PROCEEDINGS OF THE ROYAL SOCIETY .
Section A.\#151 ; Mathematical and Physical Sciences .
The Fourth Colourless Sensation i the Three-Sensation Spectrum Curves when Measured on the Centre of the Retina .
By Sir W. de W. Abney , K.C.B. , F.R.S. ; With a Note by Prof. S. W. J. Smith , F.R.S. ( Received March 19 , 1917 .
) When the image of a small disc of a spectrum colour is received on the retina of a person possessing normal sense of colour , he observes changes in the hue varying with the portion of the retina on which it falls .
On the centre the colour is that of the projected beam , and , as it travels away from the centre , the hue becomes paler as the angular distance becomes greater , and , when a certain angular distance is reached ( varying with the intensity of the beam and with its wave-length ) , the colour vanishes , and the disc appears as white of fair brightness .
On this principle the boundaries of colour fields can be mapped .
It is evident that , as the disc of light travels across the retina , the composition of the colour changes , and that an increased proportion of " white " is felt as the white boundary of the colour field is approached , whilst the colour itself gradually diminishes .
The intensity of the white at the boundary of the colour field can be measured with some certainty , but it is more difficult to obtain a measure of the white which exists in the colour when it falls on the centre of the retina .
In two papers* communicated to the Royal Society , the writer has shown that the spectrum colours can be formed by admixture of three colour * 'Phil .
Trans. , ' A , vol. 193 , p. 259 ( 1900 ) , and 'Phil .
Trans. , ' A , vol. 205 , p. 333(1906 ) .
VOL. XCIV.\#151 ; A. B 2 Sir W. de W. Abney .
Fourth Colourless sensations , but no place was found here for an admixture of the " white " sensation , as the colours were received on the centre of the retina .
Had they been received at a place outside the centre on part of the periphery ) , it is probable that the three colour sensations would have required a percentage of " white " to make a complete match .
The white sensation we are considering is not the same as is found in the spectrum colours .
It is white which cannot be analysed by refraction or diffraction , but exists as an entity by itself .
It thus differs from the white which exists in the ordinary spectrum colours , which is found to be due to the admixture of the three sensations of colour in fixed proportions , the proportions depending on the quality of the light forming the spectrum .
The " white " we are considering appears always to be of the same degree of " colourlessness " whatever may be the light used to form the spectrum .
For the sake of avoiding any ambiguity , it is proposed to define this " white " as a fourth sensation\#151 ; or the fundamental sensation of light , which is a definition that the author has employed in more than one communication which he has sent to the Eoyal Society .
In a recent paper by Dr. W. Watson and the author , it has been pointed out that , in the work by eminent foreign workers , it has been supposed that in the sensations of colour on the retina the cones are the principal agents which are active , whilst for the sensation of the " white " the rods are the sensation agents .
In that paper a slight discussion is made on these proposals .
It may be remarked that , whilst there are both in the rods and cones distinctive features , there are some processes which it would be difficult to class as possessing the features of either cones or rods .
In the same paper retinas are classified under two headings : one class , those in which at the fovea cones alone existed , as , when extinguishing a small disc of a spectrum colour , the light and colour both " went out ' at the same instant ; in the other class , the retinas on which the same coloured disc fell , when received on the fovea , first lost the sensation of colour , leaving a '''white " which was extinguished afterwards by a still further reduction of the intensity of the colour disc .
It is too early from our experience to say that in this last class both cones and rods were present in the fovea .
It may be that the processes existing there were hermaphrodite , being neither pure cones nor pure rods .
This is mentioned here because , in the subjoined account of experiments undertaken to find what very small amount of the fourth sensation\#151 ; if any\#151 ; was present when the colour sensations of the spectrum of ordinary intensity were measured at the centre of the retina , the observers both had retinas of the second class .
Sensation in the Three-Sensation Spectrum Curves .
3 In an hermaphrodite cone and rod it may be that some species are most sensitive to colour , whilst in others the greater sensitiveness may be for the " white .
