ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Final Report VARIATIONS IN SPECTRAL SENSITIVITY WITHIN THE liMAN FOVE'A H. Richard Blackwell J. H. 'i-ayior Vision Research Laboratories ERI Project 2455 BUREAU OF SHIPS, DEPARTMENT OF THE NAVY CONTRACT NO. Nobs-72038 WASHINGTON, D. C. June 1958

~2 -r I'll, K.

2455O10-F The experiments reported here represent the basis for the doctoral dissertation: "Variations in Spectral Sensitivity Within the Human Fovea" by John Hall Taylor University of Michigan, 1952

k The University of Michigan ~ Engi 2455-10-F TABLE OF CONTENTS Title List of Figures List of Tables Summary I o Introduction Io Procedures and Apparatus I., Results and Discussion References Pae iii iv 1 6 15 19 neering Research Institute I II ii

The University of Michigan T Engineering Research' Institute 2455-10-F LIST OF FIGURES Number Title 1 Spectral sensitivity data of Hsia and Graham (1952) 2 Spectral Sensitivity data of Crozier (1950) 3 Schematic diagram of optical apparatus 4 Diagram of thermopile circuitry 5 Measurements of ocular chromatic aberration 6 Spectral sensitivity data for NLT: zero eccentricity 7 Spectral sensitivity eccentricity 8 Spectral sensitivity eccentricity 9 Spectral sensitivity eccentricity 10 Spectral sensitivity eccentricity 11 Spectral sensitivity eccentricity 12 Spectral sensitivity eccentricity 13 Spectral sensitivity eccentricity 14 Spectral sensitivity eccentricity 15 Spectral sensitivity eccentricities 16 Spectral sensitivity eccentricities 17 Spectral sensitivity eccentricities data for NLT: 15 minutes data for NLT: 43 minutes data for JHT:~ zero data for JHT: 15 minutes data for JHT: 43 minutes data for DWMN zero data for DWM: 15 minutes data for DWM: 43 minutes data for NLT: all data for JHT: all data for DWM: all I iii

-L The University of Michigan ~ Engineering Research' Institute 2455-10-F LIST OF TABLES I Summary of Radiometric Measurements and Their Conversion to Units of Energy at the Cornea II Residual Errors after Correction for Chromatic Aberration (diopters) III Spectral Sensitivity for Observer NLT' Inverse Threshold Radiosity (microwatts x 10-6) IV Spectral Sensitivity for Observer JJIT; Inverse Threshold Radiosity (microwatts x 10-6) V Spectral Sensitivity for Observer DWWM Inverse Threshold Radiosity (microwatts x 10'L) iv

The University of Michigan T Engineering Research. Institute 2455-10-F SUbMMARY Spectral sensitivity curves have been measured at each of three foveal locations in each of three human eyes, utilizing a circular target subtending 1 minute of arc. Target exposure duration was 0.1 second; an artificial pupil of 6.04 mm, diameter was used, The temporal forced-choice variant of the method of constant stimuli was employed, utilizing automatic presentation and recording equipment. A total of 24,750 observations were made in all. Precautions were taken to insure that the various chromatic targets would be as well focused as possible, by the use of matching chromatic fixation lights, and by measurements of the chromatic aberration of the eye and the use of ophthalmic corrections to compensate. Spectral sensitivity was found to vary in an idiocyncratic manner from location to location and from eye to eye, to an extent far in excess of experimental uncertainties. The sensitivity curves were generally irregular, often exhibiting two peaks, In different curves, peaks were found at about 460, 480, 510, 525, 5409 555, and 610 ma. These data suggest that a large number of cone types may exist and that sensitivity for small retinal areas may represent the action of various combinations of the various classes of cones$ The spectral sensitivity data were related to the various areas which are differentiated in the entoptic image of the macula reported by the three observers, It was found that differences in sensitivity over the blue end of the spectrum were predicted very adequately by differences in the macular entoptic image, Since tritanopia does not have an appreciable effect upon the spectral sensitivity curves, these data suggest that the entoptic image of the maculawis produced by differences in the density of the macular pigment. v

I The University of Michigan ~ Engineering Research Institute 2455-1O-F I. INTRODUCTION The spectral sensitivity of the eye has been studied by any ma nvastgato. under a wite variety of conditions These investigations have revealed the existence of photoreceptors with two markedly different spectral1 sensitivity curves - One spectrPl sensitivity curve has a maxi~muv at 5bout 505 mF, and is associated with the;od pho1torecepto.rs$ which are distributed throughout the entire retina except fo:? the fovea-. The other spectral sensitivity curve has a maxiwmtm at about 555 m~. and is associated Vith the cone photo-Leceptors which are distributed throughout the entire retina with heavrr concentration in the rod-free fovea Among the most recent measuxrements of the spectral sensitivity of.:he foveal cone photoreceptors are those of Hsia and Graham (Ref- 1), which ';...e presented in Figure 1. The test target used for these measurements subtended 42 minutes of arZc-, The principle peak of the cu-rve is at 550 mv. with marked secondary peaks at about 440 mi, end j about 620 m, Z6 The ctrve presented in Figure 1 represents the ave:age of the data for five observez's The general smoothness of the tri-modal function is appsrent,- Of coutise, averetging data in this w~.y tends to smooth out small ir':egulat ities.~ However.i the datea for the individtal observers in the Hsia and Cr ahem study are generalle y smooth so thet the av.erage curve seems to be an rdequate representation of all the data, More r:ecent data by Sperling (Ref:^ 2) demonstrate that the spectral sensitivity cturve of the ht:man fovea depends to some extent upon the angular size of the test target, In gener&a.l the curve exhibits more irregularity as the size of the test target is treduced from 42- to 3< minutes of arc, The spectral sensitivity of the human fovea has been measured for test targets of essentially point-size by Croziaaer (Ref, 3Y) The tect target was a squavie,, subtending 1i6 minutes on a side; the expf3.tre duration vaw.ied between 0O,051 and 00785 seconds, These dats rare presented in Figure 2, the scales being identical with those used in Figure I,- There appear to be a number of peaks in this curve and itts general form is quite different from the curve rep!:/esenting the Hsia and Graham datea obtained with the larger test target, The dependence of the form of the spectral sensitivity function upon target size illustrated by a comparison of Figures 1 and 2 has considerable practical and theoretical rignificance,. 'Luminous signals used in military operations are frequently viewved. at such long distances that tb.ei'xr subtense is essential point-size,:, In assessing the design and use of these signals2 account most be taken of the effect of target size upon the spectral sensitivity curve. 1I i

