Leonardo

Colour as Sensation in Visual Art and in Science

Author(s): Jonathan C. Fish
Source: Leonardo, Vol. 14, No. 2 (Spring, 1981), pp. 89-98
Published by: The MIT Press
Stable URL: http://www.jstor.org/stable/1574399 .
Accessed: 12/06/2014 13:09

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.

The MIT Press and Leonardo are collaborating with JSTOR to digitize, preserve and extend access to
Leonardo.

http://www.jstor.org

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
Leonardo,Vol. 14, No. 2, pp. 89-98. 1981.
Printedin GreatBritain. 0024-094X/81/020089-10$02.00/0
? PergamonPressLtd.

COLOUR AS SENSATIONIN VISUAL

ART AND IN SCIENCE

Jonathan C. Fish*

Abstract-Contemporary colour science and colour art are, for the most part, on separate paths
in spite of strong reasons for unity and collaboration. It is argued that this schism is not based
on any intrinsic incompatibility, but on causes that are symptomatic of a disorientated culture.
These causes are recent in origin and may be classified into those that are common to art or to
science, in general, and those that are peculiar to the subject of colour. Prominent in the latter
class is the failure by artists to appreciate the importance of distinguishing colour sensations
from the external causes of these sensations and the inevitable priority given in science to the
objective, rather than to the subjective, aspects of colour.
It is suggested that material for a common philosophy, deeper than that of medium or
technique, can be found in the study of colour sensation as part of the wider problem of relating
private experience to external reality. An important example is provided by a discussion of
contemporary uncertainties about the number and nature of colour attributes and of the
relevance that this problem has for the visual arts. Some of the author's kinetic artworks and
colour demonstrations pertinent to the subject are discussed.

I.
Delacroix and Georges Seurat in the colour
Since C. P. Snow first published his 'Two Cul- science of their day have been well documented.
tures' lecture [1], the separation of science from In spite of their pioneering attempts, a reaction
the so-called humanities and the arts in western ensued leading to the wide-spread belief that the
industrial countries has been much discussed. In scientific knowledge of colour is useless, if not
the words of W. D. Wright, a pioneer of colour harmful, to visual artists. Thus this early attempt
science, 'We have in the subject of colour a quite by painters to bridge the gap between art and
specific example of the two cultures, in which science was premature.
artist and scientist speak a language the other
does not understand. We have a responsibility to
II.
put an end to this dichotomy, and if, as I now
believe, the scientist's concept of colour is incom- I believe that one of the reasons for the failure
plete, then it is high time we developed a more of artists generally to accept colour science as
adequate philosophy of colour to which both essential knowledge has been confusion between
artist and scientist can subscribe' [2]. (For the colour as it is subjectively experienced and the
view that science is not a culture in the traditional external causes of that experience, namely mate-
sense see Ref. 3.) rial substances or the light reflected by them. J. S.
In the first half of the 19th century, painters were Mill pointed out over a century ago that the
receptive to science and also to the idea, ridiculed confusion is preserved in the everyday English
today, that there might be precise rules in visual language. This is still true. Mill wrote: 'We have a
art similar to the laws of nature [4-6]. In that name for the objects which produce in us a certain
period, black and white photography was in- sensation: the word white. We have a name for
vented, which may have caused a shift in the the quality of those objects, to which we ascribe
emphasis in painting towards colour and moti- the sensation: the name whiteness. But when we
vated some artists to study its properties. The speak of the sensation itself (as we have not
interests of the painters J. M. W. Turner, Eugene occasion to do this often except in our scientific
speculations), language, which adapts itself for
the most part only to the common uses of life, has
*Painter,kineticartist,teacher,biochemist,21 CollegeRoad, provided us with no single-worded or immediate
Cneltenham, Gloucestershire,GL53 7HY, England. (Re- designation; we must employ a circumlocution,
ceived 7 Sept. 1979) and say, the sensation of white or the sensation

