Journal of the

OPTICALOf

SOCIETYAMERICA

VOLUME 48, NUMBER 9 SEPTEMBER, 1958

Description and Measurement of White Surfaces

RICHARD S. HUNTER*Hunter Associates Laboratory, McLean, Virginia

(Received March 28, 1958)

This is the report of Inter-Society Color Council Subcommittee on Problem 19 formed in 1953 to studythe color technology of white surfaces. Physically, surfaces which appear white reflect strongly and diffuselythroughout the visible spectrum. Psychophysically, whites occupy a volume without sharply definedboundaries in the top center of the color solid. Whiteness is the attribute of white surfaces which correspondsto their visual proximity to preferred white. Preferred white varies somewhat with changes of either observeror observing situation. Measurements and intercomparisons of the colors of whites are made to determineadequacies of match to standard and to determine compliance with color specifications. One-number reflect-ance measurements of whites are widely used as partial determinations of whiteness. There have been anumber of investigations to find which formulas yield the most reliable measurements of whiteness fromtristimulus values, but the whiteness scales which have resulted from these investigations have not enjoyedwidespread commercial use.

THIS is a report by ISCC Subcommittee onT Problem 19, White Surfaces.t Among the 27professional societies joined in the ISCC, there are atleast 13 actively concerned with description andmeasurement of the whitenesses of the materials withwhich they deal.

In 1953, the American Ceramic Society asked theISCC to study white surfaces because there was talkat that time of establishing a U. S. Appliance StandardWhite. Such a standard would encourage all manu-facturers of white household appliances to produce thesame white color and would permit purchasers ofappliances made by different builders to obtain units

* Chairman, ISCC Subcommittee on Problem 19.t Members of ISCC Subcommittee on Problem 19, A Study of

White Surfaces are: Richard S. Hunter, Chairman, HunterAssociates Laboratory, McLean, Virginia; Douglas Hamly,Industrial Cellulose Research, Ltd., Hawkesbury, Ontario,Canada; Eugene Allen, American Cyanimid Company, BoundBrook, New Jersey; Henry Hemmendinger, Davidson andHemmendinger, Easton, Pennsylvania; Deane B. Judd, NationalBureau of Standards, Washington 25, D. C.; David L. MacAdam,Eastman Kodak Company, Rochester 4, New York; Robert F.Patrick, Pemco Corporation, Baltimore 18, Maryland; Robert B.Hobbs, National Bureau of Standards, Washington 25, D. C.;W. J. Goodwin, Bakelite Company, Bound Brook, New Jersey;Harold E. Crosier, Colgate-Palmolive Company, Jersey City 2,New Jersey; Paul M. Fisher, American Vicose Corporation,Marcus Hook, Pennsylvania; Norman R. Pugh, Sears Roebuckand Company, Chicago, Illinois.

which matched well when put together. The CeramicSociety was interested in this proposed standardwhite because many appliances are finished in whiteporcelain enamel. Since member bodies of the ISCCdealing with paints, papers, printing, textiles,detergents, and plastics also expressed interest in astudy of white colors and whiteness, a committee wasappointed to study white surfaces. This committeeoffered in 1953 to help with the Appliance StandardWhite project, but this offer was declined, in fact, theproject was subsequently abandoned.

The attention of this committee was then directed toa study of technical methods used to identify white andnear-white colors in U. S. industry. This report is theoutcome of that study; it reviews present practicesin the evaluation of white colors and discusses whitenessas an attribute of surfaces and methods which may beused to evaluate whiteness.

WHITE COLORS AND WHITENESS

White is ordinarily perceived as a unique colordifferent from other colors of surfaces. This visualuniqueness of white is the basis of the following defini-tion: A white surface is one which appears white whenseen in its normal surroundings. There are, of course,a variety of whites and no fixed line of demarcation

597

RICHARD S. HUNTER

wh'i

FIG. 1. The surface-color solid with dotted lines enclosingapproximately that volume at the top which contains whitecolors.

between white surfaces and light-colored near-whitesurfaces. Figure 1 is a drawing of the surface color solidwhich shows by dotted line approximately that partof the top center of the solid in which are representedcolors called white.