" At the beginning of 1898 I carried out an investigation as to the form of the luminosity curves of the spectrum of very feeble arc electric light .
The measures made by myself and my assistant , Mr. Bradfield , were put on one side owing to press of other work .
It seems that these curves would show the point I wanted to obtain , as stated above , by the amount of action on the cones when the light was a disc falling on the centre of the retina .
The method adopted was as follows , the colour-patch apparatus being that which I ordinarily use :\#151 ; The various spectrum colours came through a movable slit at the focus , and a patch of the colour was formed by a lens placed in the coloured beam .
The patch of colour was caused to fall on a plate of glass ( finely ground on both surfaces ) by reflection from M ' , which occupied a position shown in the box BBBB\#151 ; a partition DD was pierced by a hole some inch in diameter .
In front of the hole discs could be placed , in which clean-cut holes of and ^ inch diameter were bored .
An aperture C in the side of the box was made , which allowed the white light reflected from the front surface of the first prism of the apparatus to fall on M. That portion of the light reflected from M which passed through the centre of the disc placed at DD passed through to the far end of the blackened box .
An eye placed at an eye-hole E at the end of the box saw the white rim of the disc illuminated , and in its centre a small disc of coloured light .
In the spectrum ray an annulus A was placed , the centre of rotation being accurately in the continuation of the slit through which the coloured ray passed .
Every 25 ' of the annulus diminished to half the initial intensity when the D light B B Fie I Sir W. de W. Abney .
Fourth Colourless was passing through the slit and the annulus was at its zero point .
The brightness of yellow beam was found to be 1-1 of a Heffner lamp at a foot distance .
To diminish the intensity of the white , a second annulus A ' was placed at the point where the reflected beam crossed and formed a line .
With this annulus every 30 ' of angle cut off half the white .
The white light passing through the annulus at its zero point was measured , and found to equal 0'2 Heffner lamp-light at 1 foot distance from a screen , so that , when the light passed through annulus at 60 ' , the intensity of the white light was reduced to 005 of the Heffner lamp-light .
For reasons which need not be given , it was determined to compare the luminosity of the colours with the white when , for each series of readings , the white light was kept constant .
It was determined to take the following intensities of white light .
The annulus was set:\#151 ; At 60 ' and the white transmitted was 0'05 H , 120 ' " " " 0-0125 H , 180 ' " " " 0-0031 H , 240 ' " " " 0-0008 H , 300 ' " " " 0-0002 H. Three sets of measures were made by my assistant and myself , and a mean value for ^-ineh disc of colour obtained by using the annulus in the coloured beam to secure equality of luminosity .
These luminosities were readily found .
The white surround to the colour disc seemed to aid the equalisation of luminosities , the small oscillation of the angle of the annulus in the colour beam showing very accordant readings .
The following Table gives the mean readings of the annulus taken by my assistant throughout the spectrum and at the four intensities of white light 240 ' , 180 ' , 120 ' , and 60 ' .
( The readings at 300 ' were practically the same as those for 240 ' , so have been omitted .
) The first column is the empiric scale of the spectrum , and the second is that scale reduced to standard scale , which has been used throughout all previous communications .
Table I gives the original readings .
The values of these intensities or luminosities are calculated from a Table which shows each 25 ' of the annulus as diminishing the light by J. For the sake of simplicity , the values shown in Table II are derived by deducting the values obtained from the highest reading given in the several columns , making the luminosity curve to show a maximum of 100 .
Table III shows the intensities when the whole numbers of the S.S.N. are employed .
The areas of these curves are taken when the scale of S.S.N. is applied .
Sensation in the Three-Sensation Spectrum Curves .
5 Table I.\#151 ; ( A ) Annulus Headings in Degrees .
Temporary scale of spectrum .
Standard scale ( S.S.N ) .
A ' Annulus in fixed positions .
240 ' .
180 ' .
120 ' .
60 ' .
