326 Colour-Blindness and the Trichromatic Theory of Colour Vision .
Part IV.\#151 ; Incomplete Colour-Blindness .
By Sir William Abney , K.C.B. , D.Sc .
, D.C.L. , F.R.S. ( Received June 24 , \#151 ; Read June 27 , 1912 .
) A Simple Test.\#151 ; In Parts II and III of this series of papers I have indicated several methods of giving a quantitative value to the amount of green or red sensation which exists in the incomplete green- or red-blind eye as compared with the normal eye .
I need not summarise what has already appeared at recent dates in the ' Proceedings ' of the Royal Society , but I will at once explain a very simple test which gives quantitative measures of the sensations in incomplete colour-blindness .
In my last paper* of the series a table is given of the luminosity of the arc spectrum and the percentage composition of the tabulated rays of the spectrum , and from the same table the amount of white which exists in these several rays can readily be calculated .
Owing to change in hue which white causes in a mixture of a pure red with a green ray I pointed out that the match of the I ) line by two such rays is open to error , not considerable perhaps , but still appreciable .
In my communication " On the Change of Hue of Spectrum Colours by Dilution with White Light , it was showTn that the change in hue at the point where the two curves of red and green sensations cut when their areas are made equal is nil .
For reasons given later it may also be mentioned that a ray at this point when mixed with a blue where the same red and green sensation curves cut , will , with proper adjustment of the widths of the slits , match the white of the reflected beam of the arc light .
The same happens for any other source of light if the curves of equal areas are calculated for such a light .
The position of the point of intersection in the yellow part of the spectrum varies with the kind of light forming the spectrum .
Looking at the table it will be seen that from S.S.N. 50 to the extreme red there is no blue sensation present in measurable quantities .
Suppose that in a beam of the arc spectrum a cell containing a saturated solution of potassium chromate of ( say ) f inch in thickness is placed ; the beam becomes yellow with very little blue present .
If a slit is caused to traverse the spectrum in the colour patch apparatus a position will be found for it which exactly matches the hue of the white light which passes through * ' Roy .
Soc. Proc. , ' A , voi .
86 , p. 44 .
t ' Roy .
Soc. Proc. , ' 1909 , A , vol. 83 , p. 120 .
Colour-Blindness and Trichromatic Theory of Colour Vision .
327 the chromate solution , and will be at S.S.N. 49'6 or \ 5828 to the normal eye .
This , like other rays , contains a fixed ratio of green to red sensations , but no blue which is measurable .
Making the red sensation unity it will be found that the green sensation present is 0*385 .
A table of the ratio of red to green sensation ( making red unity ) for the standard spectrum scale is annexed , as is also a table of wave-lengths with the like ratios:\#151 ; S.S.N. G.S. X. G.S. ' Gr .
S. corrected from diagram .
56 0 047 6300 0 -050 0-050 54 0-105 6200 0*080 0-080 52 0 -187 1 6100 0 *127 0-127 50 0*338 6000 0 -185 0-185 48 0-475 5900 0 -280 0*280 46 0 -603 5800 0-390 0-385 44 0-717 5700 0-500 0-490 42 0-830 5600 0-600 0-595 40 0-934 5500 0-700 0 -700 38 1 -15 5400 0 *805 0-805 36 1 *16 1 5300 0-910 0-910 34 1 -26 5200 1 -015 1 -015 32 1 -33 5100 1 *135 1 *120 30 1 -29 5000 1 -260 1 -225 28 26 1 -14 0-82 ; 4900 1 -340 1 *330 From these tables , a full sized diagram ( fig. 1 ) can be drawn , from which the ratios of red to green can be read off for any scale number or wavelength.* If , instead of a normal eye making a match , an incompletely green blind has to make one , the slit would have to be moved from S.S.N. 49*5 towards the red .
When he considers the match to be correct , the scale number of the ray is read off and referred to the diagram .
Supposing that the mean of several readings was 52 , the amount of green in that ray ( to the normal eye ) would be 0*187 .
By dividing this number by 0*385 ( the match for the normal eye ) , we get very closely 0*5 , and this would be the amount of green sensation ( compared with the normal ) that this incomplete green-blind possesses .
Again , if we have the reading of another incompletely colourblind at ( say ) S.S.HST .
46 , we know at once he is incompletely red-blind ; as that S.S.N. contains 0*603 green sensation , the amount of red sensation he possesses is obtained by dividing 0*385 by 0*603 , which makes him to have 0'64 R.S. as compared with unity R.S. in the normal eye .
It will be seen from an inspection of the Diagram I that the maximum * The table in ' Roy .
