Nov. 17, 1881 | 
NATURE 
65 
of a very acute-angled prism held close to the dispersing prisms 
in such a position that its refracting edge was horizontal, divid- 
ing the field of view into two equal parts. The action of this 
prism is most easily under-tood by again supposing the light to 
enter at the eye-slit, Half of the lizht proceeds as before, 
forming ultimately a pure spectrum upon the side of the box. 
The upper half of the beam, however, is deflected by the acute- 
angled prism, and the corresponding spectrum is thrown up- 
wards, so as to lie somewhat higher upon the side of the box. 
This = is also cut away, and provided with movable screens. 
By the principle of reversibility the consequence is that an 
eye placed at the first slit sees ‘wo uniform patches of colour, 
e lower formed as before by light from the lower set of slits, 
the upper, covering the acute-angled pri-m, by light from the 
upper set of slits. These colours are in close juxtaposition, and 
tay be compared with ease and accuracy. 
The great difficulty in this class of instruments is to devise 
any efficient and reasonably simple method of controlling the 
position and widths of the slits. In the present case I contented 
myself with strips of blackened cardboard cemented to the side 
of the box with sealing-wax, or soft wax, according to the degree 
of permanence of adjustment aimed at. One part of the field 
was illuminated with homogeneous yellow (about the line D) 
from a single slit. The other half was lighted with a mixture 
of full red and full green, and the observation consisted in 
adjusting the widths of the slits through which the red and 
green were admitted, until the mixture was a match with the 
simple yellow. 
‘The first trials of this instrument in the spring of last year re- 
vealed an intere-ting peculiarity of colour vision, quite distinct 
from colour blindness. The red and green mixture which to 
my eyes and to those of most people matches perfectly the 
homogeneous yellow of the line D, appeared to my three 
brothers-in-law hopelessly too red, ‘‘almost as red as red sealing- 
wax.” In order to suit their eyes the proportion of red had to 
be greatly diminished, until to normal sight the colour was a fair 
green with scarcely any approach to yellow at all. So far as 
could be made out at the time, the three abnormal observers 
agreed well among themselves, a fact which subsequent measure- 
ments haye confirmed. It appeared afterwards that a fourth 
brother was normal as well as the three sisters. 
These peculiarities were quite unexpected. After the fact had 
been proved, I remembered a dispute some years before as to 
the colour of a dichromatic liquid, which appeared to me green, 
while one of my brothers-in-law maintained that it was red ; but 
the observation was not followed up, as it ought to have been, 
each of us, I suppose, regarding the other as inaccurate. After 
the establishment of the difference I determined to carry out a 
plan, which I had tried with success some years before (October 
1877), for a colour-mixing arrangement depending on double 
refraction, by which I hoped to obtain an easily adjustable 
instrument suitable for testing the vision of a number of persons. 
In my original experiments I used a 60° doubly refracting 
prism of quartz, which threw two spectra of the linear source 
upon the screen containing the eye slit. These oppositely 
polarised spectra partially overlapped, and by svitable placing of 
the prism could be made to furnish red and green light to the 
eye. By the rotation of a small Nicol held immediately behind 
the eye slit, the red or green could be isolated or mixed in any 
desired proportion. One advantage of this arrangement is that 
the two component lights come from the same slit, so that we 
are less dependent upon the uniformity of the light behind ; but 
it is perhaps a greater merit that the adjustment of proportions 
is effected by simple rotation at the eye slit, allowing the observer 
to try the effect of small changes with ease and rapidity. 
In the new instrument, which was completed during the autumn 
of last year, separate prisms were used to effect the dispersion 
and double refraction. For the sake of compactness, a direct 
vision prism by Browning, containing two flints and three crowns, 
was chosen, in co; junction with a small achromatic double image 
prism. At one end of a long narrow box, 24" X 2" X 2", the 
light is admitted through a slit whose position and width can be 
adjusted by sliding its jaws along adivided scale. After travelling 
about 94” it falls upon the double image prism mounted upon a 
small table so as to allow of rotation, and then after two more 
inches upon a collimating lens, by which the two beams are 
rendered parallel. Next comes the dispersing prism, and then 
the focussing lens, throwing pure spectra upon the other end of 
the box, which carries the eye slit. The distance between the 
two lenses is 33”, and the entire length of the box is about 24”. 
| The eye slit is a fixture, and immediately behind it is the rotating 
Nicol, whose position is read by a pointer on a divided circle. 
