Egypt Exploration Society

An Objective Colour-Measuring System for the Recording of Egyptian Tomb Paintings Author(s): Nigel Strudwick Source: The Journal of Egyptian Archaeology, Vol. 77 (1991), pp. 43-56 Published by: Egypt Exploration Society Stable URL: http://www.jstor.org/stable/3821952 . Accessed: 29/03/2014 02:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Egypt Exploration Society is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Egyptian Archaeology.

http://www.jstor.org

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

43

AN OBJECTIVE COLOUR-MEASURING SYSTEM FOR THE RECORDING OF EGYPTIAN TOMB PAINTINGS*

By NIGEL STRUDWICK of the colours used in the reliefand paintingof Egyptianmonumentshas not The standardof documentation been as satisfactoryas that of the documentationof the decorationitself.Recent approacheshave involved the use of colour chartsas a resolutionof this problem.There is a certainlack of objectivityin this, as well as many practicaldifficulties.A MinoltaCR-22I colorimeterhas been taken into the field to attempta more objectiveapproach.This paperis concernedwith the theoreticaland practicalaspects of using a colorimeter to documentcolours;it concludesthatthe instrumentis of greatvalue if properlyused.

I. The problem SINCE 1984, the Christ's College Theban Tomb Project has been working in the Theban Necropolis. The purpose of this expedition is to record and publish some of the so-called 'Tombs of the Nobles', which are threatened with destruction from a number of sources.1 The work has two aspects, the documentation of the wall-paintings and reliefs in the funerary chapels, and the clearance and recording of the underground burial chambers. It is with the first problem that this paper is concerned. The principal subjects of work between 1984 and 1990 have been three tombs in elKhokha, those of Amenhotep (TT 294), Khnummose (TT 253), and Amenmose (TT

254).2 All date to the Eighteenth Dynasty, respectively to the reigns of Amenhotep II, Amenhotep III, and Ay. Significant amounts of colour are found in the chapels of and reliefs are recorded by means of fullKhnummose and Amenmose. These paintings size facsimile drawings (reduced for publication) and photographs.3 Such techniques permit the accurate documentation of most aspects of this decoration, with one glaring exception, that of the colours. All tombs were originally painted, or intended to be painted, whether as the basic method or as the final decorative touch to the relief, but an
*My thanks are due firstly to Minolta Corporation both in Japan and the USA. In particular, John McCasland of Minolta USA supported the original proposal and helped with the actual acquisition of the

assistancewith the detailsof the transaction. CR-22I; his assistant,Suzanne,gave invaluable Approvalto the proposal was given at MinoltaJapan by Mr Y. Kada, and my thanks are due to him. Tracey Walker of Minolta(UK) Ltd.providedthe photographof the instrumentreproducedas pl. IV, i. My greatest single debt is to those colour scientists who have kindly helped a scientificallyilliterate Egyptologistthroughthe colour maze. Max Saltzman(UCLA)has given me much of his time since seeing a first draft of this paper, and has suggested the approach taken in this article. Nick Hale of Hale Color Consultants, Inc.,has read draftsof this paperand given his advice and help freely.Bothhave saved me from manyerrors;those which remainare my own responsibility. BarryKemp and Colin Shell helped to put the instrumentthroughsome initialprovingin the University of Cambridge.I should also like to thankJohn Baines for offeringmany useful comments on a draftof this
manuscript.

1See my article'A Mortifying Case of Rising Damp', TheGeographical Magazine (November,1988),44-8. 2The resultsare to be publishedby the Griffith at Thebes. Institutein the series PrivateTombs 3The best general discussion of the problems of is in Henry G. Fischer in Caminos by Egypt epigraphy and RicardoA. Caminos,Ancient andEpigraphy (New York, 1976), I-25. Egyptian Paleography

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

44

NIGEL STRUDWICK

solution tory solution to the problem of documenting the colours has yet to be entirely satisfactory found. In the days of Norman de Garis Davies, the most prolific and finest publisher of these tombs, it was normal to note the colours in the text by 'red', 'green', etc., often with corresponding 'R' and 'G' on the drawings;4 in addition, Davies and his wife Nina produced superb colour facsimiles of many of the most beautiful scenes.5 The writing of colour abbreviations on drawings has largely fallen out of use, but colour facsimiles are still in many ways the ideal means of publishing a painted tomb; unfortunately, it requires an artist of unusual ability abilityand makes a publication prohibitively expensive. The use of colour photography has become common in recent years, but standards of reproduction vary and there is often insufficient checking (by the publisher/printer) of the printed colours against the original.6 Colour charts have been employed for many years by archaeologists as a method for recording the coloration of soils and ceramics. The first such use in Egyptology known to me dates from the I930s, when the Ostwald Colour Album was employed to describe
the colour of Predynastic sherds found in excavations at Armant.7 Today, the preferred method is the use of Munsell Soil Charts. It is only more recently, particularly with reference to the revival of interest inng Theban tombs since the 970os, that colour charts have begun to be used in epigraphy; the results, at least in terms of a The favoured recognition of the need for documentation, have been encouraging. standard in use is the Munsell Book of Color.9