Soc. Proc. , ' vol. 64 , p. 44 , gives the wave-lengths of the different standard scale numbers ( S.S.N. 's ) .
328 Sir W. de W. Abney .
Colour-Blindness and the [ June 24 , green sensation is near S.S.N. 30 , where it is about 1*32 , when the red sensation is unity .
At S.S.N. 49'6 , which matches the chromate solution , the amount of green sensation is 0-385 .
It is evident that the maximum red sensation of a red-blind that can be diagnosed quantitatively by the chromate is 0-385/ 1-32 , or an R.S. of 0*29 .
In order to ascertain quantitatively a Fig. 1.\#151 ; Showing ratio of Green Sensation to Red Sensation .
lower R.S. ( say 0"2 R.S. ) , the place of matching must be nearer the red .
The passage of the white light through a cell of bichromate of potash solution , which is at S.S.N. 51*1 , will enable such a small R.S. factor to be determined .
By matching the colour of a scarlet glass , there will be a still longer range .
On the other hand , for green-blindness of an extreme type , the chromate cell will answer , but a more open scale will be given by passing the white light through the chromate cell and a very pale solution of copper sulphate in a second cell .
The test described is so delicate that the slightest deviation from normal colour-sensation can be determined quantitatively with great accuracy to almost the second place of decimals , and it has the advantage of great simplicity .
Trichromatic Theory of Colour 1912 .
] I will give one example of the accuracy and delicacy of the test .
A partially green blind was tested by matching the chromate colour .
The normal eye made a match at S.S.N " .
49*6 , and the green-blind made it at S.S.N. 41*5 .
From the diagram the first named made a match which had 0*385 green , whilst the second match contained to the normal eye 0*855 green ( red sensation unity in both cases ) .
This gave the factor 0*45 for the green sensation in the green-blind .
The same person was tested by the luminosity method described in my previous paper , and from his observations the factor was 0*45 .
From his colour equation to match white his factor was 0*37 .
The factor from the mean of the three was thus 0*42 .
But I have no doubt that the factor obtained by the new test was nearer the truth .
In this method only one slit has to be used and is independent of the luminosity .
( I should say that before making the test the position of the maxima of the colour sensations had been determined as being the same as those in the normal curves which I have published in the ' Phil. Trans. ' ) In the match with the D light two slits are employed to match a pure colour .
In the new test only one slit is used to match a mixed colour .
When using this test it is well that the luminosities of chromate " white " should be the same as that of ray coming through the slit .
My own practice is to reduce the " white " luminosity by means of an annulus or sectors , and get a final equality by opening or closing the slit .
Four determinations are usually sufficient , two by reaching the match from the red side and another two from the blue side of a first approximate match .
A mean of the four readings is taken as the position of the correct match , though not unfrequently all four will be exactly the same .
It will be advisable now to show how the amount of any shift or displacement in the position of the maxima of sensation curves can be determined .
At the beginning of this communication it was pointed out that when a slit was placed at the ray where the two curves of equal area or stimulation cut , the addition of pure blue enabled a match to be made with the white of the light which formed the spectrum .
Let a a , bb ( fig. 2 ) , be two portions of the curves of green and red sensations respectively , cutting at O , and with an ordinate OC ; then a slit placed at C in the spectrum will allow a ray to pass which , with blue , will make white.* If the green curve were ( say ) moved bodily to the left the curves would no longer cut at O but at O ' , so that the slit would have to be moved * The same holds for the colour-blind , since the curves under consideration are " equal area " curves .
The white matched is the " colour-blind white .
" 330 Colour-Blindness and Trichromatic Theory of Colour Vision from C to C ' before a white would be produced .
Similarly , if the red curve were moved bodily the intersection of the curves would not be at O , and C would have to change position .
By obtaining observations it is Fig. 2.\#151 ; Scale of Prismatic Spectrum ( S.S.N. ) .
easy to find , if the curves remain in the position of the normal curves , whether there is any colour-blindness or not .
Should such an alteration in the place of intersection of the curves be found when making the new test described above , the difference in position of the normal intersection and that of the shift should be noted and added to or subtracted from the normal position of the match with the chromate .
If there be colourblindness the amount should be determined from this corrected position of the normal match .
The amount of any shift in either of the curves would also be found , when testing by the luminosity method , by taking the mean of factors obtained on the red side of the maximum of the normal curve and of those on the blue side of the maximum .
The mean of the two sets will give the real factor for the sensation which is in defect .
The method described for ascertaining the amount of shift , if there ever is any , is a reliable one .

