:N THE ' N o. I, a n .- Knut- .!‘ p‘. ‘ ! n... e P. .. «3.! .r. . . I :31 fish a. L i! “‘54 D'I ‘86- it s5 3.} {‘3 f ’21. s ‘2’ 2' ‘4’ ,2. 33’- ii: I $1 4U. u Id a» MEL P! H}. .. Jun T‘ ‘ f . ‘ .1 _ «.3 - W at}. "‘1": 1": '.' . U. 1 x' 1‘ ‘J ‘3 U .‘ ." 1.“: “‘1 ‘3... CL w...“ Md u u.“ II. t .I.. " .5. .l .....r v. 4. aytt PH “J um. ol the 0‘ 2b. «0.59 A”. us ”In“ AL. '0" u I. chi-88 ‘0“ h ; V“... {at t.‘ v. ... “‘m ‘vfi U.» .nd 3 I went W...“ "a; 13" .. . “hm fl.‘ :3: \ ‘_ I f: 5. 3915915 LIBRARY Michigan Stan: University ”H”; {‘1 Ti 3' - ‘55“: ‘3’“ b 9 “1“" Mn" :‘=‘--~'* ““33 F k Q 1.; ' U5 u 2 .; ..w ”x! i L“ t ABSTRACT THE EFFECT OF THEIPURKINJZ SHIFT ON THE B? GHTNESS OF HIGHHAY SIGNS AT NIGHT by John P. Fry, Jr. There is a need for increased visibility of highway signs at night. The Purkinje phenomenon, the shift in wavelength lightness sensitivity of the eye as the eye becomes dark-adapted, is known to influence perceived brightness of specific hues differently. From this it was hypothesized that the relative visibility of green hued signs in daylight would change under mesopic - incandescent light viewing conditions so that one (or more) green hue(s) would be rel- atively more visible (in terms of lightness) than other green hues at night on the highway. A theoretical psychOphysical analysis of this hypothesis was made on an initial set of five green hues. The calculations showed only slight relative lightness changes between daylight and meSOpic - incandescent viewing conditions. In view of uncertain psychophys— ical assumptions present in the mesopic calculations, it was decided to experimentally test the hypothesis. Lightness estimations of twentyafive Hunsell green hues, compris- ing four sets of hues, were made by comparison with Munsell neutral grays. The hues of each set were equally light and equally saturated in daylight. The initial set included the present Interstate green hue and hues on either side of it. The remaining sets were one step in lightness or saturation on either side of the initial set. John P. Fry, Jr. The lightness estimating took place under incandescent light at a mesopic level typical of that reflected by signs on the high- way at night. Lightness judgment training was given to the twenty- five female subjects. They were dark—adapted and viewed the hues perifoveally, conditions necessary for the Purkinje shift. Light- ness estimations were also obtained under daylight conditions for control purposes. Under mesopic - incandescent viewing conditions, the Purkinje shift was shown to have little effect on the relative visibility of green hued stinuli. Therefore the experimental results confirmed the theoretical psychophysical calculations. In view of the repre— sentative sample of hues viewed, the results, in general, also sub- stantiate the choice of the Interstate green hue presently in use on the highway. Of interest was the finding that precise lightness specifica- tion is relative to the means of measurement, i.e., theoretical psychophysical calculation, photometric measurement, and psycho- physical lightness judgments were shown to give slightly different results. i/’—\ Approved: id {/2941 M fizée» Committee Chairman Date: 22; / ,5: M47 Thesis Committee: Theodore W. Forbes, Chairman Terrence M. Allen Richard D. Hart THE EFFECT OF THE PURKIl-IJE SHIFT ON THE BRIGI‘II‘NESS OF HIGEJAY SIGI—IS AT NIGHT By John P. Fry, Jr. A THE IS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MA ' R OF ARTS U FJ Ln Department of Psychology 1967 To the memory of John P. Fry, Sr. ii ACKNOWLEDCEIEZJ T3 The author wishes to express his appreciation to Dr. T. W. Forbes, chairman of his committee and.major professor, for his guidance in this research and to Drs. T. M. Allen and R. D. Hart, whose suggestions were of great value. Thanks are also extended to R. F. Pain and R. P. Joyce, fellow graduate students, for their encouragement and help, and to Miss Pam Parsons, the author's fiancee, for her morale build- ing contributions. iii TABLE 0 F CON TEL; TS DmICAT—IONOOOOOOOOOOCO..'...OOOOOOOOOOOOOO00.0.0000... ACKNOWLEDEJENTSOO00.00000...000.00.00.00...00.0.0.0... LIST OF TABLESOOOOOOOOOCOO00.00.000.00...0.0.0.0000... LIST OF FIGURESOOOCOOOOOOO0.00.0000...0.0.0.0....0.... LIST OF APPENDICESOOOOO0.00......'...OOOOOOOOOOOOOOOOO INTRODUCTION.0.00000000000000...00000000000000.0000... The Purkinje Shiftooooooooooooooo0000000000.0.0.0 current ResearChcoccocooceocooo00.000000000000000 The PrOblemoo0.00000000000000000000000000000.0000 PsyChOthSical Consideration8....o............... Initial Stimuliooooooe000000.0000.000000000000000 Theoretical C31C111ation50000000000.... 0.000000... RGSQltS‘Of Theoretical CalculationS.............. hBILODoa0000000000000...00000000.000000000000000000000 Stimuli.coco...000.000.000.000.000.000.0000.coco. Neutral scaleoooooooooooooo000000000000...o.00000 ApparatuS.........................o.............. subjeCtSoooocoooocooooceococcocoo-00000000000000. ProcedurGOOOOOOOOOOOOOO.0..‘...OOOOOOOCOOOCOOOOOO Control MeasurementsooCocooocoooooooccecooo.coco. EXperiflental DeSignccococoo00000000000000.0000... RESULTS AND DISCUSSION................................ HueSOCOOOOOCOOOOOOOOOOO..'...COOOIOOOOOOOOO..'... subjGCtSOOOOOOCOOCOOOOO..’...OOOOOOOOOOOOCOOOO... CONCLUSION...0.0.0.0000...0.0.0.000.....OOOOOOOOOOOOOO BIBLIOGRAPHYOOOOOOOOO0.00000000000000000000000.0.00... APPENDIX Aoooooooooooooooooooo90.000000000000000000000 APPENDIX 8......OOOOOOOOOOOOOOOOOOOOOOO00.00.000.00... iv. iii vi vii \O'\)\J 42"») N HH HH HH 0 I Ht—H—s 03‘s) Q U H \O H \O 20 20 Q 2') 30 ‘J.) \r‘. kn) \O Table t. [0 LIST OF TABLES Psychologically and psychophysically determined light— ness values for five Munsell hue samples.............. Lightness and saturation values for four sets of Nun— sell hues...‘...OOOOOOOOO..00.....'...OOOOOOOOOOOOOOOO Lightness estimation means and variances across sub- jG‘CtS for the mesopic adaptation level................ Lightness estimation means and variances across sub- jeCtS for the phOtOpiC adaptation level—00000000000000. Analyses of variance for the mesopic adaptation level. Analyses of variance for the photopic adaptation level Results of‘g posteriori significance tests on ordered pairs Of meanSOOOOOOOOOOOOOOOOO0........OOCOOOOOOOOOOO Photometric measurements of the Munsell neutral gray smpleSOOOOOOOOOOO00.00.0000.....'...OOCOOCOOOOOOOOOOO Photometric measurements of the Munsell hue samples... V. k0 C'\ \ cm 10 A1 A2 Psychologically and psychophysically detennined light- ness values for five I-hmsell hue smplesooococoooccooo Apparatusooooooooooooocococooooooo00000000000000.0000. Side view of the laboratory set-up.................... r1 - . 10p VISIT Of apparatus”..........o.................... I'IindOI‘! dimensions..."................................ Lightness estimation means for set I h es............. Lightness esthnation means for set II hues and.Munsell renotation lightness values........................... Lightness esthnation means for set III hues........... Lightness esthnation means for set IV hues............ Lightness estimation means for set I, II, and IV hues. hotometric measurements of the Munsell hue samples... Comparison of the photometrically measured luminous reflectances of the Munsell neutral gray samples with 1.») Standardsoooooooooooooooooooooooooccoco-cocooooooo Page 10 H H H O\ (A C\ N N a? JV LIST OF APPSNDICES PhOtOT‘Letrj-IC I‘eadil‘lgsooooooooooooooococo...ocoocoo PrOtOCOl - RSI/11101610115000.0000...00000000000000. vii. INTRODUCTION Forbes 23 El , having reviewed research literature of factors affecting highway sign effectiveness, conclude that the greatest need for research lies in the area of attention gaining factors. The need for attention gaining characteristics is related to the fact that although signs are well within threshold visibility dis- tances, they are still not always seen by drivers. Any increase in the "target value" of signs, such as results from.increased bright- ness, then, can be viewed as a serious research goal. The Purkinje Shift Fbr such reasons as effective detection, attention value, and safety considerations, highway markers and signs have been designed in accordance with human vision factors. One factor, the Purkinje shift, although well-known, has received little attention in direct application to highway vision research. Under equal energy conditions, the human eye shifts in.maxi- mum sensitivity from.wave1ength 555 mp, during full light-adaptation (photopic or cone vision), to wavelength 510 mp, during full dark- adaptation (scotopic or red vision). This shift is called the 13 10 Purkinje phenomenon. Lagrand points out that: In 1825, the celebrated Czech physician Purkinje discovered that surfaces of sign-posts, painted in a blue and red which had about the same luminance in day; light, appeared very different at dawn, the blue being brighter than the red. This well-known and easily ob- served effect shows that the relative luminous efficiency factor increases for short wave-lengths as compared with long ones, when the darkpadapted eye observes fields of low'luminance. 1. 2. When the adaptation state of the eye falls between the fully dark-adapted or light—adapted extremes, it is called mesopic vision. This region is from about 3 x 10"]+ to 3 ftL luminance.9 Since high— way visibility researchers such as Richards16 agree that night high— way'luminances run from 3 x 10'3 to 4 ftL, it is reasonable to assert that the Purkinje shift may have an effect on the visibility of high- way signs at night. However it must be noted that the Purkinje shift does not occur unless the stimuli are observed under perivaeal or averted vision.13 This means that a given highway sign must lie 1.5 degrees outside the driver's foveal fixation point, in order for the Purkinje shift to be relevant. Current Research Recently Forbes gt gls have been investigating the relation be. tween lightness levels of Interstate green highway signs and differ- ent backgrounds. In general dark, highly saturated, green signs were seen best under simulated day conditions. It is heped that specific lightness levels for specific highway sign locations and conditions cand eventually be determined. At the same time, under simulated night conditions, bright green signs but not the lightest green signs were seen best. There- fore, given a lightness level which is correct for daytime illumina; tion conditions, it would be desirable if that lightness of green 'were also as bright as possible under night illumination conditions. Fer any given lightness level, there exists in daylight a set of equally light, equally saturated stimuli varying only in hue. However, due to the Purkinje shift, changes in their brightness 3. characteristics should result when the same set is viewed under night illumination conditions. The Problem The three conditions necessary and sufficient for the Purkinje shift to occur are: 1) a low illumination level, 2) sufficient time fer dark adaptation of the eye, and 3) stimuli seen perifoveally. Identical conditions do exist at night on the highway. Consequent- ly, a set of signs of different hue, equally light and equally sat— urated under daylight illumination conditions, should change under Purkinje shift conditions to the extent that one or more of the signs now appears lighter (brighter) than the others. The problem, then, was to simulate night highway sign viewing conditions as closely as possible in the laboratory and to determine experimentally what hue (or hues) of a set of green hues is (are) lightest (brightest) to the human eye. Since the best daylight lightness level may vary, a number of sets of hues, each set of a constant lightness value and of a constant saturation level were necessary in order to generate results generally applicable to the highway situation. The important.variables in this study were well-known; it was the purpose of this study to determine to what extent the known re- lationships exist in the region of values associated with night driving. First, however, due to the great deal of psychOphysical work that has been done relative to the problem at hand, it was felt relevant to theoretically calculate, if possible, the lightness val- ues involved. PsychOphysica; Considerations Since human eyes vary in their sensitivity to light, it has been necessary to establish generally agreed upon standard curves of lum— inosity, as seen by the so-called standard observer, for use in vision research. A luminosity curve shows as ordinates the relative amounts of energy required at each wavelength,7\ , to evoke visual sensations of equal brightness. 1'3 In general, luminosity functions show the rel- ative spectral sensitivity of the eye for brightness. Investigation at photOpic and scotopic levels of illmnination has yielded two well-known standard luminosity curves, yhand 5;, re— spectively. Comparison of these two curves shows the maadmum change in the relative spectral sensitivity of the eye to brightness as the eye is brought from a fully light-adapted state to a fully dark-adap— ted state. Since the scotopic luminosity curve has been determined by subjects observing at angles of not less than 5 degrees from the fovea, the two curves thus represent the Purkinje shift in a general sense.13 However, these two curves are relevant only to light emitted from a source equal in energy emittance across all wavelengths (day- light is close to this). menever light is anitted from an unequal energy emitting source (such as an incancescent bulb) or is reflect- ed from a surface (such as a highway sign), such factors must be accounted for in any psychophysical calculations of what the eye perceives as lightness. Therefore, in order to theoretically determine how the Purlcin- je shift affects visual perception for a specific instance, the spec- tral energy distribution of the radiant energy of the light source, 5. PA, and the spectral reflectance of the surface,f>7\ , are required. With such infomation luminous reflectance, r, can be quantified by the following equation: £95?) P“ d>\ r: ‘ 5% PA (17‘ 0 Ph = reflectance of a surface for wavelength,?\, (in percent of pure reflectance); (1) y = relative spectral sensitivity of the eye for each wave- length,?\, (dimensionless); P,‘ = spectral energy distribution of wavelengthflx , for the source, measured in spectral power per micron (arbitrary units); and r = luminous reflectance. Since luminous reflectance is the ratio of two integrals, in both of which 377,13}, appears, this product may be expressed in any units. 