application to pigment colours. by means of retinal persistence and is consequently limited in its

Transcription

1 THE PHOTOMETRY OF COLOURED PAPER. By W. H. R. RIVERS, M.D. (From the Physiological Laboratory, Cambridge.) Section I. Band Photometry.,, II. Flicker Photometry., III. Pupil Photometry.,, IV. Comparison of the three methods.,v. Comparison with other methods. I. "Band" Photometry. THE first method of colour photometry brought forward in this paper is based on a phenomenon which was first described by Jastrow'. If a disc made up of coloured sectors is rotated so as to produce mixture in the ordinary way and a fine rod passed rapidly in front of the rotating disc, the rod will be seen to have followino it a series -of coloured bands corresponding to the colours of the sectors2. If the sectors are coloured and grey respectively, similar bands occur, the grey banids being coloured by contrast. It is possible however to find for any given colour a grey which gives no bands, whatever the relative breadth of the sectors (a factor of importance in producing the phenomenon). The problem investigated was to find the grey paper which gave no bands with each of a series of coloured papers, and to compare the results so obtained with those derived from other methods of heterochromic photometry. The method can only be applied when colours are mixed by means of retinal persistence and is consequently limited in its application to pigment colours. The colours used were those of the ten papers sold by Rothe. They have been used by Hering, Schenck and others in similar work and have therefore the great advantage that comparison is possible with the results of other observers. The grey papers used were also American Journ. of P8ychology, nv. p The "band" method was demonstrated before the Physiological Society on Nov. 14th, 1896.

2 138 W. H. S. RIVERS. obtained from Rothe. They fomnied a series of 22, and the figures given in the first three tables refer to these papers numbered, no. 1 being the lightest, and no. 22 the darkest. Unfortunately they did not form a uniformly graduated series, and it was not possible with these papers to find a grey which gave no bands with several of the coloured papers, and one had to be content with the grey with which the bands were least distinct. All the observations were made by morning sunlight with the eyes adapted to brightness. The distinctness of the bands varies to some extent with the brightness of the day, but the results of different days were faund to agree with one another, and the limits of brightness of illumination did not seem to be far enough apart to lead to any appreciable influence from the Purkinje phenomenon. The observations recorded by this and by the following methods were made by Mr 0. F. F. Griinbaum and myself. The results of the "band" method are shown in Table I, the figures referring to the numbers of the papers. TABLE I. Band method. Yellow Blue Red Orange Yellow green Green green Blue Indigo Violet Purple Rivers Grunbaum One point of interest may be mentioned since it gives evidence of the accuracy of the method. The earlier observations were made with a larger series of 30 grey papers, and it was found that several colours gave no bands or equally indistinct bands with two or more of this series, thus Griinbaum matched red with 18, 19 and 20 of the original series, orange with 14 and 15. On determining their white. values it was found that several of the original series were of equal brightness, including 14 and 15, while 18, 19 and 20 differed from one another very slightly. With feeble illumination the bands are very indistinct; and it is so difficult to make satisfactory observations that it does not seem possible to employ this method for the investigation of Purkinje's phenomenon. II. Flicker Photometry. The results obtained by the "band" method were found to differ considerably from those which Schenck obtained by means of flicker 1. Schenck modified Rood's method by dividing one half of a- rotating 1 Pfluger's Arch. Lxrv