89

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
90 JonathanC. Fish
of whiteness;we must denominatethe sensation and stimuli are commonly a product of colour
either from the object, or from the attribute,by properties and illumination, it is very easy to
whichit is excited' [7]. confuse them. Three types of confusion can
Physiologists have shown that light reflected occur:
from a dab of paint and enteringthe retinaof the (1) The pigment-stimuluserror-the incor-
eye results in the dischargeof electrical signals rect identifications of the colour prop-
that are transmittedto certainregionsof the brain erties of materialswith the propertiesof
causing the occurrenceof colour sensations [8]. the light that they reflect(or transmit).
These sensations are one meaning of the term (2) The stimulus-sensationerror-the incor-
colour. But since other meanings are commonly rect identificationof colour stimuli with
given to it, two of which are discussedbelow, it is colour sensations. Ewald Hering, one of
sometimes helpful to refer to the sensation of the pioneers of colour science, wrote:
colour as psychologicalcolour. A clearexposition 'The history of physics might also be de-
of this approachis given in Ref. 9. scribedas a historyof struggleagainstthe
When light falls on a dab of coloured paint, prejudicesoriginatingfrom this linguistic
part of it is absorbed and the rest is either identification of sensory qualities with
reflected or transmitted.Usually the wavelength their physicalcauses' [12]. Thisdistinction
compositionof the light falling upon a coloured is important to artists, because colour
materialand that of the reflected(or transmitted) sensations can be evoked and controlled
light are different. Because the light fallingupon only by manipulatingcolourstimuli.
a materialcan be variedboth in compositionand (3) The pigment-sensationerror-the incor-
intensity, it follows that the composition and rect identificationof colour propertiesof
intensityof the light reflected(or transmitted)by materialswith colour sensations.Painters
it must also vary. However, the relationshipbe- commonly refer to colour sensations by
tween incidentlight and reflected(or transmitted) the names of pigments-rose madder,
light remains constant and can be used to char- chrome green, raw umber, etc. But they
acterizepaintsand other materials.I shallreferto frequentlyexperience situationsin which
this as the colour property. spatial interactionsbetween stimulicause
A tiny proportionof the reflected(or transmit- the colour sensationproducedby a parti-
ted) light reaches the retina of the eye, carrying cular pigment to be unexpected. Colour
colour informationto it. This light is termed the constancy, a psychological effect that
colour stimulus.Unlike colour properties,colour tends to maintain recognition of, say,
stimuli from an object vary not only with its black paint in differentilluminationsmay
illuminationbut also with movementsof the eye be destroyedby the influenceof stimulus
and head. The term colourstimulushardlyoccurs interactionsin human-madeobjects (see
in the literatureof art, but it should, for artists Section IX) [13]. The mechanismof col-
can control colour sensationsonly by manipulat- our vision is thus exposed to a situation
ing colour stimuli. However, although colour for which the biological evolution of vi-
perceptiondepends directlyon colour stimuli, it sion did not prepareit.
usually correlateswith colour properties.This is
due to the capacityof the brain to utilize spatial
and temporal information provided by colour IV.
stimuli about the illumination[10, 11], a subject The effect of these sourcesof confusionon the
that I shall not discuss here. Because of this visual arts is exemplifiedby the conceptualim-
correlation and because of the way in which broglioconcerningprimarycolours, a termthat is
colour sensations are experienced, humanshave frequently used without specifying whether it
come to associate colour with objects and mate- refers to colour properties, colour stimuli or
rials. Indeed, the biological function of colour colour sensations. From the time when Jacques
vision is to provide information about colour ChristopheLe Blon publishedhis ideas on print-
propertiesratherthan colour stimuli. ing in colour (1756) [5], it was widely believed
Unlike colour stimuli and colour properties, that any colour could be derived from red, blue
colour sensations are wholly private. For ex- andyellow. This made sense only when appliedto
ample, two persons looking at a painting may colour properties, however, since it was thought
agree that one area of the canvasis pale cerulean that red, blue and yellow pigmentswere sufficient
blue; however, given the present knowledge of to mix any colour, a belief sharedby manyartists,
neurophysiology,it is an untestable assumption but not borne out on their palettes [4, 5]. In 1831
that their colour sensationsare identical.It is not the physicistDavid Brewsterexpoundedhis belief
surprisingthat until recentlymost colour science that red, blue and yellow were the primarycol-
amounted to the study of colour propertiesand ours for mixing light as well, which was later
stimuli. shownto be untrue-an exampleof the pigment-
III. stimuluserror.
Influencedby Brewster, Turner assumedthat
Since coloursensationsare producedby stimuli red, blue and yellow were the principalcoloursin

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
 Colouras Sensationin VisualArt and in Science 91

sensation and used this triad as the colour struc- use in art. For example, effective compositionin
ture for some of his paintings [4]. The same paintinginvolves carefullydeterminedcolour dif-
pigment-sensation error appears in Piet Mon- ferences-sometimes subtle modulations, some-
drian's restricted use of these colours 90 years times strong contrasts. In each instance painters
later. Turner and Mondriansucceeded superbly choosingcolourswill, perhapsunconsciously,ask
in their use of colour, despite their incorrect themselves, 'In how many differentways can two
assumption. colour sensations be alike or be different?'Sur-
Perhapsit was the realizationthat 19th-century prisingly,this questionis still a subjectof debate,
colour science had not adequatelydifferentiated even amongstcolour scientists.
between stimuli and sensations that led to its Since sensations are subjective and priorityis
abandonmentfollowing its seemingly premature given to objective measurement, emphasis in
use in Impressionism and Neo-Impressionism. colour science until recentlyhas been put on the
The pseudo-scientifichypotheses of colour har- study of stimuli rather than on the study of
mony of the chemist Michel Eugene Chevreul sensations. This has resultedin the sophisticated
(1839) and of the painter Seurat may have done subject of colorimetry,which provides a way of
more harm than good to the cause that they describingcolour stimuli by combiningphysical
espoused. Henri Matisse, departing from Neo- measurementwith the knowledge, obtained by
Impressionism,wrote: 'My choice of coloursdoes research,of a typicalhumancolourresponse.
not rest on any scientific theory; it is based on This methodis commonlyappliedto an isolated
observation, on sensitivity, on felt experiences. colour stimulus, that is, one that is uniformand
Inspiredby certain pages of Delacroix, an artist has very dark surroundings.For these conditions
like Signac is preoccupied with complementary and for normalcolour vision at daylightlevels of
colours, and the theoretical knowledge of them light intensity, it has been determinedthat only
will lead him to use a certain tone in a certain three variablesof the stimulus are relevant to a
place. But I simplytry to put down colourswhich specific colour sensation. These variablescan be
rendermy sensation'[14]. expressed in different ways, but the most perti-
In 1959Naum Gabo expressedan opinion that nent triad here are the three that describe light
is probablywidely shared amongst artiststoday. obtainedby mixingmonochromaticlight (light of
He wrote: 'All textbook theories about colours a single wavelength)with white light (light, such
are concerned with their chemical and physical as sunlight, of two or more wavelengths that
propertiesand have no relationwhateverto their produces an achromatic (hueless) response) to
aestheticnature. match the given colour stimulus.Thus a stimulus
'An artistcan be absolvedfrom the necessityof can be specified quantitativelyby the three vari-
knowing all that the scientist knows about col- ables describingthe mixturethat matchesit. The
ours. Except for some practical information variablesare: (1) luminance-a measure of total
which it is necessary for him to have about the intensity of the stimulus (radiation) weighted
treatmentof pigmentsand materials,the scientific wave-length by wavelength according to the
knowledge of their chemical and physicalprop- visual sensitivity of an internationallyaccepted
erties is apt to confuse the artist more than to standardobserver;(2) dominantwavelength-the
enlightenhim in his art and to distracthim from wavelengthof the monochromaticlight required
knowingwhat his experiencesof colour are' [15]. in mixturewith white light to matchthe stimulus
The firstsentence above is untrue. Presumably and (3) colorimetricpurity-the ratio of the lumi-
Gabo had not read Ewald Hering's Outlinesof a nance of the monochromaticlight to the lumi-
Theory of the Light Sense [12] or David Katz's nance of the mixture.
World of Colour [10]. However, to the second These three variablesare psychophysicalvari-
sentence quoted, it is embarrassinglyharder to ables, and the colour they describeis here called
find more than a partial refutation, such as pre- psychophysicalcolour in order to distinguishit
sented above. from colour sensation or psychological colour.
It must be noted that Matisse was wrong to The reasonfor this is that physicalmeasurements
imply that, if colour science were to be useful, it of the light are used to calculatethree numerical
would play a role in the making of aesthetic values. The link with a typical colour responseis
decisions. Before artists can make colour deci- made by weightingthe measurementsusinginter-
sions, they must have a conceptual framework nationallyacceptedcoefficientsthat are obtained
with which to think about colour experienceand from colour matchingdata for a standardobser-
also a knowledge of the stimulus conditions by ver. These weighted numericalvalues are then
whichthis experienceis produced. used to calculatethe three psychologicalvariables
[9].
V.
VI.
While contemporary artists generally have
made little use of pertinentnew knowledgeabout I have selected for discussion here three
colour [9], it should be pointed out that there are psychophysicalvariables (luminance, dominant
certaindifficultiesin colour science that hinderits wavelength and colorimetricpurity), which are