13 luminous reflectance may be determined via equation (1) or from measuring the area under the curve of luminous enittance resul- ting from the multiplication of the luminosity curve, the energy distribution curve of the light source, and the spectral reflectance curve of the surface. Dnninous reflectance may also be measured in foot—Lamberts by a Pritchard photometer. In both cases luminous re— flectance is a theoretical quantitative value derived from psycho- physical research. Such values are supposed to be identical to what the eye perceives in qualitative psychophysical eiqieriments as light— ness (brightness). NOTE: In order to provide clarity, throughout this thesis, the term, psychophysical, will be used for theoretical quanti- 6. tative lightness calculations (as determined from the so— called standard observer) or photometric readings, while the term, psychological, will be used for perceived qualitative lightness estimations (as observed in a specific psychoohys— ical experiment). Interest at the moment is psychOphysical. Using the spectral energy distribution curve for an incandes- cent tungsten light source of 3,0250 Kelvin, representing automotive headlamp illumination, and by multiplying each wavelength in turn by each wavelength of the standard luminosity curves, Richards has shown how the standard luminosity curves,'yh_and'y{, are modified in the change from daylight to incandescent light.16 TOgether with 19 tables of relative (interpolated) spectral luminosity Weaver's curves for scotopic to photopic luminance levels and a 3,0250 Kelvin light source, Richards has also been able to plot curves showing the shift in relative sensitivity of the eye (the Purkinje shift) as a fhnction of luminance levels typical of those encountered in night highway driving. Ektending this analysis to include spectral reflec- tances of surfaces, would permit calculation of luminous reflectance, r, per equation (1). Since theoretical psychophysical data relating to the problem does exist, it was appropriate to calculate the expected lightness (luminous reflectance) differences due to the Purkinje shift for the following light sources and adaptation levels: 1. daylight - photopic 2. incandescent - mesopic 3. incandescent - scot0pic 7. Initial Stimuli Spectral reflectance curves for surfaces similar to Interstate green highway signs were obtained from the Munsell system.11 no stan— dard green background presently adopted by the Bureau of Public Roads for use in Interstate directional and informational control is ISCC~ NE no. flu or Munsell 7.56 3.0/8.0.2 ‘Iblerance limits for hue, lightness, and saturation have been determined by Munsell notation and with glossy Munsell hue samples. The hue limits range from 5 G to 0.5 33 inclusive. Since for daylight conditions, the Munsell system ostensibly contains precise sets of samples varying only in hue, psychologically equal in lightness and saturation, this system provides an ideal source of stimuli. Table 1 shows in row one and two the set of five glossy Munsell hue samples selected for theoret— ical calculation. Each hue has the same lightness level, 3/, and the same saturation level /8, according to Munsell's original notation. Theoretical Calculations For the photopic adaptation level each wavelength of the spec— tral reflectance curve was multiplied in turn by the spectral energy distribution of daylight, 6, 500° Kelvin, at each wavelength, and then by the photopic luminosity curve at each wavelength. (Actually only every five m}: were involved in these calculationS.) 'nlese products famed a new curve of luminous emittance or the products could be sum- med and used in equation (1). Similarly the scotopic adaptation level calculations consisted of the multiplication of the spectral energy distribution of incan- descent tungsten light, 2,850o Kelvin, and the scotopic luminosity curve at each wavelength. Mesopic adaptation level calculations .mcsdsoo aoaoo Haomcdz one an UoflHQQSm woeado ooeepooHMma Heppoogm Scam popmadoamo moSHm> ** .hchEoo nOHoo HHowcdz mew Scam thooeflp pocflmpno modamb * NH aa.ee Ho.oH mm.oa an.m He.m ..Hoeea ooeeeedooe eeooeeem oo.a om.a mm.a am.o em.o ..Hoeea ooeeoedeoe endoeez as.e an.e md.e we.o em.e .Heeea seaweedeoe eedoeoed Amocepooauom mdozwesfiv mmocpnwflq Heeamseddeand m.om m.mm n.0N m.om m.om AooEAOMmcwapv osHm> mmocpnwflq on o.m\ae.m a.a\mo.m m.a\mo.m mo.m\md.m m.eeme.m deepenooem\eoooenqu mmmo.m 0 OH em.a om.m om.m ode soapepocom noHoo HHomcdz on em on on on Aoesnoeeoenev eoaeo neeoeemae m\m m\m m\m m\m m\m doweenooem\eneoeemea mmmd o 3 3.x. o m Rio. 25 ooaoeeoz ooHoo Hemoodz. HeedMOHondwmm .moadsem oz: Hammad: spam you mosad> mmogpnmwfl oocflsaopoo hHHmoflmzndonoamd one hHHeOHmoHonozmm n H manna 9. consisted of multiplication of the spectral energy distribution of incandescent tungsten light, 2,850o Kelvin, and Weaver's19 0.01 ftL luminance spectral luminosity curve at each wavelength. Results of Theoretical Calculations Both equation (1) and plots of luminous emittance curves were used to estimate differences in luminous reflectance among the hue samples. Planimetric measurements yielded results consistent with those obtained by equation (1). The Nunsell spectral reflectance curves could not be read with accuracy due to the small scale and broad trace. Since there was a discrepancy between the photopic calculations and Munsell's photopic values for luminous reflectance, the calculated results for the mes- opic and scotopic levels were adjusted in proportion (-2% to +9%) to that which was necessary to make the photopic measurements coincide with those supplied by Munsell. These results are listed under Psychophysical on table 1. Dif- ferences in hue lightnesses (luminous reflectance) for each adapta- tion level are shown in figure 1. At the photopic level, the lightness differences can be attrib— uted to the psychophysical measurements being different from the psya chologically determined, equally'lighthunsell samples. Table 1, however, Shows that the more recent.Munsell renotation for these hue samples, in fact, yields small lightness differences.12 These psych- ological differences, relative to the Munsell value scale, are shown in figure 1 as curve IV. The mesopic level in figure 1 yields lightness differences quite similar to the photOpic level. Relative to the large differences cal- 10. 10. 5 10.0 I 9.5 ’4? O E g 9.00. C Adaptation Light 3 - ’ Plot 16>ng source a PsychOEhysical 5 8'50" I scotopic ZSEOOK E II mesopic 28 50°K C " III photopic 6500°K . Psychological 8'00 - IV photopic 6 5000K p a _ 0‘ g 70 50 b ‘0 II “En . :3 7.00 I '\ . D / /II 6050 p ./O A 4' ‘3 3 .7 4- m': 3.00 #__..——_—_————’ 5 > =— Iv 5 . . . . : 2-56 5 G 7.5 G 106 . 2.5a; Munsell hues Figure 1 - Psychologically and psychophysically determined lightness values for five Munsell hue samples. 