3 COLOUR PHOTOMETRY. 139 disc into six sectors, each of which was again divided into a black and white part. The proportions of the black and white were so arranged that the disc on rotating -would show a series of greys diminishing to brightness from centre to periphery of the'disc. The other half of the disc was covered with the coloured paper of which the brightness was to be determined. The disc was observed through an aperture, and the zone of the disc found in which flicker disappeared. In some preliminary experiments Sch en ck found considerable discrepancy between the values of grey papers as determined by this method and their actual brightness. It seems probable that the differences depend upon the influence of the factor recently described by Sherrington' and Grunbaum2, viz. the influence of contrast when black and white are simultaneously presented to the eye, and in favour of this suggestion is the fact that Schenck found the discrepancy considerably less when the disc was observed through a smaller aperture. For the purposes of the investigation it was necessary to determine the point of cessation of flicker for the same coloured and grey discs as were used in the "band" method. In the method employed a disc with 4 open sectors was rotated by means of an electromotor before the coloured or grey paper, which was observed through a tube having a constant circular aperture of 1-7 cm. diameter at 19 cm. from the eye. In a preliminary series the number of interruptions per second at which flicker disappeared was determined for each coloured paper, and the grey paper then found with which flicker disappeared at the same rate. The observations were made by daylight so as to be comparable with those of the "band" metbod, and it was found that the variations in the intensity of the light within a few minutes were sufficient to make this method impracticable. A method was then adopted which gave very satisfactory results. The coloured and grey discs were exposed simultaneously interleaved as when being rotated, so that the line between them divided the circular field seen through the aperture into two equal halves, one coloured and the other grey. The sectors were then rotated at such a rate that there was a juist perceptible flicker in the coloured half of the field and various grey papers tried in the other half till one was found which showed an amount of flicker equal to that in the coloured paper. The rotating sectors were covered in front by a screen. This was rendered necessary by the fact that without it on rotation of the sectors there was an alternation of light and shadow on 1 This Journal, xxi. p.' This Journal, xxi. p

4 140 W. H. R. RIVERS. the observed disc which might easily be nistaken for flicker. It had the further advantage that the observer only saw the small circular field and did not know which grey was being used, so that the results obtained could not be in any way influenced by previous knowledge derived from the "band" method. The results obtained are given in Table II. and show a very close agreement between the two observers and also with those of the "band" method. The difference between the two methods is most marked in the red and will be referred to again later. TABLE II. Yellow Flicker method. Red Orange Yellow green Green green Blue Indigo Violet Purple Rivers Griinbaum III. Another method employed for comparison with the " band " method was that described by Gorham' and Sachs2. The principle of this method is the dependence of the degree of contraction of the pupil on the intensity of the light stimulating the eye. Sachs suggested it as a method of.heterochromic photometry but did not carry out any investigations from this point of view, his papers being devoted to the relation of the pupil contraction or "motor value" of a light to Hering's "white value" of the light. In this method the size of the pupil is observed entoptically, and for this purpose I used Gorham's pupillometer. Three variations of the method were tried. In the first the pupillometer before one eye was directed towards the coloured paper for which various grey papers were substituted; a grey lighter than the colour would cause contraction and consequent separation of the diffusion circles seen in the pupillometer; a darker grey wouild cause dilatation and overlapping of the circles; while a grey of the same brightness produced no perceptible change. By this method a very small amount of light reaches the eye and the changes were consequently very slight and the results obtained indefinite, but so far as investigated they showed a fair agreement with the other two variations of the method. Blue Pupil Photometry. In the second method, which was that employed by Sachs, advantage is taken of the consensual pupil reaction. One eye (left) 1 Proc. Roy. Soc. 1884, xxxvn. p Pflilge's Arch. LII ; and Arch.f. Ophth. xxxix. Abt. 3,