Whiteness is that attribute of a white surface whichdenotes its similarity in color to some preferred white.There is evidence that preferred whites differ withobservers,",2 with observing situations, and with thefashions of time. In consumer surveys, especially thosedealing with textiles, bluish whites3 have often been pre-ferred to hueless whites. Since the preferred white is thusvariable, whiteness is not a fixed attribute of surfaces.

Physically, a white surface is one which reflectsstrongly throughout the visible spectrum (usually morethan 50%). The higher and more uniform the spectralreflectance, the whiter the surface appears. As examplesof white surfaces, a porcelain panel, a sheet of newsprint,a swatch of white cotton broadcloth, and a panel ofsemigloss wall paint were selected and measured withthe GE Recording Spectrophotometer.4 The curves ofthese specimens in Fig. 2 show that, as would beexpected, the newsprint is the least white of the four.The cotton and enamel are of about the same whitenessyet spectrally they are different. One could not decidewhich was the whiter from the curves in Fig. 2 untilhe has specific colorimetric criteria for the judgment ofwhiteness.

White materials are most likely to depart frompreferred white in the yellow and gray directions. Thisis especially true of organic materials which, in general,will have spectrophotometric curves sloping downward

I D. B. Judd, Paper Trade J., Tech. Sec. 100, 266 (1935); alsoTech. Assoc. Papers, Ser. 18, 392 (1935).

2 D. B. Judd, Paper Trade J., Tech. Sec. 103, 154 (1936); alsoTech. Assoc. Papers, Ser. 19, 359 (1936).

3 H. Hemmendinger and J. M. Lambert, J. Am. Oil Chemists'Soc. 30, 163 (1953).

4 A. C. Hardy, J. Opt. Soc. Am. 25, 305 (1935).

in the blue like the curves for the newsprint paper andthe interior paint shown in Fig. 2. This reduced re-flectance in the blue causes the surfaces to appearyellowish.

Geometrically, a white surface is one which reflectsdiffusely in all directions. Curves obtained for the samefour specimens with the Hunterlab Recording Gonio-photometer are shown in Fig. 3. It can be seen thateveryday white surfaces depart substantially from theirtraditional uniformity of directional reflectance indifferent directions. These departures are in additionto the differences in geometric reflectance centering inthe specular direction which are associated with glossand are unrelated to whiteness. This geometricalvariability of the reflectances of whites is sometimesresponsible for troublesome discrepancies betweenresults from different instruments used to measure them.

For specifications of color, three numbers are used forwhites as for other colors. Whites can be specified bytheir CIE coordinates, by dominant wavelength,purity, and luminous reflectance, or by their Munsellnotations which correspond closely with the respectivehue, saturation, and lightness dimensions of thepsychological color solid in Fig. 1. Visual uniformity ofcolor space is one requirement for a satisfactoryformula for the expressions of small color differences.If one already has specifications in terms of Munsellrenotations, then the lightness dimension and the polarcoordinates of the color solid in Fig. 1 may be used as abasis for describing whiteness, either assuming thewhite point at neutral, or assigning a near-white

IGTH, my

FIG. 2. Spectrophotometric curves for four white surfacesselected at random: (1) cotton broadcloth, (2) newsprint paper,(3) porcelain enamel, and (4) interior paint.

598 Vl. 48

September1958 DESCRIPTION AND MEASUREMENT OF WHITE SURFACES 599

TABLE I. Intercomparison of five slightly different L, a, and b coordinates systems, each based on anapproximation to a "uniform color solid." Illuminant is always CIE Illuminant C.

Equations for color dimensions in terms of CIE X, Y, and ZName, date, and White-black Red-green Yellow-blue

references Important features (L) dimension (a) dimension (b) dimension

Judd-Hunter Used with subtractive lOOkYi 700Y 1.25X Y+01+ 650ZI ( 280Y1 5Y 0 653Z(1.25X±2Y+0.6563Z/ 1.25X+2Y+0.6563Z

(19 42 ) b (visual) colorimeterb (where k varies from or in Hunter A, G, B or in Hunter A, G, B

and Hunter Multi- 0.2 to 1.2 depend- 700GI( A-sG 280Y A+2G+B)

purpose Reflectometerf ing on proximityof specimens beingcompared)

Judd-Hunter Simpler than foregoing 1OOY 7 i( 1225X-8Y+0190O7Z ) 8Y Y-0.847Z)(1.25X+2Y+0.6563Z/ k1.25X+2Y+0.6563Z