However, there are two major problems with the use of colour charts, namely,
inconsistency of the illuminant at different times of measurement in the same tomb, and inconsistency of the observer. To the latter should be added thehenatural variation between individual perceptions of colour. The consistency of the observer is not helped by the laborious nature of working through a colour chart in order to find the best match, often

in less than ideal circumstances; one tires very easily and it is easy for one's judgement to be impaired. Having suffered these problems ourselves, I resolved to try and find an electronic measuring system, which should have the advantages of objectivity and consistency, and be appropriate to the field environment; a necessary feature should be that it has its own
light source so that all readings can be taken under the same circumstances. From the

work of the Egyptian Antiquities Organization/Getty Conservation Institute (EAO/GCI) in the tomb of Nefertari, it became clear that such a device was made by the Minolta
Corporation; it is there being used successfully to document the deterioration of colour.10 An approach was made to Minolta for assistance with our project, and a CR-22I colori-

meter was made available to us on very favourable terms.

Such as Nina de Garis Davies, Scenesfrom Some Theban Tombs(Oxford, 1963), pl. i. 5For example, Nina de Garis Davies and Alan H. Gardiner, Ancient Egyptian Paintings, i-iI (Chicago, 'See my comments in a review in JEA 74 (1988), 273-4. Cf. Caminos, op. cit. 12-13. 1936). 7J. Scott Taylor in Sir Robert Mond and Oliver H. Myers, Cemeteries ofArmant, i (London, 1937), 186-7. 8 Notably in the publications of the German Archaeological Institute, such as Artur und Annelise Brack, Das 3 GrabdesIlaremheb (Mainz, 1980) and Das Grabdes Tjanuni (Mainz, 1977). In particular, Munsell Corporation, Book of Color:GlossyFinish Collection(Baltimore, 1976). 10See F. Preusser and M. Schilling, 'Color Measurements', in Egyptian Antiquities Organization and the Getty Conservation Institute, Wall Paintingsfrom the Tomb of Nefertari:Scientific Studiesfor their Conservation (Cairo/Los Angeles, 1987), 70-81; this paper will henceforth be referred to as Nefertari. It is important to emphasize that the EAO/GCI project is less concerned with absolute colour than with changes which may take place over time. My work intends to document the present colours. [Since writing this paper, I have learnt that a Minolta CR-200 was used to document the cleaning of the frescoes in the Sistine Chapel in the Vatican.]
4

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

I99I

AN OBJECTIVE COLOUR-MEASURING SYSTEM

45

It should be noted that no matter how sophisticated the instrument,one can only record the colours as they are at the time of measurement; differences of preservation,
environment and re-use of the tombs can make what was originally the same colour

appear different in two tombs, or even in the same monument. This, along with the difficultiesoutlined above, has sometimes caused the validity of colour measurementto be questioned.1" However, it would seem wrong to reject any attemptat measurement,
since an objective record of the state of the tombs today is better than no record at all. ?111

below considers some of the further theoretical and practical problems of the measurementof colour in these tombs. II. The CR-221 L.Description of theinstrument
The Minolta CR-22 I (pl. IV, i) is a two-part instrument, consisting of a measuring device and a data processor. The measuring device has a flat baseplate to which the holder for

the measuringhead is attachedand kept in tension by a spring.The head itself contains a pulsed xenon arc lamp arrangedat an angle of 45? to the subject for the purposes of
illumination, and the reflected light is received by six silicon photocells. The output from

these photocells is then passed via a cable and an RS232 link to the data processor.'2The areato be measuredshould be at least 3 mm in diameter.
This data processor is effectively a small hand-held computer. It can store up to 300

measurementsin its memory, and can perform various statisticaloperations on data. It incorporates a small thermal printer to create hard-copy output of the results. The measurementsmade by the measuringdevice can be representedin five colour notations:
Yxy, L*a*b*, L*C*H?, Munsell, and DxDyDz (the latter is a measure of colorimetric

density).Readings are displayed in the chosen notationon a small liquid-crystaldisplay.

