40 DRS. J. A. BARKER AND P. CHAPPUIS ON A 



of 8 for the diti'erent samples of platinum they examined varied greatly with their 

 purity, yet, provided that the percentage of impurity were small, the formula given 

 above held true. They found from their experiments that the T pt curve was 

 always a parabola, and that, therefore, to establish the whole curve showing the 

 divergence of the two scales, it was sufficient to know d for three fixed points. For 

 two of these, viz., and 100, d is by definition zero. For the third point, for 

 reasons indicated in their paper, GRIFFITHS and CALLENDAE chose the boiling-point 

 of sulphur, and subsequently made a new determination of this point by an air 

 thermometer, finding as their most probable value 444'53, the pressure being 

 760 millims. This value, which is nearly four degrees lower than that previously 

 obtained by REGNAU'LT, is the one which has been generally adopted in work with 

 the platinum thermometer. 



As further evidence in confirmation of this conclusion GRIFFITHS points out that, 

 if this number be taken for the boiling-point of sulphur in the calculation of the 

 values obtained by him for the boiling and freezing points of a number of substances 

 on which he experimented, the results for most of the substances concord better with 

 their accepted values as determined by other observers, than if REGNAULT'S value, 

 448'34, be adopted. 



Many of these accepted numbers quoted in GRIFFITHS' paper are given to huudredths 

 of a degree, but closer examination of the original papers in most cases reveals the 

 fact that the reductions to the normal scale and the various corrections of the thermo- 

 meters employed, if made at all, are, to say the least, veiy uncertain. Further, we see 

 no a priori reason why, in GRIFFITHS' experiments, the results with certain of the 

 substances employed should be rejected from consideration, as there does not appear 

 sufficient ground for supposing that the experimental error in these cases was higher 

 than the average. 



Substantially then our knowledge of temperatures, deduced by means of the platinum 

 thermometer, depends solely on the correctness of the conclusions of GRIFFITHS and 

 CALLENDAR : 



(1) That the boiling-point of sulphur under 760 millims. pressure is 444'53. 



(2) That the curve representing the divergence of the platinum and air scales is a 

 parabola. 



II. THE INVESTIGATION FOR THE 'KEW COMMITTEE. 



In recent years the platinum thermometer has been employed by various observers, 

 and their experience has tended to confirm the view that it could be relied upon to 

 give constant indications at a given temperature. It consequently appeared to the 

 Kew Observatory Committee that it might be possible to use this instrument as a 

 means of referring measurements of temperature to the scale of the gas thermometer 

 adopted as an International standard by the Comite International des Folds et 

 Mesures, and thus to extend their means of accurately testing thermometers sent to 