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
92 JonathanC. Fish
radius(100%saturation).A muchused represen-
tation in cylindricalcoordinatesis Munsellcolour
space (Munsellcolour system) where the coordi-
z nates are given as follows: MunsellHue, 1; Mun-
sell Value, 2; and MunsellChroma,3.
I Helen Thomas has describedin Leonardoher
use of the psychophysicalOstwaldcolourspacein
her paintings [16]. The Ostwald colour space is
cylindrical,but it is bounded by a double cone
whose axisis vertical.D. Nickersonhas presented
in Leonardoa discussionof the OpticalSocietyof
America (OSA) uniform colour scale samples
[17]. A colour space has been selected in which
coloursof suppliedsamplesare located 'according
to a space lattice in which each sample is sur-
rounded by 12 other samples, each at an equal
distance from the central sample', and equal
of colour
Fig. 1. Cylindricalcoordinatesfor the representation distancesrepresentperceptuallyequal colourdis-
space. Colourspecifiedby threevariablesis representedby a tances.
point in colourspace.
sometimes used as approximate indicators of
three particularpsychologicalvariables, or attri- VII.
butes, of colour sensation, respectively: (1) There are difficultiesthat arise in the use of
brightness,the attributethat varies mainly with three stimulus-orientedpsychophysicalvariables
luminanceon a scale from dazzlingto very dim; as a conceptualbasis for the descriptionof colour
(2) hue, the attribute that is suggested by the sensations.One arises because, when colour sen-
words red, blue, turquoise,purple, etc. (3) satu- sations are scaled equally, the corresponding
ration, the attribute that seems to refer to the psychophysicalvariablesmustbe spacedunequal-
amountof hue componentrelativeto the total of ly. (Thisproblemhas been partiallyovercomefor
the hue and the hueless componentsin the colour the purposesof measuringcolour differences[18;
sensation.Thus a perceivedred may be judgedto see also 9, p. 83].) Another arises because,
be bright,fromits intensity,and to have a red hue althoughtotal luminanceis principallyassociated
with an orange tinge. It may be judged to be with the sensation of brightness, luminance
ratherlow in saturation,meaningthat quite a bit changes also involve shifts in hue (the Bezold-
of white sensation appearsto be mixed with the Briicke phenomenon) [19]. Similarly, whilst
orange-red,producinga pinkishcolour. changesin purityof the stimulusmainlyaffectthe
These three attributesof psychologicalcolour sensation of saturation, they also involve small
are commonly described as three continuously shiftsin hue.
varying quantities that can be represented as More serious difficultiesare encounteredwhen
coordinates in a 3-dimensionalgraph, a colour it is desired to describecolour sensationswhen a
space, such that a point located in that space uniformlycolouredsurfacein view is not isolated.
designates a perceived colour. A colour space Although three variables may be sufficient to
given by cylindricalcoordinates is indicated in describe an isolated stimulus, such a stimulus
Fig. 1. Here hue is commonlygiven by an angular does not have the same appearanceas the col-
position1 (muchas the hourof the day is givenby oured surfaceswith whichmost painterswork.To
the angularposition of the hour-handon the face elicit such surface appearancesit is necessaryto
of a clock). The hues are considered to be introduce a surroundingstimulus. When this is
arrangedin a continuouslyvaryingmannerfrom done, dramaticchanges occur in the colour sen-
violet to red as they are in the continuouscolour sations that are associatedwith the stimulusthat
spectrum,and the circle is completed by joining is surrounded (termed the target stimulus) [20,
red to violet by a continuous series of non- 21]. This is most easily explainedby describinga
spectral hues (the bluish reds and purples). simpleexperimentusingtwo slide projectors.Let
Brightnessis representedby a verticalscale2 and one projector project a small disc of coloured
saturationby a radial scale 3, both increasingin light upon a white screen. This is the target
the direction indicated by the arrows. Thus the stimulus.Let the other projectorprojecta ringof
pinkish colour mentioned above would be rep- white light in such a way that it surroundsand
resentedin colour space by a point on a radiusin touches but does not overlap the target stimulus
a direction designatinghue and positioned at a (termed the surround). If, to start with, the
level designatingbrightness.The point would be surroundis very dim, then the target is isolated,
located on the radius, possibly at some point at, and its appearancecan be variedonly in the three
say, 40 or 50% of the way fromthe verticalaxisof ways that have been described. Moreover, the
the cylindrical coordinate system (where 0% target looks like a patch of light, and its colour
saturationis represented)to the extremityof the does not appear to belong to the surface of the