The University of Michigan ~ Engineering ResearchInstitute 2455 10-F The dependence of spectral sensitivity upon target size is of theoretical interest because of its relation to anatomical and physiological properties of various protions of the human fovea. These properties of the eye will be discussed in the following sections o Morphologically, the fovea comprises a roughly circular area at the retinal center with a diameter of about 15 mm.,, which thus subtends about 5 degrees of arc. 'The fovea is seen as a depression on the vitreal surface of the retinas and if the area is examined microscopically in cross section, it is apparent that certain of the tissue layers present elsewhere in the retina (Polyak's layers 5 to 9) are here absent or attenuated. The outer nuclear and bacillary layers are somewhat thickened, and it has been shown that as the center is approached there are progressively fewer rods and more cones until in the centermost 0.5 mm., only cones are present. This rod-free area subtends 100 minutes according to the recent results of Polyak (Ref. 4). The Macula Lutea Buzzi (Ref. 5), who first reported the existence of the fovea, noted the presence of a yellow pigment which permeates certain of the tissues in the region of the central retina. This area of the retina, known as the macula lutea, or yellow spot, has a maximal diameter of 5 mmO or more in man, although the pigmentation is only faintly observable in the outer 1 mm. band and it seems likely that 3 mma is the effective limit, corresponding to 10 degrees of visual angle~ The densest pigment occurs on the slopes and margin of the inner fovea, while on the floor of the foveal pit where the tissues that elsewhere carry the pigment are extremely thin and the non-pigmented photoreceptor layers are thickened, there is an area about 0.4 mm. across (80. minutes) which is relatively free of pigment. In the opinion of Polyak (op. cit.) the pigment is present in the living eye, and is not, as was claimed by Gullstrand (Ref. 6) and others, a postmortem phenomenon. Because the macular pigment is present in those layers of the retina which overlie the layer of rods and cones (Polyak's layers 4 to 10) it has long been recognized that its effect in vision must be that of a yellow filter intercepting the light reaching the macular area. The usual consequence of interposing a yellow filter in this manner is reduction of luminosity of the short-wave components of the incident light; thus, white light appears yellow; red, yellow, and green are relatively unchanged; and blue is partly or wholly absorbed, In our usual visual experience, however, we are never aware of a yellow annulus in the visual fields and it must be assumed that sensory constancy, perhaps coupled with local adaptation at the retinal or higher neural levels, operates to maintain apparent uniformity of the field. Only under certain special conditions of viewing can the macula lutea be observed entopically and its configuration in -J I 2

The University of Michigan T Engineering Research Institute 2455- 10-F the living eye estimated. Entoptic Observation of the Macula A technique for making the macular pigment appear in an entoptic image was first described by Maxwell (Ref. 7), who recommended that pieces of blue and yellow glass be alternated before the eye while regarding a uniformly lit surface. At appropriate rates of alternation, a dark spot appears in the blue phase and disappears in the yellow, the whole experience being exceedingly transient. Helmholtz (Ref0 8) ascribed the dark spot to differential rates of adaptation of the cones depending upon density of the overlying pigment~ More recently Miles (Ref0 9) has proposed a method whereby most individuals can o$server and plot the extent and character of their own maculae. His technique consists in the fixation of a smiall point on a uniformly white surface while two filters, dichroic purple and a, neutral of corresponding visual den$ity, are alternated in front of the eye. This procedure results in an entoptic image whose outline and texture may be sketched by the observer. The most commonly reported entoptic macular image by Miles' procedure has the following features: 1. A central dark spot with well-defined edges, The average angular subtense, as determined from 26 single eyes, is 32 minutes, 2. A clear annular band which appears relatively bright and has an average outside diameter of 70 minutes. The outer edge of this area is quite well-defined. 3. A broad ring lying outside the clear area, appearing to have about the same brightness as the central dark spot. The average value for the overall diameter of this area is 160 minutes, The outer edge of this area is quite variable between observers, appearing smooth, asteroid, scalloped, or like a "shell burst" to various observers, The magnitude of the elements of the macular image do not. in some respects, correspond with the histological data. The outermost dark area in the macular image subtends an average of 160 minutes instead of the 5 degrees believed to be the angular diameter of the entire foveal excavation~ The 160 minute dimension does however correspond well with the zone of greatest pigment density0 The bright annular band has outer dimensions which agree with the outer dimensions of the floor of the foveal excavation. The floor of the foveal excavation (nonpigmented) subtends 80 minutes, whereas the annular area was found to have an average outer diameter of 70 minutes. The central dark area of the macular images is of doubtful ofrigin No evidence exists from tissue studies that there is a spot of pigment in the very center of the foveal excavation which would correspond to this area. There is, however, a region of the foveal excavation which is still further indented, called the umbo.o The diameter of the umbo has not been definitely determined, but it seems reasonable to suppose that, since the umbo is probably the I 3

The University of Michigan - Engineering Research, Institute 2455-lO-F retinal fixation point, it may correspond closely to the dark central area of the macular figure0 Two anatomical possibilities then exist: either the umbo contains some yellow pigment as yet undiscovered histologically, or the sharply inclined walls of the umbo result in some refractive phenomenon which acts in the same manner as does the macular pigment. Physical Measurement of the Macular Pigment A more direct assessment of the yellow pigment has been attempted by Wald (Refo 10), who has presented spectrophotometric curves for a crude extract of human maculaeo The absorption spectrum so obtained is, as Wald points out, virtually identical with that of crystalline leaf xanthophyll, maximal absorption occurring in the neighborhood of 460 m In addition to these curves9 Wald presents some visual estimates of pigment density which show wide individual differences between observers; some apparently having no pigment whatever, and others having sufficient pigment density so that as much as 90 per cent of the light is absorbed at 436 mAo Dichromasy of the Central Fovea In 1894 Konig (Refo 11) reported the hitherto unnoticed fact that the color vision of the foveal center is dichromatic, and that in his case a carefully fixated small field could be matched by a mixture of only two primaries~ He further described the condition as similar to the uncommonly encountered variety of dichromasy known as tritanopia. Tritanopia is extremely rare' affecting an estimated 0,0001 per cent of the male population. In the past few years "small-subtense tritanopia" has been re-examined by Willmer (Ref. 12), Wright (Ref0 13), and Hartridge (Refo 14). In general, the conclusion of Konig has been borne outs that the central fovea is indeed dichromatic and that the color confusions found with sufficiently small stimuli are consonant with the tritanopic pattern, Wright (Ref, 15) has recently measured the spectral sensitivity curves of seven tritanopes and has found them to be within the range of the spectral sensitivity curves of normal observers. Thus, there is no reason to expect that "small-subtense tritanopia" will have any marked influence on the spectral sensitivity curve of the central fovea obtained with small targets0 Purpose of the Present Study From the foregoing analysis, it is apparent that the spectral sensitivity curve might well be expected to vary from point to point within the central fovea, depending upon the density of the macular pigment. Where the pigment is dense, the spectral sensitivity curve should be lowered over the considerable range of wavelengths absorbed to any appreciable extent by the macular pigment. The use of small test targets would be expected to maximize these differences, since with larger test objects there would undoubtedly be some sort of averaging process involved in the spectral sensitivity measure obtained, Presumably, spectral sensitivity variations from point to point would be related to the entoptic macular image. The spectral senstivity should be reduced over the -J

The University of Michigan ~ Engineering Research, Institute 2455-10-F short wave end of the spectrum whenever the entopltic macular image is dark, in comparison with the sensitivity in areas where the macular image is bright~ The effect of the macular pigment should be maximal at 460 mg; no effect would be expected for wavelengths longer than 570 mp, Stiles (Ref. 16) provides data which add some support to this prediction. Stiles measured detection thresholds for a 10 minute test target at several locations within and outside the central fovea. The few data which were obtained demonstrate higher thresholds for the foveal area than for the immediately surrounding peripheral area for 435 mg, 475 ma, and 580 mp test targets. Thresholds for the foveal area lower than those for the immediately surrounding peripheral area were found for a 700 m~ test target. These data show general agreement with the results to be expected from the known density of the macular pigment in these areas. The data for wavelengths 435 and 475 my conform exactly to the expected effect of the macular pigment, as do the data for 700 mL, The data for 580 mg seem surprising if the macular pigment is indeed xantrhophylX1~ Still, other factors such as photoreceptor density might be involved which would complicate interpretation of the results. This interpretation of the data wousld be greatly facilitated by having entire sensitivity curves for each retinal location studied. The "small-subtense tritanopia" mechanism would not be expected to modify the spectral sensitivity curves for small test targets since the spectral sensitivity data for tritanopes are essentially normal. However, Crozier's data shown in Figure 2 could result from the involvement of a collection of cones differing from the usual collection with regard to the number of cones of each type. From this point of view, we might expect to obtain irregular appearing spectral sensitivity curves whenever small targets are utilized, Furthermore, these curves might be expected to differ for different locations, of the test target within the central fovaa, since there is no reason to expect that cones of the various types are equally numerous at various retinal locations. The present study was designed, therefore, to assess possible variations in the spectral sensitivity curves obtained at different locations within the human fovea and to relate these variations to the presumed density of macular pigmelnt as revealed by the macular entoptic ploto Measurements were made with a circular test target with a 1 minute diameter to avoid averaging results from different areas of the retina. It was expected that irregular spectral sensitivity curves might well be obtained because of variations in the relative number of cones of the various types from point to point. J 5

