THE MEASUREMENT OF TEMPERATURE. 75 



(a) That although the values of S varied greatly, the 

 curves remained practically parabolas. 



(6) That the assumption of the parabolic form and of the 

 respective values of 8 obtained by observations in sulphur 

 vapour gave values for the boiling points of mercury, 

 benzophenone and naphthalene, which were practically the 

 same whatever the sample of the wire used, provided it was 

 of ordinary commercial purity. 



(c) That the boiling points of benzophenone and 

 naphthalene, as deduced in the above manner from the 

 numbers obtained from my experiments of 1890, agreed 

 closely with the values given by Crafts ; for example — 



and to appreciate the value of the results it must be remem- 

 bered that they depend entirely on the validity of the form- 

 ula (d) and on the correctness of the boiling point of sulphur as 

 determined by us in 1891. A reference to the original paper 

 will show that these conclusions were borne out by determina- 

 tions of the boiling points of methylsalicylate, triphenyl-meth- 

 ane, mercury and the freezing points of Sn, Bi, Cd, Pb and Zn. 

 True, the determinations of the air temperature of these 

 points by previous observers differ very greatly from each 

 other, and none appear to have been determined with the 

 care and accuracy which distinguished the work of Crafts ; 

 but the importance of the comparison lies in the fact that 

 platinum thermometers, whose constants differed greatly, 

 gave almost identical values of these temperatures, which 

 ranged from 184° to 445°C. As it was possible that the 

 departure of the t - pt curve from the parabolic form might 

 become marked at ordinary temperatures, an elaborate com- 

 parison was made between a platinum and an air thermo- 

 meter at every 5 from o° to ioo°C. A table of the re- 

 sults will be found in paper (8) (p. 155). This series of 

 experiments led to the same conclusions as the observations 

 at higher temperatures. 