72 O 60'2 80 50 23 74 58 123 95 70 30 76 55 -8 166 127 100 55 78 53 6 199 155 123 82 80 51 -4 225 178 143 103 82 49 -2 250 203 160 115 84 47 275 228 172 117 86 44-8 296 244 185 115 88 42-6 312 257 190 105 90 40 -4 318 262 190 94 92 38-2 314 252 182 80 94 36 -0 305 245 167 65 96 33-8 292 231 252 45 98 31 6 275 213 141 28 100 29 -4 260 197 118 5 110 27-2 250 182 100 \#151 ; Table Y shows the intensity values of each of the columns when the areas of the curves are made equal to that , of the spectrum of normal intensity .
When plotting Table Y , it appeared that it would be more complete if the curve for 90 ' could be added .
The original curves of annulus readings were brought into requisition , and the closely approximate annulus curve for 90 ' deduced , and from it intensity curve with a maximum of 100 calculated as shown in Table Y. This curve was reduced so as to be of the same approximate area as those shown for the 0 ' to 240 ' curves , and is also shown in Table Y. The results of the tabulated " equal area " curves are shown graphically in fig. 2 .
The diagram shows that all the curves of equal area very approximately intersect at S.S.N. 44 ( and no doubt if the measures had been more accurate the intersections would have been still more closely at the same point ) .
There must be some reason for such a coincidence in intersection.* * In the note which follows on at the end of the paper , the referee gives a mathematical reason for this intersection on a certain assumption .
The author had arrived at a similar result by a different plan , but hesitated to give it as it seems that in the " A " curve some small corrections have to be made when the intensity is altered , and also that when the radiation is reduced as it was in the experiments the white comparison is not equally reduced with the radiation .
This difference between radiation and luminosity is not known exactly , but the method given in the referee 's note makes it practical to find the difference in the reduction in the white with reduction of radiation.\#151 ; May 23 , 1917 .
Sir W. de W. Abney .
The Fourth Colourless Table II .
Table IIL Original scale .
S.S.N 240 ' .
180 ' .
120 ' .
60 ' .
S.S.N. 240 ' .
180 ' .
120 ' .
60 ' .
Normal spectrum .
72 60 -2 o-i 0*3 0-9 60 0 3 2 74 58 0-6 1 0 3-5 9-0 58 \#151 ; 1 2 14 15 76 55 -8 1 *3 2*4 8 0 20-5 56 2 4 8-5 39-1 32 78 53 -6 2 -2 5-0 15-6 38 -0 54 4 8 15 -5 57 61 80 51 -4 8 -0 7-0 26 68 -5 52 6-1 12 25 74 75 82 49-2 15 -5 19-5 41 96 -0 50 13 *3 19 35 90 91 84 47 31 39 58 100 -o 48 23 32 47 100 100 86 44 -8 58 61 *5 63 96 -0 46 42 55 61 98 94 88 42 -6 85 87 95 72 44 73 75 80 88 75 90 40 -4 100 100 95 53 42 92 96 98 74 54 92 38 -2 90 76 76 36 40 99 -5 98 96 58 37 94 36 70 62 50-5 23 -5 38 89 80 75 39 25 96 33 *8 50 42 -5 33-5 14 36 70 60 52 27 16 98 31 -6 31 26 24 8-5 34 54 * 44 35 18 9 100 29 -4 20 17 13 4*4 32 39 32 24 6 4 110 27 -2 15-5 11 4-7 30 28 24 16 2 28 19 17 10 1 26 13 12 7 640 680 700 783 665 Table IV .
Table V.\#151 ; Curves of Equal Areas Calculated from Table III and Table IV .
Deduced luminosity curve for 90 ' .
S.S.N. 240 ' .
180 ' .
120 ' .
90 ' .
60* .
0 ' .