Trl_ _ I ___....__ L_. _ r,, l,,,,t,, neilel,,, v- L i ne universiry or Iviicnigan - cngineerillg trdr[clin I utiLurt 2455-10-F II* PROCEDURES AND APPARATUS The general requirements placed upon the procedures and apparatus utilized in this study may be listed as follows: 1. spectral purity of the stimulus, of a degree attainable only with a spectroscopic instrument; 2. control of stimulus position on the retina, involving a suitably precise system of fixation and reasonably bt~af exposures; 3o control of stimulus size, by reduction of accommodative error and chromatic aberration of the eye; and 4. an adequate psychophysical method' The procedures and apparatus utilized in the study are described in detail in the following sections A. Observers Three observers were used in the experiment: a color-normal male emmetrope (JHT), a color-normal female with corrected low-degree spherical and astigmatic errors (NLT), and a deuteranopic male emmetrope (DWM). Despite the considerable tedium of long and frequent observing sessions, motivation seemed to be sustained in all three observers during the entire course of the experiement. All three were highly trained in the specific observing task, and since the present study followed closely upon another experiment of rather similar nature in which two of the observers had served, practice effects were negligible, B. Stimulus Wavelengths Initially, eleven wavelengths were chosen at approximately equal intervals throughout the visible spectrum. The exact wavelengths were chosen to match the dominant wavelength of interference filters used for fixation lights, as described below, As is well known, these filters must be coupled with suitable cut-off filters in order to eliminate secondary transmlission peaks and the stray light resulting from the failure of the transmission to go to zero at intervening wavelengths. While the physical transmissions of these filter combinations indicate reasonable purity, the spectrophotometric curves alone are deceptive in terms of their visual effect at the ends of the spectrumo In order to arrive at the true visual peak wavelength it is essential to convert the physical measurements as follows: The spectrophotometric curves were first multiplied by the energy curve for tungsten at the appropriate color temperature utilizing the data of Skogland (Ref. 17), Then, the resultant values were multiplied by the visibility curve of the eye according to the values of Hardy (Ref. 18). The resultant curves of "Ivisual effect" were plotted and the effective peak wavelerngths were determined by use of a plane polar planimeter to locate the half-area -i 6

The University of Michigan 2455 ~ Engineering Research lnstitute 5-10-F point on the wavelength abscissae. Wavelength values used hereafter in this report were specified in this way. The- test target wavelengths were set to match the fixation lights and it was found that the values read from the wavelength drum of the monochromator agreed with the values computed in the manner described above. After the threshold data were collected with the initial set of eleven wavelengths, it became apparent that additional stimulus values were required in order to define the spectral sensitivity curves adequately in certain regions. Four additional wavelengths were selected for study on this basis, Thus, a total of fifteen different wavelengths were used in all. Co Foveal Location In accordance with our intention of relating the spectral sensitivity curves to the entoptic macular images, three foveal locations were selected to represent different areas in the entoptic macular images reported by Miles (o3, cit,). The first location, designated zero eccentricity, corresponded to the fixation point and hence to the center of the entoptic plot in all cases, This Location presumably falls within the dark central area of the entoptic macular image described by Miles, corresponding to a pigmented area. The second location was at a point 15 minutes! ti'he nasal side of center. This location was intended to fall within the iftermediary bright annular band of the entoptic macular image described by Miless, corresponding to a pigment-free area. The third location was at a point 43 minutes to the nasal side of center, This location was intended to fall within the outermost dark annular area of the entoptic macular image described by Miles, corresponding again to a pigmented area. Actually, as will be discussed below, the macular entoptic images of all our observers did not conform to the classical pattern reported by Miles but these locations still represented meaningful differences in the macular images. D. PsyChophysical Method The advantages of the temporal forced-choice method of constant stimuli have been described in detail by Blackwell (Ref. 19), who has shown the superior validity and reliability obtained with this methodo With this method, the observers have to correctly identify the interval, of ~four possible intervals, in which the target was presented. Five target intensities are utilized during an experimental session which yield probabilities of correct identification from little more than chance (.25) to nearly unity. A session consists of fifty presentations of each of the five target intensitieso The intensities are presented in blocks of ten each of a given intensity, with the blocks of different intensity randomized in their order from session to session, The probabilities of correct identification are corrected for chance successes from the relation P'=P rp -.25 1 -.25 where p = corrected probability; and p= raw probability. i 7

The University of Michigan * Engineering Research'Institute 2455-10-F Fortunately, these laboratories had available the necessary specialized equipment to utilize this method with convenience for the present study. Automatic stiLmualus presentation and recording equipment were utilized, which have been described elsewhere by Blackwell, Pritchard, and Ohmart (Refo 20)o The psychophysical data obtained with this method represent values of corrected probability, p, for each of five target intensities. These data were analyzed by a variant of the probit analysis reported elsewhere by Kincaid and Blackwell (Refo 20). The analysis fits a normal ogive to the experimental data to satisfy the maximum likelihood criterion. The analysis method yeilds values of the intensity required for p0 =.50, which is designated the threshold intensity. The analysis also provides a measure of the slope of the ogive, and measures of the standard errors of the threshold and the slope as well as a test for goodness of fit of the data to the normal ogive. E. Other General Methodological Considerations The use of an artificial pupil before the eye is mandatory. in order to preclude differences in flux reaching the retina because of variations in the diameter of the natural pupil. In the present case a 6,o4 mnm, pupil was used, representing the maximur m size believed safe under the conditions of the experiment, The desirability of using fixation lights which match the stimulus in wavelength is self-evident. Reference to the published chromatic aberration curves for the human eye (for example Ref, a2)ii?) shows that use of, say, a red fixation point can result in the stimulus-4spot being as much as 3 diopters out of focus in the violet. This degree of blurring is exceedingly serious for the case of small stimuli, and intolerable in the present investigation which attempts to use targets whose angular size approaches the limits of the diffraction pattern. Additional precautions against chromatic error were taken by measuring and correcting for the chromatic aberration of each observer, as described below o F. Plotting of the Entopic Macuar I e We were we ortunate to have obtained from Dr. Walter R. Miles then of Yale University samples of the filters used by him in the study already cited, The observers were seated before a light box which was masked down so that a bright patch of high color temperature white light s$ubtending about 11 degrees was viewed through an aperture in a brow rest which controlled the eye-to-screen distance. A small fixation point was marked in the center of the bright area, and a piece of tracing paper affixed to the surface with tape. The observer was then required to practice alternating the pair of a chromatic and a neutral filter in front of his right-eye until the entoptic macular image was seen and could be reproduced at will, Either of two different sets of filters supplied by Dr. Miles could be used by the observers, the only criterion I 8

