On leaving school I trained as a dyer, and although I left practical dyeing after a few years I remained in one or another aspect of colour technology all my working life.
It has always been a touch frustrating when obliged to discuss colour with someone not familiar with the technology. For some strange psychological reason no-one wants to believe that their understanding of colour is less than complete. If the conversation starts to get beyond them they take the attitude that 'colour is as you see it' and no other person has a better understanding of it than they themselves. The fact that they have never learned anything about the subject is brushed aside.

Here is an explanation of the errors in the most popular understanding of colour, following the reasoning by which I came to doubt what I had been taught at school.

You can read it below or download as a pdf:
I retain copyright but do not place any restriction on your personal private use of it.

The second part of this discourse is not yet complete.

Perception of colour: 1


Most people utterly misunderstand the objective aspects of colour. Even those who are expert in some aspect of colour reproduction are likely to have a largely erroneous understanding of it, being based on some well-tested convention which can be successfully applied to his own technology. The problem seems to be a matter of inheritance within the education system. The practically acceptable conventions of the relationship between light and our perception of colour, together with "scientific" explanations which do not stand up to a close examination of the observed properties, are passed on from one generation to the next at academic and professional levels. Pure theory of colour is not practised as an academic subject and in most technologies gives way to workable systems. The most unfortunate outcome of this is that theory of colour perception as taught at basic school level is quite wrong and misleads people for the rest of their lives.

At an early age we learn of the division of colours into "primary" and "secondary" and accept this convention as a fundamental truth. Later we learn that red, blue and yellow function as primaries only where the mixing of pigments is concerned and this is called "subtractive" mixing; whereas the mixing of coloured lights is "additive" and involves a different set of primaries. It might be expected, on a moment's reflection, that this partition would give rise to a suspicion that that the concept of primary and secondary colours is not, after all, a fundamental property – but this is not taken up in the educational curriculum. Also we encounter the spectrum and are allowed to believe that colour is intrinsically related to light wavelength; but the relationship between wavelength and the colour circle which is taken to represent colour perception is hazily explained, if at all.

These conventions and assumptions which we accumulate during early education are totally false; but although it is a fairly simple matter to demonstrate their falsity, full acceptance of the principles of such a demonstration can only be achieved if the observer is able to break down the inhibitions which have been formed in earlier life and this is not easy.

It is due to this difficulty that this discourse must be lengthy and pedantic. If it is said that a mixture of red and green paints produce a shade of yellow, just as you may have observed a mixture of red and green lights to do, you will not believe it: but your rejection of the statement is a matter of belief arising out of indoctrination and not of logic. So the ground must be cleared before the statement can be viewed dispassionately. You might take the view that " colour is as you see it", but this confuses the objective aspects of colour vision with the truly subjective aspects of colour appreciation. Furthermore, that which you see is rarely the same as that which you think you see. Most people need a lot of guided practice before they are able to allow the eye to instruct the mind rather than the mind to obstruct the eye. This very statement will probably put the reader in a defensive mood and unreceptive to new ideas.


Let us begin by examining some familiar views of the popular understanding of light and colour. White light can be separated by refraction to give a display of colours known as the spectrum. Everyone has experienced this at least in the form of a rainbow. This can be explained in terms of wavelength, white light being a mixture of all wavelengths and refraction differentiating them, the progression from long to short wavelengths being revealed as a range of colours. This leads to the supposition that each colour is uniquely associated with a particular wavelength. In addition it can be demonstrated that the linear progression of wavelengths continues beyond each end of this visible range, through infrared at one end and ultraviolet at the other. Therefore light may be defined as the visually perceptible portion of a continuous range of energy frequencies.

colour circle

Accepting this simple though scientific assessment of one of the physical properties of light can bring difficulties to the understanding of the nature of colour. Many of its properties are often taken to be subjective but where the majority of people agree as to the observed details of a given phenomenon it may reasonably be assumed to have been objectively assessed: and it turns out that almost every such aspect of colour with respect to its divisions and mixing properties are at variance with our initial understanding of the spectral analysis of light by refraction.

colour circle The "Colour Circle" is a popular representation of the way in which we visually react to the distribution of and relationship between colours . It presents a paradox when compared with the spectrum since it is not linear and has no discontinuity between the red and blue/violet which occur at the extreme ends of the visible spectrum. In the circle, red merges to blue through magenta and other shades of purple. Furthermore, for every colour a "complementary" is defined which is diametrically opposite on the circumference. Any such pair when mixed together gives a neutral grey which is not represented in the spectrum.

The logic of the colour circle can not be applied to the linear form of the spectrum. If one assumes the apparent correlation between colour and wavelength to be fundamental, the result is nonsense. This is the cause of lack of understanding of light and colour even in some academic circles.

