the difference-coefficient d in the 

 parabolic difference-formula be determined for the British 

 Association standard thermometers by reference to the boiling- 

 point of sulphur as a secondary fixed point in the manner 

 described by Callendar and Griffiths, ' Phil. Trans. A. 1891.' 



" (Note. — It is probable that this method gives the best 

 results over the whole range at temperatures above — 100° C. 

 At very low temperatures there appear to be singularities in 

 the resistance-variation of metals which require further 

 investigation. The boiling-point of liquid oxygen would be 

 a more convenient secondary fixed point to choose for low- 

 temperature research, especially for testing thermometers the 

 construction of which did not permit their exposure to a 

 temperature as high as that of boiling sulphur.) 



" (4) That the temperature of the normal boiling-point of 

 sulphur under a pressure of 760 mm. of mercury reduced to 

 0° C., and latitude 45°, be taken for the purposes of the British 

 Association scale as 444*53° C, as determined by Callendar 

 and Griffiths (loc. cit.), with a constant pressure air-thermo- 

 meter. 



"(Note. — Until the relation between the various gas-thermo- 

 meter scales, and the expansion of glass and porcelain, have 

 been more accurately determined, it does not appear that 

 anything would be gained by changing this value to which so 

 much accurate work has already been referred.)" 



The general effect of these proposals would be to give a 

 definite and tangible existence to a practical scale of tempera- 

 ture which has been carefully tested and verified by many 

 independent observers during the last ten years. It is 

 proposed to secure this result (1) by the construction of 

 particular thermometers to serve as the standard of the 

 platinum scale ; and (2) by giving the name " British Associa- 

 tion Scale " to the scale of temperature deduced from the 

 indications of these instruments in the manner already 

 defined. It is certain that the practical standard thus defined 

 would give a very close approximation to the theoretical scale, 