The,University of Michigan * Engineering Research- Institute 2455-10-F being maximization of the macular image. After a period of practice, each observer made a penciel sketch of the image directly on the tracing paper so that its angular size could subsequently be determined, The procedure was repeated at a later time to check the consistency of the result, essentially identical plots being obtained0 One observer only (JHT) showed the three areas reported by Miles. A second (DWM) showed the three areas, but there appeared ray-like projection of the central area in both directions on the horizontal axis which extended well into the bright annular area. The remaining observer (NLT) showed no differentiation of the three areas? but sketched a more or less uniform dark spot which was slightly more dense at the center, The detailed discussion of these plots, together with their angular relationships is deferred until a later section~ G. Threshold Measurements Threshold measurements were made with the psychophysical method described above. In a control room remote from the locus of observation an array of devices was located, whose function it was to send appropriate electrical signals to the point where the observing was done, to receive return signals indicating the responses made by the observer, and, after verifying the proper behavior of all phases of the presentation, to record the information relevant to all these on a record card0 A tape-reader device and its associated relay panel cause the presentation of stimuli according to a sequence which provides trial-to-trial randomization of correct answers (i.eo, any one of the four temporal intervals) as well as stimulus intensity randomization between five values but with these held constant for ten presentations at each intensity0 Fifty single presentations constituted a single record card; five cards comprised an experimental session, An accessory counting device recorded the number of correct responses made to each of the five stimulus intensities. The geometry of the observing station is diagrammed in Figure 3o The ribbon filament of the tungsten lamp is imaged at unity magnification on the entrance slit of the monochromator by means of a lens which is stopped down to an effective aperture of f. 4 o4 to avoid introduction of stray light into the monochrom:: 'at 'Sor. Attenuation of the beam is accomplished by the interposition o6f!i Wratten "neutral density" filters of fixed values, as well as by_ similar filters in the filter wheel which are alternately placed in the optical path in accordance with the appropriate electrical impulses from the remote automatic equipment. Between the lens and the slit, at a point of small focus, are two shutters. The first of these is a rotating sector driven by a Telechron motor in synchrony with the remote tape reader which allows passage of the beam for 0.1 second per revolution. The second shutter is a flag which is removed from the optical path only on signal from the remote scheduling equipment1, and for a brief period which will let through only a single 0.1 second pulse. I 9

The University of Michigan * Engineering Research' Institute 2455-10-F The monochromator is a Hilger model D-246 with glass prism; the aperture ratio is f. 4.4~ Calibration against the lines of mercury and sodium was carried out at the outset of the experiment and checked at the conclusion with sodium0 No change in calibration occurredo The settings of the wavelength scale may be presumed accurate to 1 millimicron in the violet, and to 3 millimicrons in the red. Entrance and exit slits were maintained at.225 millimeters throughout the experiment, although the exit slit was stopped down by a tiny circular aperture subtending 1 minute of arc at the eye position, 750 mml. from the exit slit. Fixation lights were introduced by reflection from a separate system, essentially a high-brightness lamp-house which produced the parallel light necessary for proper use of the interference filters. In every case, the brightness of the fixation lights was adjusted by the use of neutral filters to a comfortable level (approximately ten times threshold). A perforated mirror allowed unimpeded passage of rays from the monochromator to the eye, while accomplishing the introduction of the fixation cross images0 Distance from the mirror to the exit slit was identical with distance from the mirror to the fixation lights. The mirror, a first-surface aluminized circle, was mounted on a nicely machined table which permitted rotation around a central vertical axis. By this means the eccentricity of the stimulus could be controlled by moving the reflected image of the fixation crosses in apparent space. A small projection system cast an image of a small incandescent filament onto a scale mounted on one wall of the observing room0 Since part of the path of the projected image involved reflection of the rays from the mirror just mentioned, it was possible to calibrate the 0, 15 and 43 minute positions with considerable accuracy by the mirror s rotation of the beam. At the eye of the observer a stray light shield was mounted and provision made for the introduction of ophthalmic correcting lenses and the artificial pupil. Proper alignment of all components was maintained by the use of rigid "bite boards' bearing previously made molds of the complete dentition of the observers, in dental impression compound. H Radiometric Measurements Total energy emergent at the exit slit of the monochromator at each of the initial wavelengths was measured by means of a thermopile so placed that all rays from the instrument fell within the limits of the blackened receiver. The associated electrical circuit is shown schematically in Figure 4 which indicates also the galvanometer, microammeter, shunt, variable resistances, switch and source of potential. It may be seen from this figure that the galvanxometer is used merely as a sensitive null instrument, and that no dependence upon its absolute sensitivity calibration is necessary. Briefly stated, the radiometric procedure consists in adjustment of the bucking potential in the right arm of the circuit to a value which will just balance the e,.mof produced by the thermopile. The current which will exactly nullify the thermoelectric potential will, of course, return the galvanometer coil to zero deflection, and its magnitude may be read directly in 10

The University of Michigan ~ Engineering Research Institute 2455-10-F microamperes. The thermopile was calibrated at the outset against a standard lamp, care being exercised to employ the identical circuitry to be used for the subsequent measurements, In the present case, it was determined that the thermopile generated one microvolt of potential for each 7,34k microwatts of energy incident upon its surface, Owing to the relative insensitivity of the thermopile, it was necessary to make some compromise with spectral purity for the energy measurements. At atmospheric pressure our thermopile required the use of slit widths as great as 1 nm, at the shortest wavelength, although in the extreme red sufficient microammeter deflection could be obtained using 0.225 mm. slits. Fortunately the energy distribution of tungsten is such that the widest slits were necessary at those wavelengths wheref dispersion of the prism is greatest, and hence purity of the beam was not seriously affected. Energy values determined by this method refer to total monochromator output at a variety of slit widths, Three further conversions are necessary before these values can be restated in terms of energy incident upon the cornea of the observer, First it is necessary to transform all values into terms of equal slit width, in the present case bringing them to 0.225 mm,, a safe minimum value to preclude obstruction of the stimulus aperture, This slit width conversion was accomplished empirically by using the 931 photomultiplier photometer with the photocathode of the multiplier tube clamped rigidly in the position to be occupied by the observer's eyeo Reduction factors from wide to narrow slits were determined from the ratio of narrow-slit to wide-slit readings at each wavelength. At this point the assumption is made that photocell response will change linearly as a function of flux reduction even though this reduction is accomplished in a manner that inevitably modifies the purity, inspection of the published spectral sensitivity curve of the photomultiplier tube (Ref, 21) shows its maximum slope to be such that even our most extreme purity change resulting from slit width reduction amounts to only 3.8 mg half width, and the average change to be a mere 1.2 m., This fact, together with the fact that the monochromator is equipped with symmetrically closing slits so that the center of that minute section of the photocell sensitivity curve important for the wavelength interval in question is approached as the slit jaws are closed, suggests that the assumption of linearity of photocell response is justified. Having applied the slit width conversion factors to the original radiometric data to obtain energy values'at uniform slit width, it now becomes necessary to reduce these values (which were measured using fulllength slits) in accordance with the physical flux reduction imposed by the circular 1 stimulus aperture. For this purpose the photoelectric photometer was again used, A reduction factor was determined from the: ratio of readings with and without the limiting aperture, (It should be mentioned parenthetically at this point that considerable care was taken to center the aperture over the exit slit of the monochromator, I I L 11