11. culated for the scotopic level, it appears that there should be no significant change in psychologically observed lightness levels in going from photopic to meSOpic adaptation levels and changing from daylight to incandescent light. There may be two reasons for this: First, calculated luminous reflectances often do not conform to the visual sensation.17 Second, Bridgman1 has challenged Weaver's procedure of interpolation of mesopic luminosity curves as only an "useful approximation". He maintains that there is "no integration between rods and cones wavelength by wavelength" and that mesopic luminosity curves based upon this integration, as above, are invalid. Bridgman appears to be correct on the basis of what he has shown. Therefore it was evident that a psychological experiment where human subjects would be tested in an environment simulating those viewing conditions present on the highway at night remained to be performed. METHOD Since the human eye can match brightnesses with great pre— cision,3 it was hOped that a lightness matching method would solve the problem at hand. In light of previous research in this area, matching hue samples with neutral gray samples by a modified meth- od of adjustment seemed most promising.17 Since viewing conditions similar to those on the highway were sought, the fellowing mater- ials and illumination conditions were arranged: Stimuli For reasons above, Mhnsell samples were used exdbusively as stimuli simulating Interstate highway sign material. In addition 12. to the initial set described above, three more sets of glossy.Mun— sell hue samples, each equally light and equally saturated in day- light and each a step away in lightness or saturation from the ini- tial set used above, were obtained from the Munsell system.12 Thus, the stimuli included a set of sthwuli a step lighter, a set of stim- uli a step darker, a set of stimuli a step less saturated, and a set of stimuli a number of steps in hue on either side of the pres- ent Interstate green, which is 7.5G 3/8. Table 2 lists these sets. Table 2 - Lightness and saturation values for the four sets of Munsell hues. Munsell Hue Set 10 GY‘ 2.5 G 5 7.5 G 10 G 2.SBG 5 EG’ 7.538 D I 4/8 4/8 4/8 4/8 4/8 4/8 4/8 4/8 II 3/8 3/8 3/8 3/8 3/3 III 3/6 3/6 3/6 3/6 3/6 3/6 3/6 3/6 IV 2/6 2/6 2/6 2/6 where: GY = Green-Yellow, G = Green, B6 = Blue-Green, 3/ = light- ness level of three, and /8 = saturation level of eight. Neutral Scale The reflection scale of Munsell neutral gray samples is essent— ially independent of the level of illumination.“ Therefore, the fol- lowing set of'Munsell 0.5 value unit step grays were selected as a neu- tral lightness scale: dark 1-0/. 1.5/, 200/, 2.5/. 3.0/. 305/, “oo/. 4.5/. 5.0/. light Such a procedure of defining lightness in terms of the luminous reflectance of neutral samples has been noted as being a far better 13. method than calculating luminous reflectances of the samples them- selves.17 In order to have lightness estimations in round numbers and to have a dark-large-down rather than a light-large—down orientation, the neutral scale as given by Munsell was linearly transformed by the equa- tion: 30 = 10 (6 - x) (2) Thus, the nine neutral grays ranged in value from 10 to 50 in steps of 5, with 10 being the lightest gray and 50 being the darkest (see appendix A). This change facilitated the subjects' task. Apparatus The Specially designed apparatus shown in figure 2 was con- structed to allow viewing of one Munsell hue at a time while at the same time permitting comparison with the neutnfl. scale in the window for matching purposes. Apparatus "A" in figure 2 was placed on a rubber mat and fastened to a table by means of C clamps. "B", a crosssection (not drawn to scale) of the area under the large win- dow, shows the grooves which were used to guide "C" and "D". "C" is the neutral scale containing the nine Munsell neutral gray samples. When "C” was placed in the right side of "A" and.moved up and down, numbers along the right hand edge of "C", corresponding to the light- ness value of the gray which was in view in the large window, appear- ed above in the small windown See figure 5. "D" represents one of the twentyefive hue samples, which were placed directly adjacent to the gray scale in apparatus "A" from the top by the experimenter. In gereral, the subjects were instructed to estimate, by com— parison with the neutral gray scale, the lightness value of the hue 14. Figure 2 — Apparatu: 15. samples. Directly above the large window was a small white triangle which was used as a fixation point by the subjects to insure peri— foveal vision. Figure 3 shows a side view of the laboratory set—up. The light source was a commercially available, high intensity, 12 volt, 15 can- dle-power, 20 watt lamp which takes a $93 automotive bulb, having a color temperature on the bulb wall of 2,7000 Kelvin.1u Although this light source is about 200° Kelvin lower than the standard highway headlamp; the difference tends to decrease the effect of the Purkinje shift, if anything. The cone-shaped shroud in figure 3 was made of black construction paper except for the tip, where a plywood snout was placed. A rec- tangular opening (3/32" by 1/4") on the snout, together with proper orientation of the bulb and the apparatus, permitted: 1. Indirect light to fall in the form of a circle (12 inches in diameter) about the apparatus. See figure 4, dashed lines. Photo- metric measurements of the black construction paper and black sur- face of the apparatus ranged from 0.01 to 0.03 ftL luminance, sim- ilar to that of dark objects viewed from.an automobile at night.15 2. Direct light to fall in a rectangular pattern over the large window of the apparatus. See figure 5, dashed lines. Photometric measurements of the Munsell hue sample equivalent to the present In- terstate green was 0.042 ftL luminance (see appendix A), a level which is well within the range of luminances reflected by standard green sheeting under varied outdoor locations and low beam illumina- 15 tion. 16. ‘1b’ lamp with shroud room 8‘ x 10' all surfaces flat black Figure 3 - Side view of laboratory set-up. I 1+3” l I ’1" ____,1 black __paper small window 1} / / Figure 4 - Ibp view of apparatus. I I II T’] l / b ‘4 ‘1 \ / _’,..apparatus large window - ‘ rubber pad direct light _. ' direct light * /’ ‘~__flZE::r‘L-%?‘ fixation point table \ t” |-<-—z’4” \ I] \\ —<———3’/+———>- f 1 Figure 5 — Window dimensions. 17. 3. Enough light for the subjects to read the numbers, made of White.§£9§§EKR§9 in the small wdndow. Since the subjects viewed the apparatus perpendicular to the surface of apparatus "A" and since the light angle of incidence was 45 degrees, glare from either the samples or the gray scale was elimp inated. The experimenter sat in the right hand chair in figure 3, where he fed the hue samples into apparatus "A" and wrote down the subjects' estimates. A 3 volt penlight with a red filter was used for his illumination. Tb avoid unexpected burnouts and to keep the color temperature as constant as possible, the bulb of the light source overhead was changed after the twelfth subject. The fixation point in figure 5 was a white, 1/8 inch, equila- teral triangle 7/8 inch from the large window; Since the subjects' eyes were approximately 15 to 17 inches from the surface of the ap- paratus, there was a visual angle of from 3 to 7 degrees subtended between the fixation point and the samples in the large window. The fovea itself subtends a visual angle of 1.5 degrees, but for most scotOpic luminosity data, 5 degrees from the fovea is the recommended angle.13 0n the highway, the distance between a sign 15 feet above the road surface and the road surface subtends a visual angle of 3 degrees at 300 feet distance. At 600 feet the visual angle is 1.50. Although the size of the samples varied, the large window lim- ited all hue samples and gray samples to the same size, 5/8 by 3/4 inches. This size is roughly proportional to highway signs and, ac- cording to others, size makes no