THE MEASUREMENT OF TEMPERATURE. 75 



has been abundantly verified by the experiments. If the 

 wire is pure to start with, its resistance is always the same 

 at the same temperature. 



(6) The relation between the platinum and the air tem- 

 perature is closely represented by the parabola— 



Vi 00 / 1 00 1 



The experimental evidence given in support of (6) would be 

 sufficient to establish the necessary relation provided we 

 could make certain that all platinum wires used should be 

 of the same degree of purity as that used by Callendar. It 

 appeared probable, however, that a greater or less degree 

 of purity might entirely alter the character of the curve. 



In the autumn of 1889 I commenced a series of deter- 

 minations of certain freezing and boiling points by means of 

 platinum thermometers. At that time I had not had the 

 advantage of reading Callendar' s paper, nor was it brought 

 under my notice until my observations were nearly com- 

 pleted ; the inquiry was therefore conducted on indepen- 

 dent lines. In order to standardise my instruments I had 

 assumed the following " fixed points " in addition to those 

 given by melting ice and steam at 760 m.m. : — 



B. P. of Naphthalene at 760 m.m. = 2i8'o6 Crafts 

 ,, Benzophenone ,, = 306*08 Crafts 



Sulphur ,, = 448"38 Regnault. 



I standardised ei^ht thermometers by means of the above 

 five points. These thermometers were of different patterns, 

 their coils were formed of different specimens of platinum 

 wire differently insulated, although the majority of the coils 

 were wound on calcined asbestos and enclosed in hard glass 

 tubes. The conclusions I arrived at may be summarised 

 as follows ((7), p. 64) : — 



(a) The readings of each thermometer arc constant 

 when the temperature is the same. 



(6) Different thermometers whose coils are formed of 

 different specimens of platinum do not give the same pla- 

 tinum temperature when at the same actual temperature. 



