JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

Spectral Energy Distribution of the International Commission on IlluminationLight Sources A, B, and C*

RAYMOND DAVIS, KASSON S. GIBSON, AND GERALDINE WALKER HAUPTNational Bureau of Standards, Washington, D. C.

(Received November 24, 1952).

Standard light sources A, B, and C were adopted by the International Commission on Illumination(C.I.E.) in 1931 for the colorimetry of materials. Source A is an incandescent lamp operated at 28540 K.Sources B and C, representative of average noon sunlight and average daylight, respectively, are producedby passing radiant flux from source A through specified Davis-Gibson filters described in Bureau of Stand-ards Miscellaneous Publication M114 (1931).

The C.I.E. published relative energy data only from 370 to 720 mu. The present paper gives data for thesesources in the ultraviolet and extends the data in the red to 780 mp. The new data were reported through theU. S. Secretariat Committee to the C.I.E. at its 1951 meeting. The values in the ultraviolet were reported tothe Optical Society of America at its March, 1952, meeting.

I. INTRODUCTION

IN 1931 the International Commission on Illumination(Commission Internationale de l'Eclairage, C.I.E.)

recommended that three light sources A, B, and C, beadopted for the general colorimetry of materials.1 Thesewere defined as:

A. A gas-filled lamp at the color temperature of28480 K (c2 = 14 350 micron degrees).t

B. The same lamp used with a filter consisting of alayer, one centimeter thick, of each of two solutions

contained in a double cell with colorless optical

0

Values of x

FIG. 1. Chromaticities of C.I.E. sources A, B, and C, togetherwith spectrum locus and blackbody locus-all plotted on the1931 C.I.E. chromaticity diagram.

* A more extended treatment of this investigation may be foundin J. Research Natl. Bur. Standards 50, 31 (1953), RP2384.

'C.I.E., Compt. rend. 19-29 (1931).t The spectral energy distribution for 28480 K with c2 = 14 350

micron degrees is the same as for 28540 K with c2= 14 380.

glass plates. The compositions of the two solutions arespecified in the resolution.

C. The same lamp in combination with a secondfilter of the same type but with different composition,also specified in the recommendation.

Source A is intended to be typical of light from thegas-filled tungsten-filament incandescent lamp, sourceB is an approximate representation of noon sunlight,and source C is an approximate representation of aver-age daylight (see Fig. 1). The relative spectral energydistributions for sources B and C are given in theC.I.E. recommendations over the wavelength rangefrom 370 to 720 mut

The two filters which, in combination with source A,comprise sources B and C are of the type designed byDavis and Gibson and described in detail in Bureauof Standards Miscellaneous Publication M114.2 Theyare two of a large number of such filters designed forthe reproduction of the various phases of sunlight anddaylight and the conversion of low to higher colortemperatures. The particular filters used to obtainsources B and C are designated in reference 2 as 28480Kto 4800'K and 28480K to 6500'K, because they weredesigned to produce exactly the colors corresponding to48000 and 6500'K, respectively, for the observer definedby the Optical Society of America (O.S.A.) excitationsextrapolated. For the 1931 C.I.E. standard observer',and the modification of it tentatively proposed in1951,5 the colors produced are slightly too purple tomatch the colors of the blackbody at any temperature.The temperatures of the blackbody most similar tothese colors (called correlated color temperatures) areas given in Table .6

From the indicated departures from the design colortemperatures of 4800'K and 6500'K, respectively, it

2 Raymond Davis and K. S. Gibson, "Filters for the repro-duction of sunlight and daylight and the determination of colortemperature," Misc. Pub. Bur. Standards No. 114 (1931), M114.

T. Smith and J. Guild, Trans. Opt. Soc. 33, 73 (1931-1932).4Deane B. Judd, J. Opt. Soc. Am. 23, 359 (1933).5 C. I. E., Compt. rend. 1, 7, 11 (1951).

Deane B. Judd, Color in Business, Science and Industry (JohnWiley & Sons, Inc., New York, 1952), p. 96.

172

VOLUME 43. NUMBER 3 MARCH, 1953

C.I.E. LIGHT SOURCES A, B, AND C

TABLE I.

