Prof. H. L. Callendar on Platinum Thermometry. 201 



varied so much as # 04 on either side of the mean, in the case 

 of their standard wire. It is, moreover, quite possible that 

 these variations may have been partly due to fortuitous differ- 

 ences at the S.B.P. and at the fixed points, in which case it 

 is probable that the Heycock and Neville method of reduction 

 would lead to more consistent results than the pure difference- 

 formula, because it does not allow full weight to the apparent 

 variations of d as determined by the S.B.P. observations. 

 It is clearly necessary, as Heycock and Neville have shown, 

 and as the above calculation would indicate, to take some 

 account of the small variations of d, at least in the case of 

 pyrometers in constant use at high temperatures. The method 

 of Heycock and Neville appears to be a very convenient and 

 practical way of doing this, provided that the variations of d 

 are small. It must also be observed that, although the indi- 

 vidual reductions by their method may differ by as much as 

 1° or 2° at 1000° from the application of the pure difference- 

 formula, the average results for the normal value of d will be 

 in exact agreement with it. 



Difference- Formula in Terms of pt. — In discussing the 

 variation of resistance as a function of the temperature, it is 

 most natural and convenient to express the results in terms of 

 the temperature t on the scale of the air-thermometer by 

 means of the parabolic formula already given. This formula 

 has the advantage of leading to simple relations between the 

 temperature-coefficients ; and it also appears to represent the 

 general phenomenon of the resistance-variation of metals 

 over a wide range of temperature with greater accuracy than 

 any other equally convenient formula. When, however, it 

 is simply a question of finding the temperature from the 

 observed value of the resistance^ or from the observed reading 

 of a platinum thermometer, over a comparatively limited 

 range, it is equally natural, and in some respects more con- 

 venientj to have a formula which gives t directly in terms of 

 pt or R. This method of expression was originally adopted 

 by Griffiths, who expressed the results of the calibration of 

 his thermometers by means of a formula of the type 



t—pt = apt + bpt 2 + cpt d + dpt 4 . ...(G) 



The introduction of the third and fourth powers of pt in 

 this equation was due to the assumption of RegnauhVs value 

 for the boiling-point of sulphur. If we make a correction 

 for this, the observations can be very fairly represented by a 

 parabolic formula of the type already given, namely, 



t-pt=dXpt/100-l)pt/l0Q=d / p(pty . . (I) 



