576 A Statistical Survey of Colour Vision .
By R. A. Houstoun , M.A. , Ph. D. , D.Sc .
, Lecturer on Physical Optics in the University of Glasgow .
( Communicated by Prof. A. Gray , F.R.S. Received April 8 , 1918 .
) It is well known that there are many people with slightly abnormal colour vision , who would not be differentiated from the normal by any of the usual tests for colour-blindness .
The object of this paper was to make a statistical study of such people ; such a study is of interest because it enables us to say whether normal colour vision passes gradually into colour-blindness , or whether the colour-blind form a well-defined class by themselves .
This is important from the point of view of colour-vision theory .
# For , suppose we test the colour vision of 1000 persons , chosen entirely at random , and assign marks to it , ranging from 1 to 30 , taking 1 for total colour-blindness and 30 for the most perfect colour vision it is possible to have .
We could thus divide them into 30 classes .
Suppose next that the number of persons in each class is plotted against the number of the class .
Then we shall get a frequency curve .
Such a curve might have various shapes ; if it were to turn out of the shape shown in fig. 1 it would show we were dealing with two classes of persons , and it would be natural to identify the large maximum with the trichromats , and the small one with the dichromats of the Young-Helmholtz theory .
La. .
10 20 30 Fig. 1 .
Fig. 2 .
If , however , the curve turns out of the shape shown in fig. 2 , and if , after the methods of biometricians have been applied , no trace can be found of the little maximum on the left , then we would have evidence that colourblindness is not due to the absence of one of three colour sensations , as maintained by the Young-Helmholtz theory of colour vision .
The point at issue then is : Has colour-blindness a maximum of its own , or is it merely an outlying portion of the main curve ?
A Statistical Survey of Colour Vision .
How are we to divide colour vision into classes ?
The method which occurs most naturally to one is that of the Edridge-Green colour perception spectrometer .
This instrument is a spectrometer , in the focal plane of the telescope * of which are two adjustable shutters with vertical edges : the shutters can be moved into the field from right and left respectively , each by its own micrometer screw .
To each of these screws is attached a drum , which gives directly in wave-lengths the position of the edge of the corresponding shutter .
Thus the limits of the region of spectrum exposed can always be ascertained .
In using the instrument the observer first ascertains the exact position of the red end of the spectrum , placing the edge of the right-hand shutter at that point and reading its position .
The left-hand shutter is then moved in and out , until he obtains the largest portion of red which appears monochromatic to him , no notice being taken of variations in brightness , but only in hue .
The position of the edge of the left-hand shutter is then noted .
The edge of the right-hand shutter is next placed at the point previously occupied by the edge of the left-hand shutter , and the latter moved towards the violet , until the next monochromatic patch is marked out .
In this way the whole spectrum is traversed , until finally its violet end is reached .
The number of monochromatic patches recorded by the observer would give the number of the class to which he belongs .
It is not possible to use Dr. Edridge-Green 's results for the purposes of my test , as they are not a random distribution , but specially selected cases .
I resolved , therefore , to acquire data of my own .
As colour perception spectrometer I used a Browning spectroscope with a single 60 ' flint prism .
The eyepiece was fitted with shutters which could be moved into the field from right and left in the focal plane , each by its own micrometer* screw .
The collimator was provided with a new slit , and diaphragms were fitted inside both collimator and telescope in order to reduce stray light to a minimum .
A scale of wave-lengths was fixed up on the wall of the room at a distance of 4 metres from the instrument ; this scale was seen in the field just below the spectrum , by reflection in the second face of the prism .
As source a 16-candle-power carbon glow lamp enclosed in a tin with a ground glass window was used .
The wave-length scale was illuminated by a glow lamp , and both this lamp and the other one were controlled by switches close to the observer , so that the spectrum and scale could be viewed either simultaneously or separately .
The method of reading the wave-length was thus similar to that employed in the old Bunsen and Kirchhoff Spectroscopes , but , as the scale was in Indian ink on drawing paper and quite 33 cm .
long , it was , of course , much more accurate .
