THE VISIBILITY FUNCTION AND VISIBILITYTHRESHOLDS FOR COLOR-DEFECTIVES

BY

MARGARET C. SHELDS

Information with regard to the luminosity sense of color-

defective eyes has seemed to the writer, interested because she is

herself a deuteranope, meagre and contradictory. No adequate

effort has been made to connect the type of color deficiency with

the form of the luminosity curve, and the question of absolute

luminosity has been, except'for the work of Sir William Abney,

untouched. The writer was recently accorded the privilege, as a

guest of the Nela Research Laboratory, of obtaining her own

visibility curve on the special equality-of-brightness spectro-photometer developed in that laboratory, and desires here to

present the results of that work, together with a few observationsupon similar data already on record.

The apparatus is essentially a Lummer-Brodhun spectro-photometer, the field of the ordinary size, and the collimators

arranged so that brightness matches are made step by step throughthe spectrum for almost indistinguishable color differences. The

writer acknowledges appreciatively her indebtedness for the use

of assembled and calibrated apparatus, and also for the deter-

mination of the brightness temperatures of the sources, fromwhich the energy distribution was computed. Both the appara-

tus and the experimental procedure were exactly as originallydescribed.'

There are on record carefully determined visibility functions for

sixteen persons recognized to be color defective. Watson2 has

offered five cases, with the thesis that on the three process theory

as represented by the Abney curves an observer, in proportion as

he fails to get the green stimulus from white, should require more

white to match in brightness a red than a normal observer, the

1 Astrophys. J., 35, p. 237, 1912; and 48, p. 65, 1918.2Proc. Roy. Soc., 88, p. 404.

362

VISIBILITY FOR COLOR-DEFECTIVES

effect being to shift his maximum visibility to the red; conversely,an observer who loses relatively more red than green, shouldhave his maximum visibility shifted to the green. He failed,however, to demonstrate for his three cases which are shifted tothe red, that they were actually found by other tests to bezmore

1. f ..F .. gg^ .. S 111111111111- --. _ -- . .- - ---. . W.--.-------r * lilile iilllilliii

W . ... i-- 44.3t1+l-\gX:1L~t IIII ! .; ! ! ̂ j l MIIIILIII:WNIIII~~tIRIIIIIIIII4-H- -. E. .g. . X W X u 111111111118111101M11144111

4F Z -- --- 4 .$Illlllll lltt1 1 l llllIT~~~~~~~~ enirv - X llllllllSII~IIIN~IIlI~

f .. .. . .R W W 4, . WIIII:I-IWMIIXIIIINIlIIIIII

S. . .g .t . gEPn-011na11./ . l . 14 gNormal MIII+klllll

FIG. Equal-area luminosity curves (Source at 2045 K).

blind to green than red, and for the two shifted to the green viceversa. Tufts3 presents three cases of color-defectives showinglashift to the red, and three to the green, with no attempt tocorrelate these shifts with the type of defect. Coblentz4 finding

a Phys. Rev., 25, p. 433.4 Bul. Bur. Std. 14, p. 167.

L44444�

S : ... _ ._

444+ - - ___ L1

_ . _ tZ

J . 1-1 1 :1 1 11 1 11 1 11 1

June, 1922] 363

MARGARET C. SHIELDS [J.O.S.A. & R.S.I., VI

four cases of shift to the red and one to the green, specifies thattwo of the former were green-blind by the Nagel test, but makesno statement as to the other three.

The finding of the writer in her own case adds one more to thetwo cases of Coblentz which definitely conform with Watson's

.0 I I . I -t -i I 1. I l I... . ..... .. ... .. ..... ... .. ....

.8

.7

'S

4

.3

.1o

.50FIG. 2

Visibility functions derived from the luminosity curves of Fig. 1.

theory. She is a deuteranope, failing completely any perceptionof green, but undoubtedly seeing the quality red except when thesaturation is fairly low,5 and her visibility curve has its maximumat 571 mg, as compared with 556-7 mg for the normal eye. (Seethe green-blind subject of Fig. 2.) It seems to the writer unfortu-

6 See Hayes, Am. Jour. Psych., 22, p. 369, case M.S. This opinion is also sup-ported by Prof. Dunlap of Johns Hopkins.

.q

11

T I

: Red--;ensirlve

' Green-blind

. NormalI.

.55

364

June, 1922] VISIBILITY FOR COLOR-DEFECTIVES 365

nate that all these visibility studies were not accompanied byadequate quantitative tests of the color sense of the subjects,but that it is none the less fairly probable that relative, blindnessto red must somehow predicate a shift of the maximum visibilityto the green, and vice versa.