difference in the ability to match luminances.7’18 Subjects Twenty—five female subjects, obtained from elementary psychol- ogy classes, were tested individually for a period of two hours. Their ages ranged from 17 to 20 years of age with a mean of 18.5 years. Of the fifteen who had vision defects, eleven were nearsight- ed and ten wore corrective lenses. All subjects reported not being colorblind; however, only eighteen had had some sort of colorblind test. Due to the low probability of colorblindness and the assumed ability to withstand the tediousness of the task, females were used exclusively. Procedure In general, the subjects were instructed in the art of making lightness judgments for a period of a half an hour. (See appendix B for complete instructions and protocol) Emphasis was placed upon their looking only for differences in the amount of light reflected, as the criterion for lightness. They were told that by focusing on the triangle, color could essentially be eliminated from the samples and therefore they could concentrate on the lightness differences. The subjects were instructed to first bracket the given sample between two grays and then to estimate in round numbers, the value of a possible gray which would be equally as light as the sample. In other words, there were no true lightness matches made, but rath- er estimated lightness matches. Even though the illumination was quite low, the subjects were able to estimate up to 0.1 value (Hunsell) steps. This was as close as those obtained by Sanders and wyszecki in a similar experiment.17 However the variance in this case was much greater. 19. The eight hue samples of set III (see page 12) were used as the practice stimuli. Subjects were given practice until the exper— imenter felt that they were competent in making lightness estimations according to the instructions and had reached a stable learning pla- teau. Three trials were then run, each trial consisting of the seven— teen samples of sets I, II, and IV. Before leaving the laboratory, set III, the practice samples, were run again as a fourth trial. Control Measurements The apparatus, samples, and subject were then moved to an office room for a trial under daylight illumination. Even though the win- dows of the room face west, no direct sunlight came into the room. It was hoped that daylight estimations would provide: 1) support for the assumption that the subjects used were a random sample of the population, 2) a source of control measurements, and 3) data that could be used to compare differences in lightness estimation ability within individual subjects. The set—up in the office room.was similar to that in the labor— atory. Glare was eliminated by adjustment of the subjects' sitting position. Here, as in previous experiments in daylight, the subjects were instructed to look directly at the samples instead of at the tri- angle. All twentyefive samples were presented once in the final trial. Each subject was then questioned about her vision and any attempts to perform other than as instructed. Egperimental Design The laboratory or mesopic trials were always run before the daylight or photOpic trial because: 1) the subjects were able to 20. dark-adapt while instructions were being given and 2) hue and satura- tion differences were minimized in the laboratory, thereby both facil- itating and keeping constant the training of lightness judging. Nonetheless, the order of presentation of samples was random- ized for each trial and for each subject. Carry-over effects with- in either mesopic or photOpic conditions should thus have been elim- inated or made equal over subjects. Carry-over effects from mesopic to photopic conditions were assumed to be negligible. Due to the large individual differences encountered in similar experiments, and the resulting need to reduce variance due to exper- imental error, a repeated-measures-on-the-sane-subject design was used. In this way each subject served as his own control, in the hope of increasing the power of the analysis. Consequently, a double-classification, mixed model (hues fixed and subjects random) analysis of variance statistical test was employed. A significance level of 0.05 was assumed to be appropriate. Because all sets of hue samples and both adaptation levels were so different to begin with, eight separate analyses of variance were made. RESULTS AND DISCUSSION 3112?. The means and variances for the lightness estimations over sub- jects for each set and adaptation level are shown in tables 3 and 4. Figures 6 through 9 illustrate the lightness estimation means, where the full line represents the meSOpic or experimental conditions and the dashed line the photopic or control conditions. The ordinate 21. Table 3 - Lightness estimation means and variances across subjects for the mesopic adaptation level. Mun sell Hue Set 100: 2.58 50 7.50 106 2.588 588 7.SBG 51' 19.53 19.01 18.80 19.19 18.69 18.83 18.53 18.95 I 52 8.72 10.31 11.52 11.08 11.95 12.91 13.98 15.40 '2 28.95 28.76 28.55 28.61 28.19 II 52 5.60 7.76 6.80 8.12 10.07 E 29.40 28.12 28.84 28.64 28.48 28.64 28.08 27.56 III s2 2.28 3.28 4.16 6.00 5.80 5.80 8.84 6.96 '2 38.59 37.49 36.96 38.01 Iv 2, s p 3.68 4.48 4.45 2.83 Table 1+ - Lightness estimation means and variances across subjects for the photopic adaptation level. Munse'Ll Hue I ‘E 19.76 19.72 20.24 20.52 18.84 19.72 18.96 19.08 52 10.11 6.52 10.60 8.56 9.50 7.24 9.56 9.84 'i 29.56 29.52. 29.20 28.96 29.56 II 32 7.92 7.44 10.50 6.12 8.80 3? 29.08 27.68 28.56 29.84 28.68 28.64 28.40 27.24 III - 52 7.60 5.96 6.33 5.88 7.88 6.56 8.56 7.73 ‘2 38.04 37.52 38.36 38.16 Iv 2 s 4.52 5.76 6.72 6.60 Munsell value light Munsell value light dark dark 4.0 H \O I Ebcperiinental lightness N o 3.9 3.2 3.0 2.9 L21 22. Control \. I ‘~ \\I all hues 4/8 1 1 .1 l l 1 1 1 10 GY 2.5G 5 G 7.5G 10 G 2.SBG 533 7.5BG Munsell hues Figure 6 - Lightness estimation means for set I hues. - 28 {3 \A) O I I Experimental lightness L31 Wermmy-_ Control ’ — — 'l /. \ Munsell renotation ./ ./ all hues 3/8 a I l J 4 2.50 5C} 7.53 10G 2.5BG Munsell hues Figure 7 - Lightness estimation means for set II hues and Munsell renotation lightness values. light Mun sell value a dark '9: b.) o w j (‘0 \) E” IL Munsell value 23. - . /', 3. /'\ .1 / ééz I. I, ' ‘<:; “in: Experimental 1’ P; - I \. /\ /. E I -‘/' 1"" '8 29- . \ / .8" \ / " " . \\// all hues 3/6 mo; ;. 5‘. $1111.; 2.5;. 