79 
1921-22. 1 New Method of investigating Colour Blindness. 
opening, the colour of the image at E changes through pink and magenta 
to purple and finally blue. The colour of the image always appears uniform 
all over, and the amounts of the red and blue in it are proportional to the 
areas of the face of the biprism covered by the red and blue parts of the 
filter. Filter 1 accordingly enables us to travel along the red-blue side 
of the triangle. The lower half of filter 2 corresponds to the point A in 
fig. 7 ; hence filter 2 enables us to travel along the line red-A. Similarly, 
filters 3, 4, and 5 enable us to travel along the lines red-B, red-0, and the 
red-green side. Filter 3 should have produced white one-third of the 
length of the median from B ; as a matter of fact, it produced it at 31 per 
cent, of the length from B. This is a striking tribute to the accuracy of 
the Wratten filters. Filters 6, 7, 8, and 9 enable us to travel along the 
lines blue-green, DE, FG, and HJ respectively. The filters were made 
specially for the investigation by Kodak 
Limited. They consist of coloured gelatine 
cemented between plates of B quality glass. 
The total coloured area on each filter 
measures 5*2 cm. by 2 cm. ; above the 
coloured area is a clear part for fitting 
into the brass holder. One filter can be 
unscrewed and another fitted in its place 
in a couple of minutes. 
The manner in which the tests were 
conducted is explained best by a full 
description of one particular case. The observer was seated in front 
of the screen E in a darkened room, and the two filters 1 were fitted and 
adjusted, so that the two images on the screen appeared pure red. Then 
the tint of one of the images was gradually altered, and the observer was 
requested to say when the difference of tint was perceptible. As soon as 
he did so, this filter was stopped and the other one altered, the observer 
being requested to state when the tints were again the same. The propor- 
tion of area of aperture covered by the blue of the first filter was then 
read. It proved to be 28 per cent. This completed the first step. The 
tint of the first image was then altered again until the difference became 
perceptible, the other image brought to equality, and the area of aperture 
covered by the blue of the first filter again read. It proved to be 39 per- 
cent. This completed the second step. The process was then repeated 
until the images were wholly blue. The ends of the steps are indicated by 
marks on the red-blue side of fig. 8. 
The pair of filters 2 was then substituted for filters I, the images 
GREEN 
Fig. 7. 