The parts of the spectrum from which the component lights 
are taken can be chosen over a sufficient range by use of the two 
adju-tments already mentioned. By rotation of the table on 
which the double image prism is mounted, the separating power 
is altered, and one spectrum made to slide over the other, while 
by moving the entrance slit the spectra are shifted together 
without relative displacement. 
It yet remains to describe the parts by which the comparison 
colour is exhibited. Between the double image prism and the 
collimating lens a small vertical reflector is mounted on a turn- 
table at an angle of about 45°. Its dimensions are such that it 
covers the lower half of the field of view only, leaving the upper 
half undisturbed, and its fanction is to reflect light coming from 
a lateral slit through the dispersing prism so as to throw a third 
spectrum upon the eye. The lateral slit is carried in a small 
draw tube projecting about 2” from the side of the box, and the 
light proceeding from it is rendered nearly parallel before reflec- 
tion by a lens of short focus. No adjustment is provided for the 
position or width of the lateral slit; all that is necessary in this 
respect being attainable by rotating the mirror and by varying 
the brightness of the light behind. As sources of light I have 
found Argand gas flames, surrounded by opal globes, to be snit- 
able, The gas tap supplying the lateral flame is within reach of 
the observer, who has thus the means of adjusting the match 
both with respect to colour and with respect to brightness, with- 
out losing sight of the subjects of comparison. The zero of the 
divided circle corresponds approximately to the complete exclu- 
sion of green, but readings were always taken on both sides of it 
so as to make the results independent of this adjustment.- The 
circle is divided into 100 parts, green being excluded at oand 50, 
and red at 25 and 75. Tenths of a division could be estimated 
pretty correctly, an accuracy of reading fully sufficient for the 
purpose, as the observations of even practised observers would 
vary two or three-tenths. 
It is evident that the numbers obtained are dependent upon the 
quality of the light by which the principal slit is illuminated. In 
order to avoid errors in the compari-on of different persons’ 
vision arising from this source, it is advisable always to take 
simultaneous observations from some practised individual whoze 
vision may be treated as a standard ; but no evidence appeared 
of any variation in the quality of the gaslight. The special 
application of such instraments to the comparison of the qualities 
of various kinds of mixed light was alluded to at the end of my 
Eee **On the Light from the Sky,” &c. (Phil. Mag., April, 
1871). 
{ ove obtained matches between simple and compound yellow 
from twenty-three male observers, principally students in the 
laboratory. Of these sixteen agree with myself within the limits 
of the errors of observation. The remaining seve1 include my 
three brothers-in-law, and two other:, Mr. J. J. Thomson and 
Mr. Threlfall, whose vision in this respect agrees very nearly 
with theirs. The vision of the other two observers differs from 
mine in the opposite direction. In one case the difference, though 
apparently real, is small, but in the other (Mr. Hart), though 
there was some difficulty in getting a good observation, the dif- 
ference is most.decided. Among seven female observers whom 
I have tried, there is not one whose vision differs sensibly from 
my own. Pes 
Although the number examined is insufficient for statistical 
purposes, it is evident that the peculiarity is by no means rare, 
at least among men, As far as my experience has gone, it would 
seem too as if normal vision were not of the nature of an average, 
from which small deviations are more probable than larger ones ; 
but this requires confirmation. In order to give a more precise 
idea of the amount of the difference in question, I have caleu- 
lated from the laws of double refraction the relative quantities of 
red and green light required by Mr. F. M. Balfour and myself 
to match the same yellow light. If we call R and G the maxi- 
mum brightnesses of the red and green light (as they would reach 
the eye if the Nicol were removed), and 7, gthe actual bright- 
nesses (as modified by the analyser) necessary for the match, then 
for Mr. Balfour— 
rlg = 1°50 (R/G), 
7/g= 3°13 (R/G). 
In other words, Mr. Balfour requires only half as much red as 
myself, in order to tum a given amount of green into yellow. 
while for myself — 