Scofield (1943)0 when used with tri-stimulus reflectometer

Adams chromatic Requires tables to con- 92v(Y) 400[v(1.02X) -v(Y)] 160[v(Y) -v(0.847Z)Jvalue (1944)d vert X, Y, and Z to

Munsell value (V)equivalents v = 0.1 V

Hunter Rd Color-Difference Meter lOOY, where 175fy(1.02X- Y) 70fy(Y-0.847Z)

(1948)6 designed so that a and fr=0.5i 11 +20y)

b compute automaticallyafter Rd (Y) is set

Hunter L Color Difference Meter lOOYi 175Y-i(1.02X-Y) 70Y-i(Y-0.847Z)(1949)6 computes all 3 auto-

matically

See reference 8.b D. B. Judd, Textile Research 9, 253 (1939); D. B. Judd, Am. J. Psychol. 52, 418 (1939).o F. Scofield, "A method for determination of color differences," Natl. Paint, Varnish Lacquer Assoc.. Sci. Sec. Circ. No. 664, Washington, D. C. (1943).d See reference 6.O See reference 7.f R. S. Hunter, J. Opt. Soc. Am. 30, 536 (1940). [See also R. S. Hunter, J. Research Natl. Bur. Standards 25, 581 (1940).]

position to which color difference may be related.(Several applications of disk colorimetry have beenmade in this manner.1 )

For other cases, rectangular dimensions of the colorsolid may be used for expressing whiteness, in fact, thisis probably the most widely used method. These

FIG. 3. Goniophotometric curves for 45° incidence of same fourspecimens used for Fig. 2: (1) cotton broadcloth, (2) newsprintpaper, (3) porcelain enamel, and (4) interior paint.

5 D. Nickerson, Color measurement and its application to colorgrading of agricultural products, U. S. Dept. Agr. Misc. Publ.No. 580 (1946).

dimensions, illustrated in Fig. 4 in skeleton form by ahorizontal cut through the center of the color solid, are(1) lightness which is the same vertical dimension asin Fig. 1; (2) a, which measures degree of rednesswhen plus (+a), gray when zero, and degree of green-ness when minus (-a); and, (3) b, which measuresdegree of yellowness when plus (+b), gray when zero,and degree of blueness when minus (- b).

The two ectangular chromatic dimensions, yellow-ness-blueness and redness-greenness, correspond closelyto visual impressions of differences between whitecolors. The rectangular coordinates of the solid6' 7 havethe advantage over the polar coordinates in thatrectangular dimensions are convenient for computingsmall color differences from the data of several widelyused colorimeters. The L, a, and b values of color maybe read directly with one type of photoelectric colorim-eter7 and may be computed easily from readings on anumber of tristimulus reflectometers.

In practice, there are several different sets of L, a,

I E. Q. Adams, J. Opt. Soc. Am. 32, 168 (1942) D. Nickerson,Am. Dyestuff Reptr. 39 (August 21, 1950); and G. L. Buc, Am.Dyestuff Reptr. 41 (June 9, 1952).

7 Gardner Laboratory, Bethesda, Maryland, Description andinstructions for Hunter color and color difference meter (June,1950); Van den Akker, Aprison, and Olson, Tappi 34, 143 A(1951); and R. S. Hunter, Direct-reading, photoelectric color andcqlor-difference meter (In preparation).

RICHARD S. HUNTER

100 HITE

-V_9 0 1 B L0"SCOK "/00°

FIG. 4. Dimensions of the rectangular L, a, b color solid: L-light-ness, a-redness-greenness, b-yellowness-blueness.

and b dimensions, each related to the CIE system byslightly different equations. These different scales existbecause different data on perceptual uniformity wereused in their various developments and because theywere designed for convenience in different use situations.Table I identifies five of these different L, a, and bcolor coordinate systems.

METHODS OF EVALUATING WHITE COLORS

The human eye, and three types of color measuringinstruments are used for color and whiteness evaluationof white materials (see Table II).

Visual Examination of Whites

In Fig. 5 are shown three arrangements used com-mercially for the visual examinations of white colors.The first two arrangements are used both for examiningadequacies of match to standard whites and for visualintercomparisons and estimates of whiteness. The thirdis used largely for whiteness evaluations.

