(Rushton), but that the equivalent background effectively adds to the real. dichotomy by using C. B. B. as subject, for, as described in the foregoing

Transcription

1 J. Physiol. (1965), 181, pp With 3 text-figures Printed in Great Britain THE ROD INCREMENT THRESHOLD DURING DARK ADAPTATION IN NORMAL AND ROD MONOCHROMAT BY C. B. BLAKEMORE AND W. A. H. RUSHTON From the Physiological Laboratory, University of Cambridge (Received 30 April 1965) In the foregoing paper (Blakemore & Rushton, 1965) we have shown that in dark adaptation the essential visual change that proceeds hand in hand with the regeneration of rhodopsin is a retinal reorganization very similar to that which accompanies a slowly fading background field. The threshold in both situations depends greatly upon the criterion used (e.g. size of test flash, etc.), but, at any stage of dark adaptation, the luminous background that has the same threshold by one criterion has the same by all. If then the threshold in dark adaptation is essentially the increment threshold to an equivalent luminance it becomes interesting to know how this internal equivalent background interacts with an external background of real light. This can be measured by obtaining dark adaptation curves in conditions where the test flash, instead of falling as is usual upon a dark background, is presented superimposed upon a background of fixed luminance. This has already been studied by several investigators (Crawford, 1947; Hattwick, 1954; Rushton, 1961). All have found for rods results that are similar to the results of the present paper, and all have interpreted them incorrectly. As we shall argue, it is not that the background stops rhodopsin regenerating (Crawford, Hattwick) nor that bleaching makes rods insensitive (Rushton), but that the equivalent background effectively adds to the real background, so that the threshold is simply the increment threshold of this total background, as Barlow (1964) has already stated. In the present paper we have eliminated complications due to the rodcone dichotomy by using C. B. B. as subject, for, as described in the foregoing paper (Blakemore & Rushton, 1965), his retina appears to lack cones entirely. METHOD The apparatus and procedure were similar to those described in the foregoing paper. The test light was flashed on for 1 sec, off for 1 sec and its intensity was adjusted by the subject to what he considered a threshold value. The criterion was either the detection of light patches of various sizes or the resolution of gratings. The flashes were presented superimposed upon a background field of determined luminance (pathway 2, Fig. 1; Blakemore & Rushton, 1965).

2 630 C. B. BLAKEMORE AND W. A. H. RUSHY'ON The difficulties in obtaining satisfactory light adaptation mentioned in the foregoing paper were overcome in the same way. The subject held his left eye steady for 2 min close to a bright pearl lamp by fixating with the other eye upon a distant point of light seen in a mirror at the proper angle. In this way about half the rhodopsin was bleached from an area of diameter 600 around the fixation spot (which was not the fovea). We wished to obtain dark adaptation curves with various kinds of test flash superimposed upon various intensities of background. Each dark adaptation takes 45 min and the programme would have been enormously lengthy unless some compresion had proved possible. Fortunately it turns out that it is not essential to maintain the background steady throughout the whole dark adaptation run. The eye rapidly adapts to one background after another and when returned to the first it gives the same threshold as when no change of background had been made. Thus in Fig. la all the results were obtained in a single run by quickly changing the background after each threshold determination. By way of a check on the method, sometimes the compression was made by keeping the background luminance fixed and changing the nature of the test flash. The two methods gave essentially the same results. Immediately at the end of dark adaptation an increment threshold curve was obtained for the full range of background intensities from zero up to saturation, using the same criteria of threshold as were used during dark adaptation. RESULTS Test flash of 60 All the points in Fig. 1 a were obtained in the course of one dark adaptation by suitably changing the background luminance. There was no time to take redundant measurements and we had to know at every stage which backgrounds were significant. In this we were helped by two preliminary observations. First, when a background is too weak to be seen by the subject it is also too weak to raise his increment threshold above the dark threshold (curve A with zero background). Secondly, when on any curve the threshold ceases to fall, it levels out at a steady value without drift or oscillation; it thus need not be examined further. Threshold measurements were therefore made as rapidly as was consistent with reliability, starting with the brightest backgrounds. After each threshold determination the background was dimmed by 1 log unit until dimming made no difference to the threshold or the subject said that the background was now invisible -both occurring at about the same level. Then the brightest background was again presented and a new threshold obtained. If this was found to have the same value as upon the previous occasion (with this background) it signified that the curve had now reached the steady level, so this field was removed from the rota of backgrounds to be studied. It will be seen that at any time the rota contained about three or four backgrounds, the highest dropping out as it reached its plateau which was always less than 3 log units above the dark threshold at that time; the lower backgrounds entering as the increasing sensitivity of the eye brought them into view. At the end of the run (45 min) the increment threshold was obtained for