60 2 60 1 -7 2 2 58 8 58 \#151 ; 1 2 7-1 13 15 56 16 56 2-0 4 7 -5 13-5 29 32 54 26 54 4-0 8 14-5 22 49 61 52 42 52 6-1 12 24 35-5 64 75 50 64 50 13-3 15 34 54 74 90 48 92 48 23 -4 35 45-5 77 87 100 46 96 46 42 -5 55 59 79 85 94 44 91 44 74 75 77 76 -5 76 -5 75 42 86 42 94 96 99 -5 72-5 69 54 40 ' 76 40 102 98 92 -5 64 50-5 37 38 60 38 90 *5 80 72-5 50 *8 33 25 36 42 36 71 60 50 -5 35 -5 23 -5 16 34 29 34 54 -7 44 34 24-5 15 -5 9 32 19 32 39 -5 32 23 16 10 4 30 12 30 28 -5 24 15 -5 10 -3 5 1 28 8 28 19 -5 17 9-6 5 1 26 5 26 13-5 12 6-7 2 Sensation in the Three-Sensation Spectrum Curves .
The theory that the cones in the retina are sensitive to the coloured rays whilst the rods similarly respond to the colourless rays is a partial answer to these results .
But there are difficulties in accepting this theory as altogether sufficient .
In a large portion of the spectrum from near D to the extreme 30 40 50 [ if ?
* I wf'hH'p 'Qgj\amp ; \#163 ; 03/ 0125/ 2 / / */ blue the sensation curves , as given in the ' Philosophical Transactions/ show that the overlapping of the red , green , and blue sensations are blended in various degrees , and the spectrum colour is a compound of the one or two pure sensations and white light .
Thus , at one spectrum green , the colour is produced by a mixture of 60 per cent , of white light and the pure green Sir W. de W. Abney .
Fourth Colourless .
sensation .
Experiment shows that by adding white to the spectrum colours , impure as they are in this region of the spectrum , the colour may be indistinguishable in the mixture and only white is perceived by the eye .
If the intensity of the spectrum is diminished , a point , varying for each wavelength , is reached in which the colours disappear , and here we have the cones in the dark-adapted retina showing no colour. .
The fact that the boundaries of colour fields are white , shows that the fourth sensation \#171 ; white " has the same effect as the mixed white of the three-colour sensation , in drowning a colour .
( In the intensities of the white comparison light and of the spectrum used in the above results there are only three which showed any signs of colours , the ordinary bright spectrum of 0 ' through the annulus , that of 60 ' , and again , partially , that of 90 ' .
In the white light which forms the spectrum coming through 180 ' and 240 ' , * the whole of the rays appear as white , or , perhaps more accurately , may be described as greyish .
Yet , at the most refrangible part of the spectrum , the cones must be receiving the same light as when the colour is still visible .
It thus appears if the cones are sensitive to colour they are also sensitive to the same white rays as the rods , but in a minor degree , whilst the rods are probably unaffected by the colour rays .
) In 'Phil .
Trans. , ' vol. 7 , 1897 , " On the Sensitiveness of the Retina to Light and Colour , " a large section is devoted to colour fields , in which measures of the extinction of colour by white light on the periphery are made .
The colours operated with are not feeble , but of a measurable brightness .
The loss of colour at a good many degrees from the axis of the retina is very sharply defined as a rule , and it is a matter for consideration whether the boundaries of the colour fields are fixed by extinction of colours by the white light formed by mixtures of the colour sensations or principally the sensitiveness of the rods to what I have called the fourth colour-sensation , which , apparently , as recently remarked , may be equally effective as the white of the mixed colour sensations extinguishes .
In any case , there is an enormous difference in the sensitiveness of the retina at the periphery to the extinction of colour to that formed at the central part .
We may take it that , when the spectrum is formed of white light of 0-0008 candle at 1 foot distance from the screen , the colourless light will be so small that it is negligible in the three-sensation curves .
In another communication it is hoped that the colour field boundaries of white may throw more light on the action of the rods and cones , and of the * When making the observations the intensity at which the colour disappeared from the rays was recorded .
See also the Author 's book on the 'Trichromatic Theory ( Longmans , 1914 ) .
Sensation in the Three-Sensation Spectrum Curves .
