1899. J on Measuring Extreme Temperatures. 105 



difference in the same direction at the F.P. of Pt as found by method 

 (4), and not a difference in the opposite direction as given by tho 

 thermo-electric method, on the arbitrary assumption of a different 

 type of formula for extrapolation at high temperatures. It is a 

 matter of some interest that the assumption of linear formulas for 

 both the specific heat and the rate of change of resistance, should lead 

 to results so nearly consistent over so wide a range of temperature in 

 tho case of platinum. 



The chief difficulty and uncertainty encountered by Paschen in 

 his experiments on radiation, was that of arranging the thermo-couple 

 so as to be at the same temperature as the radiating strip of platinum. 

 It is better for this reason to measure the temperature of the strip 

 itself by means of its electrical resistance, the method adopted by 

 Schleiermacher, Bottomley and Petavel. The same difficulty occurs 

 in the direct comparison of the scales of the thermo-couple and the 

 platinum-resistance thermometer. The simplest method of avoiding 

 this objection appears to be that recently adopted by the writer, of 

 enclosing the thermo-couple completely in a thin tube of platinum, 

 which itself forms the resistance thermometer. There can be no 

 question of difference of temperature between the two, and the same 

 tube may serve simultaneously for the expansion method and as a 

 radiating source for bolometric investigation of the law of radiation. 

 The uniformity of temperature throughout the length of the tube 

 can be tested at any time by means of potential leads, or by shifting 

 the thermo-couple to different positions along its length. The method 

 of electric heating is employed, and the central portion only of the 

 tube is utilised in the comparison. 



The methods of measurement, so far as considered, are in a certain 

 sense arbitrary in so far as they depend on extrapolation of empirical 

 formula}. If all these methods could be reduced by direct comparison 

 to perfect agreement with each other, a definite scale of temperature 

 would be attained to which all measurements could be referred, and 

 which would leave nothing to be desired from a purely practical 

 point of view. It is probable that this scale would not differ much 

 from the theoretical or absolute scale of temperature. For theoretical 

 investigations, however — without which no true scientific advance can 

 be made — it is a matter of such fundamental importance to refer every 

 measurement to the absolute scale, that no opportunity should be 

 neglected of extending the possible range of accurate observation 

 with the gas-thermometer, because this instrument affords at present 

 the closest approximation to the absolute or theoretical scale. A 

 consideration of the difficulties of the methods of gas-thermometry at 

 present in use will lead naturally to the best methods of extending 

 the range and accuracy of the instrument. 



In the ordinary method of gas-thermometry a bulb containing the 

 gas is exposed to the temperature to be measured, and the observation 

 consists in determining either the expansion of volume or the increase 

 of pressure of the gas. The principle is very similar to that of tho 



