COMPARISON OF PLATINUM AND OAS THERMOMETERS. 101 



From the different series of experiments from which a value can l>e deduced by this 

 method we have 



1st Series K.9 and glass reservoir T, = 445*27 

 2nd K.9 porcelain T, = 445'26 

 3rd K.8 T, = 445-29 



Meau = 445-27 



Although we think that the extremely close agreement of these values is to some 

 extent fortuitous, and may give an exaggerated idea of the accuracy attained in our 

 experiments, we think that until more is known concerning the expansion at high 

 temperatures of the material used as thermometric reservoir, 445'27 may he taken 

 as a close approximation to the temperature attained by the vapour of pure sulphur 

 boiling freely under a pressure of 760 millims. in the apparatus above described. 

 Whether this represents the true temperature, or whether the indications of the 

 thermometer are affected to any appreciable extent by radiation and other disturbing 

 influences, we have not attempted to consider in detail. We contented ourselves 

 with ascertaining that the form of apparatus we used is capable of giving consistent 

 results, and that the temperature attained in it by the vapour after the steady state 

 lias been reached really alters with the barometric pressure. We noticed that the 

 Irarometer we used, and those platinum thermometers which were provided with glass 

 envelopes, appeared to follow changes at very nearly the same rate. Considering 

 that the observations of the boiling-point were only made when the barometer 

 appeared to be fairly steady, we think that any error in the measurement of the 

 corresponding temperatures and pressures due to difference of lag of the two 

 instruments must have been very small. 



X I.I 1 1 REDUCTION OF RESULTS TO NORMAL SCALE. 



In view of the lack of data as to the difference between the various gas scales at 

 high temperatures ; we are unable to reduce the results of our comparisons, and the 

 value found for the boiling-point of sulphur, to what they would have been on the 

 scale of the hydrogen thermometer.* 



* [Footnote added Decomlwr 1, 1899. From the study of the different gas scales previously made l>y 

 one of us, it appears that tetwoen and 100 the point of maximum difference between the hydrogen 

 and nitrogen scales is at 40, where the nitrogen thermometer reads higher by 0O1. At 100 the 

 difference between the two scales Ixjcomes zero by definition, and above that temperature it changes sign 

 and has a value which appears not to exceed 0'l l>elow 600. 



The scale of the constant volume nitrogen thermometer appears not to be independent of the initial 



pressure ; if we may judge by the variation of the coefficient -f , which approaches that of hydrogen 



pa at 



as the pressure diminishes, we may assume that the difference between the scales of the nitrogen and 



