MAR. 4, 1923 CRITTENDEN: MEASUREMENT OF LIGHT 79 
melting point of platinum on the Nela Research Laboratory scale. 
It would appear, however, that agreement of the photometric results 
could be obtained by assigning such a temperature well within the 
range of the present uncertainty of the true melting point. 
While more refinement will be necessary before the practical stand- 
ards of candlepower can be based on black body measurements, these 
two investigations seem to justify hope that this can eventually be 
done so that we shall have a really reproducible primary standard 
depending only on the melting point of platinum and the complete 
radiator, and otherwise independent of the properties of materials. 
Flux Standards.—Before passing from the subject of standards it 
may be well to revert to the question of measurement of flux. All 
the standards mentioned serve to furnish values only of candlepower 
or brightness, whereas now-a-days all lamps are preferably rated in 
lumens, so that some sort of transformation of values is necessary. 
This is accomplished in two steps. Standards of flux (or of mean 
spherical candlepower) are established by measuring successively the 
candlepowers of electric lamps at various angles from the axis of the 
lamps. The total flux can then be calculated, and these lamps are 
then used as standards in an integrating sphere for the measurement 
of other lamps. 
A full discussion of the theory of the sphere as actually used is 
hardly practicable here, but the basic principle is exceedingly simple. 
The interior of the sphere is supposed to have a perfectly diffusing 
surface; every element of this surface when illuminated reflects light 
of an intensity which is at a maximum normal to the surface and falls 
off in proportion to the cosine of the angle as one departs from the 
normal. If we consider where this light falls on another element of 
the surface, we find by a simple calculation that each element of the 
surface illuminates every other one equally. The result is that, if we 
do not count in the original light which fell on the surface to make it 
luminous the illumination of all parts of the interior is the same. 
Furthermore if the wall is uniform, the illumination of all parts will 
be the same no matter how the original light was distributed. If we 
therefore put a lamp in the sphere, add a small screen to shade a spot 
from the lamp, and then measure the brightness of that spot, we 
have a measure of the total flux from the lamp. ~ 
In actual practice the theoretical requirements of the sphere can 
be approximated so closely that the systematic errors can be kept 
below one per cent except in extreme conditions. All lamp factories 
and important laboratories now use spheres, and a large proportion 
of the standard lamps called for are flux standards. 