Additionally, it can measure colour differences against preset values in four notations. Statistical functions available ae are maximum, minimum, mean and standard deviation of a range of values. It is possible to control many of the different options offered, for example, whether automatic printing is turned on, whether single or multiple (three) measurements of the sample are made, and the specification of the light source. One extremely useful feature is the existence of an RS232 output port so that the processor can be connected to a computer and the stored data transmitted instantaneously elsewhere (but see ?11.2below). Whenever data is listed, it goes to this output port as well as to the printer. The instrument is accompanied by generally adequate documentation. Power for the instrument can be supplied from mains current or from batteries. A small transformer supplied with the unit can be used to provide gv from a regular mains supply, or six AA type i.5v batteries can be used. In the field, batteries will generally be the preferred method; we used rechargeable NiCad cells. A note is in order on the conversion between the colour notations offered by the CR221. Data from the measuring head is always converted into the Yxy system when the reading is made, and the readings in the other notations are mathematical conversions, often complex, from this; the only exception to this is the Munsell readings, which also
E.g., C. Beinlich-Seeber and Abdel Ghaffar Shedid, Das Grabdes Userhat (TT56) (Mainz, 1987),
I 21-2.

12RS232 is an 'industry standard' interface for serial communication between devices. The dimensions of The data processor measures 50 X220 X 200 the measuring head are 100 X 8 i X 243 mm; weight 1225g. mm; weight I 300 g without batteries.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

46

NIGEL STRUDWICK

JEA 77

employ a look-up table. While it is possible to print measurements in all colour notations one by one as they are taken, they are stored in the memory only in the one notation currently chosen. An extra option I would like to see added to the instrument would be the

ability to convert stored readings to any of the other notations, so that one could, for example,printa list of readingsfirstlyin Munselland then in Yxy notation.
2. Use of the CR-22I The instrument is very simple to use. It must first be calibrated using a white reference

tile supplied with the unit. To maintain accuracy, it is advisable to recalibratethe The preferredcolour notationshould then be chosen, along with instrumentfrequently.13
setting the various indexes regarding printing, single/multiple measurement and so on. A suitable area should then be selected for measurement and a 'measure' button pressed,

eitherthaton the measuringhead or thaton the dataprocessor. to use in the field. Batterieslast for The instrumentwas generallyvery straightforward
and 300 readings, an adequate number for a day's work, particularly since between 200oo

the maximum capacity of the memory is 300 readings.With simultaneousprinting of

results in all colour notations, one reading can be made every ten seconds or so; without printing it is much faster. We printed in all notations, since it was desired to keep as complete a record of the measurements as possible and because the instrument can only store data in one colour notation. One minor difficulty is that the thermal printer has low paper traction and will only feed the paper through when the data processor is horizontal; it had been hoped to carry the processor vertically in a small shoulder bag leaving one hand free for other purposes, but in actuality the processor either had to be held in the measurer's hand or placed on a table. An Apple Macintosh Plus computer was to be taken into the field for a variety of purposes during the season, and it was logical that the CR-22I be linked to it. Our only technical problem arose when attempting to link the instrument to the computer, since despite wiring the cables in the conventional way according to the data supplied by the manufacturers, it was not possible to make the instruments communicate. After examination of the port with an oscilloscope, it was found that the various pins of the RS232 port on the CR-22I needed to be wired in a somewhat unusual fashion. Specifically, the RTS and CTS pins are active high, whereas normally they would be expected to be low. The diagram below shows that only two wires are necessary at the CR-22I end.14 Once activated, this link worked very successfully and saved a great deal of time. Data was transferred into a text file which was then read into cards of a Hypercard stack. Notes were then added to this database concerning the location and colour of the original. These notes were made in the field with a hand-held cassette tape recorder as a more rapid alternative to writing. Analysis of the data is being undertaken with the aid of the database program Reflex Plus and the spreadsheet Wingz.
13 The instrument's manual recommends recalibration at the beginning of each measuring session; I have tried to recalibrate when moving from one scene to another in the tombs, perhaps on average every twenty to thirty readings. For future work Nick Hale has recommended that the instrument be calibrated against a reference tile visually close to each colour to be measured. 14 I am indebted to Graham Louth (Statistical Laboratory, University of Cambridge) for his help in solving this particular problem.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

I991

AN OBJECTIVE SYSTEM COLOUR-MEASURING

TXD-GND - _@ 1 ? 8 @ 3

47

3. Modifications
The design of the CR-22 I is such that the measuring unit is attached to a baseplate, the purpose of which is to facilitate correct alignment of the measuring head with the sample.