There is this quantitative difference between the point of viewof Watson and the measured cases, that the totally green-blindshould have their maximum definitely in the red, (the Abneycurve, assuming the temperature of the arc to be 3500TC, wouldbring the maximum at 590mtL) whereas the largest displacementfound is one of Coblentz' cases at 578my. Moreover, contraryto the statement of Watson, it is to be noted that the integralluminosity of the green-blind may be arbitrarily made equal tothe normal without giving rise to an abnormally high maximumluminosity in the red. (See Fig. 1.) This is true for the presentcase and for all four of Coblentz' cases having the maximum in thered. The form of the curve is indeed closely comparable with thatof a normal red-sensitive subject, as shown in Fig. 1. The integralvisibility derived from equal-area luminosity curves is of coursebelow normal for any curve with its maximum in the red, but it isto be remembered that these curves represent luminosity for anequal energy spectrum, not the conditions of actual vision.

Further interest attaches to the correspondence of the presentcase with previous observations in that all the latter were takenby a flicker method in which the observer measured the bright-ness of the color in terms of white as he saw it, whereas this curvewas taken by an equality-of-brightness photometer using red at650 mg as a standard, since each wave length was m6asured interms of a shorter, starting at this point. Ives and others haveraised the question as to whether the two methods measure pre-cisely the same thing, and have suggested that in flicker the conesare relatively more important. In such case, if color-blindness beat the same time rightly ascribed to a deficiency of the cones, therecognized divergence between the two methods might be foundmuch more striking for color-defective eyes. The qualitativeagreement between the two methods may therefore be taken asindicating that the two are none the less as nearly equivalent for

MARGARET C. SHIELDS [J.O.S.A. & R.S.I., VI

color-blind as for normal eyes, or that the observed shifts are notattributable to the method.

On the basis of the simple three process theory which makes thetotal luminosity sense at each point in the spectrum the sum of theordinates of three color sensation curves, Abney was led to draw

. ... . . . . .. .. . .. .. .. .. .. .... .. ... ... .. ,-TT__ TI T 1I_ TT 1 iI, Red-seansirive - --

8 4 t Gren-blind i

±4+~~~~~~~~~1

3

2 G11r11111Er1i X0X t

i

..50o .55 .60

FIG. 3Visibility thresholds.

curves giving for a person totally lacking the green sense an inte-gral luminosity only .7 the normal, and for a red-blind persononly .3. This extreme lowering of the absolute brightness sensi-bility of color-defectives he justified partly on the basis of thresh-old determinations. It was therefore a matter of considerableinterest to test this by comparing the visibility thresholds for a

t

366

i � i N"f- .

t

June, 1922] VISIBILITY FOR COLOR-DEFECTIVES

few persons whose visibility functions had been determined inthe original investigation in the Nela Laboratory with that of theone available color-blind case. This was accomplished with thesame spectrophotometer, using only one lamp, operating it at alow temperature, narrowing the slit, then allowing the observer

52 *5.3 5531 55 S ub

FIG. 4Detail of Fig. 3.

to set the rotating disc so as just to extinguish the pattern, hiseyes being in a stable state of dark adaptation. The energy distri-bution was determined as before from the equivalent black bodytemperature of the filament, 13750K, with the correction ap-plied for dispersion. The results are given for three typicalcases in Fig. 3, showing in arbitrary units the minimum energyperceptible as a function of the wave length; Fig. 4 shows the

+H+ ±I±H444+1

Vii

444+ + ......

i

367

SO -.07 For.1

MARGARET C. SHIELDS [J.O.S.A. & R.S.I., VI

detail of the short wave length portion of the same curves.Judgments of what constitutes extinction are difficult to makewith precision, but it is at least certain that in the case of thewriter there is no evidence of marked loss of brightness sensi-bility. As compared with this one deuteranope, the red-sensitivesubject of Figs. 1 and 2 with normal color vision has a somewhatlower threshold throughout; the third, a subject, whose visibilityis close to the average has one somewhat higher throughout. Stillanother normal observer required the slit width to be doubledbefore he could see the pattern at all, obtaining then a curvenearly identical in form with that of the green-blind subject. Thethree threshold curves lie in the red in the inverse order to thecorresponding visibility curves, but in the same order in the blue,so that they cannot be correlated absolutely. It would of coursebe desirable to measure the least discernable brightness differencefor these subjects from the threshold up to ordinary levels; buteven without this, if, contrary to Abney's assertion, a green-blindmay actually perceive a smaller minimum of energy than a normal,not only in the spectral region where the relative visibility is highbut beyond the green where his relative visibility is low, there isthe possibility that his integral luminosity at ordinary brightnesslevels may not be markedly below normal.

The statement made by Tufts and by Coblentz is amply justi-fied, that an abnormal visibility function is not necessarily asso-ciated with color-defective vision; but it is certainly equally truethat there is no case on record of a color-defective with a normalvisibility function. The existing evidence indicates, rather, thatcolor-defective vision does condition a perfectly definite modifica-tion of the visibility function; it would therefore appear that atheory of vision should interrelate brightness sense and colorsense to the extent of accounting for this. It seems doubtful,however, if color-blindness does involve a lowered brightnesssense in the extreme fashion in which the Young-Helmholtztheory has been interpreted to involve it.

MouNT HOLYOKE COLLEGESOUTH HADLEY, MASS.

368