VIZIO 1016
`
`

`
`/
`
`\//4
`/////
`
`// ”////
`\\
`
`\\
`
`W///////%
`
`OLOUR
`
`Princeton University Press
`Princeton, New Jersey
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`
`
`Copyright © 1983 by Hazel Rossotti
`All rights reserved
`First Pelican original edition, 1983
`First Princeton Paperback printing, with corrections, 1985
`DOC 84 — l I451
`ISBN 0 -691 — 08369 — X
`ISBN 0 - 691 —02386 -7 (pbk.)
`
`Reprinted by arrangement with Penguin Books Ltd.
`Made and printed in Great Britain
`by Richard Clay (The Chaucer Press) Ltd,
`Bungay, Suffolk
`Set in VIP Times
`
`Clothbound editions ofPrinceton University Press books are printed
`on acid-free paper, and binding materials are chosen for strength and durability.
`Paperbacks, while satisfactory for personal collections, are not usually suitable
`for library rebinding.
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`CONTENTS
`
`List of Text Figures
`Foreword
`Preface to the Princeton Edition
`Introduction
`
`Part One: Light and Dark
`
`I\)h—I
`
`. Light Particles
`. White Light on Clear Glass
`
`Part Two: Lights and Colours
`
`ONUI-AU)
`
`. Steady Colours
`. Shimmering Colours
`. Special Effects
`. Lights
`
`Part Three: The Natural World
`
`. Air and Water
`Earth and Fire
`. Vegetable Colours
`The Colours of Animals
`
`.°‘°?°“
`
`Part Four: Sensations of Colour
`
`11.
`12.
`13.
`14.
`15.
`
`16.
`17.
`
`Light and the Eye
`Anomalous Colour Vision
`Colour Vision in Animals
`The Eye and the Brain
`Sorting and Recording
`
`Part Five: Technology
`
`Colour Reproduction
`Added Colour
`
`11
`I2
`13
`
`19
`26
`
`37
`44
`48
`55
`
`65
`77
`84
`91
`
`109
`122
`126
`130
`143
`
`169
`185
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`Contents
`
`Part Six: Uses and Links
`
`18. Imparting Information
`19. Communicating Feelings
`20. Colour, Music and Movement
`21. Words and Colours
`
`Index
`
`Acknowledgements
`
`203
`209
`220
`222
`
`231
`
`239
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`x////2?//,
`
`7"///
`
`SORTING AND RECORDING
`
`To what extent can we record a colour? Can we impose any order on our
`rich variety of colour sensations? If we can, would our scheme be entirely
`personal, or could we use it to communicate information about a colour?
`How can we best tell someone the exact colour we should like the new door
`to be?
`As we shall see in Chapter 21, there are many difficulties in trying to
`describe colours with words. A request to paint the door turquoise would be
`likely to produce a fairly bright door in the greenish—blue (or might it be
`bluish-green?) range, neither very pastel nor very murky. Perhaps we could
`get nearer to the colour we have in mind by a request to paint the wall the
`same colour as the curtains. But, though the two may seem a good match in
`one light, they may clash horribly in another (see page 152). And as they
`will have different textures, they will have different highlights; so although
`the colours of the two may ‘go’ very satisfactorily, they will certainly not
`match all over, even if the light falling on them happens to be identical. It is .
`safer to choose from samples of the actual paint, on the manufacturer’s own
`colour card. But maybe the exact colour required is not available. ‘Please
`mix me. . .’ How does one continue? ‘Something between these two’;
`‘Something like this, only lighter’; ‘A more subtle shade of that’; ‘A bluer
`version of this one.’ It is difficult to know how to specify, or how to produce,
`the colour required. And it may even be difficult to envisage precisely any
`colours which are not on the colour card.