It should be noted that , as the telescope was focussed on this scale at 4 metres , the collimator had to be put Dr. R. A. Honstoun .
slightly out of focus for parallel light , and the rays did not go through the prism parallel , but this was not an appreciable source of error .
To compare the colours at the edges of a patch , the eye has to travel backwards and forwards from the one edge to the other .
Lord Rayleigh has pointed out* that , while , when working in this way , he marks out as monochromatic the region 586-595 yaya , or 15 times the distance between the sodium lines , yet , when the colours bound each other sharply , he can detect a difference of colour between the sodium lines themselves .
In the former case the gradual character of the transition is an obstacle to the recognition of the difference .
Dr. Edridge-Green takes exception to Lord Rayleigh 's results , f stating that the difference the latter obtains between the two halves of his field is not one of colour , but is due to stray light or to a difference in intensity .
If a double-image prism is placed between the eye and eyepiece , so as to form two images of the rectangular monochromatic patch , and these images are arranged side by side , so that the more refrangible side of the one touches the less refrangible side of the other , then , according to Dr. Edridge-Green , the double patch appears monochromatic the whole way across .
I have repeated this experiment with the double-image prism , using both a calcite Wollaston prism with a separation of 1 ' and a simple calcite wedge with a smaller separation , and find that by this means I can reduce the width of the monochromatic patch in the neighbourhood of the D lines to | , and in the neighbourhood of 480 yaya to J. These were the only regions in which I tried the experiment .
The result is quite decisive , and , to me at least , admits no manner of doubt .
The reason why Lord Rayleigh was able to reduce the difference of colour to y1^ was doubtless owing to the more favourable conditions in his experiment .
He was working with two fields , which were probably as broad as they were deep ; with the double-image prism and colour perception spectrometer I was comparing two very narrow rectangles , in one case the apparent width of each being only 15 ' .
I am therefore unable to agree with Dr. Edridge-Green that the monochromatic region is a " fundamental physiological fact .
" Its width depends to some extent on the dispersion of the spectrometer and magnification of the eyepiece .
Thus , in Dr. Edridge-Green 's instrument , the average number of patches is about eighteen ; in mine it is 14*86 .
In my instrument the whole length and height of the spectrum subtend apparent angles of 27 ' and 8 ' at the eye .
* 'Scientific Papers/ vol. 5 , p. 621 ; ' Roy .
Soc. Proc./ A , vol. 84 , p. 464 ( 1910 ) .
t 'Roy .
Soc. Proc./ B , vol. 84 , p. 116 ( 1911 ) .
A Statistical Survey of Colour Vision .
The brightness of the spectrum has also an influence on the number of patches visible , but this can be neglected as far as the slight alterations of intensity that occur under ordinary working conditions are concerned .
For example , the brightest part of my spectrum has an illumination of 500 metre-candles .
I then see 17 patches .
When the illumination of the brightest part of the spectrum is reduced to 64 metre-candles by putting a fogged photographic plate in front of the slit , I see 12 patches .
When it is reduced to 2 metre-candles , I see 7 patches ; at this illumination the violet is so faint that I cannot see it at all .
From measurements made at different times , it seems that the voltage of the lamp did not vary 5 volts on 250 .
This would cause only a 10 per cent , change in the intensity , and would not affect the number of patches appreciably .
The slit was always kept at the same width ; when illuminated with sodium , it appeared 10 nn wide .
The following Table gives my results .
Of the 79 observers , 74 were students ; the remaining five included two professors of physics , a lecturer on physics , a head science teacher , and myself .
Of the students 38 wrere women ; the men and women are indicated respectively by the letters M and W in the Table .
The number immediately after this letter gives the number of patches .
The other entries give the limits of the patches in 10-6 cm .
When the decimal is not printed , it is understood to be 0 .
Of the students , 8 were in their Honours course , 12 in their Higher course and the remainder in their Intermediate course , so they were all well experienced in physical measurements .
The distribution was purely a random one ; I knew nothing of the colour vision of any of the observers before asking them to make a set of readings , and no one who was asked declined .