3 I.NCREMENT THRESHOLD IN DARK ADAPTATION 631 each background and gave the values shown plotted in Fig. la at time 45 min. The fact that these final values correspond so closely to those where the curves appeared to level out justify us in our omission of the intermediate points, earlier experiments having established that thresholds remained steady throughout. The increment threshold measurements : 4 Qc) tc n Dark Time (min) Log background (trolands) Fig. la. Dark adaptation curves, A against dark background as usual, dotted curves when test flash was presented upon a fixed background of luminance (log td) indicated by numbers at right of curves. b. Increment threshold curves, B after full recovery from bleaching, dotted curves7at 5, 10,...min of recovery as indicated by numbers at left of curves and by corresponding ordinates in Fig. lb. Test flash was a circular patch subtending bs to 2 I. 1 A BDark. r_i Dark Time (min) Fig. 2a. Dark adaptation curves, b, increment threshold curves similar to Fig. la, b except that now test flash subtended 5' of arc.

4 632 C. B. BLAKEMORE AND W. A. H. RUSHTON obtained after 45 min are replotted as the lowest curve (B) of Fig. lb where some further high level points are added to define the exact course of the curve in the region of saturation (Aguilar & Stiles, 1954). Test flash of 5'. Figure 2a gives the results of a single dark adaptation run conducted and plotted precisely as in the experiment of Fig. 1 except that here the test flash subtended 5' whereas there it subtended 60. The lowest curve, B, of Fig. 2 b shows the increment threshold with a 5' test in the same way. The lowest curves A and B of Figs. 1 and 2 simply repeat the experiment of Fig. 2 a and b of the foregoing paper (Blakemore & Rushton, 1965) and show very similar results. In particular the results with a 5' field of test flash are almost identical to those with a 60 test but with ordinates all scaled down by a factor of about 2/3. And this same scaling applies pretty well to the whole family of curves in Figs. 1 and 2. Interpretation Photochemistry. It has long been believed that the fall in rod threshold during dark adaptation is due to the regeneration of rhodopsin. It was therefore natural to interpret curves such as those of Fig. 1 a in terms of photochemistry. In the presence of a bright background, the dark adaptation curves are seen to stop their improvement in sensitivity, and it was natural to conclude that bleaching by the light from the background had stopped the regeneration of rhodopsin (Crawford, 1947; Hattwick, 1954). But this view cannot possibly be correct, for as Rushton (1961) pointed out, not only is the background light in fact thousands of times too weak to alter the rhodopsin equilibrium as supposed, but changes from one background brightness to another approach equilibrium in a few seconds. Indeed the results of Figs. 1 and 2 display a striking contrast between the rapid and reversible changes of threshold as the backgrounds are quickly altered, and the unaffected slow fall in dark threshold (curve A) that keeps pace with the quite uninterrupted regeneration of rhodopsin. It is plain therefore that the family of curves in Fig. la is not to be explained by changes in rhodopsin level produced by background bleaching. Retinal organization. Rushton (1961) showed that a clearer insight into the curves such as those of Figs. La and 2 a might be obtained if the same experimental results were differently displayed. The ordinate drawn in Fig. la at 45 min cuts all seven curves of the family and hence gives the increment threshold against these seven background luminances. These points are plotted in Fig. 1 b and curve B is drawn through them. This of course is the increment threshold after 45 min of dark adaptation. The increment threshold after, say, 20 min of dark adaptation may likewise be obtained by erecting an ordinate in Fig. 1 a at 20 min and noting where this cuts each of the seven dark adaptation curves. In general the ordinate will