Standardized artificial light sources and inspectioncabinets are popular but natural daylight, preferablyfrom the north sky, is still used for much if not mostvisual color examinations. As is shown in Fig. 5, it ispossible with nonrigid white materials to enhance hueby multiple reflections within cavities of the materialbeing studied. In practice, inspectors use all cluesavailable to them and all the techniques they havedeveloped from past experience in making colorexaminations of white materials. They are frequentlyunaware of the mechanisms by which they make these

judgments even though these judgments are, in general,surprisingly acute.

The perceptibility of color differences betweenwhites varies with illuminant differences. Differences inthe yellowness of near-white surfaces are due to differ-ences in absorption of blue light. Relatively, daylightilluminants have two to five times as much energy in theblue as do incandescent illuminants. The eye is thereforemuch more sensitive to yellowness differences indaylight than in incandescent light. In other words,differences in the b dimension of Fig. 4, which arereadily seen under illuminants rich in blue, are not seenunder yellowish illuminants. This is one of the reasonsthat bluish illuminants are selected for the visualinspection of whites where yellowness differences areimportant. (Examples: 7500'K artificial daylight isnow standard for color inspection of raw cotton andcolor printing on white paper.)

Instruments Used to Measure White Surfaces

By comparison with the eye, instruments for ratingwhite surfaces have a great advantage in that theyfurnish recordable and repeatable numbers rather thanmere mental impressions. However, instruments havelittle if any advantage in chromatic sensitivity over theeye. With many of them, moreover, readings have to betransformed by more or less complex computationalprocedures to provide numbers corresponding to visualappearance.

Instruments used to measure color of whites are ofthree types: (1) spectrophotometers, (2) tristimulusfilter reflectometers, and (3) trichromatic colorimeters.Figure 6 shows by block diagrams the basic differencesbetween these three types. Table III lists today's mostpopular color-measuring instruments of each of thesethree types. All are photoelectric.

None of these color measuring instruments nowavailable is importantly better than the human eyein ability to detect certain small color differences.Although differences in the vertical dimension of thecolor solid can easily be measured with precisionexceeding that of the eye, horizontal differences cannotbe measured with the same precision. For the red-greendimension, it is necessary that tristimulus refiectometers

LE II.

Whiteness evaluationsTrichromatic color evaluations Partial Complete

Compliance Adequacy of evaluations (trichromatic)with color match to of whiteness evaluations

Reasons for study of whites specifications standard by reflectance of whiteness

ApparatusHuman eye X XReflectometer with blue or

green filter XTristimulus Colorimeter or

Tristimulus Reflectometer X X X XSpectrophotometer X X X X

600 Vol. 48

September1958 DESCRIPTION AND MEASUREMENT OF

ARTIFICIAL DAYLIGHTINGUNIT

NATURAL NORTH-SKYLIGHT

USING MULTIPLE REFLECTIONSIN CAVITIES TO ENHANCE

SATURAT ION

FIG. 5. Three lightingarrangements used forvisual examination ofwhite materials.

DAYLIGHT'BLUE FILTER

-FROSTED GLASS

LIGHT GRAY

and clorimeters detect signal changes of only 0.03%of the signals being measured to equal the best eyes inability to see color difference.

A spectrophotometer uses a prism as suggested inFig. 6 or a grating to isolate wavelengths of the visiblespectrum. It gives curves, as shown in Fig. 2 ofreflectance as a function of wavelength.

A tristimulus reflectometer or colorimeter uses filtersinstead of the prism or grating used in a spectro-photometer to isolate different wavelengths. It is notpossible with filters to isolate narrow spectral bands.However, the tristimulus instruments widely used forcolor measurements8 require filters transmitting ratherbroad spectral bands corresponding to the spectralresponse functions of the CIE-standard observer.Figure 7 shows the spectral character of most of thesource-filter-photocell combinations used for themeasurement of white surfaces.

Colorimeters give not values of reflectance, butnumbers corresponding to position in the color solid.For example, Hunter's photoelectric color and colordifference meter 7 gives results directly in the L, a, andb dimensions of Fig. 4.

