76 SCIENCE PROGRESS. 



It thus appears that a complete standardisation of a pla- 

 tinum thermometer can be made by observation of its resist- 

 ance at three temperatures only, and the ones selected by 

 us for the purpose are o°, ioo° and 444 '5°C. (sulphur vapour 

 at 760 m.m.). Great care must, however, be taken that the 

 thermometer, when in sulphur vapour, is thoroughly screened 

 as described by us ((8), p. 143). Our experiments with 

 various specimens of sulphur indicate that the sample used 

 need not be of extreme purity, the ordinary impurities not 

 affecting the temperature of the vapour when the boiling has 

 been continued for some time. Thus R s , R x and R being- 

 known pt s can be obtained, and hence the value of $ for the 

 particular sample of wire used can at once be deduced from 

 formula (d). The standardisation can now be regarded as 

 complete, and we have an instrument whose fixed points do 

 not change, provided that the wire has been carefully 

 annealed, no matter what temperature it has been pre- 

 viously exposed to, and whose readings are independent of 

 position and external pressure. Its capacity for heat is 

 small, it can be made of almost any size, so as to give the 

 mean temperature of a space, or the temperature "at a 

 point," it can be placed interior of any apparatus and read 

 at any convenient distance. Also, if, as should invariably 

 be the case, the stem is supplied with double electrodes, the 

 readings are uninfluenced by the temperature of the stem 

 or leads. I derived such confidence from the experiments 

 performed with platinum thermometers, whose constants had 

 been determined in the manner I have described, that I 

 standardised the mercury ones used during my determination 

 of the mechanical equivalent of heat by platinum thermo- 

 meters only. At the close of that investigation, those mercury 

 thermometers were carefully compared with the standard sup- 

 plied by the International Bureau. The results of the com- 

 parison are given in detail ((11), p. 430) and show that in 

 actual elevation a difference of 0'005°C. was found but that 

 the value of temperature ranges was practically identical. 



Again, a further comparison was made by Callendar and 

 myself during the summer of 1893, between a new form of 

 air thermometer, whose indications were independent of 