The observers included all the laboratory students available , except one or two foreigners with a defective knowledge of English .
The Monochromatic Patches Marked out in the Spectrum by 79 Different Observers .
1 M. 22 73.5-69-66-64-62-60.7-59.8-58.5-57 * 8-56.3-54.5-52.5-51.3-50- 49'2-48-47-46-44 * 7-43.3-43-42-41 .
2 M. 18 76.5-70'5-66-63-61-59.8-58-57.5-55.5-54-52.2-50.8-50-49.5- 48.5-47'5-46'5-45-42.7-42.2 .
3 M. 20 76.5-65-62.5-60.5-59.8-58'1-57-55.8-54.4-52.6-51.4-50.6-49.7- 48.9-48.2-47.4-46.8-46'2-45-43-1-41-6 .
4 M. 17 74.7-67-64.6-62.1-60.8-60-58.7-57.2-55.7-53.3-5l.5-49.8-48.7- 47 .
5-46 .
4-45 .
3-43 .
6-42 .
0 .
5 W. 12 77-65-60*2-54 .
3-51 .
2-50-48 .
5-47 -5-46'8-46-45 .
6-43-41 .
5 .
6 W. 26 74-65.6-61.2-59.4-58.5-57.6-56.5-55.5-55-53.5-51-49\#166 ; 5-48.7-48-47.3-46.7-46.2-45.8-45'4-44 '8-43.8-43.3-42.6-42.2-41.7-40.6-38.8 .
7 W. 13 71-65-61-59-57 * 2-56.2-54.5-52.4-48-47-46\#166 ; 2-45-43.5-41.5 .
$ $$$$ Dr. R. A. Houstoun 8 W. 18 9 W. 20 10 W. 12 11 M. 17 12 M. 14 13 W. 9 14 M. 18 15 M. 20 21 M. 17 22 M. 16 23 W. 24 24 W. 19 25 M. 12 26 M. 18 27 M. 11 .
28 W. 13 29 M. 13 30 M. 12 31 W. 9 32 W. 15 33 W. 13 34 W. 12 35 W. 16 36 W. 13 37 M. 14 38 M. 5 39 M. 16 *0 M. 16 73-65-63-62-61-60 * 5-59.5-59-58-56.6-54 '9-52.5-50.8-49.4-48.t47-45-8-42-8-41-8 71'3-65 '5-62.7-60.5-59 * 1-57.8-56'1-54.8-53.4-52-51.2-50.1-49\#166 ; 248 .
8-48 .
1-47 .
6- 47 .
2-46 %5-45 .
6-43 '6-41 .
6 .
75-64-60.3-57 * 8-56-52 * 8 a1-49.5-48.2-47.2-45.8-43.5-41.5 .
73_64.8-61.4-60-58.9-57.5-56.8-55 .8-51.2-49'4-48.8-48.1-47.4- 46.8-45.8-44.6-42.6-41.2 .
72- 66-62.9-60.8-59 * 2-57 * 6-54.8-51.6-49 * 8-48.7-47.4-46.1-44 '242-0-40-8 .
71.3-63-58.8-56-53-49-47.4-45.4-41.8-41 .
74.2-66-63.8-61-59 * 7-57 *8-56.6-53.8-51.1-49.9-48.8-47.9-46.9 % 46-3-45-1-43-7-42-5-41-8-41-5 .
73- 66-64 -2-61 * 1-60.0-59-57.8-56'1-54-51.9-50.7-50-48.8-48 .
347.5-46.4-45.4-44-42.7-42.2-41.5 .
73-65-62-59-56-51.8-50-48.5-47.2-45 *3-43'1-41'9-40 * 8 .
77.5-64.8-59.9-57.6-55.8-53-50.8-49.2-47.6-46-43.8-42.8-41-2 .
72- 65-61-59-57-54.3-51.7-50*2-48 '7-47.5-46 *2-43.8-41.7 .