`Perhaps it would help if we could arrange colours in some rational order
`and label them appropriately. We could then refer to them in much the
`same way as we can pinpoint a place by a map reference. Many child-hours
`must be passed in just this way, rearranging crayons, pastels or embroidery
`threads. It is easy enough to make a line of the rainbow colours, and join the
`ends, through purple, to give a circle. But problems soon arise. Do black,
`grey and white count as colours? If they are to be included, where should
`they go? What should be done with pale colours: primrose, duck-egg blue,
`salmon? And what about the browns? One can imagine a cross-roads at
`yellow, with primrose leading to white on one side, and ochre leading to
`khaki, brown and black on the other. Perhaps, however, black and white
`should be joined through the greys, to give another circle, perpendicular to
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`144
`
`Sensations of Colour
`
`the first. It seems we cannot arrange our coloured objects on a flat surface,
`but need some three-dimensional scheme.
`
`Maybe it would be better to seek a more ‘scientific’ classification than
`any subjective arrangement of coloured materials? We might try to specify
`the colour of a sample by irradiating it with the light of a large number of
`very narrow bands of wavelength and measuring the percentage of each
`which the sample reflects. These measurements can be made extremely
`easily, given the appropriate equipment. But the light which enters the eye
`depends on the lighting as well as on the sample, so we would also need to
`know the composition of the illumination. Even this does not tell us the
`colour of the sample unless we know how the eye reacts to light of different
`wavelengths. Two materials may match exactly under one type of illumina-
`tion even if they send light of totally different composition to the eye; we
`know that many yellows can be matched by mixtures of red and green light.
`We might, however, combine, for each narrow band of wavelength,
`measurements of the reflecting powers of the sample, and the composition
`of the light source with our knowledge of the response of the retina.
`Although this procedure still needs laboratory equipment, it relates the
`scientific measurements of the light reaching us to our perceptions of
`colour, assuming that the observer has normal colour vision, adapted for
`daylight viewing and uninfluenced by the effects (such as after-images,
`contrast of near-by colours, memory, expectation) which we discussed in
`the previous chapter. So we are attempting to chart, not just the stimulus of
`the light entering the eye, but a normal observer’s response to it. The idea of
`attempting to measure colour in this way sounds attractive, if somewhat
`complex. But do its advantages always outweigh those of map references
`with an ordered arrangement, albeit subjectively chosen and represented in
`three dimensions? Since both systems are used in practice, we shall look at
`each in more detail.
`
`If we are to arrange a number of colours in any systematic order, we must
`decide what qualities we shall use to sort them. Let us first recall the ways in
`which light can vary. The sensation of colour depends primarily on the
`composition of the light, and partly on the intensity; and the composition
`may be usually described as a mixture, in a certain proportion, of white light
`with a ‘coloured’ light of a particular dominant wavelength within the
`visible spectrum. (For purple light, the ‘coloured’ component is itself a
`mixture.) We can describe the primary sensation of colour in terms of hue,
`which refers to the greenness, blueness and so forth, and varies with any
`change in the dominant wavelength. The extent to which this wavelength in
`fact dominates the light is known as saturation (or chroma). As the domin-
`ant wavelength is diluted with white light, the saturation decreases. An
`increase in intensity in light of a particular composition increases the
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`
`
`Sorting and Recording
`
`145
`
`brightness (or value). The so-called ‘natural’ or ‘achromatic’ colours, black,
`grey and white, are of zero saturation, and differ from each other only in
`brightness. A series obtained by adding one hue, say blue, to white differs
`only in saturation, as does a series obtained by adding a blue pigment to a
`grey one. If the same pigment were added to a white one in the same
`proportions, the dusky, pale blue would differ from the clear pale blue only
`in brightness.
`There are many three-dimensional arrangements of colours, the best
`known being those devised by Munsell and by Ostwald. Both are based on
`the colour circle formed by joining the two ends of the spectrum through
`purple (see Figure 58). So the hue changes around the circumference of the
`circle, much as the hours progress around the face of a clock. Through the
`centre "of the clock face, and perpendicular to it, like the axle of a wheel,
`runs the line representing the neutral colours, usually with white at the top,
`changing, through deepening greys, to black at the bottom. Radially, like
`spokes on a wheel, the saturation increases towards the rim, to give a space
`which can be filled in by different colours, according to which system is
`being used.