As will be seen from Table III and Fig. 6, there aregeometric as well as spectral differences betweeninstruments used to measure color. For reflectancework, commercial instruments use either a near normal-diffuse or a 450 0° geometry. The 450 0° geometry isadvantageous where it is necessary to avoid specularreflectance (responsible for gloss) in the measurementof color. The diffuse geometry is more efficient in lightutilization. Results with it do not vary with rotationof a textured specimen in its own plane as do thosewith the 450 0° geometry.

However, such variation may occur, even withdiffuse geometry, if the light incident on the sample is

8 R. S. Hunter, Photoelectric tristimulus colorimetry with threefilters, Natl. Bur. Standards Circ. No. C429 (1942).

polarized, as in the General Electric Recording Spectro-photometer, or if the efficiency of the collection systemdepends upon the plane of polarization of the lightreflected from the sample.

Matches to Standard

In the commercial production of many whites, coloruniformity of product is as important a goal as white-ness. Thus, all the paper used in a single book, thecollars and body pieces used to make a man's shirt, orthe white paint and porcelain enamel used for thehousehold appliances which will be assembled in akitchen should, in each case, have a uniform whitecolor.

Industrial color matching of whites is still donelargely by eye, although instruments and numericalcolor tolerances computed from their readings nowenjoy some use. With most materials of commerce, thetolerances are small. For example, a mail order housewhich issues a standard white panel for the color ofkitchen appliances, requires that the color of this panel

-A\

L 2.PRISM IN FILTERS ON DISK IN

SPECTROPHOTOMETER TRISTIMULUS REFLECTOMETER

(a)

0° DIFFUSE(b)

3.FILTERS USED SIMULTANEOUSLYIN TRISTIMULUS COLORIMETER

450 O0

FIG. 6. Block diagrams showing (a) wavelength selectingdevices and (b) geometric conditions used in instruments for thecolors of white surfaces.

t

. = . . _

8

I. I

1J

WHITE SURFACES 601

RICHARD S. HUNTER

TABLE III. Commercial instruments regularly used for color measurements of whites.

Instrument and manufacturer Geometry Illuminant color Remarks

(1) SpectrophlotwnetersG. E. Recordings Near normal Plots curves automatically. Tristimulus(General Electric Company, Schenectady, diffuse Any integrator may be added.

New York)Beckman Model DUb 450 O° Any Wavelength and standard value set be-(Beckman Instruments, Fullerton, California) fore each reading

(2) Tristimilus filter reflectontetersHunter Multipurpose, 450 O Daylight Convenient only for whites and light(Gardner Laboratory, Bethesda, Maryland) colorsPhotovolt Model 610 450 O° Daylight Low cost. Small search unit movable to(Photovolt Corporation, New York City) test surfacesColor Eye Near normal Good comparator, particularly for dark(Instrument Development Laboratories, diffuse Daylight or lamplight specimens

Needham, Massachusetts)Colormaster Differential Colorimeterd 450 O° Daylight Good stability, Precise comparator(Manufacturers Engineering and Equipment

Company, Willow Grove, Pennsylvania)(3) Trichlrornatic calorimeter

Hunter Color-Difference Meter" 450 00 Daylight Reads color directly on L, a, b "uniform(Gardner Laboratory, Bethesda, Maryland and color scales"

Hunter Associates Laboratory, Falls Church,Virginia)

Nickerson-Hunter Cotton Colorimeterf 450 O° Daylight Reads luminous reflectance and +b(Gardner Laboratory, Bethesda, Maryland) (yellowness) automatically

a See reference 4.b H. H. Cary and A. 0. Beckman, J. Opt. Soc. Am. 31, 682 (1941).o R. S. Hunter, J. Opt. Soc. Am. 30, 536 (1940). See also R. S. Hunter, J. Research Natl. Bur. Standards 25, 581 (1940).d L. G. Glasser and D. J. Troy, J. Opt. Soc. Am. 42, 652 (1952).o See reference 7.See reference 14.

be matched to tolerances which average AY=40.015,Ax= 40.0025, Ay= 0.003. Instrument measurementshaving the precision necessary to establish conformityto these small tolerances must be made with care.

There is still no agreement on how best to obtainand maintain permanent standards for the visual colormatching of impermanent materials. Ceramic porcelainenamels and some plastics are quite permanent incolor and may therefore be retained as standard whites.

FIG. 7. Spectral character of five source-filter-photocell combina-tions used for color measurements of white surfaces.