73- 67-61-59-57.8-55.2-52.5-49 * 7-49-48-47.2-46-44.8-43-41.8 .
78-72-66.6-63-60.2-59-57.8-56.2-54.2-52-50-48.5-47.3-46.1-44 .
2- 42-1-40-2 .
71-65-61-59-58-56-6-55-53.1-51.1-50-49-48 * 2-47.5-46'-8-46-45.242-9-41-7 .
71-63-60 .
6-59 ' 7^58-56-53 - 7-51 * 9-51 .
4-50 ' 4-49 .
3-48 .
4-47 '2-46 .
244-4-42-5-41-7 .
71- 63.8-60.8-59.5-58.5-57\#166 ; 8-57-56-54.2-52.5-51.3-50.5-49.9-49.3- 48 .
8-48 .
3-47 .
4-46 .
7-45 .
8-44 .
5-43 '7-42 .
8-42-41 .
2-39 -7 .
67-63-60.5-59.7-58-57.4-56.5-55 '3-53-51'1-50-48.8-48-47'3-46'446-45'5-45'1-44.6-41.8 . .
70- 64'5-62-59'8-58'7-57'6-52-49-47 * 6-46'5-45'4-42.8-41.5 .
77- 64-61-59.7-58.2-56.8-55'1-53'3-51'7-50'2-49.2-48 *2-47.4- 46'7-46'1-45 *5-44.8-43.2-41.3 .
72- 64-61-59'2-56-53-51'2-49'8-48.2-46'8-45'2-41.2 .
75- 61-59'7-58-56.6-53.4-51-50-49'3-48.3-47-45'8-43.8-41.5 .
72-62'4-60.4-58.8-57.8-56'4-53.6-51.1-49.2-48.5-47 * 1-45 * 5- 43-2-41-8 .
76- 67-62.2-59.7-57.6-54-6-51'4-49.2-47 * 7-46'2-44.9-43-2-42 .
74.2-64-59.2-57-54.8-49.3-47.8-46.8-45-40.8 .
74- 66-60-58 * 7-57.8-57.1-56.0-53.8-51'7-49'1-48-47 * 1-46'3-45'343'6-41'5 .
72-63'8-61-59.5-58.1-56.6-52-50 '7-49 * 1-48.1-46'9-45 * 6-42'6-41 .
78- 65-61-59.7-58.7-57.3-56'1-54'1-51.1-48 '2-47.3-46 * 2-41.6 .
76-66'5-63.1-61.1-60-58.8-57'6-55 *4-52 '5-50 *8-49'3-48'4-47.646-7-45'5-44'4-40'8 .
71- 63-60-59-57.8-56.8-55'2-52'3-49'6-48'6-47 * 8-46'3-43'7-42.2 .
70-64-61'6-60-58.8-57-54.6-51'7-49'9-48.7-47.7-46'8-45 '842-9-41'2 .
70-62-57-51'8-48.4-42.5 .
75- 63-60'6-58.5-55-52 '9-51 '6-50.6-49 *9-48.8-47'5-46'6-45'344-2-42'9-42-2'41-2 .
70-65.4-61-60-58.8-57.2-54-52-50.7-49.8-48.8-47.7-46.8-45.6- 43-6-42-4-41-5 .
41 W. 11 42 M. 9 43 M. 16 44 M. 15 45 W. 12 46 W. 16 47 W. 15 48 M. 12 49 M. 9 50 M. 13 51 M. 13 52 W. 16 53 W. 5 54 w. 13 55 M. 15 56 M. 21 57 W. 16 58 M. 19 59 M. 15 60 M. 13 61 W. 18 62 W. 19 63 M. 16 64 M. 14 65 W. 12 66 W. 16 67 M. 15 68 W. 19 69 M. 13 70 M. 13 71 M. 8 72 W. 14 73 W. 17 74 W. 10 75 W. 16 A Statistical Survey of Colour 74- 63-61-59 .
2-58 \#166 ; 1-56'6-54 .
5-51 .
5-49 .
7-46 .
2-44 .
6^2 .