`
`White
`
`Green
`
`
`Vivid colour
`
`um” calm
`
`Sauraflon (chroma)
`
`Blue
`
`Red
`
`Hue
`
`Figure 58. Colours in space The skeleton ofa colour solid. The ‘achromatic’ colours
`form the vertical backbone from which the different hues radiate: red in one direction,
`green in the opposite one, and the others in between. For any one hue, the colour
`becomes more vivid the farther it is from the centre.
`
`Ostwald arranged his colours in a double cone, based on twenty-four
`different hues, arranged around the circumference (see Figure 59). Each
`hue is combined, in a number of fixed proportions, with each of eight
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`146
`
`(a)
`
`Sensations of Colour
`
`white
`
`Planes of constant dominant
`wavelength (hue)
`
`Full saturation at edge
`(complementaryto (1))
`
`Hue (2)
`
`Grey axis
`
`Figure 59. Ostwald’s colour solid (a) Exterior view, (b) Vertical section.
`(Adapted, with permission, from G. J. Chamberlin and D. G. Chamberlin, Colour: Its
`Measurement, Computation and Application, Heyden, London, 1980.)
`
`equally spaced neutral colours from white to black. The resulting colours
`are arranged so that brightness decreases vertically towards the bottom of
`the diagram, while saturation decreases towards the centre. Thus Ostwald’s
`colour solid consists of twenty-four triangles (one for each hue), arranged
`radially so that a vertical section through it gives two such triangles, for
`complementary hues, fused at the centre. Each position within the solid is
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`Sorting and Recording
`
`147
`
`numbered on a grid system, so that any colour contained by the solid can be
`~5pecified by a map-reference.
`1n Munsell’s arrangement, saturation is increased by a series of visually
`equal steps rather than by adding a fixed proportion of pigment; and there
`are nine neutral colours, rather than eight. As the number of equal steps of
`saturation at a particular hue and brightness depends on the hue, the arms
`are of different length in different parts of the solid. Munsell’s solid is
`therefore much less regular than Ostwald’s, and on account of its untidy
`appearance is known as a colour ‘tree’. A typical vertical section through it
`is shown in Figure 60. Each position in the tree, as in Ostwald’s cone, is
`encoded, thereby allowing colours to be specified. And the tree has one
`great advantage over Ostwald’s solid: whenever some new, dazzling pig-
`ment is made, it may be incorporated by extending an existing branch.
`
`Maximal
`saturation
`
`Maximal
`
`saturation
`
`(complementary to (1))
`
`Figure 60. Munsell’s tree Vertical section (cf. Figures 59(b), page 146, and 83. page
`226).
`
`The two systems resemble each other in that the circumference is divided
`arbitrarily into hues, and the vertical axis is graduated into visually equal
`steps, which are obtained by asking large numbers of observers to estimate
`equal differences in brightness. But the two solids are based on different
`ways of varying saturation. While Ostwald used the ratio of neutral pigment
`to saturated pigment, Munsell again invoked visual assessment by the
`average observer.
`When Ostwald devised his colour solid, he was an old man whose sensi-
`tivity to blue was doubtless declining, which accounts for the slight com-
`pression in the blue region of the circumference. Swedish workers have
`attempted to correct this defect in Ostwald’s system by placing each of
`the four psychological primaries at right angles to one another on the
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`148
`
`Sensations of Colour
`
`:8»_uEo~toI
`
`
`
`.2oom_aw>mm:_>o_2
`
`_mo_to>
`
`cmom
`
`uaucsm
`
`0:<55.
`
`2mom_oo>w
`
`2....
`
`.2”.
`
`
`
` 8.3.:8 \.um::_a.co..unmcE»
`
`.9W)E
`a_aEmm.06.£am.uuwxfi
`
`
`uwow_am>wuoxi
`
`VIZIO 1016
`
`
`
`VIZIO 1016
`
`
`
`

`
`Sorting and Recording
`
`149
`
`circumference and dividing the segments between them into visually equal
`steps, constituting a more ‘natural’ colour circle for use as the basis of a
`three-dimensional colour solid.