Special paint-like pastes are sometimes compoundedwith nonreactive oils and pigments to serve aspermanent standards of color for the production ofpaints. For papers, textiles, foods, and other im-permanent materials, careful instrumental measure-ments are used to follow the color drift of standardsamples so that the users of standard samples of thesematerials know how best to maintain them and howlong to trust them for color permanence. Permanentceramic standards are sometimes sought to match thecolors of these impermanent materials. However, thetexture differences in appearance between ceramic andnonceramic materials frequently make visual color com-parisons difficult. Where instruments are used to measureimpermanent white materials by reference to permanentwhite standards, the problems of intercomparingsurfaces of different texture largely disappear. It is forthis reason that the use of instruments for color match-ing of whites is growing.

Color Specifications

Unfortunately, there exist no well-standardizedassortments of white chips like the colored chips of theMunsell Color System and Container CorporationColor Harmony Manual. There is thus no organizedsystem with which to specify white colors by closestvisual match. One who wishes to specify a white coloris limited to specifications based on instrument measure-ments unless he is in a position to prepare and supplyhis own standard white chips.

Color specifications for whites may employ the

602 Vol. 48

September1958 DESCRIPTION AND MEASUREMENT OF WHITE

TABLE IV. Reflectance measurements of white colors.

Paints Papers Textilesand and and

Applications plastics printing detergents Ceramics

Luminous reflectance Efficiency of surfaces used to re- x x x(Green light) flect lighta(ye, Fig. 7) Cleaning efficiency measured by x

gray soil removalOpacity by contrast-ratio x x x

methodb e

Blue-light reflectance Degree of bleach and chemical x x(B and iA, Fig. 7) purityd e

Brilliance attainable by dyeing x xor color printing

Effectiveness of fluorescent x xwhiteners (Measure with UVin illuminant)

a See reference 9. b See reference 10. c See reference II. d See reference 13. e See reference 12.

conventional standard coordinates of the CIE system.Thus, American flags purchased by the Federal Govern-ment are required by Federal Specification TTC-591to visually match a white color specified by Y=0.57,x=0.32, y=0.33 relative to magnesium oxide underCIE illuminant C. However, many whites are specifiedin the L, a, b and other coordinate systems. Thus, theQuality Control Department of a soap manufacturerrequires that a white detergent powder which it marketshave the color specified by L= 93 min, a= 04t 1.0,b=3.0H41.0. This sort of specification is, of course,intended for use with a tristimulus colorimeter orreflectometer. Tristimulus instruments are widely usedfor whites because the spectral errors to whichtristimulus instruments are subject are minimized bythe lack of spectral selectivity characteristic of whites.

Reflectance partial measurements of whiteness arewidely obtained with one or another of the tristimulus-filter-photocell combinations shown in Fig. 7. Differentfilters and different instruments have been adopted bycustom and use in different industries. The. mostcommonly used reflectance methods for whites areidentified in Table IV together with the materials forwhich they are used and the aspects of whiteness eachis designed to measure. A number of these industry-wide, one-number reflectance methods for whitematerials have been used for many years and arefirmly established as partial measures of whiteness.None of them provides a complete measurement ofcolor because none involves the three determinationsnecessary.

Luminous reflectance measurements with the yc(green) combination in Fig. 7 are used for detergencystudies of soaps, of cleaning agents and of the machineswhich perform the cleaning. Textile swatches artificallysoiled to a gray color are measured before and aftercontrolled cleaning operations. The gain or percent gainin reflectance of each swatch is used as a measure of

the effectiveness of its detergent, cleaning process, or

combination.

For paints used on interior walls, and ceramics andplastics employed in lighting units, luminous reflectancemeasurements are used to measure light-conservingefficiency. Paints are widely specified by minimumacceptable luminous reflectance. There are ASTM andFederal standard methods of test for their measure-ment9,10

Luminous reflectance measurements are employedfor opacity evaluations of paper, paint, and porcelainenamel. Films of these materials are measured foropacity by the contrast ratios of their luminous re-flectance when backed by black, divided by that whenthe same films are backed by white. The paint, paper,and ceramic industries, and the Federal Governmenthave standard reflectance methods for opacitymeasurement. 0 ,1

Blue reflectance is widely used as a partial measure ofwhiteness in the paper industry for two reasons:

(1) Since white papers normally have poorestreflectance in the blue region of the spectrum, bluereflectance gives a combination measure of the freedom-from-grayness and freedom-from-yellowness com-ponents of whiteness.