5 INCREMENT THRESHOLD IN DARK ADAPTATION 633 pass between two experimental points on each curve and to estimate the threshold at 20 min exactly we take the intersection of the ordinate with the little line joining those two points. This is the threshold value transferred to Fig. 1 b to give the series of points lying on the curve labelled 20'. In the same way the ordinates erected at 5, 10, 15 and 30 min in Fig. 1 a give the threshold values that lie upon the curves labelled correspondingly in Fig. 1 b. Except for backgrounds above 2 log td where saturation sets in, all the curves of Fig. 1 b have the same shape and may be fitted rather well simply by sliding curve B (without rotation) up the 450 Fechner line to the right. Now a displacement of the curve up the 450 line is precisely what would be seen if the effect of bleaching was to reduce the sensitivity of the rods alike to test and to background. But this rather plausible explanation (Rushton, 1961) is quite inconsistent with the results of Fig. 2b that are constructed from 2a just as Fig. lb is from Fig. 1a. Here again all the curves of Fig. 2 b may be fitted by sliding curve B up the linear part of the curve. But the line is now not sloping 450 and will no longer bear the interpretation that rods have become insensitive to field and flash alike. They become very much more insensitive to fields, so the phenomenon cannot be explained just by rod insensitivity, it must depend upon rod organization. How it does so is given by the answer to the following analytical question. If curve B, Fig. 2, follows the generalized Fechner relation fl+ log AI = a log (I+ ID) (1) (which is true within the accuracy of the measurements) what change in the equation will cause the curve to slide up this Fechner line of slope a? Analytically we are just shifting the origin of the curve; all values of log I are increased by some value c and all values of log AI by ac c. Define IB by the relation c = log (IDI'B) (2) After the shift of origin log I becomes log I + c = log (I. IDI'B), hence I becomes I.ID/IB. Thus the new form of eqn. (1), obtained by increasing log I by c, and log AI by acc (where c is given by eqn. (2)) is l+ log, ai + log (IB/ID) = a log (I.-+ ID) = a log (I + IB) + a log (ID!I'B) or fl+log I = a log (I-+IB). (3) Now eqn. (3) is the same as eqn. (1) except that 'D has changed to IB. This answers very simply our analytical question. 'The condition for the curve

6 634 C. B. BLAKEMORE AND W. A. H. RUSHTON to slide up the Fechner line with no change of shape is that ID be simply increased to a new value IB.' Now ID is the Eigengrau of Fechner (1860) or 'receptor noise' of Barlow (1957), so the increase from ID to IB may be interpreted as an increase in receptor noise from that corresponding to the Eigengrau to that corresponding to IB, the total 'dark light' of bleaching, and indeed Barlow (1964) has so interpreted it. From eqn. (3) we see that the log threshold is just the increment threshold corresponding to a background of (I+IB). Thus real (I) and equivalent (IB) backgrounds simply add together. We need not depend upon the foregoing analysis, however, to reach this conclusion, for the experimental points of Figs. 1 and 2 allow us to test directly the accuracy and range of the idea that real and equivalent backgrounds are additive. The lowest curves, A and B, in the two halves of each figure are ordinary dark adaptation and increment threshold curves similar to those in Fig. 2 of the foregoing paper (Blakemore & Rushton, 1965). It was there shown how the 'equivalent background' of bleaching could be obtained at any point in dark adaptation by drawing a horizontal line from that point on curve A, till it meets curve B. The background corresponding to this intersection is the equivalent background of bleaching. For example, in Fig. 1 the horizontal that cuts curve A at the 10 min ordinate cuts B at the 25 td ordinate, hence the equivalent background of bleaching at 10 min is 25 td. Thus to each real background plotted as abscissa in Fig. 1 b must be added 25 td to give the total background and the increment threshold appropriate to this sum may be read off from curve B. The dotted line labelled 10 min shows the increment threshold at 10 min of dark adaptation obtained in this way, and all the dotted curves of Fig. l b were obtained likewise. In Fig. I a the curves were similarly found by adding a fixed real background to the continually fading equivalent background, instead of adding the fixed equivalent to a range of real backgrounds. The whole set of dotted curves in Fig. I a and b is derived from curves A and B with nothing arbitrary in any operation. It rests upon the concept of equivalent background of bleaching, and on the additivity of all backgrounds real and equivalent. The dotted curves fit the observed thresholds well enough, not only in Fig. 1 but equally in Fig. 2 where the test flash only subtended 5' of angle. The additivity of real and equivalent backgrounds If we measure the increment threshold by flashing a test upon a screen illuminated by a steady light I1, no complication will arise when the experiment is extended to the case where the screen is illuminated in two lights I, and '2* Obviously the threshold will be the same as when I1 is increased