`Although a colour solid is a useful concept, and may even be constructed
`as a display object of great visual and intellectual appeal, either swatches or
`books are more convenient for everyday use. Colour atlases, such as the
`Munsell Book of Colour, often represent vertical sections of a colour solid,
`cut through each hue represented. For more specialist use, a restricted
`range of colours, varying by smaller gradations, may be reproduced as in
`collections for those who wish to specify the precise colour of a rose petal or
`a sample of human skin or tooth.
`Such visual matching of colour is a stepwise process, a placing of the
`sample of unknown specifications between two standard colours whose
`specifications are known. But how can we try to measure the colour specifi-
`cations of a material if we have no standard colour which matches it? What
`can we do to try to measure ‘colour’, to produce specifications of hue,
`saturation and brightness?
`Since colour is a sensation, there is a lot to be said for the measurements
`being made- by the eye. The human eye is, in fact, an excellent detector of
`differences of hue, and many people can assess the percentage of red, blue
`and yellow in a pigment with surprising precision. But human estimates of
`saturation, and of brightness, are much less reliable.
`‘The most precise visual methods of attempting to specify colours, like the
`use of a colour atlas, involve matching. The simplest are those devised
`merely to measure saturation, as in the determination of the concentration
`of a single coloured component in a liquid (see Figure 61).
`If the two solutions appear to be the same colour, the ratio of their
`concentrations is simply related to the ratio of the length of the two
`solutions through which the light has passed; and this may easily be found
`using a simple comparator involving either a plunger or a wedge.
`
`
`Figure 61. Comparison of saturation A fixed depth of the sample is viewed with one
`eye and a variable length ofstandard with the other. The geometry ofthe instrument is
`adjusted until the same depth of colour is observed by each eye.
`(a) Sample.
`(b) Variation of depth of standard by means of transparent plunger.
`(c) Variation of depth of standard by use of wedge.
`(d) Split field, one half from each eyepiece.
`In the top two diagrams, the same depth used for sample and standard gives a darker
`' field on the right. In the lower diagram, the length (I) ofstandard has been adjusted so
`that the two fields are indistinguishable. In this example, d =21, indicating that the
`saturation of the sample is half that of the standard.
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`150
`
`Sensations of Colour
`
`For many purposes, however, we need to know the hue, as well as the
`saturation. We again compare our sample with a colour which we can
`specify. But how can we vary this colour whilst still being able to specify it?
`One way is to mix coloured lights of known wavelength in known pro-
`portions. A simple arrangement is shown in Figure 62. The amount of red,
`blue and green light can be varied by horizontal and vertical movement of
`the filter assembly over the source of light, and the three lights are then
`mixed by diffusion and multiple reflections. More sophisticated devices,
`used mainly for
`research on colour,
`involve six lights of varying
`wavelengths. In each case the mixture of coloured lights, shone on to a
`white background, is matched with the unknown sample, illuminated from
`a standard source of white light.
`Alternatively, the sample may be matched by a patch of coloured light
`which has been obtained by passing white light through three filters, one
`
`L“““V
`nufiuttv
`
`Green filter
`
`Figure 62. Mixing lights The required mixture ofred, blue and green light may be
`obtained by adjusting that area ofeach filter which lies over the aperture to the mixing
`box.
`
`.:|I
`
`-
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`Sorting and Recording
`
`151
`
`magenta, one yellow and one blue-green (see Figure 63). Sets of such filters
`are available commercially for use in an instrument equipped with a stan-
`dard light source and known as the Lovibond Tintometer. The full range of
`250 filters of different depth for each of the three hues allows nearly nine
`million different colours to be obtained,
`including the full range of
`achromatic colours from white to black. The colour of the sample is readily
`specified in terms of the three filters used to match it.