9 S.S.N. I. II .
III .
IV .
Luminosity of spectrum with maximum of 100 formed by white light Absolute luminosity of spectrum in II compared with I. Percentage of III compared with I. Of 1 candle at 1 foot distant from screen .
Of 0 *0008 candle at 1 foot distant from screen .
56 32 2 0 -0016 0-00005 52 75 6-1 0 -0048 0 -000017 48 100 23 0 -0184 0-00018 44 75 75 0 -0600 0-0008 40 37 99-5 0 -0796 0 -0021 36 18 70 0 -0560 0-003 32 4 39 0 -0312 0-008 28 1 19 0 -0152 0-015 26 0-6 13 0 -0104 0-017 hermaphrodite processes which more or less replace them in the retina .
So far , we have from the present enquiry gathered that the fourth sensation is very small in the spectrum colours of an ordinary bright spectrum when falling on the centre of the retina\#151 ; so small , indeed , that the sensations are 100,000 times more intense than any cone effect , and so do not appreciably affect the sensation curves which are given in the ' Philosophical Transactions .
' This paper is sent in to the Royal Society after some delay .
Its communication was delayed by the fact that Dr. ( now Lieut.-Colonel ) W. Watson was commandeered as scientific adviser at the Front .
Up to the time of his leaving the country , the author and Colonel Watson were in the thick of some interesting researches .
Amongst others was a repetition of the experiments , with slight alterations , which are given above .
In this repetition , instead of leaving the reflected white light used in the apparatus unaltered and altering the intensity of the colour to match the luminosity , the intensity of the spectrum colours in each set of observations remained unaltered , and the matches of luminosity were made by alteration in the intensity of the white .
The colour-patch apparatus was used in its usual way , and without any external reflection of beams , as shown in the above observations .
We hope that , when the war is over , these new results will be published , showing that similar results were obtained to those above , the difference being that the intersection of equal area curves took place slightly nearer the most refrangible part of the spectrum .
Again , we had made a partial examination of some congenital night-blind cases .
The remarks as to the night-blind apply to the central part of the retina .
10 Sir W. de W. Abney .
Fourth Colourless One point of interest connected with these examinations may be mentioned .
We found that , without any restriction to the fovea in diminishing the intensity of a spectrum beam , the colour and light were lost in them at the same time , the colour and light disappearing at the same intensity as the colour disappears in the normal eye .
This is similar to the effect of receiving the spot of light on the fovea of Class I retina , which is supposed to be coneless .
If this he true , then , the night-blind , according to the foreign theory , possess only cones in the whole retina .
We propose at some future date to give some account of what happens at the periphery of the retina of these congenital night-blind cases .
Appendix .
S.S.N. X. S.S.N. X. 60 6728 40 5270 58 6521 38 5172 56 6330 36 5085 54 6152 34 5002 52 5996 32 4924 50 5850 30 4848 48 5720 28 4778 46 5596 26 4707 44 5481 24 4639 42 5372 22 4578 Note by Prof. S. W. J. Smith , F.R.S. The important result illustrated by fig. 2 that the equal area curves have , approximately , a common ordinate at the scale reading = 44 admits of a simple interpretation , which may be formulated in the following manner .
Suppose that the dotted curve ABREF is obtained by plotting the sum of the ordinates ( at each abscissa ) of two curves ABC and DEF .
If these curves ( ABC and DEF ) overlap it is obvious that , whatever the respective areas included between them and AF may be , an abscissa S can be found which satisfies the condition SP _ SQ = SP + SQ _ SR area ABC area DEF area ABC + area DEF area ABREF ' Suppose now that a second curve AB'R'E'F is obtained by plotting the sums of the ordinates of two curves AIVC and DE'F .
Let AB'C be such that each ordinate is a times the corresponding ordinate of ABC , and let DE'F be such Sensation in the Three-Sensation Spectrum Curves .
11 i that each ordinate is / 3 times the corresponding ordinate of DEF .