The opticalhead must be at right-anglesto the specimen surfaceso thatthe incidentlight does not escape and sufficient light reflected from the colour reaches the sensors. However,the presence of this baseplatecreatestwo problemswhen the device is used on a tomb wall: (i) it is highly undesirablethat such a metal plate come into direct contact
with the painted plaster or rock; (2) given that Egyptian tomb walls are usually far from perfectly flat, the rather large baseplate sometimes prevents the measuring head being as

close to the subject as one would like. Ideally,the baseplateshould be removed, but with
the CR-22I this is not practical. The obvious solution to (i) is to provide a soft surface of some type between the baseplate and the wall. (2) can only be solved by careful selection of subject areas, although it is to be hoped that Minolta will enable the baseplate to be removed from future versions. A layer of transparent plastic acetate was used for the protective surface, a material easily available as it is the same as used for making facsimile drawings. Since providing this protective surface for the wall increases the distance between subject and measuring head, the optical conditions vary slightly from the design condition and some measurement error may result. I examined the effect of this on the CR-22 I before going into the field and found that the presence of the acetate does not have a serious effect on the readings.15As reference colours, a British Standards Institution colour chart which we had been using in the field before turning to the CR-22I was employed.16 Readings were taken on seventy-five randomly chosen colours, one without the padding and one with; the readings were made in Yxy colour spaces, so that any calculation of the differences could be easily quantified. The following presents the results in the summary form of the maximum and minimum differences and also the standard deviation for the range of values:
Y Max Min Standard deviation i.88 - 0.64 0.40 x 0.04 o.oo
o.o0

y o.o I o.oo 0.00

15A study of the effect of the addition of a protective cushion on the CR-I2I carried out by the GCI reached a similar conclusion. Their approach was different in that the baseplate of the CR-12I could be easily removed, and a padding was only needed immediately around the measuring head (Nefertari, 70-2). The problems are, however, essentially the same. 16Framework for colour co-ordinationfor buildingpurposes(BS 5252: 1976).

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

48

NIGELSTRUDWICK

JEA 77

From these figures, it will be apparent that the difference is indeed minimal and can safely be disregarded. Putting it in the visual terms of the Munsell Book of Color as given by the CR-22 I, I give below three typical examples: Bookof Color 5B 7/8 IoR 4/6 2.5Y 8/6
CR-22I unmodified CR-22I modified

5.iB 7.0/7.4 Io.oR 4.0/6. I i.7Y 8.0/4.7

5.oB 7.0/7.4 i o.oR 4.0/6. I i.7Y 7.9/4.3

These measurements should be considered in the light of the following discussion of the relationship of the systems. However, it shows that the difference produced by the modification has an inconsequential effect on the Munsell colour space. III. Systems and problems of colour measurement I think it is safe to say that, although the basis on which colour is being measured in Egyptological fieldwork is improving, Egyptologists have undertaken this work without really understanding the basic theoretical issues and problems involved. We were equally guilty in our first season in the field with the CR-22I. Of the available literature, the introduction I would recommend to readers is Fred W. Billmeyer, Jr. and Max Saltzman, (New York, 1981 ). Principlesof Color Technology2 It is important to begin by understanding the component elements of what is called colour, whether it be seen by the human eye and brain or measured by an instrument. These elements are three in number: the light source under which the coloured object is viewed, the coloured object itself, and the response of the observer. These are more technically known respectively as the Spectral Power Distribution of the light source, the Spectral Reflectance of the object, and the Spectral Response of the observer. If the wavelengths of the three components are multiplied together, the resulting product (the perceived colour) can be termed the Stimulus for Colour.17Therefore, it can be seen that the variability of the light source and the observer can make crucial differences to the perceived colour, even with the same sample. Most readers of this paper will have had the disconcerting experience of finding that two items, such as clothes, which match under artificial light, do not match in daylight. This phenomenon is known as 'metamerism', and two such colours as a 'metameric pair', and is due to the fact that the spectral reflectance of the two objects is different, but thatthe product of the calculation described above happens to be the same for a particular light source and observer.18 Metamerism is a potential problem when colour charts are used to quantify colour. Even when using an instrument with its implicit objectivity and consistency, the light source must remain constant for the readings to have any meaning. The CR-22 I solves this by having its own in-built source which is powerful enough to overcome the influence of the local light source. An extremely important issue is the question of the colour standard to be used when referring to colours in the publication, given the ability of the CR-22I to measure in a
17See the figure in Billmeyer and Saltzman, op. cit. I 7.

"Illustrated in ibid. 54-5, there using the calculation of CIE Tristimulus values. See ibid. 44-7 for an explanation of this term and its calculation.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

i99i

AN OBJECTIVE COLOUR-MEASURING SYSTEM

49

number of colour spaces. The instrument makes two types of colour space available, numerical and visual, the former being typified by the Yxy system (the base system of the CR-2 21) and the latter by the Munsell Book of Color.

which may be summarizedas follows: Each has its advantagesand disadvantages, Numerical (Yxy) Advantages:
2.