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
 Colouras Sensationin VisualArtand in Science 93

screen. Now suppose that the luminanceof the indicatedabove. (A demonstrationof this can be
white surroundis increasedto about half that of given with the use of my kinetic light box de-
the target. The target now begins to look like a scribedin Section IX.)
brilliantlycoloured surface that is emitting light By itself the additionof lightnessas a variable
(it gives the illusion of being fluorescent).As the of colour is not sufficientto invalidatethe use of
luminanceof the surroundis raised further, the three variablesto describesurfacecolours,if they
target becomes less brilliant and resembles a are assumedto be viewed under the same illumi-
normalcolouredsurface.As the luminanceof the nation, for in this case brightnessis irrelevant.
surroundis made higher than that of the target, The same cannot be said about brillianceor, in a
two new attributesof the colour sensationappear more limited sense, greyness (as indicated be-
that are not presentin isolatedstimulibut that are low). (I think the word blacknesswould be more
normally present in the colour sensations pro- appropriate.)One can readily be convincedthat
duced by coloured areas in paintings.Therefore, this is an attributeof non-isolatedcolours, if one
a stimulus-basedconcept about the attributesof notes that the specific colour sensations neutral
colour sensation as they are experienced by a grey, ochre, brown, olive green, maroon and
personviewinga colouredarea in a paintingmust navy blue are not elicited by isolated stimuli.
be based not on three psycho/physicalvariables Those who insist on the use of three attributesto
but at least on six, three for the stimulusassoci- describepsychologicalcolour must explainthese
ated with the coloured area and three for the distinctivesensationsby the assumptionthat they
surround. (In Leonardo, Ref. 22, a method has are simplydarkeror less saturatedversionsof the
been discussedfor extendingthe luminancerange corresponding colours, white, yellow, orange,
of colours in a paintingby selective illumination green, red and blue. Whetheror not brownreally
of parts of it. A striking 3-dimensionalappear- looks like a mixture of orange and black or
ance may be producedby so doing.) whetherit is 'unique'is anotherquestion[24].
In a fascinatingand provocativebook, Evans One important point made by Evans is that
has suggested that, except when the stimulusis lightness and saturationare perceivedseparately
isolated againsta darkbackground,there are not from greyness. Another is that separate colour
three but five variablesof coloursensationrelated attributesare not necessarilyindependent from
to the six stimulusvariables [21]. In additionto one another. In the past this has caused confu-
the attributesof brightness, hue and saturation sion. For example, the series of coloursextending
already described, there are two additionalvari- fromblackthroughgrey to white has usuallybeen
ables, lightnessand brilliance.Both lightnessand conceived as the variation of a single attribute.
brilliance are related to the luminance of the Accordingto Evans, however, two attributesare
stimulusrelativeto the luminanceof its surround, concernedthat vary simultaneously.As the series
but brilliance varies with dominant wavelength progresses from black to white the colours
whereas lightness does not. The term lightness become lighter (lightness increases), but sim-
correspondsto the artists'termstonalityand value ultaneously they become less grey, or more
of a surface colour. While the attributeof light- brilliant,(Fig. 2, top).
ness may be indicated by the psychophysical The evidence that lightness is separate from
luminancefactor, the relation is not a linear one greyness (brilliance) is based on absorbing re-
(Section IX). search to which justice cannot be done here.
Brightness in related (that is, non-isolated) Briefly, Evans discovered that, when the lumi-
colours is the impressionof illuminationand is nance of a very high puritystimulusis decreased
relatedto the total luminanceof the stimulusplus from a high value and the luminanceof a white
that of the surround.A piece of coal in strong surroundis held constant, the relativeluminance
sunlightlooks both dark and bright, whilstwhite (the ratio of one to the other) at which greyness
paper in a dark room looks light but dim. This first appearsconstitutesa thresholdquantitythat
distinctionwas made long ago (1911) by David varies greatlywith the dominantwavelength(and
Katz, who demonstratedthat the visual mechan- hence the hue) of that stimulus.Lightness,on the
ism is able to separate informationabout illumi- other hand, varies only with the relative lumi-
nationfrom that about the reflectingpropertiesof nance. To comprehend this, compare in 'the
surfaces[10]. Part of the fascinationof paintingis mind's eye' the sensations evoked by viewing a
the way in which the visible lightnessof pigments chrome yellow and an ultramarinepaint sample.
may generate paradoxicalillusions of brightness. Both sensations can be produced without any
Faber Birren in an article in Leonardo has pre- discernablegrey content and of equal brilliance,
sented illustrationsshowingseveral differentillu- yet, when matched, the blue is darker than the
sions of illuminationthat may be producedwith yellow. The sensation of greyness of a colour is
the use of paint [23]. On the other hand, artists not the same sensationas its darkness.Evansused
who work with coloured light rather than pig- the word brilliance to refer to two mutually
ments know that it is possible to elicit the illusion opposite attributes, greyness and fluorence, the
of pigmentedsurfacesfrom the projectionof light apparentfluorescenceof very brilliantstimuli.
on the back of a translucentscreen. In fact, all Unfortunately, Evans died before the publi-
surface colours can be produced this way, as cationof his book and, since muchof the research