Then we shall obviously have SP ' _ SP , SQ ' _ SQ areaAB'C area ABC area DE'F area DEF whence SF + SQ ' = SP + SQ area AB'C + area DE'F area ABC + area DEF ' SR ' _ ' SR area AB'R'E'F area ABREF* Hence if we plot the curves to such scales that the areas AB'R'E'F and ABREF are equal we shall at the same time make SR ' = SR .
In other words , the curves will have a common ordinate at S. This result suggests that the luminosity curves of fig. 2 arise from the fact that , at each scale reading s , the sum of two luminosities xa and xi is recorded .
It suggests also that the curves representing the variations of xa and * with s remain always of the same type so long as the measurements are made at constant total illumination .
When this constant illumination is decreased continuously it would appear that one of the luminosities , say xa , becomes continuously less important than the other , say xj , .
Finally , the luminosity curve practically coincides with that of the " B sensation " to which , we may suppose , is due .
Conversely , when the illumination is as bright as possible , Xf , becomes negligible in comparison with xa and the luminosity curve exhibits the typical variation , with s , of the luminosity xa due to the " A sensation .
" If this interpretation be correct ( and it seems to be what is intended ) the B sensation would be what the Author calls the " fourth " ( colourless ) 12 The Three-Sensation Spectrum Curves .
sensation , while the A sensation would be , presumably , the sum of the other three .
Further it should be possible to deduce , from the two extreme curves , the forms of all observable intermediate curves .
Taking the curves of fig. 2 marked 0 ' and 240 ' , respectively , to represent sufficiently nearly the forms of the curves of the A and B sensations , the dotted curves shown in fig. 4 ( attached ) can be calculated .
30 40 50 !
/ / / 4 / / t r i i / \#174 ; : : !
io ' \i \ .
\ \ v 7 / / r v / / o 7 , / if They lepresent results obtainable by combining the A and B sensations in different relative proportions .
The assumed ratios of the areas of the A and B curves are , respectively , 1:0 , 9:1 , 2:1 , 1:1 , 1 :2 , 1:9 and 0:1 .
I Periodic Disturbance of Level .
It will be seen that the calculated results are not very dissimilar from those obtained experimentally and give reasonable support to the interpretation put forward .
These curves would serve also to show , more exactly than is shown in the text , the relative insignificance of the fourth-sensation-contribution to luminosity curves of ordinary brightness .
They form also a more concrete presentment of the argument in favour of the existence of this fourth sensation .
Periodic Disturbance of Level arising from the Load of Neighbouring Oceanic Tides .
By K. Terazawa .
( Communicated by Sir Joseph Larmor , F.R.S. Received April 17 , 1916 .
) ( Abstract .
) In Hecker 's observations on the lunar deflection of gravity , the force apparently acting on the pendulum at Potsdam is a larger fraction of the moon 's direct attraction when it acts towards east or west than when it acts towards north or south .
A similar result has been found by Michelson in his observations of the lunar perturbation of water-level at Chicago .
A calculation is here made to ascertain to what extent the tilting of the ground caused by the excess pressure of the tide in the North Atlantic is important for the explanation of this geodynamical discrepancy .
Replacing the North Atlantic by a circular basin of radius 2000 kilom .
, taking the position of Chicago to be 1000 kilom .
from the coast , and the rigidity of the earth to be 6 x 1011 C.G-.S .
, it is found that the attraction effect of a uniform tide per metre of height is about 0*0024 " , while its tilting effect is as much as 00069 " , the maximum of the direct lunar attraction being 0'017 " .
If the surface of tide is ellipsoidal , shelving towards the coast , nearly the same result is reached for the same mean tidal height .
Note by Communicator .
[ July 16.\#151 ; The following proposition , which is implied in the paper , deserves explicit statement : it is readily proved for a load localised at a point , * and follows generally by superposition .
A load is distributed anyhow on the earth 's surface , over a region not too large to be considered as plane ; for * Cf .
Love 's ' Theory of Elasticity , ' para .
135 .

