HIGH TEMPERATURE MEASUREMENT 511 



but variations in the time of exposure to the high temperature 

 and also in the rate of heating make considerable differences in 

 the results. 



For measurements in the region above the range of the gas- 

 thermometer (say 1,200" C.) some extrapolation scale of tempera- 

 ture is at present the only available provisional standard. Two 

 chief methods have up to now been employed in the estab- 

 lishment of such a scale. The first — the thermo-electric 

 method — depends on the hypothesis that the formula, which 

 represents the relation between electromotive force and 

 temperature in a thermo-couple up to 1,200° C, holds at higher 

 ranges. The work of the National Physical Laboratory has 

 shown that a formula of the usual parabolic type, which 

 represents this relation very closely for almost any thermo- 

 couple formed of the platinum metals, gives a consistent 

 extrapolation scale up to the melting-point of platinum. 

 On this " thermo-electric scale " the melting-point of nickel 

 is 1,427° C, that of pure iron 1,502° C, and that of platinum 

 1,710° C. It is to temperatures on this scale that the readings 

 of couples are almost always referred. 



The second extrapolation scale is based on the assumption 

 of a definite value for the so-called constant in the Wien-Planck 

 equation, expressing the relation between intensity of radiation 

 and temperature. The usually accepted value for this constant 

 is about 14,500, but it is still appreciably uncertain. On this 

 optical scale, which should agree with the thermo-electric and 

 gas-scales at 1,200°, the melting-point of nickel would become 

 about 1,450° C. and that of platinum 1,750° C, whilst the " black- 

 body " temperature of the crater of the electric arc, which is 

 nearly independent of the current and voltage employed, is 

 about 3,500° C. 



A word in conclusion may be devoted to the standardisation 

 of the various types of instruments which have been mentioned. 

 This can be accomplished in this country at the National 

 Physical Laboratory, which was opened at Teddington in 

 1902. Into the methods by which these standardisations are 

 carried out, it is beyond the province of this article to enter. 

 During the past eight years it has been found possible at 

 that institution gradually to extend the scope and range of such 

 standardisations so that methods and apparatus are now avail- 

 able for tests on practically all types of instruments. Mercury 