The University of Michigan ~ Engineering Research Institute 2455-10-F This was done with the help of the photocell; its maximum response indicating optimal centering. Once achieved, this center position was rigidly maintained throughout the experiment.) By this technique, measuring at 499.0 m,, the factor converting energy at.225 mmo fulllength slits to the 1 minute circular aperture was found to be 4.85 x 10'3o Up to this point in the radiometric procedure we have found values to express total flux passing through the 1 minute aperture. The final conversion required to obtain the quantity of energy reaching the eye is based simply upon the geometric optics of the system. From a knowledge of the slit-to-eye dis$tanace and of the angle at which the rays from the slit diverge, one can readily compute the proportion of the total flux passing through an artificial pupil of known size, and hence accessible to the eye0 In the present case, since the monochromator has an effective aperture of f. 4.4, the emergent cone of light has a half-angle of 6.48 degrees and the whole cone illuminates therefore an approximately circular area of 228 square centimeters0 At the center of this round area is placed the artificial pupil whose diameter is.604 centimeters and whose area is.286 square centimeters0 Strictly speaeking, a direct areal conversion using these values is inadmissable owing to the cosine law of illumitnation. This law, which is commonly stated in the form-: E = cos 0 D expresses the fact that the circular patch is not uniformly bright, but rather that the peripheral illumination is always less than that on the axis of the cone0 In the present instance, however, the peripheral value is within about a third of a percent of the axial one, and the inaccuracy introduced by assuming uniformity of flux across the area is inconsequential, This final reduction factor which permits computation of amounts of flux passing through the artificial pupil was found to be 1.26 x 10-3. Table I shows the results of the radiometric measurements0 For each of the eleven wavelengths, the following information is given: the slit width necessary for suitable galvanometer response, monochromator output in microwatts at.225 mm. s lits, amount of energy at the eye, and an estimate of spectral purity. It should be stated that the half-width purty figures are derived from a crude empirical curve of dispersion, and represent only an approximation to the true values. As indicated above, after collecting threshold data with these eleven wavelengths, it was considered desirable to study four additional wavelengths to assist in the definition of the spectral sensitivity curves, Radiomnetric measurements were not made for these wavelengths Instead, radiometric values were estimated from smoothed curves relating radiant output to wavelength for the wavelengths actually measured. Since actual data points existed within less than 15 mg on either side of each of these four wavelengths, it is considered that this procedure cannot result in errors of appreciable magnitude. 12

T h. I.... In;,.'a.. I AA I;. L.' -;_ -~~~ L 1. _-,ie universi ty ovl viicnigan Cnglneerling Kesearcn, instirure 2455-10-F I. Chromatic Aberration Measurement The oculometer, an instrument of outstanding usefulness in measuring the refractive condition of the eye under normal conditions of accommodation, has been described elsewhere by Ogle (Ref. 22), The observer looks through the artificial pupil at fixation crosses exactly as in the main part of the experiment but with the single difference that the test target is replaced by the reflected image of a bright stigma seen reflected in the half-silvered mirror. Thus the eye simultaneously sees the fixation crosses at 750 mm, in real space as well as a point of light in the optical space created by the mirror and field lens. The lamp-house assembly, which carries the stigma, may be moved along the optic axis by the observer, changing the position of the brtight point in apparent space. Now if the focal length of the field lens i's known, the distance from it to the eye may be made equal to this focal length, and a linear scale may be affixed which can be calibrated directly in diopters. The scale modulus is determined from the dioptric strength of the field lens, and its zero point occurs where the stigma-to-lens distance is equal to the focal length of the lens and the image is at infrinity In use, the observer views the two fixation crosses through the mirror and the reflected image of the stigma adjusted to appear between them0 To obviate the possibility of using the stigma rather than the fixation crosses as an accommodative cue, the oculometer lamp is flashed continuously during the adjustments. Maintaining careful fixation, the job of the observer is to adjust the position of the stigma so that it appears in best focus. At this point the index of the oculometer scale will show the extent of accommodation in diopters, The average of a number of such settings is taken to represent the refractive state of the eye. As ordinarily used for the measurement of spherical error, the stigma as well as the external fixation device are illuminated by white light. Under these conditions the average scale reading is compared with the normal dioptric accommo:dation at a given distance and the discrepancy taken as indicative of spherical refractive erroro If, however, white fixation cros usesd in conjunction with chromatic stigmata, the instrument is converted into a sensitive device for the measurement of chromatic aberration~ The observer adjusts the position of the stigma for best focus as before, but now the readings will deviate fromn the corresponding Lwhite-a;against-whiIte ones by amounts depending upon the chromatic abertation curve of the eye, This system was used in the present study. Approximately monochromatic stigmata were produced by the insertion of interference filters, together with their appropriate cut-off filters, between the oculometer lamp and the stigma. Brightnesses of the fixation crosses and stigma were adjusted for maximum ease of observation. Twenty settings of the oculometer were made by each observer at each of the eleven initial wavelengths under normal observing conditions (6~04 mm, artificial pupil and 750 nmm eye-to-object distance). Observers with 13

The University of Michigan T Engineering Research Institute 2455-10-F normal vision and free from persistent accommodative habits will tend to set the instrument at -1o33 diopters, the theoretical normal refractive state for a viewing distance of 750 mm, Two observers (DWM and JLT) were approximately normal emmetropes, and their readings were therefore closely clustered about -l.33 diopters in the yellow-green. The third observer (NLT) displayed both spherical and astigmatic errors which had to be corrected by placing ophthalmic lenses (+.62 sphere and +1.00 cylinder) just in front of the artificial pupil. (In this position, correcting lenses of relatively weak power have a negligible effect upon the oculometer readings, although rigorously considered there will be some shift of the zero point of the scale as well as disturbance of the modulus and linearity.) Results of the chromatic aberration measurements are plotted in Figure 5 (Values on the ordinate can be converted into terms of refractive error at each wavelength by subtraction of -1.33 diopters from the obtained readings.) It must be emphasized that these values represent the comblined errors resultling from uncorrected spherical as well as chromatic errors For our purposes this seeming contamination-of the curves is of no concern since the subsequent corrections will aim at removal of all errors regardless of cause It was not possible to obtain interference filters with peak wavelengths cortresponding to the four wavelengths added after data on the initial eleven wavelengths had been analyzed0 In each case, the fixation lights of wavelength most nearly equal were used. In no case was there more than a 15 mpl differencec:in wavelength between the test target and the fixation lights. It is of course possible to determine from the data in Figure 5 the magnitude of refractive error to be expected from such a wavelength differenceo In no case was there as much as a 0.1 diopter refractive error introduced in this way. Jo Corrections for Chromatic Aberration Having measured chromatic aberration, it is now possible to provide ophthalmic corrections, provided certain other aspects of the system are taken into account. Crudely, one might simply introduce lenses of powers corresponding to the deviation from -1.33 diopters at every wavelength. A somewhat more elaborate procedure was followed in this study in view of the precision desired for later computations. The first step in this procedure was to approximate the indicated ophthalmic prescription by introducing lenses before the eye. For this purpose the lenses comprising a conventional ophthalmic test kit were used; hence we were limited to approximate corrections which, in the worst case, might fail as much as +0.0265 diopters to meet the desired value. One further complication arose from the ~impossibility of placing the lenses at the usual distance before the eye, and the consequent change in effective power. The usual test lens is calibrated for a specific distance from the eye (13o75 mm, from anterior pole of the cornea in the Tillyer system). Changes from the nominal dioptric values were made to allow for this Ln 14

