Dr. P. Chappuis on Gas- Thermometry. 435 



From these data it is a simple matter to calculate the diver- 

 o-ences of the nitrogen thermometer from the normal scale 

 within the interval covered by our comparisons. 



Differences of Ssale. 

 Temperature. Actual Scale — Normal Scale. 



P = l metre. 



100° o-ooo 



150 -0-008 



200 -0017 



250 -0-026 



300 -0034 



350 -0-043 



400 -0-051 



450 -0-060 



500 -0-068 



From this it would appear that the pressures indicated by a 

 constant-volume gas-thermometer are a little too low, but 

 the errors are small and appear to be of less account than 

 other sources of uncertainty which enter into the measure- 

 ments. Besides I would remark that in our comparisons of 

 the platinum-thermometer with the nitrogen-thermometer, the 

 initial pressure was always less than one metre. (In the 

 comparisons between 100° and 200° it was 793 millim , and in 

 those between 250° and 500° it was always less than 550 

 millim.) Under these conditions the systematic divergences 

 of the nitrogen scale from the normal are probably diminished 

 in the same proportion as the initial pressures. They will 

 then be less than o- 04 for measurements at temperatures near 

 the boiling-point of sulphur. 



II. Determination of the Sulphur- Point. 



, The mean result of our experiments for the determination 

 of this important point is 



T s = 445°-2 



under the barometric pressure of 760 millim. Messrs. Callen- 

 dar and Griffiths, employing the constant-pressure air- 

 thermometer, have found for the same point 



T s =444°53. 



The difference between these two results — 7 — is, I believe, 

 to be attributed to the joint action of several causes : — 



(1) The scale of the constant-pressure nitrogen-thermometer 

 diverges more widely from the normal temperature-scale than 