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
94 JonathanC. Fish
(possibly unconscious?) of stimulus conditions,
whichmost personsmay not possess, is assumed.
Does a light surface appear to reflect a higher
fractionof light than a dark surfaceto painters?
This method of defining attributesof perceived
colour using the verbsperceive,appearandjudge
(Hunt) may work better for some attributes,such
4 as brightness,but it does suggest an obstacle to
understandingbetween the objective and subjec-
4I
-t
tive approach.Those who are not concernedwith
.4
psychophysicalmeasurementmay also feel that
the representationof hue in colour space as if it
were a dimensionalattributewith angularmagni-
tude fails to describe the associated qualitative
changesof subjectivehue. Even more puzzlingto
most artists must be an hypothesis of colour
sensation that describes the positive colour ex-
WH1.rE YESS >
perience black as a colour with zero lightness,
zero saturationor colourfulnessand zero hue.
< SAA.ACANEIS ORq7.jEA#YMjIS
The most convincingalternativeframeworkfor
the descriptionof colour sensationwas suggested
by EwaldHeringover 100yearsago [12]. As early
ULAC- as 1865, H. Aubert had stated that all colour
W14IT-
sensationscould be conceived of as if they were
mixturesof, or related to, six fundamentalsen-
sations [12]. These psychologicalprimarycolours
Fig. 2. (Top) The simultaneous change of lightness and bril-
liance in the black-grey-white series of colour sensations. (commonly termed unique colours) are those
referredto by the words black, white,red, green,
(Bottom) Hering's schematic representation showing the pro-
gressive change in the proportions of white and black sensations blue andyellow. They can be imaginedas forming
experienced in the black-grey-white series. [Modified from the vertices of an octahedron of which the line
Ref. 12].
joining two vertices represents the possible
that he had initiated needs independent evalu- psychological (visual) mixturesbetween the two
ation and continuation, his interestingideas are corresponding primaries (Fig. 3, left) (Hofler
not yet universallyaccepted. More recently, R. 1897, see ref. 5). Hering developed the idea of
the six primarycolours in what he describedas
G. W. Hunt and his colleagueshave reintroduced the Natural Colour System (N.C.S.), having de-
his (Hunt's)term colourfulness[25]to referto the clared his intention to classify colour sensations
apparent absolute chromatic (hue) content of a with completedisregardof stimuliandproperties.
colour as opposed to its achromatic(black-white) He pointedout that the black-grey-white(achro-
content [18]. Hunt's definitionsdistinguishattri-
butes of colour appearance(his term) that have matic) series should be regardednot as a single
absolute magnitude, such as colourfulness and magnitudechange, but as a progressivechangein
proportionof two distinct and equally powerful
brightness, from those that have only relative sensations. As any grey approaches black, it
magnitude,such as saturation,lightnessand per-
ceived chroma. appearsmore blackish,as it approacheswhite, it
appearsmore whitish(Fig. 2, bottom).
VIII.
The variety of ways in which colour sensations
are either described or defined in the scientific
literatureon colour is confusingfor artists.While
it is most unlikely that a colour scientist will
confusestimuliwith sensations,in my opinionit is
possiblethat the convenienceof the colour-space
concept for specifyingmathematicallythe stimu-
lus correlates of sensation may prejudice the
definitionand choice of sensation attributes.For
example, in the C.I.E. (1970)InternationalLight-
ing Vocabulary,lightnessis defined as 'that attri-
bute accordingto which an area appearsto reflect
diffuselyor transmita greateror smallerfraction
of incident light' [26]. By the use of the verb Fig. 3. (Left) Octahedral arrangementfor visualising the mix-
appear the task of defining lightness wholly in ture of unique colours proposed by the author. (Right) The
termsof its correlatesis avoided, but a knowledge Swedish N. C.S. Colour Space after A. Hdrd [Ref. 33].