I The University of Michigan T Engineering 'Research- Institute 2455-10-F the placement of the ophthalmic corrections, With approximate corrections in place, we repeated the oculometnrc measurements in reverse0 That is to say, in place of white fixation lights and chromatic stigmata, the color of the fixation crosses was controlled with interference filters while the stigma remained white. In the absence of correcting lenses this -technique may be expected to yield a chromatic aberration curve which is reciprocally related to the already obtained one. Assuming that perfect correction of chromatic error had been achieved by the use of lenses, the refractive state of the eye would be constant, since chromatic differences in focus would have been neutralized. In the present case, owing to our inability to achieve exact correction, we would expect residual refractive errors, Algebraically added to these errors are undoubtedly errors in refraction arising from accommodative habit together with errors of measurement these deviations from perfect correction combine to yield what we have called the residual error. Table II presents the residual errors at each wavelength for each of the three observers, computed according to the formula. R.E. = aeC - (A + BR + D)] where A is the mean oculometer scale reading with chromatic stigma, B' the mean reading with white stigma and correcting lens, C the corrected lens power in diopters, A the correction for chromatic aberration of the oculometer lens, and D = 2.66 diopters. It is evident from these data that the maximum error occurs at 475 mg for observer JHT. Under this, the worst condition, the retinal image is blurred to a size of approximately 4 minutes of arc. This raises the question of how much this degree of blurring will increase the energy required for visual detection, It may be inferred from the facts of spatial summation that blurring of this magnitude will not influence the detection threshold greatly, since the energy required for detection does not vary as a function of retinal image size to a considerable extent within these limits. It would perhaps seem a simple matter to revise the ophthalmic corrections by a second approximation, to produce zero refractive error. Unfortunately, this is impossible owing to the accommodative habits of the observers, who seem simply to accommodate out the additional lens power and revert to about the same refractive error. This instance of observers failing to accommodate so as to insure adequate ocular focus is of interest ese, It presumably is related to the fact that with chromatic stimull to accommodation, best focus represents an accommodative condition which differs from that usually associated with the ocular vergence required to maintain binocular fusion, Rather than accept an accommodation-vergence relation which departs from the one which the observers usually adopts, the chromatic stimulus is allowed to blur somewhat. III. RESULTS AND DISCUSSION The psychophysical data, representing 24,750 observations, were subjected to the variant of the probit analysis described in Section II i 15

The University of Michigan T Engineering Research Institute 2455-10-F above0 The original probit estimates of the thresholds were multiplied by the radiometric data and corrected for attenuation by the neutral filter used in each case. Owing to the selective transmission of our so-called 'neutral" Wratten filters, it was necessary to apply corrections to their white-light transmission values for each different wavelength target separatelyo This correction was applied for each of the fixed filters as well as for the psychophy$sical filter~ The corrections were based upon spectrophotometric data for each filter involvedo These manipulations result in threshold values in terms of total energy incident on the cornea, within the area of the pupillary aperture. Sensitivity values are represented by the inverse of these values. Sensitivity values are presented for each observer in Tables III, IV, and V for each of the retinal locations studied. In a number of instances, the values represent averages of two experimental sessions0 The data are plotted in Figures 6-8 for observer NLT, Figures 9-11 for observer JHT, and Figures 1214 for observer DWMo The scales used for these figures are identical with those used in presenting the Hsia and Graham data in Figure 1 and the Crozier data in Figure 2P Consider first the data for observer NLT, a color-normal female with a rather undifferentiated entoptic macular image. Although there are not a great many data points, it seems obvious that the spectral sensitivity curves are markedly different in the three locations. The standard errors of the sensitivity measures are well represented by the size of the data points, so that the manifest-i.differences are highly significant from a statistical point of view, The curve for zero eccentricity is the smoothest, the curve for 15 minute eccentricity is next most smooth, and the curve for 43 minute eccentricity is most "irregular"~ There is a suggestion of bimodality in the data for the 43 minute eccentricity, with peaks at 555 and about 510 mto The data for observer JHT, the color-normal male with the classical entoptic macular image, are entirely differento The curve for 43 minute eccentricity is the smoothest, the curve for zero eccentricity next most smooth, and the curve for 15 minute eccentricity is most "irregularto0 There is a definite suggestion of bimodality in all three curves0 The peaks at zero eccentricity occur at about 525 and 575 mwo those at 15 minutes eccentricity occur at about 480 and 540 m4; and those at 43 minutes eccentricity occur at about 460 and 555 mio The data for observer DWM, the deuteranopic male with the nearly classical entoptic macular image, are still differento The curve for zero eccentricity seems to be bimodal as does the curve for 43 minutes eccentricity0 All the curves appear to be smooth but this may well be due to the fact that there are comparatively few data points. The curves for zero and 43 minute eccentricity appear to be bimodal with peaks at about 550 and 610 mg in each case.o The curve for' 15 minute eccentricity has a single peak at about 550 mgo 16

The University of Michigan E Engineering Researchl Institute 2455-10-F Apparently, the most that can be concluded from these results is that the spectral sensitivity curve for point-size targets is by no means the same for the two different color-normal eyes and the deuteranopic eye., nor is the same.for different retinal-locations in the same eye. Thus, the spectral sensitivity curve measured for small retinal areas is highly idlocyncratic and there seems to be no simple way to characterize it. It is true that insufficient data exist to define these curves with complete adequacy; thus, there could be small irregularities of the ty"e shown by the Crozier data which would have been missed. However, the differences among the curves are extremely large as examination of Tables III, IV and V will reveal. There can be little doubt that these differences are real. It seems reasonable to interpret the differences among the spectral sensitivity curves as the result of different numbers of cones of various types in the different retinal locations studied~ From this point of view, the peaks in the various curves presumably represent various types of cones, Peaks occur at 460, 480, 510, 5259 540, 555, and 610 m Thus, these considerations would suggest the existence of seven cone photoreceptor systems. These peaks do not agree at all well with the peaks shown in the Crozier data, Unquestilonably further measurements are needed to provide additional data points to define these peaks. As noted in Section II, it was our intent to evaluate the extent to which variations in the spectral sensitivity curves from point to point correspond to differentiations in the entoptic macular images reported by the various observerso The data in Figures 614 have been replotted in Figures 15, 16, and 17 to facilitate this evaluation. The graphs for different locations have been replotted so as to agree at 575 mg by shifting the curves for zero and 15 minute eccentricity to match the curve for 43 minute eccentricity at that wavelength. Adjustment to agreement at this point Was selected since there is no appreciable absorption of wavelengths this long or longer by the macular pigmento Let us examine the relations shown in these figures in relation to the characteristics of the entoptic macular images reported by these observers.% As noted in Section II, observer NLT reported an entoptic macular image in the positive phase consisting of a uniformly dark center and a somewhat less dark surrounding area. The central area of the greatest darkness had a radius of 32 minutes$ the outer zone of lesser darkness had an outer radius of 61 minutes. The pigment interpretation of the macular image would predict that the sensitivity for the blue end of the spectrum would be equivalent for zero and 15 minutes eccentricity but would be highe for the 43 minute eccentricity The data presented in Figure 15 agree with this predicti'on exactly. The macular image for observer JHT in the positive phase consisted of a small dark area of radius 11 minutes, surrounded by a bright annulus extending out to a radius of 28 minutes, with an outermost dark annulus extending out to a radius of 82 minutes. The pigment interpretation of the macular image leads to the prediction that blue-wavelength 17

The University of Michigan T Engineering Research, Institute 2455-10-F sensitivity for zero and 43 minutes eccentricity will be the same whereas the sensitivity for the.15 minute eccentricity will be highero The data presented in Figure 16 agree with this prediction also0 The macular image fors observer DWM (a deuteranope) in the positive phase consisted of a central dark area of radiusa 12 minutes9 a dark surrounding annulus of radius 33 minutes, surrounded by a dark annulus of radius 61 minutes. However, there was an extension of the central dark area out entirely through the bright annular area on both sides along a generally horizontal meridian. Thus, we would expect to find equal sensitivity in the blue wavelength for all retinal locations. The data presented ina Flgure 17 agree generally with this prediction, at least in comparison with the differences in sensitivity found for the other two observers. These data certainly lend considerable support to the hypothesis that blue wavelength spectral sensitivity varies from point to point in agreement with the location of dark and bright zones in the entoptic macular image. Since there is no reason to believe that "smallsubtense tritanopia" would affect the sensitivity curves, this correspondence supports the view that the macular image is produced by differ. ences in pigment density rather than differences in receptor populations as Walls (Ref. 23) has suggested~ 18