(2) Blue reflectance measures the freedom of finishedpapers from the yellowish impurities which are removedby bleaching. It therefore correlates with the com-pleteness of bleaching and with paper permanence.

Blue reflectance measured under specified standardconditions is termed "paper brightness" in the industry.It is used commercially for the specification of manypaper raw materials as well as for finished papers.A test method, which is now standard for the Technical

9 Am. Soc. Testing Material Method E97-53T, 45-deg, 0-degdirectional reflectance of opaque specimens by filter photometry,Am. Soc. Testing Materials (1953).

10 Federal Specification TT-P-141b, Paint, Varnish, Lacquer andRelated Materials; Methods of inspection, sampling and testing:Method 612.1 directional reflectance, Method 412.1 dry opacity)Method 613.1 yellowness and yellowing.

11 Tech. Assoc. Pulp Paper Ind. Method T425m-44, Opacity ofpaper (1944).

SURFACES 603

RICHARD S. HUNTER

FIG. 8. Block diagram of Hunterlab Reflectometer used formeasurement of fluorescence contribution to blue reflectance:L-lamp, N-ultraviolet absorbing filter before specimen, S-specimen, F-same filter after specimen absorbing reflectedultraviolet, but not ultraviolet converted to blue by fluorescence,B-blue filter, and P-photodetector.

Association of the Pulp and Paper Industry,' 2 uses theblue-light spectral distribution labeled B in Fig. 7.Calibrated standards for this test are regularly circu-lated through the paper industry by the Instituteof Paper Chemistry of Appleton, Wisconsin. Thespectral distribution specified by the ASTM methodfor paper brightness' measurements is somewhatdifferent, being the CIE z function for illuminant a(curve A in Fig. 7).

Because it also deals largely with cellulose materials,it is suggested that the textile industry might find bluereflectance a useful measure of approximate whitenessfor the same reasons the paper industry has found ituseful. Blue reflectance measured under the properconditions could be made to evaluate still an additionalfactor of considerable commercial interest to the textileindustry. The so-called optical bleaches are widely usedto enhance whiteness by ultraviolet-stimulated fluo-rescent blue light which supplements the normallydeficient natural blue reflectance. A measurement of thecontribution of near-ultraviolet to blue reflectance maybe used to rate these optical bleaches using a reflec-tometer such as is shown by diagram in Fig. 8. It willbe seen that this reflectometer will measure bluereflectance with the filter which absorbs near ultravioletin either of two positions: (1) before the specimen sothat ultraviolet-induced fluorescence does not occur,and (2) after the specimen where the ultraviolet-induced fluorescent blue light is added to the naturallyreflected blue light.

1 2 Tech. Assoc. Pulp Paper Ind., Method T425m-48, Brightnessof paper (1948).

13 Am. Soc. Testing Material Method D985-50, Method of testfor 45-deg, 0-deg directional reflectance for blue light (brightness)of paper (1950).

Combined Luminous Reflectance and YellownessRatings of Whiteness

With many white materials of commerce, departuresfrom neutral in the red or green direction are normallyless pronounced than those in the yellow or bluedirection. Organic materials, in general, have spectralcurves sloping downward toward the blue. Manyinorganic materials have the same tendency whichproduces yellowish colors. Thus, the a dimension ofcolor difference can frequently be ignored in evaluatingwhiteness, but the b dimension is usually important.

By adding procedures for yellowness to those forreflectance just shown, it is possible to obtain whitenessdeterminations adequate for many purposes. Differencesbetween reflectances in the yellow and blue portions ofthe spectrum form the basis for some measurements ofyellowness. One plastic manufacturer uses the followingratio computed from readings of the spectrophotometriccurves:

(R 700-R 450)/R7 00.

The Federal Specification'0 method for yellownesscalls for photoelectric tristimulus measurements ofblue, amber, and green 450 0° reflectances and com-putation of the ratio:

(A-B)/G as yellowness.

B, G, and A are, respectively, the source C reflectancesfor the Zc and Yc functions, and the long-wave portionof the xc function (XCA), see Fig. 7. These may bemeasured with any of the tristimulus reflectometerslisted in Table I.

