MAR. 4, 1923 CRITTENDEN: MEASUREMENT OF LIGHT is 
admitted that such a drift is possible. Differences between the vari- 
ous national laboratories may develop and a common primary stand- 
ard for reference would be highly desirable. Tradition and national 
pride make it difficult for any country to give up its old standards 
and especially so as to adopt those of some other nation, but so far 
as the United States is concerned the field is entirely open, and inter- 
nationally the time appears favorable for the consideration of new 
proposals in this field. 
In order to receive serious consideration, however, any proposed 
primary standard of light must be capable of reproducing values with 
an accuracy better than one per cent, and one-tenth per cent should 
be within the range of its possibilities. To attain such accuracy will 
require the most careful application of all our knowledge of radiating 
bodies, but all recent discussion trends toward defining the unit of 
light in terms of the established properties of radiators, rather than 
of trying to devise other special standards of candlepower. The 
properties of the complete radiator, or black body, are well established 
both in theory and in experiment, and this line of development appears 
to hold out most hope for the attainment of a light source exactly 
reproducible from specifications. The very rapid variation of bright- 
ness with change in temperature, which makes optical pyrometry 
practicable and precise, is, however, a serious and fundamental 
difficulty in using the black body as a standard of candlepower. At 
the temperatures which would be most suitable, a change of 1 per 
cent in temperature makes about 12 per cent change in brightness. 
Consequently the temperature must be known to about 2 degrees 
or one-tenth per cent in order to determine the light output to one 
per cent. The attainment of such accuracy does not, however, 
appear impossible. . 3 
Two comparatively recent experimental investigations bearing 
directly on the possible establishment of the black body as a primary 
standard of candlepower have been reported. One of these made by 
Hyde, Forsythe, and Cady in the Nela Research Laboratory was a 
determination of the brightness of a carbon-tube furnace over a 
considerable range of temperatures, including the temperature at 
which the light from the furnace matched in color that given by the 
carbon incandescent electric lamps which are now our basic candle- 
power standards. This temperature was found to be about 2077 
degrees K., and at 2077 degrees the brightness was determined as 
70.2 candles per square centimeter. This value has therefore been 
proposed tentatively as an absolute standard of light. 