71- 63-59.9-58.1-51.3-49.7-48.3-46.6-44.3-42 .
72- 66 .5-63.8-61.2-60-58.8-57.3-55.8-53.2-51-49 '9-48.8-48-46 * 945-7-43 -7-41-9 .
72- 68-63-61.8-59.5-58.5-54.9-51'4-50.4-49.3-48.2-46.9-45.943-8-43-2-42-3 .
75- 64-60-56.5-52.2-50.7-49.4-48.4-47.4-46.2-44.8-43.3-41.8 .
73- 65-63-60.8-59.5-58.2-56.8-53'4-51.1-49.3-48.2-47-45.6-44- 42-4-41-5-40-7 .
69- 65.6-61-59.7-57.8-56.6-54.8-52.4-50-48.6-46.6-45.8-44.9-4442-41 .
71-66-63.5-60.8-59-57-51-49 -2-48.4-47 * 2-46-43.2-42.2 .
66- 58 .
8-54 .
5-52-50 '5-49 '2-48 .
4-47 * 5-45 * 2 .
75-69-65-62-60-58.8-57 * 1-54.8-51.8-49.7-48.4-46-42.5-41.7 .
71- 63-60.8-59.9-58.2-56.4-54 * 5-52-50-47.6-46.8-45 * 4-43.9-42.7 .
77_65_62 .
2-61 .
5-59 ' 3-58 .
8-58-57-55-52 .
2-50 .
8-49 '4-48 .
5-47 .
346'2-44-5-40-8 .
73-62-57.2-50.2-48-41.4 .
75-69.5-66-62.8-60.3-58\#166 ; 2-55 '2-52-50.8-49-47 '2-45'7-44-42 .
73-66-63'5-61'8-60'2-57'2-56-54'5-52'5-51'4-49'9-48'8-47'445'5-43'4-41'7 .
73-67-64-62'5-61'2-59'7-58'8-58-57'2-54.2-51'3-50'1-49'6-49'2- 48'8-48'5-47'6-46'7-45'7-43'2-42-41'4 .
70- 65-62-61'4-59'9-58'6-58-57-52'7-51'4-49'5-48'5-47'6-46'544-8-42'4-41'6 .
73-65-62'1-60'5-59-58'4-57'8-56'4-54'9-52'3-51-49'8-49'1-48'447 '5-46'5-45'9-44 .
8-42'9-42'3 .
67- 63-62-59'8-58'2-57-55-52 '8-51'0-49'8-49'2-48'2-46'6-45'1- 43'5-42'3 .
73-66'5-62-61-59'8-58'8-56'5-52'4-50'7-49'9-48'8-47'1-44'7-41'5 .
73-63-60'5-59'5-58'8-58-54.8-52'2-50'3-49'6-48'8-48'2-47'246'2-45'2-44-42 '8-41'6-40'5 .
70- 65-62'8-61-59'7-58'2-57'2-56'2-54'6-52'5-51'4-50'1-49'272- 66-63-62-61-60-59'1-57'8-56'2-54-50-49-48'2-47'2-46-42'8-41'8 .
77-65-62-60'3-59-57'7-56-52'4-50'6-49'2-47'8-46'7-45'8-44'3-42'5 .
73- 66-61-60-58-55-51 '8-50'2-48'8-47'7-46'4-44'8-41'2 .
71- 64-5-61-59'7-58'2-57-56-53 '8-51'7-50'2-49.1-48'2-47'0-46-45- * 43'2-41'4 .
75-63-61'2-60-58-57-55'8-53'9-52-50'3-49'2-47'9-46'6-45'7-43'7-42 77-67'5-65'5-63'5-61'2-59 '5-58-56'9-54'9-53-51'9-51-49'8-48'747'5-46'4-45-43'2-41'8-40'9 .
72- 63-60-59-58-54'3-51'8-50'1-48'8-47'8-46'9-45-42'8-41'8 .
73- 62-59'5-57'8-56'5-54'8-52'4-50'1-49-48-47-45'8-43'5-42 .