The Nickerson-Hunter Cotton Colorimeter 4 is uniquein that it not only computes but plots automaticallyboth the lightness and yellowness values of near-whitecolors. Raw cotton is color graded on the basis of thesetwo dimensions. It is interesting to note that the gradingscale for raw cotton differs from the usual whitenessscale in that within the limits of what are called the"white grades," a light creamy color may be preferredto an equally light color that is more free of yellowness.This departure from the usual criterion of whitenessprobably corresponds to the fact that freshly openedbolls of cotton have a yellowish or creamy color whichbleaches out if they are left too long in the sun andweather. (This preference is not, however, universal andit falls down completely when the creaminess is suffi-ciently deep to change the grade from a "White"classification to one that must be called "Spotted" or"Tinged.")

Complete (Trichromatic) Measures of Whiteness

Whiteness is an important competitive attribute ofmost white materials in commerce because of itsassociation with cleanness and purity. Most commercial

4 Nickerson, Hunter, and Powell, J. Opt. Soc. Am. 40, 446(1950).

604 Vol. 48

September1958 DESCRIPTION AND MEASUREMENT OF WHITE SURFACES 605

ratings of whiteness and relative comparative rankingsof whiteness are based on visual judgments.

The problem of how to obtain numerical values ofwhiteness from tristimulus values has received attentionfrom a number of investigators during the past twenty-five years. MacAdam,'5 Judd,' 2 Harrison, 6 Hunter,8

and Selling and Friele'7 have all proposed formulas orcolorimetric procedures for reducing calorimetric specifi-cations of white colors to a single-valued measure ofwhiteness. Asmussen and Buchmann-Olson,'8 Sellingand Friele,'7 Hutchins, Sterns, and Sundstrom,'9

Hemmendinger and Lambert,' and MacAdam,"1 areamong the many who have intercompared whitenessformulas for different applications. The formulas whichhave appeared to date have not enjoyed general usage,probably because they seem too complex for routineapplication.

A formula for whiteness which has enjoyed some useis based on the assumption that equal distances betweenpoints in the L, a, and b color solid of Fig. 4 representdifferences of equal perceptibility 8 :

W= 100-[(100-L)'+ (a'+b')]i.

Sometimes Lp, the lightness of the preferred white ofwhatever material is under study, is substituted forthe second "100." This has the effect of changing froman ideal to an actually attainable preferred white. If onestudies comparative whiteness rankings made understandard observing conditions with tristimulus valuesfor the same white surfaces, he finds that such anequation for distance in Euclidian color space does not

1D. L. MacAdam, J. Opt. Soc. Am. 24, 188 (1934).16 V. G. W. Harrison, Patra J. Nos. 2 and 3 (London, 1938,

1939).17 H. J. Selling and L. F. C. Friele, Appl. Sci. Research B1, 453

[Vezelinst. T.N.O. (Delft), 1950].18 R. W. Asmussen and B. Buchmann-Olsen, Trans. Danish

Acad. Tech. Sci. No. 9 (1949).19 Hutchins, Stearns, and Sundstrom, Tappi 35, 342 (1952).20 D. L. MacAdam, Tappi 38, 78 (1955).

0 5 tOb, YELLOWNESS

FIG. 9. Lightness-yellowness diagram showing by dottedquarter circles colors of equal whiteness by the Hunter equation, 8

and by solid lines, colors of equal whiteness according toMacAdam.15

accurately represent visual whiteness rankings. Figure9 is an L, b diagram on which are drawn: (1) concentricquarter circles corresponding to whiteness ratings bythe foregoing equation and (2) loci of equal whitenessranking obtained by MacAdam.15 It will be seen thatthe relation between yellowness (b-scale value) andwhiteness is nonlinear. It would appear from Mac-Adam's experiments that when yellowness (b) exceedsa certain minimal value, it interferes strongly withwhiteness acceptability. Because it can eliminate thisyellowness, the use of small amounts of blue dye canfrequently improve whiteness, even though a blue dyealways decreases lightness.

It is hoped that further work with direct-readingwhiteness instruments like the Cotton Colorimeter,4

and with whiteness formulas like those by MacAdam,Judd, -and others will lead to more widespreadfamiliarity with, and acceptance of, single-numberwhiteness rankings derived from tristimulus data.