I. It is possible to obtaina readingfor all colours. For the same reason, it is possible to quantify arithmetically the difference between

colours, and also to performstatisticalcalculations.19 3. In the present context, a numerical system lends itself to measurement by instrument. Disadvantages:
i.

It is impossible to illustrate the colour concerned without reference to a visual

standard,such as Munsell. Visual (Munsell) Advantages:

2.

i. A rapidmethod for seeing what colour the readingrepresents. The visual nature of the system is more accessible and perhaps more likely to gain

initialacceptance. Disadvantages:
I.

It is impossible to represent visually all possible colours, since the colour chips in

the Bookof Colorare but a subset of all possible colours.

2. Spectral reflectance was noted above as one of the three important elements of

colour. Therefore, it is only correct to compare a glossy colour, such as in the

Glossy Finish Edition of the Book of Color, the usual set of Munsell charts taken into the field, with the rougher, more matte finish of tomb painting when the gloss is the of the observer. Metamerism will also be a problem as deflected from the eyes outlined above, especially since the Munsell charts are designed to be matched using the CIE (CommissionInternationalede l'Eclairage)illuminant C, which approximates to average north daylight, and CIE 2 degree Standard Observer;20neither of these is likely to be precisely what obtains in a Theban tomb. 3. It is difficult to relate the hue components except within the confines of the same chart, since although Munsell space is three-dimensional, and can be represented as a solid, the charts are two-dimensional, and show the value and chroma for a set hue. The choice between the two is difficult, since it seems that numerical systems are inherently more accurate, and yet visual systems are far more likely to gain acceptance in
19This is the purpose of the EAO/GCI investigation, where the variation of the colour is more important than the actual visual hue. 20 These standards are explained in Billmeyer and Saltzman, op. cit. 34-44.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

50

NIGEL STRUDWICK

JEA 77

the context of why an Egyptologist should want to study the colours of tomb paintings. I will describe in ?IV a solution which uses aspects of both. If the Munsell system is to be used with the CR-22 , it is necessary that users know Book of Color. As pointed out above, the thise is relative to how accurateent is a from the made via conversion process from the Yxy system. I CR-22 Munsell reading At the time of the first use of our instrument in the field, we had no independent information on the reliability of this conversion. However, since that time, I have learnt that the instrument is not i 00 per cent accurate in this respect; the internal conversion from Yxy to Munsell is performed by looking up the values in a table, and certain corners have been cut to enable this process to be fitted into the storage of the CR-22i . As described at the end of this section, I have now obtained corrected Munsell readings. However, the question of what degree of error is acceptable is a very important one for the archaeologist. The instrument, and much of the colour technology being discussed, is emphatically not designed with the circumstances of the archaeologist in mind, but rather for colour measurement and matching under more ideal, almost laboratory, conditions. So many potential secondary factors are likely to creep in while in the field that aiming for total accuracy is unfeasible, and thus, the innate error factor is, for our purposes, not a major problem. To be able to make comparisons, we were most fortunate to have the loan of a copy of the Book of Color: GlossyFinish Collectionwhile in Luxor in late i988.21 The introduction to this publication should be consulted for an explanation of the Munsell system; it is assumed in this paper that readers are familiar with its basic tenets. Three sets of tests were carried out with the aid of this publication. Firstly, some CR221 readings were made of the colour chips and then compared.22 This showed that the readings were, with some differences, substantially accurate. This view was subsequently confirmed by Max Saltzman and Nick Hale, with the proviso, as noted above, that any errors innate to the machine would not really be significant given the less than ideal conditions in which the instrument was to be used. The second was in circumstances more relevant to our purpose, namely, a rather rapid initial test in Theban tomb 178, with the cooperation of Eva Hoffmann of the Agyptologisches Institut Heidelberg. The results were compared with her estimations from the Munsell Book of Color. The results were very variable, and the two following examples show typical difference and agreement. Bookof Color
ioBG 6/4 2.5YR 7/6

CR-22I
i.gBG 6.3/2.0 2.7YR 6.8/4.3

Colour
Turquoise Pink

This variability can be partly attributed to the unavoidable rapidity with which this test was carried out, but there are also two other important factors. Firstly, some of the
colours are faded or dirty. Although the human eye and brain may take much longer than
211 I am indebted to Karl Seyfried, Field Director of the University of Heidelberg 'Ramessidischer Beamtengraber in Theben' project for the loan. 22It should be pointed out that the chips in the Book of Color are the closest possible approximation to the colour described within the tolerances accorded that publication at the time it was produced.By the time the book has been in use for a few years the chips may have changed slightly. None of this is critical to the practical use in the field where tolerances are large, but theoretically it is of the highest importance.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