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
 Colouras Sensationin VisualArt and in Science 95

Tryggve Johansson (1905-1960) and later

others, notably A. Hard of the Swedish Colour
Centre Foundation, developed Hering's sugges-
tion into a quantitativemethod of colour sen-
sation specification [28]. The Swedish N.C.S.
colour space (Fig. 3, right) and the new atlas of
colour based on it embody much importantre-
searchinto the subjectivescalingof coloursensa-
A uniquehueor tions and are of realvalue to artists.However,the
a mixtureof unique Swedish N.C.S. also involves the colour space
hues
concept, a concept that, as far as I am aware,
Heringdid not employ. It has been shownthat in
Constantblack content manyrespectsthe SwedishN.C.S. colourspace is
similar to the much-used Munsell colour space
[29]. R. W. G. Hunt has concluded that 'for a
given set of viewing conditions, uniform colour
Uniqueblack spaces are useful for systematizingthose attri-
butes that are independentof the absolute lumi-
Fig. 4. A colour specified by a point (Q) in a N.C.S. constant nance level' [18].
hue triangle.
So, was Evansmistakenin his convictionthat at
least four variables are necessary to describe
In the same way, changesof saturatedhue can surfacecoloursunderconstantillumination?I do
be thought of as changes in proportion of two not think so. Across the pages containingcolour
uniquehues but not in any combination.Red and chips of constant hue in the Munsell Book of
green cannot be combinedtogether nor can blue Colour can be seen sloping lines of constant
and yellow. Hering called these two pairs oppo- brilliance,which is a variablenot representedin
nent colours. The four possible hue ranges (red- Munsell notation. Similarly,the SwedishN.C.S.
yellow, yellow-green, green-blue and blue-red) colourspace accommodatesbut does not describe
can be imagined to form the sides of a square the variable lightness. According to Hard and
whose corners are occupied by the four unique Sivik [30], 'In the N.C.S., lines for constantvisual
hues red, yellow, green and blue, althoughHer- lightness are straight and converge at a point P
ing himself preferred to position these ranges outside the N.C.S. color triangle,the position of
along the four quadrantsof the hue circle. The P being dependenton hue'. In my opinion, these
advantageof the circleis that it makesthe hues on two examples show that colour sensationcannot
the perimeterequidistantfrom the centre where be described by one set of three variablesand,
an achromaticcolour can be represented. The hence, that the colour space concept is mis-
disadvantageof this circulararrangementis that leading.
there is not a geometricalrepresentationof the In his originaldescriptionof the N.C.S., Her-
sharp change that occurs at the positions of the ing pointed out that, in additionto those colour
unique hues. changes that are attributableto varyingpropor-
Research has confirmedthat hues can indeed tions of the primaries,colours can be either light
be describedconsistentlywith four colour names or dark [12]. He used the word brightnessin a
[19, 27]. Further, Hering described how colour context that makes it equivalentto Evans' light-
sensations of a single hue may appear to be ness. He stated that primarysensationshave an
'veiled'with black (that is, look blackish),'veiled' intrinsiclightness (brightness)that is invariant.
with white or with any mixtureof black andwhite White, yellow and red are intrinsicallylighter
(grey). For a given hue, this means that a colour than black, blue and green. It is, therefore, a
can be specified by a point Q on or within a dependentvariable, since the lightnessof a com-
triangle, the corers of which represent the un- pound sensation would be intermediatebetween
veiled hue C, black S and white W (Fig. 4). the lightnesses of its constituent primaries. To
Crossing the triangle in three directions are illustratethis, I have drawn an N.C.S. constant
dashed lines of constant black content, constant hue triangleinside an imaginaryprismthe vertical
white content and constant hue content. When axis of which represents lightness (Fig. 5). The
the positionsof the dashedlines are moved in the triangles correspondingto different hues tilt at
directions indicated by arrows, the contents of different angles, although data given by Hunt
black, white and hue increases.Of the six primary suggestthat they might be curvedratherthan flat
colour sensations, not more than four (black, planes [18]. It may be true that three variables
white and two unique hues) can be combinedin a suffice to specify colour sensationsagainsta con-
homogenous colour sensation. Only relative stant background,but it does not follow that they
proportionsof these four primariesare meaning- are sufficientto describethem. This distinction,so
ful. Thus, the N.C.S. requiresthe specificationof importantto painting, is made too rarely. Evans
three quantities and not more than two hue maintained that the four variables of surface
qualities. colour are independant of each other, but his

