ON STANDARDS OF WHITE LIGHT. 63 
terms of the wave-length when made out for the different sources of light, 
are of the same form. ‘Thus a second research must be undertaken to 
discover whether the infinite number of spectrum radiation curves, which 
can be obtained from a carbon filament by varying the current, are 
identical in form when the filament is changed, but the material remains 
so far as possible of constant composition. 
It will be an object for a later research to determine whether, when 
the radiation spectrum curve of any source of light has been mapped, a 
similar curve can be found among the infinite number of curves which 
can be obtained from a single filament. 
The next step proposed is to examine a large number of carbon or of 
platinum-iridium filaments, and to find whether the radiation spectrum 
curve of different specimens of the same material is identical when the 
resistance is changed in all to « times the resistance at 0° C. If this 
law be true, a measurement of the resistance of the filament would be a 
convenient statement of the nature of the radiation curve. If, then, a 
number of filaments were thus tested to give the same radiation spectrum 
curye, their luminosities would in all cases be proportional to their 
radiations, or (if there be no loss in convection, conduction, absorption, 
and reflection) proportional to the electrical energies consumed, 
Thus it might be hoped to establish a standard of white light, and to 
deiine it somewhat in the following manner :—A unit of light is obtained 
from a straight carbon filament, in the direction at right angles to the middle 
of the filament, when the resistance of the filament is one-half of its resistance 
at 0° C., and when it conswmes 10° C.G.S. wnits of electrical energy per second. 
Since Mr. Swan has taught us how to make carbon filaments of 
constant section by passing the material of which they are composed 
through a die, it is conceivable that another absolute standard should be 
possible—viz., a carbon filament of circular section, with a surface, say, 
+ ip 8q. centimetre, and consuming, say, 10° C.G.S. units of energy per 
second. 
Whether such standards are possible or not depends upon the experi- 
ments of the Committee. The probability of success is sufficient to render 
these experiments desirable. 
Proposed Later Experimental Researches. 
Should these hopes be realised, and an absolute standard of white 
light thus obtained of a character which would commend it to the civilised 
world, it would then become an object of the Committee to find the ratio 
of luminosity when the radiation spectrum curve of the standard filament 
is varied by varying the current, and consequently the resistance of the 
filament. 
Thus, by a large number of subjective experiments on human eyes, a 
multiplier would be found to express the illumination from the standard 
lamp, with each degree of resistance of the filament. 
A research, previously hinted at, would then be undertaken—viz., to 
find whether the radiation spectrum curves of all sources of illumination 
agree with one or other of the curves of the standard filament. It is not 
improbable that this should be the case except for the high temperature 
of the electric arc. 
Should this be found to be true, then photometry would be very 
accarate, and the process would be as foliows:—Adjust the standard fila- 
ment until its radiation spectrum curve is similar to that of the liyht to be 