5I an instrumentto make readings, and be much less consistent, it can more easily take a
I99I

AN OBJECTIVE COLOUR-MEASURING SYSTEM

sample on a much smaller area than the CR-22 I.23Secondly, the angle of measurement is critical. The CR-22I, being designed more for laboratory conditions than anything else,

has problems with an uneven surface.The above readingswere made with the baseplate
in place, the purpose of which is to ensure that the head is held at the correct perpendicular angle to the sample; given the large size of the baseplate, it is possible for an uneven surface to distance the sample from the wall slightly (as with the protective The important lesson learnt from this test was clearly that it is necessary to select one's sample with great care, choosing an area that is sufficiently large, clean and level. This was applied in our third and major test of the instrument, in the tombs which I am

or to change the measuringangle. modification,see ?11.3),

publishing.The tomb of Amenmose (TT 254) is the best preservedof the three,and also
has the freshest colours. We hoped that it would thus be a good test of the CR-221.

We began by takinga numberof readingsand comparingeach with our opinion of the colour by use of the Book of Color.Areas were selected following the criteriadiscussed above. An initial rounding up or down was made of the reading to the nearest Book of Colorchip, which was then comparedwith the wall, under lightingconditionswhich were a combinationof daylightand tungsten bulbs. We worked from this first approximation
until the best Munsell match was found. It will be seen that the CR-22I was usually

extremelyclose to the readingdecided visuallywith the Bookof Color,althoughthis is not

really an unbiased test, in that our Book of Color starting point was defined for us. Ten typical results are given below: Munsellfrom
CR-22 I in the Reading
I
2

More accurate
Munsell 8.57YR 6.95/6.32 7.24G 6.19/1.47 7.-iB 6.22/2. I0 8.87YR 8.86/i. I I
I.I

Colour and subject Yellow-Background to text on border of stela Green-Face of god Osiris god Osiris White-White crown of god Osiris Red-brown-Stomach of Amenmose

Book ofColor field

ioYR 7/6
7.5G 6/2 7.5B 6/2 I oYR 9/ I ioR6/6 2.5YR 7/4

8.3YR 7.o/6.2 7.2G6.3/I.5

7.IB 6.3/1.9 9.3YR 8.9/ .i

3
4 5

Blue-Blue areaon the throneof

o.9YR5-7/5.5
2.2YR 7.3/3.8 4.7YR 6.3/6.5 9. IYR 7.8/3.2 5.iGY 7.2/2.2 8.4R 4.7/6. I

IYR 5.63/5.69

6 7 8
9 Io
23Or

Pink-White kilt over the red

legs of Amenmose

2.43YR 7.23/4.08 4.84YR 6.26/6.77 9.o9YR 7.74/3.24 5.o2GY 7.14/ I .99 8.42R 4.70/6. I O

Yellow-brown-Round loaves
among offerings

5YR 6/6
ioYR 8/4 5GY 8/2 ioR 5/6

Yellow (very light)-Female skin colour

Green-Block border to right of stela Red (dark)-Pot held up by serving woman

Likewise, the observer may think he can make allowances for the dirt on a colour sample when choosing his chip, but this is again improbable.

The humanobservermerely believes he can, but is rather,it thinksit can take such a measurement. liable to be heavily influencedby the surroundingcolours.Therefore,his result is almost certainlygoing to whateverthe problemsinvolvedin measuring. thanthe 'objective' more inaccurate readingof the instrument,

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

52

NIGEL STRUDWICK

EA 77

It was indicated above that the internal conversion of the CR-22I from its native Yxy readings to Munsell was slightly inaccurate. The original Yxy data has been converted to more accurate Munsell readings using a computer program developed at Teeside Polytechnic:24this data is presented in the fourth column above. Overall the results are very satisfactory, provided suitable care is taken in choosing the locations for the readings. After these initial tests, a total of 273 readings was made. The CR-22I was recalibrated regularly, and periodic comparisons were made with the Book of Color. The accuracy remained good, only certain colours showing occasional differences. These were most noticeably blues and greens; the former may perhaps be attributed to the tendency of Egyptian blues to discolour and decay. The tests just described were performed purely for our own satisfaction as to the instrument's behaviour. Given the likely variability of the human observer, the light source and the problem outlined above with regard to the different reflective characteristics of the tomb walls and the Book of Color, there is very little objective value in these comparisons, but we hope that they will help to convince Egyptologists of the trustworthiness of the CR-2 2 I.