Colour pair mixtures

Consider the effect of mixing two wavelengths in terms of the colours which they apparently represent. Take a red pigment and mix it in turn with pigments of other colours (i.e. supposedly wavelengths) successively farther removed from it along the spectrum:

  • Red + orange            > scarlet
  • Red + yellow            > orange (but not the brightest shade)
  • Red + yellow-green  > khaki
  • Red + blue-green     > grey (neutral)
  • Red + cyan             > dull purple
  • Red + blue              > brighter purple
  • and if we take red with a touch of blue, the result is bright magenta
colour circle

From the first two mixtures of this series it appears that the result must be a colour which represents a wavelength lying between the two constituents; but as the pair becomes more separated, as red and green, the resultant becomes less identifiable. When red is mixed with blue-green or cyan a neutral grey occurs. In other words there is no tendency to represent any intermediate wavelength. In fact, since the term “neutral” includes the range from black to white through all shades of grey, then this mixture apparently bears some relation to "white light" which is also neutral – but white light is commonly described as a mixture of all wavelengths, which clearly can not occur when only two colours/wavelengths have been added together.

One might reasonably expect that after passing this "neutral zone"where mixed with cyan, a mixture of two even more separated components – red and blue – might continue to lose its identity and become, at the far end of the spectrum, black. But surprisingly this does not happen. Instead, another colour emerges - purple, which we associate with violet situated at the extreme short-wave end of the spectrum. Different proportions of red and blue give rise to a range of purple shades, forcing the viewer away from the spectrum and back to the colour circle

Adding to the confusion, it appears at first sight that a different result is obtained when coloured lights are mixed instead of paints/pigments. Ignore for now the popular convention of "additive/subtractive" colour mixing and consider that when two beams of light of different colours overlap on a screen, the amount of light – regardless of its colour - reflected from the overlap must equal the sum of the two components. The amount of light reflected from a mixture of paints, however, can only equal the average of its components. Now consider the above range of colour pair mixtures where red was paired with other colours, this time viewing the result of overlapping beams of coloured light on a white screen instead of mixing pigments:

  • Red + red            > a brighter shade of red
  • Red + yellow            > orange
  • Red + yellow-green  > yellow
  • Red + blue-green     > pale yellow (neutral)
  • Red + cyan             > white
  • Red + blue              > light purple/magenta
(Please note that it is not possible to correctly display in print
the appearance of spotlights shining onto a screen)
colour circle

Notwithstanding the differing degrees of intensity, observe the similarities between the two systems for each pair of "wavelengths": as the complementary pair of red + blue-green is approached, the identity of the hue of the mixture decreases though in the second series it is as though the resultant hue were mixed with white, which maintains the illusion of a bright colour, and when the complementary pair is reached the result is pure white. White is neutral, i.e. no hue can be identified, as is the case with the mixture of pigments where the neutral tone produced is grey. Therefore in terms of hue, the two systems are identical.

The wrong explanation

One common explanation of the "averaging" effect of colour mixing in terms of wavelengths goes as follows:
If one mixes, say, a blue pigment with a yellow one, the resulting effect is green. This is because the blue pigment reflects a little of the wavelengths on either side, i.e. green and violet, and the yellow reflects a little green and orange. The mixture absorbs all wavelengths which are not common to both the constituent pigments and reflects only that which is common, i.e. green.

This is a nonsensical explanation for several reasons. In a mixture of particulate pigments where the different coloured particles are lying side by side it is to be expected that each reflects its own characteristic wavelengths and there is no reason why either should absorb reflections from the other. In considering this suggested mechanism with regard to mixtures of other spectrum colour pairs, it leads to the conclusion that a mixture of red and blue pigments could result in nothing but black since the colours are so widely separated that they could have no common wavelength and hence each would absorb the whole of the other.

It offers no explanation of the way in which complementary hues act in mixture. Generally, there is no reason to suppose that any chosen mixed pair of wavelengths should act differently to any other pair or that there should be some periodic division of wavelengths giving rise to "primary" and "secondary" colours. Neither would one expect that a mixture of wavelengths could transform to a single average wavelength – and spectrographic analysis proves that it does not. A mixture remains a mixture, though the eye can never perceive anything other than a single colour.

Consider how our ear reacts to a mixture of sound frequencies. We know we are hearing a mixture, in fact a trained person can define the frequencies involved in a complex mixture or "chord". The eye can not differentiate. It is only by means of physical experiments that we learn that mixtures exist and further that a single perceived colour can arise from a number of different mixtures of wavelengths.

A conclusion:

Since there is clearly such a division between the logic of the physical analysis of light (the continuous band of wavelengths) and our perception of it (the colour circle) we must conclude that two fundamentally dissimilar systems exist, each conforming to its own set of rules.

The spectrum obtained by refraction, in so far as that it presents an analysis of a band of wave frequencies, illustrates a physical property of light: but the colour which is associated with it is more objectively described by the colour circle and is purely a product of the brain, stimulated by signals received from the eye which itself is stimulated by light energy frequencies.

So we conclude that light is a physical phenomenon and colour a physiological one. Colour is in no sense whatever a physical property of light. Colour does not exist in the material world, only inside our heads.

The second part of this discourse examines the above principles in greater detail and as a result looks at possible mechanisms by means of which our eyes might translate the range of light wavelengths into the colour circle.
Sorry, it's even longer.
(Under preparation.)