75-69-65'4-62-58-53-48'5-44'5-42 '3 .
73- 62-60 '2-58'8-57'2-55'4-52'7-50'2-48'6-47 * 5-46'7-45'6-44'743'5-42'0 .
69-65-63-62-60'5-59'5-58'6-57'3-56'2-53'7-51'3-50'3-49'4-48'547'7-46'4-45'2-42 .
74- 62-60'5-59'5-57'6-53'5-50'2-48'8-47-45'6-42'5 .
74-66-61'5-59'5-58-57-55'2-54-52-51'2-49'7-48'8-48-46'7-45'243-41-8 .
Dr. R. A. Houstoun .
76 W. 15 72-63.2-61-60-58.9-58-56.3-54.5-51.6-50.1-49.3-48\#166 ; 1-46.7-45.7- 43 9-41-4 .
77 W. 16 71-66-64-62-60.5-59-57.6-56\#166 ; 5-55.1-53.6-51.7-50-48.8-47.4-46- 44-2-42 .
78 M. 17 73-64-61.8-60- 5-59 * 7-58.8-58-57.2-56-54.6-52.9-51.5-50 '2-49.2- 47-8-46-44-3-42-4 .
79 M. 24 73-65-63-61-59.9-59-58.1-57.2-56 *2-55-53 '7-52-51-50-49.5-49.2- 48.8-48'2-47.5-46.7-45.5-44-42.9-42-41.3 .
Before making the test the method of taking readings was explained by means of a water-colour drawing of the spectrum and two sliding card-hoard screens .
Then a start was made from the red end .
Only one reading was , as a rule , taken for each patch , but generally one or two readings were repeated in each set , after the set was completed .
It was found they almost always agreed .
No preliminary practice was given .
The first two or three observers read their own wave-lengths , and set the telescope in each fresh position themselves .
Afterwards I did this , as I found it saved time .
Each complete test required 30 or 35 minutes .
I supervised every test personally , and watched the fingers of the observer on the micrometer screw , to make sure he was testing the patch from both sides .
Only in 3 of the 79 cases was there any misunderstanding .
One of the three I discovered , when about halfway through , was marking off differences of intensity as well as of colour , so we had to start again .
Another was inclined to ignore degrees of yellow at the brightest part of the spectrum , regarding them as changes of brightness , while a third in one case ignored a thin strip of colour close up to the edge of a patch " because it was a bit of the last patch accidentally left over .
" The question arises as to the reliability of the numbers .
If the observer were to make a second set of readings , would he obtain exactly the same number of patches ?
Probably not .
No. 4 in the Table is my own set .
About a year previously I made a set , obtaining 76-64.7-62.7-61-59\#166 ; 2-58 '2-57.2-56-53.3-52.2-50.7-49.8-48.7-47.6-46.7-45 * 543-7-42 , i.e. , 17 patches , but with different limits .
Three weeks later I made two sets on the forenoon and afternoon of the same day , when I was very tired , obtaining 76- 65-62-60-59-58-56.6-55-52-49.8-49-47'8-46.6-44.2-41 *6 , and 77- 65-62.5-60 '2-58.7-57.7-56-53-50.3-49.3-47.8-47-45.5-43'5-41.5 , giving 14 patches each time .
In taking these two sets the left-hand shutter was moved only in one direction , in the first set into the field , and in the second set out of the field , the object of the experiment being to see whether A Statistical Survey of Colour^ Vision .
583 the direction of setting had an appreciable effect on the number of patches .
Apparently it had not .
Four months later than this I made a set when feeling very fit , and obtained 19 patches , namely 76-66-63-61-59 * 7-58 * 8-58 * 1-57-56-54 * 5-52 * 5-51-50-49.1-48 * 4-47 * 5-46 * 5-45 * 443*2-41-3 .
Also seven of the other observers made a second set four weeks or so later than their first set , when No. 9 got 24 instead of 20 , No. 1 got 26 instead of 22 , No. 6 got 30 instead of 26 , No. 38 got 6 instead of 5 , No. 2 got 16 instead of 18 , No. 18 got 15 instead of 12 , and No. 17 got 16 instead of 12 .
