Prof. H. L. Callendar on Platinum Thermometry. 199 



between 0° and 100° C, it is quite possible that this might not 

 always be the case. 



For work at low temperatures it would be preferable, from 

 every point of view, to make use of the boiling-point of oxygen 

 as the secondary fixed point. There appears to be a very 

 general consensus of opinion that the temperature of liquid 

 oxygen boiling under a pressure of 760 mm. is — 182°*5 C, 

 on the scale of the constant-volume hydrogen or helium 

 thermometer*. It is quite possible that, as in the case of 

 water and most other liquids, the temperature of the boiling 

 liquid would be different from that of the condensing vapour 

 at the same pressure ; but the boiling liquid is the most con- 

 venient to employ, and it appears that its temperature is 

 steady to two or three tenths of a degree, and reproducible by 

 different observers to a similar order of accuracy. I have 

 found it convenient for purposes of distinction to employ the 

 symbol d° to denote the value of d deduced from the boiling- 

 point of oxygen, and the symbol d" to denote that deduced 

 from the boiling-point of sulphur. The formulae for the 

 pressure correction in the case of oxygen are approximately 



t= -182-5 + -020 A; p(t) = 5-16--00093A. 



The Resistance Formula. — I have shown in the paper 

 already referred to that the adoption of the parabolic differ- 

 ence-formula for the relation between pt and t is equivalent 

 to assuming for the resistance- variation the formula 



R,/B° = l + at + bt* (3) 



The values of the coefficients a and b are found in terms of c 

 and d, or vice versa, by means of the relations 



a=c {l + d/100), b=-cd/10fl00. 



Graphic Method of Reduction. — The quickest and most 

 generally convenient method of reducing platinum tempe- 

 ratures to the air- scale is to plot the difference t— pt in terms 

 of t as abscissa, and to deduce graphically the curve of differ- 

 ence in terms of pt as abscissa, as described and illustrated 

 in my original paper. This method is particularly suitable at 

 temperatures up to 500° C, as the difference over this range 

 is relatively small and accurately known. It is also very con- 

 venient if a large number of determinations are to be made 

 with a single instrument. It is not so convenient in the case 

 of a number of different instruments with different coefficients, 



* The experimental evidence for this number is not quite satisfactory, 

 owing to differences in the atmospheric pressure and impurities in the 

 oxygen. It must be understood that the adoption of this value is 

 provisional and subject to correction. 



P2 