7 INCREMENT THRESHOLD IN DARK ADAPTATION 635 to the value (I + I2). It seems natural to regard the summation of the bright light from the background and the dark light of bleaching in the same way. This may, however, be a very misleading over-simplification, and we should not be led by the 'naturalness' of the concept and the pleasing fit of the curves in Figs. 1 and 2 to accept the principle of additivity without more careful examination. Below the level of rod saturation, what the experiments of Figs. 1 and 2 show in fact is that when log I exceeds log IB by a unit or more, IB is without appreciable effect; when log 'B exceeds log I by a unit or more, I is without appreciable effect; when I and IB are about the same size their combined action raises the threshold above that due to either alone by about 0 3 in Fig. 1 and 0-2 in Fig. 2. All this would be expected from the principle of additivity but only the third relation goes any distance to justify it, and that relation was not securely established because of the difficulty in measuring simultaneously thresholds with different backgrounds. We therefore performed a number of experiments to test additivity at its most sensitive point, namely when I = IB, hence log (I+ IB) = log 2 IB = 03+logIIB. (4) The experiment was to find the threshold at some moment during dark adaptation, to deduce from it the value of IB, to add a background of this value and to measure the rise in log threshold produced, which should ideally be 0 3 log unit with a 60 test flash. These experiments never contradicted the additivity principle, but they were too imprecise to support it strongly. One difficulty is that it takes 30 sec or more to reach full equilibrium after the background I is introduced. At the end of this time some regeneration has occurred and the dark light is no longer IB, the background therefore no longer matches it and the threshold also is not comparable with the earlier figure. To overcome this difficulty we decided to compare not one retinal region in two successive states but two retinal regions simultaneously, one steadily illuminated and one dark. This required fixation more reliable than that of C. B. B. and the subject of these confirmatory observations was W. A. H. R. whose cones are normal, and consequently limit the range of rod adaptation to less than 3 log units. A confirmatory experiment Principle. The eye, fixating upon the cross of the inset, Fig. 3, observed two fairly large oval test patches P and Q that flashed on and off together and were situated 50 either side of the fovea. Flash P was seen against a black background, and was attenuated by the interposition of a fixed neutral density filter of 0-3. Flash Q was seen directly, but superposed

8 636 C. B. BLAKEMORE AND W. A. H. RUSHTON upon a red background which in the fully dark adapted state was adjusted to be ID. This was found (see Discussion in the foregoing paper (Blakemore & Rushton, 1965)) by plotting the log increment threshold curve for flash Q against log background luminance and producing the Fechner line downwards to cut the absolute threshold horizontal, which it does at the value log ID* The experiment consisted of 3 parts. (a) The red background was adjusted to have scotopic luminance ID, (b) the whole retinal region was uniformly bleached, (c) flash P was brought to threshold simply by adjusting the wedge W, (Fig. 1 of Blakemore & Rushton 1965), situated immediately in front of the eye. This altered flash P, flash Q, and the red background to Q all in the same proportion. The threshold setting of W, for flash P (with its interposed 0 3 density) was always found to be also the threshold setting for flash Q (on its varying red background). This is what would be expected if the real red background light I added to IB, the equivalent total background of bleaching. Equation (3) expresses the dependence of threshold AI upon I and IB. We may put a = 1, since the flashes P and Q had large areas and the Fechner line had a gradient of 0-95, and if we express AI in suitable units we may also put i = 0. Thus eqn. (3) becomes log (Al) = log (I+IB) (5) In the dark adapted state the threshold AIp for flash P is putting I = 0, IB = ID and adding 0-3 for the interposed filter. obtained by Hence log (AI.) = 0 3 +log ID- For flash Q, IB = ID and I has also been made ID. Hence log (AIq) = log (ID + ID) = log 2 +log ID = log (AI.). Now after bleaching, when the equivalent total background is IB, the threshold for flash P is again obtained from eqn. (5) giving log (AIp) = 0 3+±log IB The common wedge which brings flash P always to threshold must increase transmission above the dark adapted conditions in the ratio (AIp)/(AIp) = IB/ID. Consequently the red background that in the dark was ID is now IB and from eqn. (5) the threshold for flash Q is given by log (AIQ) = log (IB +IB) = log 2+log IB = log (AIp). Thus with this arrangement of flashes and common wedge we secure that all through the course of dark adaptation the red background I is equal to 'B' the equivalent total background of bleaching. If these two backgrounds simply added in their effect on the increment threshold, then the background would always raise the log threshold by 0 3. It should