Correlated color temperature, degrees Kelvin Departure from designC2 = 14 350 C2 = 14 380 color temperature prd

Source Source Source Source Source SourceObserver B C B C B C

O.S.A. "Excitations" 4800 6500 4810 6514 0.0 0.0

1951 C.I.E. Tentative 4815 6560 4825 6574 0.7 1.4

1931 C.I.E. Standard 4860 6726 4870 6740 2.6 5.4

will be noted that these designations are only approxi-mate if the 1931 C.I.E. standard observer is used toassess the colors. But if the tentative 1951 modifiedobserver is used, the designations may be consideredprecise for ordinary observing conditions under whichdifferences in reciprocal color temperature in the neigh-borhood of 1.0 jird (micro-reciprocal-degree) are im-perceptible.

Regardless of this, however, the spectral distribu-tions of C.I.E. light sources B and C define thesesources in unambiguous terms irrespective of observeror color temperature; and these distributions give ap-proximately the colors of noon sunlight and averagedaylight.

It happened that charts illustrating these particularfilters and the resulting spectral distributions were notgiven in reference 2 among those chosen for illustration.The chart for the 28480-to-4800'K filter and source waslater published7 but that for the 28480-to-65000K filterhas not been published. Accordingly, as a matter ofconvenience, these are both reproduced in slightlyrevised form in the present paper, Figs. 2 and 3.Hereinafter the sources (and filters) will be designatedas sources B and C (or filters B and C) as they weredesignated by the C.I.E. in 1931.

Reference to Figs. 1-3 will give considerable infor-mation about these filters and the resulting lightsources. Many other details relating to the filters maybe found in the original publication.2

In the (U.S.) Secretariat questionnaire and reportprepared for the 1948 meetings of the C.I.E. in Paris8

the question of standard sources was raised anew, andthe following question was asked:

"In view of the possible influence of ultraviolet

irradiation on the colours of fluorescent samples, doyou believe that the definition of standard illuminantsA, B, and C should be made more precise in regard tothe ultraviolet?"

The authors of reference 2 offered to extend the datafor sources B and C below 350 mpi if that seemeddesirable.

Replies to the above questions were received fromseveral countries and after discussion at the meetingsthe following recommendation was adopted:8

"Ultraviolet Content of Standard Illuminants-(a) Itis recommended that the offer made through the United

7 Raymond Davis and K. S. Gibson, Bur. Standards J. Research7, 791 (1931), RP 374.

8 C. I. E., Compt. rend. 16, 238 (1948).

States Committee by Messrs. Davis and Gibson todetermine the relative energy distributions of illumin-ants B and C below 370 mAt, including the spectraltransmission factor of the cells, be accepted."

It was also recommended that the National Com-mittees study the question of a standard source havingrelatively more ultraviolet energy than sources A, B,and C, for the colorimetry of fluorescent materials.

In compliance with the first recommendation thespectral transmittances of the B and C filters have beenredetermined and extended in the ultraviolet, and thesetransmittances and the resulting spectral energy dis-tributions were reported to the 1951 meetings of theCommission in Stockholm through the (U.S.) Secre-tariat Committee.9

In addition there are given herein some publishedand unpublished data for the long-wave end of thevisible spectrum. In the charts of reference 2, as illus-trated in Figs. 2 and 3 of the present paper, the datacovered the visible range only to 720 ma, this beingthe long-wave limit of the O.S.A. excitations used inthat work. However, the spectrophotometric data onthe component filter parts had been determined to750 m/12 so that extension of the transmission andenergy data could readily be made.

When the C.I.E. distributions defining the newstandard observer were adopted, they were extended asfar as 780 m,4.' At the request of Smith and Guild'values of relative energy for sources B and C wereextended to 780 m,4. These values were published inTable V, page 102, of their paper. They were also pub-lished by Judd.4

II. SPECTROPHOTOMETRIC DATA

The spectral transmittances of the filters as given inFigs. 2 and 3 in the columns headed "T" were derivedby computation from spectrophotometric measure-ments made on certain standard solutions2 including astudy of Beer's law for the respective components overthe range of concentrations of interest. This previousspectrophotometric work on the standard solutions wasvery extensive. It involved, for each standard com-ponent of the filter, measurements with several differentsolutions, at cell thicknesses from 1 to 10 cm and withthree or four spectrophotometers. A detailed discussionof these measurements has been given.2 All data wereobtained at 250 C. Check measurements were also madeon the complete filters.