5 COLOUR PHOTOMETRY. looked through a tube at the coloured surface, while the- right eye with the pupillometer was directed towards a uniform grey surface. When the field of the left eye changed in brightness, the pupil of the right eye was observed to contract or dilate, and as in the first method various grey papers were placed before the coloured paper till one was found which caused neither contraction nor dilatation. The third method was a combination of the first and second, both the left eye through the tube and the right with the pupillometer being directed towards the coloured surface, so that advantage was taken both of the consensual and the direct reactions. The chief practical difficulty of these methods is the contraction of the pupil on accommodation. There seems to be a tendency for a momentary accommodation change to take place when any change is made in the visual field and this tendency has to be overcome, as otherwise contraction of the pupil may occur with change from light to dark as well as with change from dark to light. Secondly, if there is- accurate accommodation for the surface this will change on going from one colour to another, or from a coloured to a grey surface. To overcome this difficulty the coloured surface was placed beyond the far-point (both observers being myopic). It was found that the two observers differed as to the method witb which they obtained the most satisfactory results. Griinbaum worked best with the second method, and ascribed his difficulty with the third method to his tendency to vary his accommodation when the colour of the diffusion circles changed. I obtained more satisfactory results with the third method, and this may be due to the fact that I am accustomed to make observations without altering my accommodation as is often necessary in ophthalmoscopic work and in experiments on binocular vision. The results are given in Table III.; as in the flicker method the observer did not know which grey paper was being used, so that the results cannot have been prejudiced by previous knowledge. TABLE III. Yellow Pupil method. Red Orange Yellow green Green green- Blue Indigo Violet Purple Rivers Grrunbaum The agreement between the two observers is not so close as with the other methods, but is still close enough to show that this method is capable of giving consistent results, except perhaps with the darker Blue 141

6 142 W. H. R. RIVERS. colours, as indigo and violet. The observations made with these colours were noted as difficult and indefinite. It may be mentioned that contraction of the pupil took place more rapidly than dilatation and was consequently more easily observed. IV. Comparison of the three methods. In order to compare the three methods more exactly with each other and with those of other observers, the white values of the grey papers were determined by matching them with a mixture of white and black on the rotating colour-mixer. This was rendered somewha.t difficult by the fact that the grey papers used had a distinct bluish tinge. Consequently two methods were adopted. In one ten matches were made with each grey. The average of the ten was adopted, the mean variation being found in order to test accuracy. These results were controlled in a certain nuinber of cases by the second method, in which a small blue sector was added to the inixture of black and white so that an exact match became possible. The white value was then calculated by substituting for the blue an equal sector of the grey which had been found to be of the same brightness as the blue. It was found that the results obtained by the two methods agreed closely. In both methods a further correction had to be made to allow for the light reflected from the black sector. Hering has found that Rothe's black paper reflects Zth of the light reflected from his white paper and this figure was made the basis of the correction. TABLE IV. Comparison of three methods. Method Bands Flicker Pupil Observer R. G. R. G. R. G. Red A4 Orange Yellow Yellow Green Green Blue Green Blue Indigo Violet 26 28, Purple , In Table IV. the three foregoing tables have been put together and the white values of the grey papers substituted for their numbers. In cases when a coloured paper was put between two grey papers the mean of their white values has been taken. It will be seen that the results of the-three methods show a very close agreement. The exact

7 COLOUR PHQTQMETIRY. corrtespondence in many cases depends on the limited number of grey papers, and would doubtless be less close if a larger series were available. This agreement is especially marked between the band and flicker method and for medium and short wave-length colours. The most marked discrepancy between the band and flicker methods was in the case of red, and that this is not accidental is shown not only by- the fact that it occurred for both observers, but also by the fact that the discrepancy existed in the case of orange and purple, the difference being in both cases in the same direction, though of less amount. Red. appears to be a brighter colour by the flicker than by the band method. The only explanation which can be suggested for this is that in the former method the retinal area stimulated was more limited and was confined to the macular region, while in the band method no aperture was used and a much larger area of the retina stimulated. It is consistent with the peculiarities of macular and foveal vision that red should be a relatively bright colour for this region of the retina. The figure obtained by the pupil method showed greater differences. Here the change was most inarked for the long-wave colours; red, orange, yellow and yellow green were darker colours than by the other two methods, while blue green, blue, indigo and violet were slightly brighter. The deviations were approximately the same for the two observers, and it seems probable that they were due to the influence of the Purkinje phenomenon. In the pupil method the eyes were protected from the light to a much greater extent than in the other methods and would consequently become adapted to a feebler illumination. The relative darkening of the long-wave and brightening of the short-wave colours are the changes to be expected from this cause. V. Comparison with the results obtained by other methods. 143 Several other investigations of the relative brightness of different colours have been carried out with Rothe's coloured papers and the results are given in Table V. In the first column are the flicker results obtained by Schenck. The most marked deviation from the results previously given is in the great brightness which he found for yellow. It seems probable as already stated that the difference may be referable to the special feature of his method, viz. that the grey he used was itself produced by intermittent stimulation, and that its brightness