5.; 7...: Figure 8 - Lightness estimation means for set III hues. I 2.3 E" N j Experimental lightness 37 - . Experimental7/ #- '\ .I’ / \ 38 F- 0/ \ 0 r’. Control )\” P ' all hues 2/6 1 J l 1 2.5 G 5 G 7.5 G 10 G Mun sell hues Figure 9 - Lightness estimation means for set IV hues. 24. shows the transformed lightness scale and its corresponding Munsell lightness values. Of interest in figure 7 is the additional plot of the Munsell renotation lightness values which are from table 1 and are shown as plot IV in figure 2, page 10. The similarity between the Munsell re- notation plot and the control plot is striking. This coincidence, except for a constant lightness difference, offers evidence that the subjects chosen were performing similar to those subjects used by the Optical Society of America subcommittee in their spacing of the Mun- sell colors.12 In addition, it attests to the validity of the light. ness judgment training methods employed. Unfortunately, renotation values fer the other hues were not readily available. On the other hand, the means for the experimental conditions do not appear to be very different from the Munsell renotation val- ues either. Therefore, lightness differences under the experimen- tal condition can be attributed to the lightness differences exist— ing between the hue samples prior to their being taken into the lab- oratory. Fortunately, analyses of variance results in tables 5 and 6 show'that under neither adaptation condition were set II hues sig- nificantly different in lightness. However, this finding points out that for other sets of hues, the lightness differences obtained, even if significantly different, are suspect because of possible lightness differences prior to exper- imentation. Since the control means in figure 7 are quite similar in shape to the Munsell renotation values, it is reasonable to assume that control means for other sets of hues are representative of what the actual daylight lightness differences (Munsell renotations) are. Table 5 - Analyses of variance for the mesopic adaptation level. 25. Set Source of Variation df SS M3 F Between Hues 7 51 7.29 0.93 Between People 24 3854 160.58 31.92** I Interaction 168 1324 7.88 1.57 Within Cells 400 2011 5.03 Total 599 7240* Between Hues 4 23 5.75 1.47 Between People 24 1585 66.04 17.99** II Interaction 96 375 3.91 1.06 Within Cells 250 917 3.67 TEtal 374 2900 Between Hues 7 71 10.14 4.72** III Between People 24 714 29.75 13.84"“.l Interaction 168 362 2.15 Total 199 1147 Betlfeen Hues 3 109 36. 33 100“?“* Between People 24 440 18.33 7.83** IV Interaction 72 250 3.47 1.48* Within Cells 200 467 2.34 Tbial 299 ‘1266 Table 6 - Analyses of variance for the photopic adaptation level. Set Source of Variation df SS HS F Between Hues 7 64 9.14 1.79 I Between Peeple 24 942 39.30 7.71** Interaction 168 856 5.10 Total 199 1862 Between Hues 4 7 1.75 0.30 II Between People 24 466 19.42 3.37** Interaction 4__96 .554 5177 Tbtal 124 1027 Between Hues 7 111 15.86 3.22** III Between People 24 635 26.46 5.38** Interaction 168 778 4.92, Tbtal 199 1524 Between Hues 3 10 3.33 0.86 IV Between PeOPle 24 313 13.04 3.39** Interaction 72 277 3.84 Tbtal 99 600 * p‘<_.05 *1! p'<_.01 26. In table 5 only sets III and IV are shown to have overall sig- nificant lightness differences. However, set III hues in figure 8 have almost identical lightness means for both experimental and con- trol conditions. This plus the fact that set III hues were observed for only one trial (the fourth), casts doubt on the validity of the significance test. This cannot be said of set IV lightness means. Table 7 presents results of significance tests made on ordered pairs of means. 8:232221 tests on the non-significant overall F test means were not justified because differences that did occur were shown to be irrelevant. A_posteriori tests, using Tukey's pro- cedure,20 were made on the overall significant F tests of sets III and IV. Significant lightness differences for'means of set III, meSOpic, were between the bluest hues, 7.58G, SBG, 2.580, and the yellowest hue, ioGY. For set III, photopic, significant lightness mean differ— ences are accounted for in general by 7.50, which is quite deviant from.the rest of the means. Alone, such differences, even though widely spaced, could be of interest. However, even these differen- ces are probably due to differences inherent in the samples themsel- ves and not due to the experimental and control conditions. Significant lightness differences for set IV, mesopic, means show general support for choosing 7.5G from a set of green hues of Munsell lightness level 2 as the lightest hue under simulated night viewing conditions. However, where tested, interaction variance was significant only for set IV, mesopic, conditions. Although eich findings lend credence to the other sets as being free of confound- ing influences, it does render questionable the results of set IV no. V a a. Re I e 3 ..i I o m 0.382: >H ** a; I om.m em.m 00H m m em.e I I I .. .. mm m . ... .. u .. .. .. mm.~ 038er HHH m z. - - - - - - 83 RN $3 02 8nd 0 m men Re Bug .. I .. .. 02 H H H H H . “mum 0382 HH .1... I I I I I I amen. 32 e n 8m.m emé. 8H and 8m 8mg. wed: HHemqsz Hebea pom coaeflewee .mcees.mo enema pegopuo so mpmep ooceofimficwflm fihofluepmom.m mo mpadmom I A manna 28. significance tests. Since the same people had no interaction with other sets of hues, it would appear that set IV, being quite dark and consisting of only four samples, is unique in its evocation of lightness estimation differences. It should be noted that the lightness differences shown in fig— ures 6 through 9 are exaggerated by the use of only a small part of the lightness scale. Figure 10, which shows half of the Munsell lightness scale, gives a more realistic View of the observed light. ness differences. Set III hues are not shown because they would ever- lap those of set II. Statistically, the assumptions of homogeneity of variance and normality of distribution were met. A number of histograms (not shown) were made showing normal distributions of hue scores over 20 subjects. Since kurtosis was not present, F max tests were per— formed, showing no significant heterogeneity of variance to be present. Subjects For each of the eight analyses of variance, there was a signif— icant difference between means of the subjects over hues. Although this was an expected result and was not of concern, it does bear out the knowledge that large individual differences are found in experi- ments of this kind. Training of the subjects went smoothly for the most part. 'hx3 subjects who could not or would not perform the task adequately were excused after a critical amount of time had elapsed. All of the subjects stated that.they had performed only as in- structed. Lightness estimations given for samples on previous trials were not remembered on subsequent trials. Although half of the sub- 5.0V. 10g- light “'OHF 20* 3£~~~.---./ 29. Eocperimental ——c\.’7:\\ ::”:——-3 Control Set I (4/8) Experimental c—-——o.——-——c o——————. .———.¢—-""".— --.\‘~. Control ._________. ’0 ——.———I—. Munsell renotation ° Set II (3/8) EXperimental .rc\ :‘7‘ §\o—-""'--"': Control Set III (2/6) 3.0? 3:30P m . :2 :3 r :0 g e p H :1 3 I :3 . 