9 C.I.E., Compt. rend. 3, 116 (1951).

173March 1953

DAVIS, GIBSON, AND HAUPT

2848IK TO 4800'K

FIG. 2. C.I.E. source B.Spectrophotometric data andother information on this lightsource and the filter used toobtain it. This chart is similarto the charts of references 2and 7. Values below 350 andabove 720 mu are given inTable TI.

No such .elaborate series of measurements was con-sidered necessary for the present purpose, which wasprimarily to extend the values of spectral transmit-tance and energy below 350 m as far as there wassignificant transmittance. Since the filter cell contains

three borosilicate glass plates of total thickness about7 mm, this transmittance could not in any case extendmuch below 300 m/.

Two sets of solutions were prepared, one in February,1950, and one in January, 1951. Spectrophotometric

[

174 Vol. 43

C.I.E. LIGHT SOURCES A, B, AND C

' 2848° K TO 6500°K

M0)LaC5,1

350 400 450 500 550 600 650 700 7Wave Length--millimicrons (m,u)

(mFll) T E" * E'

350 0.23 1 7,0 .07260 .330 12.9 ./2 770 .42? 21.4 20880 52 9 33.0 .31290 *6.5 I 7. t /ii4

400 L675 63.3 .5790 .71 i g0.6 .129

20 .733 9 1. .i1030 .7 1/ I 1 2.Jf I .00240 .663 121.5 J20. 1

450 .517 1 24.0 1.10360 .511 123. 1 1.09770 .4S3 123. 1 .10ff80 .402 1/23.9 /./ /690 .35/ 12.'7 /.097

500 .293 112.1 1.;02910 .Z*3 / 02.3 *5020 .210 q 9.9 .9130 .19U 91.0 .93540 .116 /02.1 .911

550 .1Z. /0I.2 I .03S60 .16.5 1 05.3 1.05370 .I'97 1X02.3 I.04'280 .1338 .7.1 1.0,90 .1199 73.2 .916

600 .1090 1I. 7 .7671 0 I0 1 6 11.ff .974'20 .0961 8 .1 .9230 fff1 ?.° 0 40 0Y7/1 .17. 1 033

650 .0138 1 1.z I .062.60 .0101 I7.9 /±093-7gn

80A1r_ IV A

.07 10 1.019 I'90 .06S54 10.2 /.06 0

700 .060' 76.3 1,03510 .05551 72. 1.00520 .0509 6 &.3 .972

Bureau of Standards. Davis-Gibson Filters

_ _ _ _ _ .00

T -- Spectral Transmission of Filter at 5°CV--Relative Visibility FunctionE--Relative Energy of 2848 KE'--Relative Energy of 6500'K

E" (=T xE )*-- o o -- Relative Energy of2848 K and Filter Combination

Light Transmission of Filterfor 2848 K = 0.155

These dta are for a one centimeter layereach of solutions A and B in a double cellwith three plates of borosilicate crown glass(refractive index. D line,= 1.51), each 2.5 mm thick.

* Adjusted to make sum of E- E' from 400to 720 m,u equal practically -to zero

** Factor to be used to multiply the candle-powerof the light source to obtain the candle-powerof the source-and-filter combination

measurements were made on each set of solutionswithin one to four weeks from the date of preparation.In each case two methods of determining the filtertransmittance were used: (1) The spectral transmit-tancies of the four standard filter components were

measured and the transmittances of the B and C filterscomputed for the appropriate concentrations, as was donefor all of the filters of reference 2, (2) direct measure-ments of spectral transmittance were made on B and Cfilters prepared in accordance with the specifications.

March 1953 175

CHART 27

FIG. 3. C.I.E. source C.Spectrophotometric data andother information on this lightsource and the filter used toobtain it. This chart is similarto those of reference 2. Valuesbelow 350 and above 720 muare given in Table II.

FILTER FORMULA

A

Copper Sulphate (CuSO4.5H20) 3.412 gramsMannite (C6H8(OH)6) 3.412 gramsPyridine (C5 H5 N) 30.0 mlWater (distilled) to make 1000. ml

B

Cobalt Ammonium Sulphate(CoSO4'(NH4 ) S04 -6H 20) 30.580 grams

CopperSulphate SCuSQ4 5H20) 22.520 gramsSulphuric Acid (sp.gr. 1.835) 10.0 mlWater (distilled) to make 1000. ml

- , .- -- 1 1 . .- - - 1 .