9 INCREMENT THRESHOLD IN DARK ADAPTATION 637 therefore match the ordinary dark adaptation threshold (against a dark background) with a 0 3 neutral density interposed. The observation that these thresholds in fact match well substantiates the principle of additivity of real and equivalent backgrounds at the value where a discrepancy should be most easy to detect. Experimental details. Figure 3 (inset) shows the arrangement of the visual field used in conjunction with our increment threshold equipment (Fig. 1, Blakemore & Rushton, 1965). The cross was a small bright fixation point. The test flashes P, Q were oval areas subtending 10 x 30 situated 50 on either side of the foveal centre. Both lay in the same test beam and flashed together. The equivalence of the two flashes and of the two retinal regions was verified in preliminary trials. In the experiment, P was made half as intense as Q by the inter-position of a 0-3 neutral density. The luminous background upon which Q but not P was projected was made deep red (Ilford filter no. 206), whereas both flashes were green (Ilford 624) to bring rods into prominence and to suppress cones. The log increment threshold curve was determined using flash Q against its red background, and the gradient of the Fechner line found to be The background was found corresponding to 'D where the Fechner line produced cuts the horizontal through the absolute threshold for Q. The background was set here and the flash brought to threshold. It was found to be 03 long units above absolute value and flash P (seen through the 0-3 filter) was also just at threshold in these conditions. For the Qualitative Experiment next described, all settings were left in this position except for the common wedge, Wc, which was moved to compensate for the effects of bleaching and subsequent recovery. Qualitative results. Bleaching naturally raised the threshold, and the density of the common wedge W, had to be reduced by 3 log units or more. It was then found that the threshold for P (in spite of the 0 3 density interposed) was a good deal lower than the increment threshold for Q. This inequality, however, only persisted so long as the thresholds referred to cones. With the transition to rods (as shown by the break in the dark adaptation curve and the change in colour and definiteness of the test flash) the thresholds for P and Q became practically identical. At every subsequent stage of dark adaptation while X was steadily fixated, either P and Q were both seen flashing together or neither of them was visible. There was never a stage where one could assert 'as you move W0, P always comes in before Q does' or the reverse. That is found while cones determine the threshold and it can be seen if the test flashes P, Q are 5' pin points, instead of subtending rather large areas. It is easy to see why the equality of threshold between P and Q that holds for rods breaks down for cones. We used a deep red background for

10 638 C. B. BLAKEMORE AND W. A. H. RUSHTON Q to raise the increment threshold for cones much more than for rods so as to ensure that when the threshold at P lay upon the rod branch, the stronger flash Q could not be exciting cones. It is precisely this high cone threshold at Q which causes the inequality observed when the threshold lies with cones. With 5' point flashes at P and Q we confirm what was already seen in Fig. 2, that during dark adaptation, rods do not improve their sensitivity equally to the flash and to the background. In order to test the principle of additivity with point flashes where the gradient of the Fechner line is 2/3 (as in Fig. 2b) all that is needed is to have, instead of the common wedge Wc, two wedges W2, W3 arranged to move rigidly together. T4" has a density of 2 log units per 150 cm and is interposed in the flash. W3 has a density of 3 log units per 150 cm and is interposed in the background. The experiment could then be performed precisely as before. But we have not done this. By a more clumsy method, however, we have satisfied ourselves that point flashes do not obviously contradict the principle of additivity of bright and dark lights, as they obviously do contradict the principle of change in receptor sensitivity. Quantitative results. The foregoing experiment where the flashes P and Q are seen to come in and out of view together as the wedge W, is moved, is very convincing to the subject who has his hand on the wedge, but the reader will wish to know something of the precision of these threshold measurements. To obtain this, the procedure was modified very slightly. As before, the bleached eye, dark adapting, measures the threshold at P by moving the common wedge W, which also adjusts the red background of Q so that it is always equal to the equivalent background of bleaching IB at the moment. In the beam of the test flash there is a second wedge W, (Fig. 1, Blakemore & Rushton, 1965) which in the Qualitative Experiment was left at some fixed position y. In this experiment also it is always set at position y whenever the threshold of P is being measured by moving W, But now, immediately after each such measurement, W, is left at that value and the threshold of Q against the background IB is found by moving W1, which was displaced from position y by the operator and returned to the proper threshold position by the subject. Since the retina receiving Q is not subjected to any sudden change of adaptation by this procedure, the threshold position of W1 is set in a few seconds, and the place is always very close to y. The operator then replaces the wedge exactly at y for the next threshold determination at P. The results of one experiment are shown by the curves of Fig. 3. The dark adaptation, as measured by flash P, is seen in curve P where the black points indicate thresholds of rods. Curve Q shows by how much the

