288 Mr. H. F. W. Burstall on the Measurement 



about 20 ohms was used, and even after many hours' work 

 was found perfectly constant. 



There now remains the question, when the resistance of the 

 measuring-wire at any point has been determined, what tempe- 

 rature on the air-scale corresponds to the resistance found. 

 Callendar has shown that if temperature, in degrees Centi- 

 grade, be denoted by t, and pt denotes what is called the 

 platinum temperature, then 



*/ * 2 t \ 



where 8 is a constant which has to be determined for any 

 particular platinum wire ; pt is defined as equal to 



100 



R — R 



where R is the resistance of the wire in melting ice, R^ is 

 the resistance of the wire in steam at atmospheric pressure, 

 and R the resistance of the wire at the temperature t*. 



As my thermometers consisted of very fine naked wires 

 which were, of necessity, exposed to the erosive action of high- 

 temperature gases, it could not be expected that the fixed 

 points, i. e. the resistance in ice and steam, would remain 

 constant after the wire had been in the cylinder for any length 

 of time. Also, owing to the irregular working of the gas- and 

 air-valves, the temperatures reached in the cylinder sometimes 

 far exceeded the mean. When one of these abnormal explo- 

 sions occurred the wire was sometimes completely fused, and 

 more frequently melted on the surface. These two causes 

 necessitated a frequent determination of the fixed points and 

 also of the value of 8. 



Three sizes of measuring-wire were used : they were all 

 about | inch long, their diameters being 0*003, 0*0025, and 

 0*002 inch respectively. The weights of these wires are 

 approximately 0*00182, 0*0013, and 0*00081 gramme respec- 

 tively, whence, taking the specific heat of platinum as 0*037, 

 the thermal capacities are 0*0000673, 0*0000481, and 0*00003 

 gramme-degrees respectively. 



I had intended to measure the temperatures by means of 

 all three wires, and from the results obtain a curve giving 

 the lag of the wires; this was, however, not possible owing to 

 want of time and apparatus, but the two finer wires gave 

 temperatures which were nearly equal, while the coarser wire 

 gave results somewhat lower than the two others. This in- 

 duced me to work with the 0*0025 wire for the bulk of the 

 experiments, as the 0*002 wire had so little mechanical 



* Phil. Trans. A, 1887. 