2: 3 2.0L 5:40? i n 1. G - SO- .x ' ' g a 'U 0, GI. 60. —k 10GY 2.3 G 51G 7.5G CG 2. BG H} . Figure 10 - Lightness estimation means for set I, II, and IV hues. 30. jects performed the photopic task under rainy, overcast skies and the other half under clear skies, no differences were observed. Most subjects felt the daylight trial to be more difficult than those in the laboratory because of the marked increase in perceived hue saturation, which interfered with lightness judging. Also several of the subjects showed signs of fatigue toward the end of the experiment. Two interesting features of the experiment shOuld be noted. First, if the subjects focused too intensely on the triangle, neg— ative afterimages of the triangle and loss of perifoveal vision re— sulted. The subjects were therefore warned of these well-known phe— nomena13 and told to keep their eyes moving in the vicinity of the triangle. See appendix B. Second, the experimenter had to insist that the subjects allow a few seconds to elapse after they had moved from one gray sample to another. Even though the lightness differ— ences were small, their eyes still had to have time to adapt to this change before accurate lightness estimations could be made. CONCLUSION This experiment was designed to investigate the effect of con- ditions relevant to the Purkinje shift at one level of illumination typical of that encountered on the highway at night. Of the four sets of equally light hues used for stimuli only one, set IV, gave any evidence of enough lightness differences to support the conten- tion that one or more hues of green are lighter than others under night viewing conditions. However, these results were confounded 31. with subjects, rendering the evidence suSpect. If the other sets showed similar results, it would be warranted to investigate further the interaction and/or lightness differences. r t since set IV con- sists of only four hues which are quite dark (Munsell value two) the pursuit is not justified. Rather, it seems reasonable to conclude that, in general, with— in the green hue limits set by the Bureau of Public Roads, evidence is shown to support the conclusion that no differences in lightness levels among hues, perceived to be equally light in daylight, occur due to the Purkinje shift under night highway viewing conditions. Un— doubtedly, between different hues, such as red and green, the Purkin- je shift does have a significant effect on night lightness perception. Within one hue, such as green, however, this seems not to be the case. At first glance, the differences between standard luminosity curves, the extremes of fully light and fully dark-adaptation, appear quite great. But when incandescent light replaces daylight at the scotopic level, the curves are much closer together than before as Richards has shown.16 Likewise, the lighter the adaptation level in the meSOpic region, the closer the two curves move toward coincidence. In addition, viewing stimuli placed within 5 degrees of foveal vision, lessens the effect of the Purkinje shift. The above three factors adequately account for the results of this study. The infermation is relevant since it eliminates one more factor from consideration among possible hig way sign attention fac- tors. In addition, the methodology employed for this study may have applicability in other related work. Results of theoretical psychOphysical calculations, although 32. subject to uncertain assumptions, suggested that there would be no significant lig.tness differences between set II hues under a mes- Opic adaptation level and incandescent light. Although these sug— gestions were questioned, they were shown to be correct. In fact, the Munsell renotation lightness values for set II hues displayed greater lightness differences than those obtained under the mesopic - incandescent experhnental conditions (see fig— ure 7). Not to be overlooked, however, are the even greater light- ness differences displayed by the theoretical calculations them- selves (see figure 1). If such differences of this magnitude had been observed under the experimental conditions, most likely hue, 7.5 G, would have been significantly lighter than the other hues of set II. That this was not the case, can only be explained by the differences between the two methods of measuring lightness, the psychophysical being a general quantitative approximation, the psye chological being a specific qualitative estimation. In addition, photometric measurements of set II hues displayed yet a third and different set of lightness values (see appendix A). Therefore, precise lightness specification is shown to be relative to the means of'measurement. For general and approximate results, psychophysical calculations and photometric measurements are ade— quate; but for specific and critical situations, such as simulated here, psychological experimental specification (by means of a psycho- physical experiment), even though subject to large individual dif— ferences, are preferred. 12. 13. 14. 15. BI BLIOG..APHY Bridgman, C.S. The luminosity curve as affected by the re- lation between rod and cone adaptation. J. Opt. Soc. Am,, 221 1953. 733. ‘ Bureau of Public Roads, Washington, D.C. Evans, R.M. An Introduction to Color. Wiley & Sons, New York, 19%. Evans, RM. Variables of perceived color. J.Oot. Soc. A111,, 22, 195“. 75. Forbes, T.W., Pain, R.F., Fry, J.P. and Joyce, R.P. Effect of sign position and brightness on seeing simulated high- way sigqs. Highway Research Board, Night Visibility Com— mittee. Jan. 1966: In press. Forbes, T.N., Snyder, T.E., and Pain, R.F. Traffic sign re— quirements. Highway Research Record, _7_Q, Highway Research Board, Washington D.C. , 1955,78. Graham, C. (ed) Vision and Visual Perception. Wiley & Sons New York, 1965. IES Lightingjiandbook. (3rd ed) New York, 1959. Judd, D.B. Color in Eisiness, Science, and Indust_ry. Wiley & Sons, New York, (2nd ed) 1963. LeGrand, Y. Light, Colour and Vision. Translated by Hunt, R.W.G., Walsh, J.W.T. and Hunt, F.R.W., Wiley 8: Sons, New York, 1957. Munsell Color Company, Inc., Baltimore, Maryland. New’nall, S.M., Nickerson, D., and Judd, D.B. Final report of the O.S.A. subcommittee on the spacing of Munsell colors. J. Opt. SOC. m1, 223 19143! 3850 Optical Society of America. The Science of Color. Thomas Crowell 00., New York, 1953. Personal communication, General mectric Miniature Lamp Dept. , Cleveland Ohio. Powers, L.D. Effectiveness of sign background reflectori- zation. Highway Research Record, 29, Highway Research Board, Washington D.C. , 1965, 71+. 33. 17. 18. 19. 20. 3L: 0 Richards, O.V. Vision at levels of night road illumination. Bulletin 5E, Highway Research Board, washington, D.C., 1952. ’5 ;r )0. Sanders, C.L. and Nyszecki, G. Correlate for lightness in tenns of CIE tristimulus values. Part II. J. Opt. Soc. Amé, Elf—Z, 1957 9 3980 Straub, A.L. and Allen, TZM. Sign brightness in relation to position, distance, and reflectorization. Bulletin 146, Highway Research Board, washington, D.C., 1955:713. Weaver, K.S. A provisional standard observer for low level photometry. J. Opt. Soc. Amy 32, 1949, 278. Winer, B.J. Statistical Principles in Experimental Design. McGrawbHill Book Co., New York, 1962. APPENDIX A — PHOTOMETRIC READINGS Tables A1 and A2 give the luminance values (in foot-Lamberts) for each of the Munsell gray and hue samples under the meSOpic il- lumination condition. These photometric readings are plotted in fig- ures Al and A2. In general, the photometrically determined plots of figure A1 are not quite similar to the psychologically determined plots of figures 6 through 9 above. This gives evidence that photometric readings, in addition to psychophysical calculations, do not give an accurate representation of what the eye sees. This tends to further corroborate the rationale for using a psychological method rather than a psychophysical one fer experiments of this type. Figure A2 shows that the photometric readings do agree perfect- ly with an independent source, the Illumination Engineering Society Handbook.8 Such similarity insures that the photometric readings are reliable and valid. 