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
96 JonathanC. Fish

I.HTtA 5vS

(HS.II qs5
4'1X/^rm^&5;ES

Fig. 6. Maxwell discs for demonstrating colour phenomena,

metal discs, electric motors, 30 x 120 cm, 1970. The discs are
Fig. 5. The representation of a N. C.S. constant hue triangle in a
shown at rest (top) and in rotation (bottom).
prism to include lightness.
phenomenon described by Rayona Scina (1847)
grounds for making this claim are not wholly [3] (I have modifiedthe design slightly).A black
convincing(to me, at least) [21]. and white painted design is arrangedso that half
However, I suggest that Hering's original of it (showing black squares) is seen through a
hypothesis is in agreement with Evans' central sheet of blue Perspex,whilstthe otherhalf (show-
contentionthat at least four variablesare necess- ing larger black square bands) is seen by its
ary to describe surface colours under constant reflection on the blue Perspex surface. In the
illumination. If Evans was right, then much of explanations,attentionis confinedto what is seen
what had been written about colour as it is throughand on the blue Perspexsheet.
actually experienced has been confusingor mis- The stimulusreferringto the backgroundis a
leading. mixture of two stimuli: blue light, produced by
It is unfortunate that early attempts to use daylightpassingthroughthe blue Perspex,reflect-
colour as a bridgebetween science and the visual ing fromthe white backgroundof the coveredhalf
arts did not result in the unqualifiedbelief by of the picture and passingback throughthe blue
painters that colour science is relevant. The Perspexagain, and whitelight, producedby day-
stimulus-orientatedconcepts that were used for light reflectingfrom the backgroundof the un-
those early attemptswere inadequate.Sincethen, covered half of the picture to the blue Perspex
however, there has been much new knowledge where it is reflected. The backgroundis seen as
about colour. With the revival of interest in lightblue.
colour sensation among scientists, it is time for The stimulusreferringto the squaresis a mix-
anotherattempt to apply colour science in paint- ture of white with a trace of blue light; the blue
ing. A failure to do this would be a sign that light componentis reduced by absorptionby the
perhaps Colin Blakemore was justifiedwhen he blackpaintin the coveredhalf of the picture.The
remarkedthat 'society could collapse, because it squaresare seen as yellow.
no longer comprehendsits own culturalinherit- The stimulus referring to the black square
ance' [31]. bands is a mixture of blue with a trace of white
Figures 6 and 7 (see colour plate) show two light; the white light component is reduced by
teaching devices I have designed to demonstrate absorptionby the black paint in the uncovered
some of the colour phenomenamentionedabove. half of the picture. The squarebands are seen as
In one device, Maxwell discs (steel discs with darkblue.
sectorspaintedin two colours)are rotatedrapidly Why are the squaresyellow?The humanvisual
by independentlyactivated electric motors (Fig. system behaves in such a way that information
6). The colours are mixed (averaged)by the eye aboutcolouraspropertyis retainedandutilizedin
and the appearanceof surfacetexturedisappears. spite of naturally occurring changes in the
Magnetictape on the back of each disc permitsits wavelengthcompositionof the illumination;col-
easy attachmentto a motor shaft. A collectionof ours of surfaces in view are assigned mentally
sets of discs of different colour pairs has been after compensationsare made for the colour of
accumulatedfor a varietyof demonstrations.For the illumination.Thus, in effect, a pictureshow-
example, the rotating black and white sectored ing black squares surroundedby black square
discs in Fig. 6, bottom, demonstratethat, when bands is being viewed in blue illumination.Sub-
the luminance factor (luminance in relation to tracting blue from the blue and white stimulus
pure white) is increasedby ten equal steps from a mixture(background)leaves white and from the
black (4% pure white) to a white (88% pure blue with the negligible trace of white (square
white), the lightnessincreasesin steps of decreas- bands) leaves blue. Since white light (daylight)
ing magnitude. minus blue light normallyproduces yellow (the
The colour adaptationbox (Fig. 7, see colour complementarycolour), the white light with the
plate) is used to demonstratea stimuluscontrast negligible trace of blue (squares) leaves yellow

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
 Colouras Sensationin VisualArt and in Science 97

Fig. 8. Untitled,kinetic artwork,a series of boxes in which

colours are produced by additive mixture. Perspex screens
(15 x 45 cm), electroniccontrolsystem,1978.
when the blue is subtracted.The yellow colour
occurs as the result of misappliedcompensation
by the visual process; it does not cope with
surfacereflectionas a separatephenomenon,one
that is seldom of consequence to humans in
nature but frequently encountered in viewing
Fig. 9. Kineticdisplaypanel. Movingdiscs of colourshaving
objects made by humans. Similar explanations the appearanceof surfacecoloursareproducedon translucent
can be applied to many colour illusions experi- screens.Perspexscreen,draughtingfilm, electriclights,electric
enced in paintingand sculpture. motor, 120x 120cm. 1969.
Figures8 and 9 illustratemy interestin kinetic
art, a likely mediumfor the applicationof colour identicalto that of a surfacecolour.The turquoise
science. Equipment is shown for the display of field consists of a sheet of transparentturquoise
kinetic colour compositions that can be written Perspex backed by plywood painted white. The
down with the use of a notation or recordedon ambientlight reflectedfrom the turquoisefield is
tape. Figure 8 shows eight (there can be more) of high enoughluminanceto producea targetand
colour mixing boxes in which red, green, blue- surroundconditiondiscussedabove.
green and blue light are mixed and the result I have used the above artworksand devicesfor
displayed.The colour is controlledin each box by demonstrationpurposesin a courseon sciencefor
an electronicsystem that can be operatedmanu- artistsat the GloucestershireCollege of Art and
ally whilstcomposing.The coloursof a total of 36 Design. I shall be pleased to supplyto interested
channels can be recorded simultaneouslyon a personsmore detailedinformationon theirdesign
channel of an audio cassette player. Each dim- and operation. I wish to call attention to other
ming circuit is controlled by a single audio fre- devices used in teaching colour to art students
quency that is separated from the mixture of discussedin an earlierarticlein Leonardo[32].
frequencieson the tape by an electronic filter. I
am indebted to Robin Pyewell for technical
assistancewith the electronics. Recently I have REFERENCES
realized that a more powerful method of pro-
1. C. P. Snow, The Two Culturesand a Second Look
grammingis availableusingmicroprocessors,with (Cambridge:CambridgeUniv. Press, 1964).
which I am now experimenting. 2. W. D. Wright, The Rays Are Not Coloured(Bristol,
Figure 9 shows an earlier kinetic displaypanel .ngland:Adam Hilger, 1967)Lecture2.
that is viewed in moderate daylight. Black lines 3. P. Auger, Contemporary Anthropocentrism:On Science
made on a long band of draughtingfilm that is andTraditionalCultures,Leonardo13, 223 (1980).
drawnby a rollerpast light-sensitiveswitchesthat 4. J. Gage, Colourin Turner(London:StudioVista, 1969).
5. F. Birren,A Historyof Colourin Painting(New York:
control the illuminationof the panel. The panel Reinhold,1965).
displaysa 10 x 10 arrayof smalltranslucentcircles 6. C. E. Gauss, The AestheticTheoriesof FrenchArtists
in a turquoisefield. Although the colours of the (Baltimore:JohnsHopkinsUniv. Press, 1949).
circles (green, blue or blue-green) are produced 7. J. S. Mill, A Systemof Logic (London:Longman,1843)
Bk. I, Ch. 3.
by projected light (a green and a blue electric 8. D. Jameson and L. M. Hurvich, From Contrast to
light is enclosed in a tube below each circle), the Assimilation:In Art and in the Eye, Leonardo8, 125
appearance of the colour produced is almost (1975).