`MB-G Y
`
`
`
`Sample
`
`Figure 63. Labelling with filters The colour ofthe sample can be matched with that of
`light which has passed through three Lovibond filters, magenta (M), blue-green (BG)
`and yellow (Y) ofspecified strength. A purplish blue sample, for example, would need
`a deep magenta (to absorb most 0fthe green), a medium blue-green (to absorb some,
`but not all, ofthe red), and yellow ofappropriate depth to reduce the intensity of the
`colour to that of the sample.
`
`Instead of matching the light which reaches us from a sample with that
`from an unknown, we can exploit the phenomenon of persistence of vision
`and match only the sensations. Split discs, coloured in saturated blue, green
`and red, are placed on a revolving platform in such a way that the pro-
`portions of the three colours can be varied (see Figure 64). When the
`platform is spun at high speed, the sensations merge, just as if three
`coloured lights were superimposed. The areas of the three colours are
`adjusted until the colour of the spinning disc exactly matches that of the
`sample, which can then be expressed in terms of the proportions of primary
`colours used.
`Nowadays, there is an increasing tendency to use photoelectric instru-
`ments for colour matching, instead of the eye of one or two individual
`observers. Such instruments monitor narrow bands over the whole visible
`spectrum, first detecting how much light in each band reaches them from
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`152
`
`Sensations of Colour
`
`the sample, and then converting this information into.the size of the
`stimulus which bombards each of the three cone systems in a normal human
`eye. Finally, the responses to these three stimuli are combined to give the
`colour experienced by the ‘standard observer’. Inescapably, these instru-
`ments give results based on sensations experienced as a consequence of
`human vision. But the vision is that, not of a few individuals, but of the
`‘standard observer’ built up from observations made by a large number of
`individuals selected for their normality of‘vision. But even when the match-
`ing is done, rather approximately, by eye, it is seldom left for a single
`individual: more often two, or even three, observers are used.
`
`(6)
`
`(D)
`
`
`
`Figure 64. Tops for colour measurement (a) Split circular disc ofpaper ofstandard
`colour. (b) Standardgreen, blue and red interlocked, exposing known areas ofeachfor
`colour mixing when the disc is rotated.
`
`Matching, whether by eye or machine, often gives different results with
`different sources of light; and as matching implies identity only ofresponse,
`this is no surprise. Two extreme examples of identical colours produced by
`light of very different composition were given in Table 3 (page 119).
`Imagine a pair of yellow pigments, one reflecting light of only 580nm and
`one reflecting only light of 540 nm and 630 nm, each at half the intensity of
`the first pigment. If both were illuminated with light containing equal
`intensities of the three wavelengths, the two pigments would match exactly.
`But if they were illuminated with light containing a slightly higher pro-
`portion of longer wavelengths, the second pigment would look redder than
`the first. Such ‘metameric’ pigments, which match only under one light
`source, are the bane of those who try to match clothes and accessories, or
`paint and fabrics (see Figure 65). Metamerism accounts for much of the
`popularity of ‘colour coordinated’ goods sold by the same manufacturer
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`
`
`refieehnoe
`
`Figure 65. A good match? The reflectance spectra oftwo fabric: which match perfectly
`in daylight. (Reproduced, with permission, from W. D. Wright, The Measurement of
`Colour, 4!}: edition, Adam Hilger, London, I 969.)
`Green
`
`Lm2.a-Imabw g.1-eenish-1.-allow
`
`Etna-:;,II'ean
`
`fellluw
`
`B|ue—violet
`
`Magenta
`
`Orange—red
`
`Figure 66. Maxwell’s triangle Shows how many, but not all, colours can be represented
`as a mixture of three primary coloured lights. The nearer a point is to an apex of the
`triangle the higher is the proportion oflight ofthe colour represented by that apex. The
`point X (50 per cent green, 35 per cent orange-red and 15 per cent blue-violet)
`represents an unsaturated greenish yellow.
