€4 Messrs. Paterson and Dudding on the 



2. The six sets of electric substandards of candle-power 

 are described, which vary in the colour of the light radiated 

 from that of the pentane lamp (red) to that of a tungsten 

 vacuum lamp operating at 1*5 watts per candle. 



3. The differences between the values obtained by different 

 observers in the process of stepping from one colour to the 

 next are discussed and compared with those obtained by the 

 direct comparison of the first and last sets of substandards. 



4. The probable errors of the determination are discussed 

 and shown to be of the order of 0*2 5 per cent. 



5. The absolute value of the unit of candle-power has been 

 redetermined from the pentane lamp, and found to agree 

 with the determination made by one of the authors ten years 

 ago to within less than 0*1 per cent. 



6. The corrections to the pentane lamp for humidity and 

 barometric changes were also redetermined, and found to 

 agree within narrow limits with those previously found by 

 one of the authors. 



7. The relation between the humidity and temperature 

 corrections for the pentane lamp is discussed, and the fact 

 that these two effects may act together is suggested as the 

 reason for a discrepancy which has been noticed between 

 the humidity constants of the lamp as determined in London 

 and Washington. 



General Discussion. 



The principal object of this paper is to describe and discuss 

 the methods which have been used at the National Physical 

 Laboratory, and the results obtained in the standardization 

 of high-efficiency metal filament lamps in terms of the unit 

 given by the 10-candle pentane lamp. Such a standardiza- 

 tion entails the comparison of two sources whose radiation in 

 the visible spectrum, though continuous, differs appreciably 

 in hue. The amount of this difference may be gauged by 

 remembering that if the voltage on an ordinary carbon fila- 

 ment lamp is lowered until the hue of its light matches that 

 from the pentane lamp, the carbon lamp will be found to be 

 burning at a rate of consumption of about 7*7 watts per mean 

 horizontal candle. The ordinary consumption of carbon fila- 

 ment lamps is in the neighbourhood of 4 watts per mean 

 horizontal candle, and of tungsten lamps between 1*0 and 

 1*5 watts per mean horizontal candle. If the Hefner lamp 

 were matched on the same basis the specific consumption 

 would be still higher, viz., at the rate of about 11*0 watts 

 per candle. 



In approaching this question some criterion is wanted in 