The University of Michigan T Engineering Research, Institute 2455-10-F REFERENCES 1. Hsia, Y, and Graham, C. l. "Spectral Sensitivity of the Cones in the Dark Adapted Human Eye"0 Proc. Nat. Acad. Sci., 8 80-85 (1952) 2 0 Sperling, H. G. "Some Comparisons Among Spectral Sensitivity Data Obtained in Different Retinal Locations and with Two Sizes of Foveal Stimulus". Ph.Do Dissertation, Columbia University, Univ0 Microfilms No0 6711, 23 p (1953) 3 Crozier, W. J. "On the Visibility of Radiation at the Human Fovea"0 J. Geno Physiol., I 3, 87-136 (1950) 4. Polyak, S. The Retina. Chicago: Univ. Chcago Press, 607 p. (1941) 5 Buzzi, F. "Nuove sperienze fatte sull'occhio umano". In Opuscoli scelti sulle scienze e sulle arti v. 5 Milano: G. Marelli ( 1782F --- -.. 6. 0 1Gulstrand, A. "Die Farbe der Maciula centralis retinae" v. Graefes Archh Opt 0 6_2 378 (1905) %7 4MtMaxwell, J. C. "On the Unequal Sensibility of the Foramen Centrale to Light of Different Colours"0 Athenaeum, No0 1505, 1093 (1856) 8. Helmholtz, H. vonTreatise on physiological optis.(Trans. J.P.C. Southall) v,. 2, Menasha; Ban'ta, 1924+ ) 9. no Miles, WG Ro "Entopic Plotting of the Macular Area". Proc. Army_-Nvy-OSRD Vis. Commo0 14th meeting, 13-26 (1945) 10. Wald, G. "Human Vision and the Spectrum"' Science, 101o, 653-658 (1945) 11. Konr"ag,A. In So B. Akado Wiss. Ber l. 577 (1894) 12. Willmer, E. N. "Colour Vision in the Central 2Fovea"0 Documenta Ophthalmologica, 3, 194-213 (1949) 13 Wright, W. D, Researches on Normal and Defective Colour Vision. St Louis: Moxby, 383 p. (1947) 14. Hartridge, H0 "The Colour of Small Objects", J. Physiol, 104, 2 (1946) I 19

The University of Michigan E Engineering Research, Institute 2455-10-F REFERENCES (cont d) 15. Wright, W. D. "The Characteristics of Tritanopia". J. 0pto Soc. Amer., 4, 509-521 (1952) 16. Stiles, W. S. "Increment Thresholds and the Mechanisms of Colour Vision", Documenta ophthalmologica _ 138-163 (1949) 17. Skogland, J3, Fo Tables of spectral energy distribution',.~anld relative brightness from spectrpphotometric data, Washington: N.B.S. Miscellaneous publication 86 (1929 18. Hardy, A. C. Handbook of Colorimetry. Cambridgeo The Technology Press, -p. 16s 19. Blackwell, H. R, Psyghophysical Thresholds: Experimental Studies of Methods of Measurement. University of Michigan, Engineering Research Bulletin No- 36, 227 p. (1953) University of Michigan, Engineering Research Institute Report 2144-283-T (in press) 21. Zworkyin, V. K* and Ramberg, E. G. Photoelectricit and Its Applicatio, New York: Wiley (1949 22. Ogle, K. N. Notes on the Stigmatoscopy Method of Refraction. Unpublished paper. 12 po (1947) 23. Walls, G. L. and Mathews, R. W. "New Means of Studying Color Blindness and Normal Foveal Color Vision: With Some Results and Their Genetical Implications" Univ. of Calif. Publications in Psch 7, 1-172 (1952) 24. Wald, G. and Griffin, D. R. "The Change in Refractive?ower of the Human Eye in Dim and Bright LightiY JO Ot.o Soc. Amer,4 L, 321-336 (1947) -- 20

4.' 4.m Q Lo ce I00 W'' a 4l f 00 c cc m to Q,:F SUMMARY OF Wavelength (mI) Slit Width ((mn) Microamperes with Re = 0=,4 oh, Microvolts Microwatts 2455-10-F TABLE I RADIOMETRIC EASUREMENTS AND THEIR CONVERSION TO 423,5 459.0 475.0 499.0 526.5 553.0 1.000 0.750 0.750 0.500 0.375 0.375 UNITS OF 571.0 0.250 0o824 6065 o.~4E 6 ~05 606.0 0,225 3.08 1.23 9.03 AT" THE 635.0 0.225 3.28 9 31 9.62 669o.0 0. 225 4.29 1.72 12.60 1.o89 0.756 5 -55 2.87 1.15 8 44 698.0 0.225 5 o13 2 05 15 04 4.04 1.62 11.89 2.90 1.o16 8.51 2.51 1 00 7.34 3 80 1.52 11 16 Empirical Conversion to.225 mmn. slit Microwatts with.225 mm. slit Microwatts with 1 minute aperture Microwatts through 6.04 mmn, pupil Estimated Half -width purity (mv) ~ 0393.0682 O.871.188 ~357 o343.830.775 1 o000 1.000 1.000.218.576 1 036 1,600 2.62 3.83 5.02 7 00 9 62 12.60 15.04 1,06 x 2.79 x 1o-3 10'3 1.34 x 3,52 x lo-6 10o6 03 1-3 5.02 x 7.76 x lo-3 10-3 1.2 x 1086 x 243 x 34: x 4.6 x 6.11 x 729 x 10o" 10: 2 lo2 1 0 0 10-2 io 2 6.3 x 9.78 x 160 x 2,34 x 3.06 x 4. 28 x 5.88 x:10o 106 1io-5 10o5 io5 l0o5 105 7.70 x 9.19 x 10o5 10-5 1.7 2,3 2 7 3 3 3 07 4,5 2 5 3.0 3 5 i..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The Universitv of Michigan * Engineering Research Institute 2455-10-F TABLE II RESIDUAL ERRORS AFTER CORRECTION FOR CHROMATIC ABERRATION (diopters) Observer Wavelength JHT NLT DWM 423 o5 - 342 -.001 +.900 459 o — 354 -~009 +.281 475 Q0 -.568 -.180 +o018 499 o0 -, 318 - 132 +.3 12 526 5 - ~377 + 139 +.214 553 o0 -.201 -o026 +.282 571.0 -.223 -.023 +.190 60600 -154 t o229 +.301 635.0 -.174.t179 - o19 669.0 -.220 o.212 +,o90 698.0 -o093 - 037 +,093 22

I The University of Michigan ~ Engineering Research Institute 2455-10-F TABLE III SPECTRAL SENSITIVITY FOR OBSERVER NLT; Inverse Threshold Radiosity (microwatts xlO6) Wave length 423 o5 459 o0 475. o 499.0 514.0 526 5 540 00 553.0 571 0 606 >0 635o0 669 o0 698.0 5%4, Eccentri 15 minutes 0 5 05 1. o20 2.48 6 O41 2 05 3.40 3 o56 3.23 2 o16 5.65 8.93 1.44 x 106 x 107 x loQ x 10 x 108 x 108 x 108 x 108 x 108 x 1o7 x 106 x 106 6 33 9.57 2.67 5.00 1o76 x 3.12 x 3.69 x 2.78 x 1,52 x 4.88 x 5.18 x 9?80 x lo8 8 10Q 108 108 107 106,105 x 106 x 106 x 107 x 107 Lcity 43 minutes 3.77 X 106 2 39 x lo7 3.22 x 107 8.93 x lo7 9 10 x 107 1.09 x 108 1.75 x 108 2,68 x 108 1.57 x 108 1.00 x 108 3.18 x 107 7.63 x 106 8.15 x 105 1 23