`
`VIZIO 1016
`
`Sorting and Recording
`
`1 S3
`
`0 uses dyes such that any change in illumination causes almost the same
`ange in colour for the different materials.
`Colour solids and atlases can give us no information about the composi-
`- n of the light which causes a particular colour. For charts relating colour
`o 8
`
` Relative
`
`I
`
`0 R3
`
`-lfllll
`
`EIDIJ
`
`E-I3I.'ZI
`
`':'IZHl]-
`
`Wavelength (nm)
`
`VIZIO 1016
`
`

`
`154
`
`Sensations of Colour
`
`to composition, we turn to the second, more ‘scientific’ approach. Just as
`many colour solids are based on a ring of spectral colours, joined through
`purple, a triangle usually forms the basis of attempts to chart the colours
`produced by the mixing of lights. As early as 1855, Maxwell found that a
`great number of colours could be produced by mixing lights of only the
`three ‘primary’ colours: orange-red, green and blue—violet. The colour
`resulting from a particular mixture can be represented by a point on a
`triangular grid (see Figure 66). Many colours can be specified in this way:
`but not all. Whichever three primary sources we choose, there are always
`some colours (including many pure spectral ones) which cannot be rep-
`resented by a point in, or on, the triangle; which confirms that we cannot
`always match one colour by a mixture of three others unless we allow
`ourselves the option of mixing one of the primary colours with the sample
`and then matching the result with a mixture of the two other lights. Figure
`67 gives the recipe for obtaining a match for every visible wavelength with
`three primaries. Thus vivid yellow (570nm) (cf. page 119) can never be
`exactly matched by red (700 nm) and green (546 nm); but if a little blue
`(436 nm) is added to the yellow, a perfect match can be made. We can
`express this algebraically by stating that vivid yellow can be matched by red,
`green and a small negative amount of blue. But since there is no scope for
`plotting negative contributions on a Maxwell triangle, colours such as vivid
`yellow cannot be represented on it.
`It is too bad that we cannot choose any three wavelengths which, when
`themselves mixed together, will produce all visible colours. But there is
`nothing to stop us imagining that such ideal primary colours might exist;
`and if they did, they could be mixed in such a way as to produce three
`convenient real primaries, such as Maxwell used. So we could draw a
`mathematical modification of Maxwell’s triangle, with our three imaginary
`primaries at the comers. Points for the real ones, and for all other colours,
`would then be within it. Three such imaginary primaries have indeed been
`devised, such that any real colours can be represented as a mixture of the
`appropriate amounts of the three of them, and plotted as an idealized
`version of the Maxwell triangle. But not everyone is used to triangular
`graphs and the ruled paper may not be easy to obtain. Could we not use
`squared paper instead? How many variables do we need? Since we have
`decreed that a real colour R can be matched by a mixture of our three
`Imaginary Primary Lights— say, I units of one, P units of the second and L of
`the third — it might look as though we need the three variables, I, P, and L, to
`specify R. But we could also say that the brightness of the light is the sum S
`of the three primaries (so S=I+P+L), mixed in the ratio of I/S:P/S:L/S.
`And if we know IIS and P/S we already know L/S, because
`I/S+P/S+L/S=1. Since we are often more interested in colour than in
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`Sorting and Recording
`
`155
`
`Red (700)
`
`I
`
`These ‘positive’ colours
`when mixed together
`
`+2-5
`
`'-are
`
`+1-5
`
`+10
`
`+045
`
`9 '
`
`‘T?
`
`MATCH
`
`mixture of pure patch and
`these ‘negative’ colours
`
`400
`
`500
`
`600
`
`700
`
`wavelength of pure patch of light (nm)
`
`Figure 67. ‘Negative’ colours Many pure spectral colours can be exactly matched with
`three primaries only by use of ‘negati ve’ colours. When one primary is mixed with the
`‘pure’ patch 0flight, this mixture can be matched by some mixture ofthe other two. For
`the primary lights used here, the only colour which can be matched directly by the three
`primaries is a greenish yellow in the region of550nm. (Reproduced, with permission,
`from F. W. Billmeyer and M. Saltzman, Principles of Color Technology, Interscience,
`New York, 1966, p. 33.)