35. Moonpcdn mmH nod oocwpooHMon mafia mo pcooamm ** mCOflpflpcoo COHpmcflsdaafl 00QOmmE no :5 popoEopoxd phagopwpm nod mpaopsdqlpoom CH * 0000.0 0000.0 0000.0 0000.0 0\0 >H 330.0 330.0 030.0 330.0 330.0 030.0 030.0 0a30.0 0\0 HHH 030.0 330.0 030.0 330.0 030.0 0\0 HH 000.0 300.0 000.0 000.0 000.0 000.0 000.0 *300.0 0\3 H 000.0 000 000.0 000 00.0 00 00.0 0000 «aopgo\msamp 000 30 33:0: gm .monEmm on: damn: E_o£p mo mpCogopswmoE oaapoEoponm I N< oHQNB 00.00 00.00 00.00 00.0 00.0 00.3 00.0 00.0 00.0 ..00300000000 0300005; 000.0 000.0 000.0 300.0 330.0 300.0 000.0 000.0 000.0 .000000250 00 00 00 0 00 00 03 03 00 games: 00300000 0.0 0.3 0.3 0.0 0.0 0.0 0.0 0.0 0.0 osaa> 0000032 60.3080 09% H.053: ”8032 05 .00 3085.03.38 0.0503330” u «4 0.3.09 Luminous reflectance (in foot-Iamberts) 0.09% 5813 I 0.080. \/\/ \. 0.075 I l l l l l l l 10GY 2.56 5G 7.5G 10G 2.533 58G 7.SBG Munsell hues 0.050!- Set II 0.0435 . .\ o \./ \ 0.040 L n 4 n o 2.5G 5 G 7. G 10G 2.530 Munsell hues 0.050 L ./°\ Set III 0.01-lr In ’ ‘ 5 \o—o\./o—o 0.040 4 n 1 l l l l n 10GY 2.5G 5 G 7.5G 10G 2.58G 5K: 7.533 Munsell hues 0.025 I- Set IV 0.020 - \'/ \ 0.015 ‘ ‘ ' ‘ 2.5G 50‘ 7.50 106 Munsell hues Figure A1 - Photometric measurements of the Munsell hue samples. ’30. J0 o 20 I- ‘ +0.15 18r- 16 3 d D. 12 ’3 o A 5 {a O 0 1,4, I- (D a "E 0 s. 0 +3 3 8 a. m o . 4, I35 v 5 standards 8 g; 5 3 1° ' 8 3 i=5 5 2 0 o 8 .. O. 06 o u " ‘ 0 ° 5 H H 3 . 13 o k h "g 6 "' x Photometric a ..q / measurements 3 2 f3 / H 1+ h- / JO. 03 g/ 0 l l l A J L l l l L. 00 1.0 1.5 2.0 2.5 3.0 3.5 “.0 “.5 5.0 Munsell value Figure 12 - Comparison of the photometrically measured luminous re— flectances of the Munsell neutral gray samples with IE3 standards. APPEI‘IDIX B: PROIOGDL-EQS‘I‘RUCTIONS The subject enters laboratory and is seated. The spotlight is turned on and the room light is turned off. The door is closed and the penlight is turned on. Instructions are begun. YOU ARE ABDUT TO TAKE PART 131 A STRAIGHTFOR’.~L\RD PSYCHOPHYSICS EXPERIMENT WHICH CONCERNS THE EFFECT OF ION ILLUMINATION ON THE VISUAL PROCESS. YOUR PART CDNSISTS ENTIIUiLY OF MAKING A NUMBER OF LIGHTNESS EBTlI-IATIONS. I HAVE ALREADY SPENT MUCH TIME AND EFFORT PREPARING FOR THIS EDG’EREIHVT AND HOW I AM ASKING YOU TO USE YOUR EYES AS A TESTING DEVICE---‘IO TEST MY HYPOI‘HESIS. BUT SINCE THE TASK I AM ASKING YOU TO PERFORM IS A TEDIOUS ONE, I MUST ALSO ASK YOU TO BE AS HONEST AND CAREFUL IN YOUR JU {BITS AS I mULD BE MYSELF. OTHERWISE YOU WILL BE UNABLE TO HELP ME. I WANT YOU TO UNDERSTAND THE SERIOUSNESS OF YOUR TASK. The dust cover of viewing apparatus is removed. The table is moved into position so that the spotlight projects on the large window of the apparatus. Resume instructions. THE LIGHT AHDVE YOU IS TO SHINE IN THE LARGE WINDOW AT ALL TIMES. IF IT IS NOT, PLEASE MOVE THE TABLE SO THAT IT IS. IN THE APPARATUS IN FRONT OF YOU IS A GRAY STANDARD SCALE. OBSERVE THAT AS YOU MOVE IT UP AND DONN, WHICH YOU MAY DO, DIFFERENT NUMBERS APPEAR IN THE SMALL WINDOW. EACH NUMBER IN THE SMALL WINDOW REPRE— SENTS OR (DERESPONDS TO THE LIGHINESS VALUE OF THE GRAY WHICH IS IN THE LARGE WINDOW. THERE ARE NINE GRAYS, M~IGING IN VALUE FROM 10 TO 50 IN STEPS OF 5. NOTICE HOW THE HGHTNESS STEPS DIFFER ON THE GRAY SCALE IN 39. bro. THE AICOUNT OF LIGHT REFLECTIE. THE AMOUNT OF LIGHT REFLECTED IS THE MADI CRITERION OF WHAT LIGHTNESS IS. I AM TRYING TO TRAIN YOU TO JUDGE LIGHTNESS DIFFEIEE‘JCES. mIORED SAI’IPLES WILL BE PLACED IN THE IAR‘GE WINDOI'I (one is put in) AND YOU WILL BE ASKED TO JUDGE THEIR. LIGHTNESS IN RELATION TO THE GRAY SCALE. TO HELP YOU I'EAKE YOUR LIGHTNESS JUDCMENTS, IT IS BEST THAT THROUGHOUT THE EXPERI- MENT, YOU KEEP YOUR EYES FOCUSED IN THE VICINITY OF THE SMALL TRI- ANGLE. YOU MAY IOOK AT THE NUMBERS IN THE SMALL WINDOW BUT NEVER IOOK AT THE GRAYS OR SAMPLES IN THE LARGE WIN‘ W. ’H'IIS IS BECAUSE WITH PARAFOVEAL VISION (THAT VISION BETWEEN I‘DVEAL AND PERIFERAL), YOU CAN FSSB‘ITIAILY ELIMBIATE (DIOR FHDM YOUR VISION AND THEREBY CONCENTRATE ON THE LIGHTNES DIFFERENCES. EACH TIME I PUT A SAMPLE IN THE APPARATUS, YOU ARE TO DO THE FOLLOWING: FIRST ADJUS'D’IENT: QUICKLY ZERO IN OR BRACKET 'H‘IE GIVEN SAMPLE BE- MEN Th0 GRAYS; SUCH THAT ONE GRAY IS DEFINITEY LIGHTER AND THE OTHER GRAY IS DEFINITELY BAKER THAN THE SAMPLE. TO MAKE SURE OF THIS, CHECK THE LIGHTER AND DARICER GRAY AT LEAST TWICE. (This may have to be restated in different words.) IT IS POSSIBLE THAT THE GIVEN SAMPLE MIL EXACTLY MATCH ONE OF THE GRAYS IN LIGHTNESS-—- HOWEVER THIS SHOULD NOT OCCUR VERY OFTEN. YOU MAY NOW BRACKET THE SAMPLE WHIQI IS IN THE APPARATUS. (Wait until subject announces her bracket.) SECDND mm: NOW, BY LOOKING AT THESE Tm GRAYS AND BY TAKING YOUR TIME, YOU ARE TO ESTIMATE, AS CLOSE AS YOU CAN IN ROUND NUMBERS, THE VALUE OF A POSSIBLE GRAY WICH MOULD BE EQUALLY AS LIGHT AS THE GIVEN SAMPLE (this may have to be re- stated in different words). PLEASE ESTIMATE THE LIGHTNESS OF THE A1. SAIleLE. HOWEVER MIFNEVER YOU MOVE THE GRAY SCALE DURING THE SEC- OND ADJUSTMENT, IT IS ESSENTIAL THAT YOU KEEP ONEIAND ONLY ONE GRAY IN VIEN AT ANY ONE TIME. IT IS ALSO ESSENTIAL THAT YOU AL- LOW'A FEW SECONDS FOR.YOUR EYES TO ADJUST TO EACH NEN'LIGHTNESS LEVEL, EVEN THOUGH THE CHANGE IS A SMALL ONE. OTHERWISE YOU WILL BIAS YOUR ESTIMATE. FOR EXAMPLE: IF YOU ARE LOOKING AT THE DARKER GRAY AND THEN GO TO THE LIGHTER ONE, FOR THE FIRST SECOND OR SO THE LIGHTER GRAY WILL APPEAR LIGHTER TRAN IT REALLY IS. HOVRJER, IF YOU ALLOW A FEW SECONDS FOR.YOUR EYES TO ADJUST, IT WILL RETURN TO ITS TRUE LIGHTNESS VALUE, (Obtain estimate from the subject and record it.) REMEMBER, THE RESULTS OF THIS EXPERIMENT DEFEND COMPLETELY UPON YOUR ABILITY TO MAKE ACCURATE LIGHTNESS ESTIMATIONS. YOU WILL NOW’HAVE A PRACTICE TRIAL. P F “E ASK QUESTIONS ABOUT ANY; THING THAT IS UNCLEAR. ALL OF THE SAMPLES WHICH I AM GIVING YOU FOR PRACTICE ARE FAIRLY CLOSE IN LIGHTNESS. LATER THE SAMPLES WILL BE mTH FAR APART AND CLOSE TOGETHER IN LIGHTNESS. HERE I WANT YOU‘TO GET USED TO THE SUBTLE DIFFERENCES THAT D0 EXIST. The practice trial is run. If the subject appears to be mak- ing fairly accurate lightness estimations and has no further ques- tions, the regular trials are begun. A After each trial repeat the following: REIvIEIvIBER TO KEEP YOUR MIND ON LIGHTNESSu-IN TENTS OF THE AMOUNT OF LIGHT REFLECTEDuu AS OPPOSED ‘IO COIOR DIFFERENCES. IGNORE (DIOR IF YOU CAN. DIS- CIPLINE YOUR MIND IO REmGNIZE ONLY LIGHTNESS DIFFERENCES. Before each regular trial, be sure samples are lined up accord— ing to the assigned random order. After three regular trials repeat 42. the practice trial. NOW LETS SEE HOW'YOU CAN ESTIMATE THE LIGHTNESS VALUE OF THE PRACTICE TRIAL SAMPLES—--NOW’THAT YOU HAVE HAD PLENTY'OF PRACTICE. (Run trial) NOW WE ARE GOING TO GO TO MY’OFFICE AND LOOK AT EACH OF THESE SAMPLES ONCE MORE IN THE DAYLIGHT. The door is Opened and the room lights are turned on. The apparatus is broken down, taken to the office and set-up again. The subject is seated and instructions be- gun. YOU ARE NOW TO TOOK DIECTLY AT THE LARGE 'wDIDON. IT MAY BE MORE DIFFICULT TO DISRmARD THE COLOR NOW, BUT SINCE YOU ARE NOW AN EXPERT ATZMAKIHG LI HTNESS JUDGMENTS, YOU SHOULD BE ABLE TO DO ALP RIGHT. JUST OONCENTRATE ON THE LIGHTNESS DIFFERENCES---IN TERMS OF THE AMOUNT OF LIGHT REFLECTED. After the subject has finished, questions are asked about: corrective lenses, vision, colorblindness, age and general comments about the experiment. Specific questions about "hunches" the subject might have had and about remembering numbers previously given to specific samples are asked. The experiment is explained to the sub— ject if she is interested. mumynulmgtmuWmumuwmH