750

3FA , I JFIF

176 t~~~]AVIS, IBS~ON, ANE FIAtPT VL4TABLE I. Relative spectral efiefgY distributlotts E of Cj.LE.

light sources A, B, and C, and spectral tansmittances 7' of CILE.filters B and C, and of the glass-wvater cell, W.

wavelength

(mA)

30010203040

35060708090

40010203040

45060708090

50010203040

55060708090

60010203040

65060708090

70010203040

750607080

0.931.361.922.663.584.746.147.829.79

12.0914.7117.6821.0024.6728.7033.0937.8242.8748.2553.9159.8666.0672.5079.1385.9592.91

100.00107.18114.44121.73129.04136.34143.62150.83157.98165.03171.96178.77185.43191.93198.26204.41210.36216.12221.66227.00232.11237.01241.67

0.00.00.02

0.52.45.69.6

15.222.431.341.352.163.273.180.885.488.392.095.296.594.290.789.592.296.9

101.0102.8102.6101.099.298.098.599.7

101.0102.2103.9105.0104.9103.9101.699.196.292.989.486.985.284.785.487.0

EC

0.00.00.010.42.77.0

12.921.433.047.463.380.698.1

112.4121.5124.0123.1123.8123.9120.7112.1102.396.998.0

102.1105.2105.3102.397.893.289.788.488.188.087.888.287.986.384.080.276.372.468.364.461.559.258.158.259.1

7 'B

0.0000.0000..0030.0480.1750.3010.4020.4980.5880.6660.7210.7570.7740.7610.7230.6630.6010.5510.5070.4600.4040.3530.3170.3000.2900.2790.2640.2460.2270.2090.1950.1860.1780.1720.1660.1620.1570.1510.14400.13600.12840.12100.11340.10630.10070.09640.09380.09260.0925

0.000.000

0:,0010.0250.1200.2310.3300.4290.5290.615-0.6750.7140.7330.7140.6630.5870.5110.4530.4020.3510.2930.2430.2100.1940.1860.1780.1650.14970. 13380.11990.10900. 10160.09610.09140.08710.08380.08010.07560.07100.06540.06040.05550.05090.04670.04350.04090.03920.03850.0384

TW

0.0100.1020.3140.5360.7020.7980.8490.8710.8850.8890.8910.8920.8930.8940.8950.8960.8970.8980.9000.9010.9020.9030.9040.9040.9040.9040.9040.9040.9040.9030.9020.9020.9020.9010.9000.9000.9000.9000.9000.8980.8970.8930.8830.8710.8630.8610.8620.8640.866

The Beckman quartz (DU) pectrophotometer wagused for all of the present measurements, with tem-peratures of filters and components kept at 25T'. The1950 and 1951 measurements were equivalent in everyrespect, the repetition being made to obtain greatercertainty in the results.

In the complete Bureau paper* will be found theaverage 1950-51 data for (1) the standard filter com-ponents, including the double glass cell filled withwater, (2) the spectral transmittances of the B and Cifiters, and (3) the spectral energy distribution ofsources B and C, together with (4) discussion of thereliability of the data and comparison with the previousdata from 350 to 400 mA. Reference should be made tothat paper,* if information is desired beyond that givenin Table II and Figs. to 3 of this paper.

As a matter of convenience there are given in TableII over the spectral range froml 300 to 780 m/. the re-lative spectral energy distributions of C.I.E. sourcesA, B, and C, the spectral transmittances of filters Band C, and the spectral transmittance of the doubleglass cell filled with distilled water.

For source A the values from 380 to 780 m areidentical with those published by Smith and Guild'and by udd.4 From 300 to 370 m the values havebeen derived by the authors from previous publica-tions,'"'" with allowance for differences in C2 f Planck'sequation.

For sources B and C the values from 380 to 720 m.are identical or consistent with the published val-ues,'1-4 7 except for an erroneous value at 380 m forthe source C as published by the C.I.E. From 730 to780 m the values are as published by Smith and Guildand by Judd.

For filters B and C and for the glass-water cell thevalues from 380 to 720 m are unchanged from theprevious data.

10 R. Stair and W. 0. Smith, J. Research Natl. Bur. Standards30, 449 (1943), RP 1543.

11 J F. Skogland, Misc. Pub. Bur. Standards No. 86 (1929),M86.

176 Vol. 43