62 Estimation of High Temperatures by Colour Identity. 



Comparisons were made up to an efficiency o£ about 0*75 

 watt per mean spherical candle for the vacuum lamp, or about 

 0*9 watt per mean spherical candle for the gas-filled lamp. 

 The results are shown in fig. 5, where watt per mean spherical 

 candle for the vacuum lamp is plotted as ordinate and watts 

 per mean spherical candle for the gas-filled lamp as abscissa. 

 The considerable difference between these curves must be 

 ascribed to differences in the amount of gas in the bulbs and 

 to varying diameters of filaments and spirals. 



The curves have been extrapolated to pass through the 

 zero of the diagram, and it will be seen that all the points 

 lie on a straight line which passes through the origin except 

 at the comparatively low values of efficiency. 



The ordinary working efficiency of gas-filled lamps is at 

 the present time about 0'7 watt per mean spherical candle, 

 corresponding in identity of colour of radiation with the 

 vacuum lamp at 0*5 watt per mean spherical candle — 

 i. e. 25 lumens per watt approximately. 



From the equation to the curve shown on fig. 3 this gives 

 a temperature of 2800° C. abs. for the ordinary working 

 temperature of tungsten in gas-filled lamps. 



If the half-watt lamps are overrun they will be found to 

 burn for a short time satisfactorily at 0'40 watt per mean 

 spherical candle, corresponding to a vacuum lamp at 0*32 watt 

 per mean spherical candle, or approximately 40 lumens per 

 watt. Using the same formula this value of lumens per watt 

 corresponds with a temperature of 3100° C. abs. (2833° C.).. 

 It is clear, then, that the melting-point of tungsten would be 

 ahove this value, but it cannot be said with certainty how 

 much higher. The comparison lamps used in the determi- 

 nation were of the vacuum type and themselves had tungsten 

 filaments. At 3000° C. abs. the blackening of the bulbs of 

 such lamps is rapid, and liable to lead to error if the tempe- 

 rature of the filament is pushed up to the melting-point. 

 The determination of the melting-point of tungsten, there- 

 fore, by this method could only be undertaken if precautions 

 were taken against errors due to blackening of the bulb. 

 The experiment will be more successful when gas-filled 

 colour standards are made by comparison direct against the 

 black body. Burgess and Le Ohatelier* give the melting- 

 point of tungsten as (3000 ±100)° C, a value with which 

 the authors' observations by the colour-identity method are 

 not inconsistent at the highest temperature at which they 

 could safely make measurements. 



* Loc. cit. p. 492. 