But the most interesting case was that of No. 5 .
She obtained 12 patches originally , 60*2-54*3 being only one patch .
This observation was repeated at the time and verified .
Six weeks later she started a second set , obtaining seven patches between the red end of the spectrum and 60 , where she had obtained only two before .
The test was then discontinued , and another test made a week later , when she obtained 24 patche^for the whole spectrum , four of these being between the red end and 60*4 .
As the laboratory work of this student on Fresnel 's biprism , polarimetry , etc. , was exceptionally accurate and reliable , I have no alternative except to assume that her colour vision had undergone a decided change .
Excluding her case as an abnormal one , I believe the number of patches recorded by the same observer can vary within a total range of about 35 per cent. , but if the two sets are made on the same day , the agreement will be much better .
I am inclined to think the general increase in the number of patches at the second determinations recorded above was due to the fact that they were all made on two very dark December days , when the appreciation of colour might be consequently more acute .
It should , of course , be remembered that variations of individual observers from day to day compensate one another , and do not affect the results obtained from a large number of observers over a long period . .
Fig. 3 gives the results , each observer being plotted , the men as circles and women as crosses , above the number of patches obtained .
It will be observed that this varies from 5 to 26 .
The two observers with 5 and the observer with 8 patches were decidedly colour-blind .
The observer with 26 was one of the four students who " saw indigo."* The other three have left the University , and so I could not get their number of patches .
The two observers with 5 patches did not know they were colour-blind ; they had difficulty in naming yellow , calling it red or green .
The observer with 8 patches knew ; he was colour-blind , as he could not distinguish a blue * R. A. Houstoun , " Newton and the Colours of the Spectrum , " 4 Science Progress , ' vol. 46 , p. 250 ( 1917 ) .
JDr .
R , .
A. Houstoun .
from a green tramcar at a greater distance than 100 yards .
He called green a dirty yellow , and , as there was some doubt as to his confusing intensity with hue in his first set of readings , a second set was made immediately afterwards , when he obtained 5 patches .
x X XX x XX X XX x XX X XX XX X O XX o o X X o X X X o o X o o o X X X o o o o o o o X X X o X o o o o o o o X o X o o o X o o o o o o o o O Q O O o X _JU -i .
JL 5 10 15 20 25 Fig. 3.\#151 ; Circles , men ; crosses , women .
The observations recorded in fig. 3 show that normal colour vision has quite enough " scatter " to include colour-blindness as an outlying portion of itself .
It is true , of course , that the two " fives " are separated by a gap at the one end , but so are the two " twenty-fours " and the " twenty-six " at the other end .
But it would require a very much greater number of observers than 79 to positively disprove the existence of separate maxima for the dichromats and monochromats .
I have been much puzzled by the minimum at 14 in fig. 3 .
It seems to divide the observers into two groups , though two members of the mathematical staff of the University to whom the results have been shown declare that this division is purely fortuitous .
If the observers are plotted against the width of the patch they obtain at 589 / -\amp ; // , , or the width of the patch they obtain at 480 yu , / / , , we get the same two groups with practically the same number of observers to the left of the minimum in all three cases .
If the observers are plotted against the width of the patch they obtain at 535 there is not the same well-defined minimum .
If in the case of an even number of patches the middle boundary is called the spectrum middle point , if in the case of an odd number of patches the centre of the middle patch is called the spectrum middle point , and if the observers are plotted against their spectrum middle points , then we get only one maximum .
Thus the essential difference between the two apparent groups lies in a greater power of colour discrimination at the points of the spectrum , where the colour changes most rapidly with wave-length .
I was under the impression that there were two groups forming before I had supervised 20 tests , and formed various hypotheses to explain it , e , g. , that the one group was confusing hue A Statistical Survey of Colour Vision .
with intensity , etc. , but to no purpose .
It will be interesting to know if the same division shows itself when tests are made with another apparatus .