IV. Presentation and evaluation of results In the course of our 1988-9 and 1989-go90 seasons a total of 360 measurements was made of the colours in the group of tombs under study. Much thought has subsequently been given to how the results should be presented in the publication and to possible analyses of the data. The former is the more important at the present moment, since the production of adequate documentation of the tombs is our primary concern. The two principal systems of measurement for our purposes are the Munsell and Yxy colour spaces, and the data was recorded using both systems; the advantages and disadvantages of both have been discussed in ?III. Because of the requirement to be able to compare our readings with other published accounts, which use the Munsell system, it seems desirable to present the basic data in that format. The first problem is whether our readings should be corrected to the nearest colour chips in the Book of Color. As noted, that publication can only represent i ,500 or so of the millions of colours which can be been apparent to the reader that distinguished by the human eye, and it will have already already our readings fall between those neat divisions. Correcting the hue and chroma to the nearest whole values is visually not difficult,particularly since they appear on the same page of the Book of Color, but attempting to interpolate the hue, except in the context of the colour solid, is not so easy. Also, too direct a comparison with the Book of Color makes no allowance for the different reflective characteristics of those colours and those in the tombs. It was thus decided to leave the CR-22I readings (corrected as described in as they stand, permitting readers to make their own approximations. ?111) How should the data be arranged? There are two basic possibilities-arrangement by location or grouping by colour. Given the need to be able to find quickly the colour of a particular item in the visual documentation of the paintings, it seems that arrangement by
24These programs have been developed by T. W. A. Whitfield and Neville S. Smith, and I am indebted to them for making these conversions for me. A detailed description of the program has been published in ColorResearchand Application (December, 1990).

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

I99I

AN OBJECTIVE SYSTEM COLOUR-MEASURING

53

location, wall by wall, tomb by tomb, is the best, and it is my intention, publisher permitting, to give a list of the 360 measurements. It is important, nonetheless, to be able to group together readings of like colour so that the range of colour can be realized. Immediately, the problem of how to define verbally the general coloration of a sample arises. The subjectivity of individual descriptions has already been alluded to above; in the context of Egyptian wall-paintings, one person's 'dark red' is another's 'red-brown'. There are several ways in which this difficulty may be handled, of which I shall consider three. In all cases, detailed discussion of the results will be reserved for the final publication. i. Categoriesof objects The coloration of similar categorites of objects can be studied: good examples of this would be comparison of the male or female skin colours, each of which might be expected to be reasonably consistent. There will, of course, be some variation within the Munsell readings assigned to a particular object category. This is not easy to visualize except within the context of the Munsell colour solid, and very difficult to quantify, since the Munsell system, being principally visual, does not lend itself easily to other methods of representation. This is where the readings in the Yxy system become useful, in that colour, as described in the CIE system from which the Yxy notation is derived, can be plotted on a two-dimensional graph or chromaticity diagram using the x and y components of that notation.25 Chromaticity diagrams are a very good way of visually representing the spread of any particular grouping of colour readings. By plotting two or more sets of values on one
+ TT253 A Tr254

0.3800

~
_ A

0.3700-'A
Y-

A ^
_4A

A
A
A

A A A

0.3500

A A

0.3400-

A
I i i

0.3300

I i

T l l

i i il l 0.4000

i i il l

' ' ' l l 0.4400 0.4600

1 0.4800

0.3400

0.3600

0.3800

0.4200 x

FIG. I. Chromaticity diagram of readings of the male skin colour in TT 253 and 254. " For more detail see Billmeyer and Saltzman, op. cit., in particular 47-52. The spread of the names of the principal colour divisions over a chromaticity diagram is illustrated in the figure on their p. 50.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

54

NIGEL STRUDWICK

JEA 77

graph it is possible to compare how the same groupings of data vary between tombs. For example, fig. i shows the distribution of the measurements of male skin colour from both TT 253 and TT 254. More detailed discussion of the implications of this will be reserved

for the published report of these tombs, although it will be immediately clear that althoughthere is quite a wide degree of variationin the same tomb, the two chapels are using distinctlysimilarversions of the same intendedcolour.
2. Systematic useof colour names The very wide range of Munsell colours can be reduced by the use of a systematically organized range of colour names. On the advice of Max Saltzman, I have used the scheme

put forward by Kelly and Judd in their I976 National Bureau of Standards (NBS) and Dictionary publication Color:Universal Language of Names,specificallythe charts on In this the Munsell colour pp. I5-34.26 publication, space is divided into 267 colours,
named according to a fixed set of conventions. Munsell colours can be given their

colour name either by using the charts,or else by the use of a purpose-built appropriate computer program.For example, the NBS colour name 'greyish reddish orange'covers
Munsell readings in the hue range 8. iR-3YR and the value/chroma range 4.59/ 5.29-6.37/5.85, and also the following ten verbal descriptions given by myself at the time

of takingmeasurements:
Brown Pale reddish-brown Red Brownish-red Pink Red (dark) Red-brown Red-pink Reddish-brown Yellow-brown

This variationcan be illustratedby the chromaticity diagramin fig. 2.