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
98 JonathanC. Fish
9. G. A. Agoston, Color Theoryand Its Applicationin Art the Brain,Leonardo9, 105 (1976).
and Design (New York:Springer-Verlag,1979). 24. C. J. Bartleson,Brown, Color:Research& Application
10. D. Katz, TheWorldof Colour(trans.by R. B. Macleod 1, 181 (1976).
andC. W. Fox) (London:KeganPaul,Trench,Trubner, 25. M. R. Pointer,Colourfulness:A New Concept,in Color
1935). 77, Proc. ThirdCong. Int. ColourAssoc. (AIC), F. W.
11. J. Beck, SurfaceColourPerception(Ithaca,NY: Cornell Billmeyer,Jr. and G. Wyszecki,eds. (Bristol,England:
Univ. Press, 1972). AdamHilger, 1978).
12. E. Hering,Outlinesof a Theoryof theLightSense(trans. 26. InternationalLighting Vocabulary.PublicationNo. 17
by L. M. Hurvichand D. Jameson)(Cambridge,MA: (Paris:CommissionInternationalede l'Eclairage.1970).
HarvardUniv. Press, 1964). 27. R. M. Boynton, W. Schofer and M. E. Neum, Hue-
13. H. von Helmholtz, Treatiseon Physiological Optics Wavelength Relation Measured by Color-Naming
(1866), J. P. C. Southall,ed. (New York:Dover, 1962). Method for Three Retinal Locations,Science 146, 666
14. J. Flam,Matisseon Art (Oxford:Phaidon,1973). (1964).
15. N. Gabo, Of Divers Arts (A. W. Mellon Lectures) 28. A. Hard, QualitativeAttributesof ColorPerception,in
(London:FaberandFaber, 1962). Color 69, Proc. First Cong. Int. ColourAssoc. (AIC),
16. H. Thomas,Applicationof the OstwaldColorSystemin (Gottingen:Musterschmidt,1970).
My Painting,Leonardo13, 11 (1980) 29. D. B. JuddandD. Nickerson,RelationbetweenMunsell
17. D. Nickerson,Optical Society of America (OSA) Uni- and SwedishNaturalColor SystemScales,J. Opt. Soc.
formColorScale Samples,Leonardo12, 206 (1979). Am. 65, 85 (1975).
18. R. W. G. Hunt, The Specificationof Colour Appear- 30. A. HardandL. Sivik,Distinctnessof BorderlineRelated
ance. I. Concepts and Terms, Color: Research and to the NaturalColor System, in Color 77, Proc. Third
Application2, 55 (1977). Cong. Int. ColourAssoc. (AIC), F. W. Billmeyer,Jr.
19. R. M. BoyntonandJ. Gordon,Bezold-BriickeHue Shift and G. Wyszecki,eds; (Bristol,England:Adam Hilger,
Measured by Color-NamingTechnique, J. Opt. Soc. 1978).
Am. 55, 654 (1965). 31. C. Blackmore,Mechanicsof Mind (B.B.C. Reith Lec-
20. H. Wallach, BrightnessConstancyand the Nature of tures, 1976)(Cambridge:CambridgeUniv. Press, 1977)
AchromaticColours,J. Exp. Psychol.38, 310 (1948). Lecture4.
21. R. M. Evans, ThePerceptionof Color(New York:John 32. M. L. Meixner.Instructionon Lightand Colorin Art at
Wiley, 1974). the Iowa State University,Leonardo9, 52 (1976).
22. A. A. Mills, Intensity-Modulated Illuminationof Paint- 33. A. Hard, The NaturalColourSystemand Its Universal
ings andan Unexpected3-DimensionalEffect, Leonardo Applicationin the Study of EnvironmentalDesign, in
11, 213 (1978). Colourfor Architecture,T. Porter and B. Mikellides
23. F. Birren,ColorPerceptionin Art: Beyondthe Eye into (London:StudioVista, 1976)p. 112.

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions
 in

....
....
i:

Top left: Jonathan Fish. Colour adaptation box, perspex, wood, 30 x 20 x 20 cm, 1970. (Fig. 7, cf. page 96)
Top right: Pavel Kuznetsov. 'Kirghizian Woman', oil on canvas, 85 x 59 cm, 1919. (Fig. 3, cf. page 141)
Center: Paul Re. III-9: 'Front and Back', acrylic paint on Masonite 34.5 x 60.5 cm, 1975. (Photo: K. Cornyn,
San Francisco, CA, U.S.A.) (Fig 7, cf. page 107)
Bottom left: Roberto Donnini. 'Immagina in Progress', collage, paper, wood (mahogany), 65 x 135 cm, 1974 (unfinished).
(Fig. 2, cf. page 122)
Bottom right: Tariffe Raslain. Untitled, acrylic on canvas. 114 x 162 cm, 1979. (Fig. 5, cf. page 132)

This content downloaded from 188.72.96.55 on Thu, 12 Jun 2014 13:09:09 PM

All use subject to JSTOR Terms and Conditions