`
`brightness, we could concentrate on the two quantities I/S and P/S which
`specify the colour. We could then use ordinary squared graph paper. If we
`ever needed to know L/S, it would be easy enough to calculate it. And if we
`decided that, after all, we wanted to specify the brightness, we could
`represent it on an axis rising vertically, out of the paper. The final diagram
`would be much like a plan, with the two specifications of colour running
`north-south and east-west; or, if brightness is added, like a map which also
`shows heights, or isobars (or any third variable) superimposed on the plan
`as a series of contours.
`
`So all that we need in order to specify the colour of an object is a
`knowledge of the composition of the light falling on it, of the modification
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`156
`
`Sensations of Colour
`
`of the light by it, of the response of the normal human eye, and of the
`quantities of three imaginary primary lights which would produce the same
`response; and a piece of ordinary graph paper. The Commission Inter-
`nationale de l’Eclairage (Cl E), in 1931, defined the standard observer and
`three possible standard sources; and they produced tables showing the
`relationship between the observers’ response and the quantities of the
`imaginary primary lights which would in theory be needed to produce them.
`The tongue-shaped curve (see Figure 68) in the graph shows the specifica-
`100% primary P
`‘ideal green’
`In
`
`p
`
`0-8
`
`0-6
`
`0-4
`
`0-2
`
`'-
`
`_ 620650
`
`" 770nm
`
`.
`
`0
`‘ideal blue‘
`100%primary L
`
`0-2
`
`0-4
`
`mm
`08
`/f
`100% primary I ‘ideal red’
`
`0-6
`
`Figure 68. CIE tongue diagram The tongue encloses all visible colours, with the pure
`spectral ones lying along its curved edge. The inner triangle encloses those colours
`obtainable by mixing real primaries 436nm, 546nm, 700nm (i.e. those enclosed in
`MaxweIl’s triangle of Figure 66, page 153).
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`
`
`Sorting and Recording
`
`157
`
`08
`0-6
`04
`0-2
`0
`Figure 69. Purple and white The point Eforp = 0-33, i = 0-33 (and so I =1S==0-33) is
`‘equal energy’ white. Purples lie along the base line (see text).
`
`tions of the pure spectral colours as their values of i=1/S and p=P/S.
`(Modified definitions of the standard observer and the standard sources,
`introduced by the CIE in 1967, change only-details on the graph.) The
`diagram, known as a CIE chromaticity curve, has been used to depict
`colours, and the relationship between them, in a wide variety of situations.
`The enormous usefulness of the CI E diagrams arises from the fact that
`we can represent a mixture of two lights as a point on the line joining the
`points which specify them. So purples, formed by mixing red and blue, lie on
`the base-line of the ‘tongue’ (see Figure 69). A mixture of two parts red
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`158
`
`Sensations of Colour
`
`(770 nm) and one part violet (380 nm) would lie at point X, twice as near to
`the red point as to the violet point. Since all the colours formed by mixing
`real lights lie inside the area enclosed by the tongue, it is only those colours
`which are represented by points inside the curve which are visible. As the
`area outside the -curve represents only imaginary stimuli, we need not
`consider it further.
`
`We can also use CI E diagrams to specify a colour in terms of its dominant
`wavelength and its saturation, as well as in terms of the contributions of
`responses to imaginary primaries. The point E on Figure 69 represents an
`equal mixture of the three primaries and hence its position is p=0-3, i=0-3
`(and so l=0~3). It is known as equal energy white. Suppose now that we
`want to find the dominant wavelength of the colour represented by point X
`in Figure 70a. We can think of this colour as being some mixture of a
`
`0'8
`
`0-4
`
`02
`
`560
`
`
`Dominant wavelength 570nm
`
`650
`770 nm
`
`0
`
`02
`
`0-4
`
`0-6
`
`0-3
`
`VIZIO 1 01 6
`
`VIZIO 1016
`
`

`
`Sorting and Recording
`
`159
`
`0
`
`0-2
`
`04
`
`06
`
`0-8
`
`Figure 70. Dominant wavelength Specification ofcolour as a mixture of white light
`with:
`
`(a) One spectral wavelength.