+ TT253 0.3800 A TT254

0.3700 .3 0.36000.3500
-

A
A0
iA

A A+ : A
+

A A

0.34000.3300 11 T I i

+
I ,I ,

+
i 0.4300
x I , , ,

, , [ [ [ I , I , I [ , I ' [1 0.4500 0.4600 0.4700

0.3900

0.4000

0.410O

0.4200

0.4400

FIG.2. Chromaticity diagramof NBS colour 'greyishreddishorange'in TT 253 and 254.

26 Six

levels into which colours can be groupedare summarizedin the tablein Billmeyerand Saltzman, op.

cit. 32.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

199I

SYSTEM AN OBJECTIVE COLOUR-MEASURING

55

This method has a disadvantage insofar as fine differences between colours can be obscured in the grouping. This might be a serious objection when working under ideal

conditions,but given that some inaccuracycan creep in when workingin the (laboratory) field (as described above), such grouping can actuallyhave the advantageof permitting
some leeway in the observations and preventing one from becoming convinced by what

may be a misleadingdegree of accuracy. In connection with these figures,the readermay wish to see how the range of readings used for fig. i were described at the time of recording (column i in table below) and in terms of the NBS standardcolours (column 2):
Colourdescription Brownishred Pink Pinkish-red Red Red-brown NBS colour name Greyishreddishorange Greyishyellowishpink Lightreddishbrown Moderateyellowishpink Moderateyellowishpink Greyishreddishorange Moderatereddishorange Brownishorange Greyishreddishorange Greyishreddishorange/Lightbrown Greyishreddishorange/Moderatereddishorange Lightbrown Lightreddishbrown Moderatereddishbrown Moderatereddishbrown/Light reddishbrown Moderatereddishorange Moderateyellowishpink Lightbrown

Red-brown

Yellow red-brown

Both tables illustrate the variability of one's terminology when faced with colours in a tomb, although it does also show the range of shades of colour which can be disguised by one standardized term. Then again, the first table effectively reveals a bias in the observer-for the most part the male skin colour is characterized by Egyptologists as a reddish-brown, and thus that is the colour noted by myself when taking a reading on a normal example. Yet, as can be seen from the selection of colour names corresponding to 'red-brown' above, there is far more variation in the male skin colour than the observer realizes without reference to some more objective method.

aspects 3. General
The colours can be referred to in the broadest terms, such as the five basic hues of the Munsell system: red, yellow, green, blue, and purple. More relevant to Theban tombs might be a division using the broad divisions of the New Kingdom palette, defined most recently by Baines as black, white, red, green, yellow, strong blue, light blue, grey, brown, and pink.27 This method has the disadvantage of adding an element of subjectivity, insofar as it is necessary to make visual and verbal classifications of even these broad
27 American

Anthropologist87 (1985), 287.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

56

NIGEL STRUDWICK

JEA 77

divisions. For many colours, this should not present a problem; however, the variation

between blue and green and also between brown and red is such that objectivityis very difficult.Using a carefulcombinationof these subjectiveapproacheswith those outlined
in Sections I and II above, it should be possible to collect readings for each of Baines' ten

of the colours of a particular tomb. colours, and thus produce a characterization V. Conclusion
This paper has tried to show the advantages and disadvantages of an instrument such as

the CR-22I over methods currentlyin use in Theban tombs. The increases in speed and
objectivity vastly outweigh some small practical difficulties, and if the instrument is used carefully, it can greatly improve our ability to document coloured scenes. With careful consideration of the nature of the data, it is possible to present it in such a way that more complete documentation can be offered than is currently the case, and, if a computer is used in conjunction with the CR-22I, a wider range of possibilities is available for analyzing the results. It has always been my intention to make this equipment available to colleagues who might wish to experiment with it on their own colour recording problems. I should be interested to hear from anyone who might wish to use the instrument.

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions

PLATE IV

I. The Minolta CR-22I Colourimeter (courtesyof Minolta (UK) Ltd.) AN OBJECTIVE COLOUR-MEASURING SYSTEM (pp. 43-56)

2. P. BM 0o40I

REQUISITIONS FROM UPPER EGYPTIAN TEMPLES (pp. 79-94)

This content downloaded from 182.178.239.129 on Sat, 29 Mar 2014 02:53:53 AM All use subject to JSTOR Terms and Conditions