8 144 W. H. R. RI VERS. would be influenced by the contrast effect the importance of which has been shown by Sherring.ton and Griinbaum. In column II. are given the results obtained by Martius" with his after-image method. In this method the brightness of a coloured surface is determined by finding the grey patch which when placed upon the surface becomes neither lighter nor darker on fixation. The method is very easily applicable but requires much practice for exact observations. The results obtained by Martius differ considerably from those obtained by three methods of this paper and also from those obtained by Schenck, but there does not seem to be anv obvious explanation of this difference. TABLE V. Method Flicker After-image, Direct comparison White value Observer Schenck Martius Schenck Hering Hering Red Orange Yellow Yellow Green Green Blue Green Blue Indigo 12? Violet 12? Purple The next two columns give the figures obtained by Schenck and Hering by the simple method of direct comparison. Schenck's figures were given in his paper on the flicker metho(d (p. 618), and some, viz. those for yellow, indigo, violet and purple, differ considerably from those obtained by him with this method. On the whole they agree with my results by the flicker method, the most marked exception being blue. Hering's figures are given in his paper on total colour blindness3. They differ considerably both from Schenck's figures and my own. The last column gives the white values of the colours as Beitriige zur Psychologie u. Philosophie, I. p Schenck's co-observer, Pabst, obtained even lower values for indigo and violet, viz. 7-2 and 6-6, while by a modified flicker method Schenck matched indigo with a grey containing only 30 of white. In a paper published since the above was written (Pfuger's Arch. LiVIII. p. 43, 1897) Schenck has suggested that the discrepancies found by him between the methods of flicker and direct comparison might be due to the difference in the size of the retinal area stimulated in the two cases, and in further experiments he found that the indigo paper was matched with a brighter grey by the flicker method when the stimulated area was larger; with the red paper he found no difference. 3 Pflfger's Arch. XLIX

9 COLOUR PHOTOMETRY. seen by the case of total colour blindness and differ fundamentally from all those obtained by the other methods, the brightest colour being the yellowish-green, and the results given in this paper confirm Schenck, in showing that the various methods of colour photometry measure something different from Hering's white value. As regards the right of these methods to be called photometric, one is only justified in saying that in all of them the same attribute of sensation is being measured, and that from the approximate agreement with the results of direct comparison it is most probable that this attribute is the brightness of the colour. Of the three methods especially brought forward in this paper the band and flicker methods depend on the same principle, viz. the dependence of retinal persistance on intensity. The band method is unfortunately not applicable to spectral colours, but in simplicity and accuracy it is fully equal to the flicker method. The pupil method requires much more practice than either of the other methods, but the figures given show that it is possible to obtain with it consistent results. The nature of the "band" phenomenon is somewhat obscure and will be considered in a later paper. The result of this investigation shows that one of the chief factors upon which it depends is intensity. It seems probable that it does not depend on wave-length. If not, two colours found to be of the same intensity should not give bands. Among the colours used, there were none which gave no bands, but there were probably none which would have given no bands with the same grey if one had had a larger series. Colours such as orange and blue green, which differ greatly in wave-length and slightly in intensity, gave much less distinct bands than colours such as yellow and blue, which differ greatly in brightness. In conclusion, I should like to express my indebtedness to Mr 0. F. F. Grunbaum for his observations and for other help, especially in connection with the flicker method. 145 PH. XXII. 10