232 



Mr. H. L. Cailendar. 



applicable or unsatisfactory, since even the best observers using all 

 possible precautions disagree.* 



The platinum wire used was the purest obtainable, with an unusually 

 high temperature coefficient, and very unalterable. If Ri be the 

 resistance of a given coil at 100° C, and R the resistance at 0° C, 

 and R the resistance at any other temperature, the experiments 

 abundantly prove (1) that with proper precautions R is constant 

 to 0"01 per cent., even if the coil has been subjected for hours 

 to a temperature of 1300° C. (2) For several different pieces of 



wire from the same reel it was found that the ratio 5l was constant 



R 



to the same degree of accuracy, and equal to 1*3460. (3) That for 







two different wires of pure platinum, though the values of — 1= — _ 



Ro 







may differ 3 per cent., yet the values of — ~ 100, that is to say, of 



Ri — R 



the temperature Centigrade by platinum wire, agree to 0*1 per cent, 

 through the range — 600°. 



The method of comparing the temperature variations of different 

 wires is very accurate. The wires to be compared are interwound 

 symmetrically so as to form a double screw thread on non-conducting 

 material (e.g., clay), and are heated in a vacuous porcelain tube in a 

 gas furnace regulated to produce steady temperatures. When the 

 temperature is steady we may assume that the mean temperatures of 

 the two wires are accurately the same. The simultaneous resistances 

 are measured by the Wheatstone bridge method with a sensitive mirror 

 galvanometer and resistance boxes accurate to 0"01 per cent. The 

 comparison of two similar pure platinum wires might be effected 

 with even greater accuracy than 0*01 per cent, by using a wire bridge 

 (Carey Foster's method), since the ratio of their resistance is very 

 nearly constant (variation perhaps 2 per cent, between 0° and 600°), 

 and all sources of error affect both wires equally. The resistance of 

 the connecting wires at each steady temperature is measured and 

 allowed for, and thermoelectric effects are carefully eliminated : as a 

 precaution the insulation is also measured. 



It is evident then that by adopting one wire as the standard and 

 comparing others with it, temperature measurements may be made 

 strictly comparable through a very wide range, since the platinum 

 thermometer is almost universally applicable, and is usually more 

 convenient, sensitive, and accurate than any other kind. The chief 

 objection to its use is the need of auxiliary apparatus for accurately 

 measuring the resistance : its great superiority lies in its wide range, 

 its freedom from zero error, and its adaptability. A length of wire is 



* See article by W. N". Shaw, "Pyrometers," 'Encycl. Brit.,' 9th Edit., vol. 20, 

 p. 129. 