`(b) A non-spectral mixture.
`(See text.)
`
`dominant wavelength D with white light of composition represented by E.
`To find which is the dominant wavelength, we remember that X must lie on
`a line joining E and D. So we draw a line from E to X and continue it until it
`meets the curve, at the dominant wavelength D. So X is an unsaturated
`version of the pure spectral colour D. How unsaturated? We know this
`from the distance between the mixture X and the dominant wavelength D.
`In this example, X is exactly h'alf—way between pure colour and pure
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`160
`
`Sensations of Colour
`
`white and so can be specified as a 50-50 mixture of light of 570nm and
`white light.
`What happens when the line joining E and the point for a colour does not
`meet the curve of pure spectral colours, but cuts the purple base line? For
`point Y in Figure 70b, the ‘dominant colour’ is point D, on the line joining
`the blue and red ends of the curve. Since D is half as far from the red end as
`from the blue end, it represents a purple mixture oftwo parts red to one part
`blue. The colour Y can be obtained by mixing this purple with white; and
`as the distance between white and Y is three times that between Y and
`the dominant colour, Y is a mixture of three parts of D with one part of
`white.
`We can use CIE diagrams, not only to represent colours, but also both to
`
`08
`
`0-6
`
`041
`
`02
`
`o
`
`770 nm
`
`-
`
`0-4
`
`0-6
`VIZIO 1 01 6
`
`are
`
`VIZIO 1016
`
`

`
`
`
`Sorting and Recording
`
`161
`
`650
`770 nm
`
`0
`
`___.
`04
`0'6
`08
`
`02
`
`Figure 71. Colours and complementaries (a) The use of a CIE diagram ‘to find the
`colour of a mixture and its complement. (b) Pairs of complementary pure spectral
`colours (see text).
`
`predict the result of mixing coloured lights and to locate complementary
`colours. Suppose we mix four parts of 490 nm with three parts of 550 nm;
`we obtain a colour represented by point Z in Figure 71a. But what colour is
`it? We find out, as before, by drawing a line from E, to Z, and beyond, till it
`cuts the curve, at 500 nm. So this is the dominant wavelength. The distances
`of Z from pure colour and pure white tell us that Z could be made from six
`parts of 500 nm and five of pure white. And the complementary colour?
`This is the colour 2,, which, when mixed with Z, gives white. So Z, Z, and E
`must lie on the same line; and the points 2 and Zc must be the same distance
`
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`162
`
`Sensations of Colour
`
`from E. We can therefore find Z by extending the line from Z to E by an
`equal distance beyond E. The colour complementary to point Z (unsatu-
`rated green) is unsaturated reddish purple.
`The colour Yc which is complementary to a pure (saturated) spectral
`colour Y is itself saturated and can easily be located by drawing a line»from
`the point Y through E and beyond until it cuts the other side of the curve at
`the required point YC (see Figure 71b).
`'
`So far, we have ignored any change in brightness, even though we could
`add this information to the CIE diagram by using a vertical scale. We know
`(page 120) that the brightness of a light depends both on its intensity and on
`its wavelength, since the eye is more sensitive to some wavelengths than to
`others. So the brightness of an object depends both on the object itself and
`
`520
`
`0-8
`
`0.5 .
`
`500
`
`0-4
`
`02}-
`
`l_
`
`0
`
`0-2
`
`J__
`0-4
`
`_____1____________i
`06
`0'8
`VIZIO 1016
`
`VIZIO 1016
`
`

`
`Sorting and Recording
`
`163
`
`0'8
`
`0‘6
`
`500
`
`0-4
`
`
`0
`0-2
`0-4
`0-6
`08
`
`Fig