

1905.] On Colour-Vision by Very Weak Light. 203 
the physical stimulus varies differently with the intensity for red and for 
green, and that it was desirable to make quantitative measurements.* 
With respect to the violet end of the spectrum another difficulty arises. 
Paper of all kinds fluoresces strongly in the violet, giving out a light 
consisting largely of green, and is hence unsuitable as a reflector before the 
slit of the spectroscope to reduce the light. Porcelain, which does not 
fluoresce, does not seem to reflect much of the violet near H and K. 
- Powdered sodium bicarbonate, pressed into a flat cake, seems as good as 
anything if pure. 
This part of the work was done from 1873 to 1878 in my father’s works at 
Cheshunt. 
Later Quantitatiwe Experiments. 
I repeated these experiments in 1893-5 with precisely similar results, in 
the Physiological Laboratory, Oxford, but on referring to the literature of the 
subject and finding how many experiments had been published with results 
diametrically opposed to mine, I determined to investigate the subject yet a 
third time, and to obtain also quantitative measurements in regard to: 
(a) the sensitiveness of the eye to red and to blue by daylight and in darkness, 
and (b) the time-relations of the subjective phenomena which accompany a 
prolonged period of darkness. 
This final revision of the work was begun in 1901 and, after various 
unavoidable delays, was completed about a month ago. 
The source of light was a 16 candle-power glow-lamp placed some distance 
from a piece of white paper fixed at an angle of 45° outside the wall of the 
dark room, in which was a circular aperture 2°5 cm. in diameter. The 
arrangement is shown diagrammatically in Fig. 2. A is the glow-lamp, 
B the paper reflector, C the wall of the dark room, D a polarising 
prism, E the spectroscope, with F a double-image prism over the 
eye-piece. 
In the focal plane of the eye-piece I placed a stop with two narrow slits, 
the distance between them being such that if the red about B was brought 
under one, the blue-violet about G was visible through the other. With the 
double-image prism the order in which the bands appeared was V,, Vz, Ry, Ry. 
Blue-violet was chosen to contrast with red because the light available was 
not strong in violet and I considered the relative intensity of the red and 
this part of the spectrum might be less affected by changes in the strength of 
_ * T was not at that time acquainted with Purkinje’s work, with which this is in entire 
agreement. : 
p 2 