It occurred to me that the one group might differ from the other in other ways , for example , in the amount of red it added to green to make yellow .
For testing this , I required a colour-mixing apparatus of the type used by Lord Rayleigh .
I was able to improvise one very satisfactorily in the following manner:\#151 ; The Cornu prism consists of two 30 ' prisms of right-handed and lefthanded quartz , with their optical axes at right angles to the plane of symmetry of the prism .
I separated the two halves of a Cornu prism and reversed them , i.e. , mounted them with the hypotenuse surfaces in contact .
When set at minimum deviation in this position , they gave two spectra with the red of the one falling on the green of the other , these spectra being practically plane polarised ( really very slightly elliptically polarised ) at right angles to one another .
The telescope of the spectrometer possessed a slit in its focal plane .
Between this slit and the eye was a nicol eyepiece , the lenses of which were removed .
An eye looking into the telescope saw the surface of the prism illuminated with whatever colour happened to fall upon the slit .
The instrument was set so that the green of the one spectrum and red of the other fell upon this slit ; thus , the colour of the prism could be changed from pure green through yellow to red by rotating the nicol .
To produce the comparison colour , a 60 ' crown glass prism , which had approximately the same deviation as the Cornu prism , was clamped on the top of the latter , so that it appeared illuminated in sodium yellow .
It was then itself not far from minimum deviation .
The red and green employed with this apparatus were respectively of wave-lengths 620 yy , and 560 yy .
I tested 15 of the observers with this apparatus , each making four settings with the nicol , taking them in turn from the top of each of the groups of fig. 3 , but found that their settings all agreed within the error of observation .
The only previous attempt which I have been able to find to apply statistical methods to colour vision is in a paper by Schuster.* He inclines to the belief expressed by Lord Rayleigh , that differences from normal vision do not seem to follow the lawT of error .
We have not , however , as much faith in the Gaussian law of error now as 25 years ago .
The experimental psychologists have introduced methods for measuring the different sensations , although they do not seem yet to have established many results , and the past 20 years have been noted for the development of statistical methods .
I think some of this activity might be profitably directed to the question of * " Experiments with Lord Rayleigh 's Colour Box , " 4 Roy .
Soc. Proc.,5 vol. 48 , p. 140 ( 1890 ) .
586 A Statistical Survey of Colour Vision .
the theory of colour vision , so as to settle the present deadlock ; the time is now past for the description of special cases of colour-blindness .
It might be advisable to institute a test in colour vision as part of the students ' course in certain of the larger physical laboratories in the country , and appoint a committee to deal with the results .
It is , of course , very likely that objection may be taken #to the nature of the test described in this paper .
One has to trust to the judgment of the observer as to when the difference of colour appears .
It would be better if the reading were to follow automatically ; for example , if upon a deep red background we were to make an orange-red letter appear , keeping the luminosity of the letter the same as that of the background , but gradually increasing the admixture of yellow .
The amount of yellow would be controlled by the supervisor , who would note the reading as soon as the observer recognised the letter .
Observers with good colour vision would require less yellow .
We would have the difficulty here that the luminosity of colours to different observers varies in a different way with wave-length , and it would not be possible to maintain the condition of constant luminosity for all observers .
I believe , however , that contrast due to difference of luminosity would be much less effective in causing a letter to appear than contrast due to difference of colour , and might be ignored .
I have attempted to construct a colour perception spectrometer depending on this principle .
It consisted of a spectrometer with the eyepiece removed , and a slit substituted .
Against the last face of the prism was placed a photographic plate , which had been fixed without being exposed , and so was covered with a uniform film of gelatine .
On this film a letter was painted in an aniline dye , the idea being that the dye would be quite transparent to certain colours of the spectrum , but that as the telescope was gradually rotated , a point would be reached at which the letter would stand out .
In practice the dye itself was transparent enough , but when it was applied the gelatine swelled out and acted like a lens , and so the letter remained visible when it ought to have disappeared .
Another attempt was made by photographing a letter with a Lumiere three-colour plate , but I have no time at present to follow the matter up .