The University of Michigan * Engineering Research Institute 2455-10-F TABLE IV SPECTRAL SENSITIVITY FOR OBSERVER JHT Inverse Threshold Radiosity (Microwatts x106) Wave length Eccentricity n.... 423.5 459 oO0 467 o 475.0 487 00 499.o0 514.0 526 5 540.0 553 00 571 oO0 606 0 635 o.0 669,0 698 0:"~o 0 15 minutes ", 6.94 1 34 x 106 x 107 7 o04 1 39 x 106 x 10 to6;to'1.43. minutes 2.48 x 106 6 9.09 x 10 8o44 x 106 8~70 x 10o6 2o14 x 107 3.40 3 44 4 o27 5.99 x 107 3.26 2,87 2 o81 3 88 2 ~27 1.70 1 o39 2.04 8 10 x 108 x 108 x 10 8 x 107 x 106 1 50 2,15 1 o01 5.52 2 o00 7 o14 1o48 x 1Q7 x lo7B x 108 x 108 X 108 x 1o? 8 x 107 x 106 x 106 x 10 3 5-0 x 5.74 x 8.93 x 1,17 x 1.49 x 1,51 x 6090 x 1:145 x 4,81 x 5.21 x 10o 107 107 108 108 108 107 o107 106 105

The University of Michigan * Engineering Research Institute 2455- 10F TABLE V SPECTRAL SENSITIVITY FOR OBSERVER DWMI Inverse Threshold Radiosity (microwatts x10?) Wave length Eccentricity 423 o5 4i59.0 475.0.499.0 526 o5 553 o0 571 o0 606. o0 63590 669.o0 698.0 0 8093 x 106 1 o 10 x 107 2 05 x 107 7.63 x 107 3oll x 108 3.85 x lo8 3.03 x 108 3o18 x 0o8 2.54 x o10 4102 x a07 2o42.x 106 15 minutes 5.92 x 106 1o08 x 1i0 2.62 x 107 3.88 x o17 2o21 x 108 3o51 x 108 2.29 x 108 1.58 x 108 1o34 x o10 1o49 x 106 7.58 -10 2.20 x 107 7 2 19 x 107 5~43 x 107 1.64 x 108 2038 x LO8 1.82 x 108 2,86 x 108 7 4o72 x 107 1o03 x 107 9.26 x 105 25

U1 OD rG 1599 to 1 1 1 1 1 1 I 4 0.6 Ite 0.02 Z4 w 0.006: oo.. 8 4 0.8.-._I 2 -,.o 0.2 0.1.... IW () hi, v.,0 A' I0\ 0.02 K 0.006 OD04L)O~~~~~\ I OOl _ - 00046 0 0 0.04 0.02 0..01 0.004 - 4_ _ 0.00 0.002 B0.001...6 400 450 500 550 600 )( MLLIMiICRONS) 650 700 400 45 X 500 550 600 X(MILLIMIcRONS) 650 700 Fig. 1. Data of Hsia and Graham. Fig. 2. Data of Crozier.

GA- 19Gb ) 1000 WATT TUNGSTEN LAMP (RIBBON FILAMENT) FIXED FILTERS PSYCHOPHYSICAL FILTERS LENS ROTATING SHUTTER FLAG SHUTTER - ARTIFICIAL PUPIL EYE Fig. 3. Optical schematic drawing of the stimulus presentation equipment. 0-0K) MICROAMP r - v0 THERMOPILE.04 OHM SWITCH I.5 v. loo))oon Sz MEG n I~po Fig. 4. Circuit for the radiometric equipment. -0.4 -0.6 c -0.6 -1.0 F -1.2 z - -1.6 tt- -1.0 0 -2.0 0 - -2.2 -2.4 -2.6 0 9, U (:] a o a 0 o o a ^ o A o 8 0 Ub. A,LT O Jlrr 400 440 480 520 560 600 640 60 WAVELENGTH (m/) Fig. 5. Chromatic aberration measurements. 27

AP ~ S c I. c IN C C c ) I' C.C A a LA C 2 X3 1189 1000 600. 400 2- 0 -_ _ _ _ _ 200 C40 20 Li LO OBSERVER: NLT o' t000 800 200..... -100 i...... 40 to 60 1-1 < 4.2 400 450 500 550 600 650 700 X (MLLUIMIGRONS) IM 400 450 500 550 600 X (MILLIMICRONS) U50 - 7U Fig. 6. Threshold data. Fi g. 7. Threshold data.

&A 187 tw -o too,, 40 to 4 \ OBSERVER: NLT 43'.2 ____ ____~ o a1 Ro I x 0 2 w I 1800 - -_.. SW 600.: U- -7 - 400 goo 40.2 OBSERVER: JHT OF.. _....-,,b *YW q~~~~~~~.j X (MILUMICRONS) 500~~~~~~~~~~~~~~~~~,........~~~~~10 400 450 500 550 60b X (MILLIMICRONS) 650 700 I w 4mV 500 550(Go 60 50M X (MILLIMICROIN) 700 Fig. 8. Threshold data. Fig. 9. Threshold data.

r. a93 CA 1194 200 I.40 20 a 6, I I I I I I 0 b 'I x 4 3 I. IC c 4 (I a 'a ) 400 820_ 40o ____ 20 10 8 d2.6 ___,. ~~~.4a~~~~~~43 0 A I I I I I OBSERVER: 15' l JHT -- 400 450 500 550 600 X (MILLIMICRONS) 650 700 400 450 500 550 600 X (MILLIMICRONS) 650 700 Fig. 10. Threshold data. Fig. 11. Threshold data.

01.192 800 200 60 40 2.0 00' 2, OBSERVER: OWM 0O 400 450 500 550 600 650 700 X (MILLIMICRONS) Fig. 12. Threshold data.! 3.I I 4 ~ 1 k I 1000.., Sooc'600 400. 200..100, 80 o 6 r. 40, n 20 E 0 6 OBSERVER: DWM 15'.2 Ae~ft Af% O-W! IWO 1400 500 550. 600 650 700 X (MILLIMIRONS) Fig. 13. Threshold data.

0 x cn 4 0 Ar 0 \Iu:;A 1190 200. 40 20 10 L / OBSERVER: DWM 43' 2Z.-, I00 o.~~~~~~~~~~~100 ~ 60 0 20 4r6o l e, - wt a:C a O 2,. I0 a:. m2 A2 AIe 400 450 00 550 600 X (MILLIMICRONS) 650 719 X (MILLIMIGRONS) 650 Fig. 14. Threshold data. Fig. 15. Adjusted threshold data.

3A 3202 600 Soo 400 200 x 40 0' I s F \ ~.2 I. > 4 I OBSERVER: JHT A. a: i kJ Z 2 a U II z /1203 600 so.~ 80 00 _ _ r I _ _ 4C a I OBSERVER: DWM 2'A....... 400 450 500 550 600 X (MILLIMICRONS) 650 700 V0 45o 0500 550 600 X (WLLIMICRONS) 650 7W0 Fig. 16. Adjusted threshold data. Fig. 17. Adjusted threshold data.

UNIVERSITY OF MICHIGAN 3 111015 02514 78541111111 3 9015 02514 7854