11 INCREMENT THRESHOLD IN DARK ADAPTATION 639 log increment threshold at Q exceeded the log dark threshold at P. If in the experiment the subject had always set W1 to its initial position y, all the points would lie upon the 0 3 horizontal. This indeed would have been the expected course if the Fechner line had shown a gradient of P Q 2 R ri~ ~~~~Dr 0 I~~~~~~~~~~~~~~ Minutes Fig. 3. Inset, cross is the fixation point, P and Q oval test areas that flash together. Q is seen upon a red background, P through a 0 3 density upon a dark background. Curve P shows the dark adaptation curve measured by flash P (white circles, cones; black circles, rods). Curve Q shows by how much the long increment threshold Q exceeds the log threshold of P at each stage of adaptation, when the red background to Q is kept equal to the total equivalent background of bleaching. The black circles, Q, should fall on the theoretical slightly curved line if real and equivalent backgrounds simply add. Actually the gradient was 095 which means that in dark adaptation the curve slides down the 095 slope and in compensation is pushed up the 1.0 slope. This leads to the expectation that the points of curve Q, Fig. 3, should fall not upon a perfectly horizontal line but on the slightly curved computed line Q drawn in Fig. 3. It is clear that the points fulfil this expectation and thus demonstrate that even at its most sensitive point the threshold is entirely consistent with the principle of additivity of real and equivalent backgrounds. SUMMARY 1. In the previous paper (Blakemore & Rushton, 1965) it was shown that bleachings and backgrounds raised the threshold in similar fashion no matter by what criteria the thresholds were judged. For any state of bleaching there was a certain 'equivalent background' that could be measured in trolands.

12 640 C. B. BLAKEMORE AND W. A. H. RUSHTON 2. In this paper we ask 'If after bleaching, the test flash falls upon a luminous background so that both equivalent and real backgrounds are present, in what way do they combine to define the resulting threshold?' 3. This was investigated on the rod-monochromat over a millionfold range of rod thresholds. It was found that, independent of the criterion of threshold used, equivalent and real backgrounds added together, and the observed threshold was the increment threshold to a real background equal to that sum. 4. The most sensitive observation in testing the additivity of real and equivalent backgrounds is when they contribute in equal parts to the total background. This was tested in the normal eye by a special arrangement. During dark adaptation, as the equivalent background decreased, the real background was also decreased so that the two remained equal. The threshold throughout had the value corresponding to the sum of these equal real and equivalent backgrounds. Our thanks are due to Mr Clive Hood who built the equipment and assisted in the experiments. W.A.H.R. acknowledges with thanks a grant from the U.S. National Institute of Neurological Diseases and Blindness (NB ). REFERENCES AGUnAR, M. & STILES, W. S. (1954). Saturation of the rod mechanism of the retina at high levels of stimulation. Optica Acta, 1, BARLOW, H. B. (1957). Increment thresholds at low intensities considered as signal-noise discriminations. J. Phy8iol. 136, BARLow, H. B. (1964). The physical limits of visual discrimination. In GIESE, A. C., Photophysiology, Chap. 16. New York: Academic Press. BLAKEMORE, C. B. & RUSHTON, W. A. H. (1965). Dark adaptation and increment threshold in a rod monochromat. J. Phy-iol. 181, CRAWFORD, B. H. (1947). Visual adaptation in relation to brief conditioning stimuli. Proc. Roy. Soc. B, 134, FECENER, G. T. (1860). Elemente der Psychophyaik. Leipzig: Breitkopf und Hartel. HATTwicK, R. G. (1954). Dark adaptation to intermediate levels and to complete darkness. J. opt. Soc. Amer. 44, RUSITON, W. A. H. (1961). Rhodopsin measurement and dark-adaptation in a subject deficient in cone vision. J. Phy8iol. 156,