What is yellow

What is yellow?

On a fine summer day you may have the opportunity to enjoy and contemplate the beauty of the variegated flowers on a meadow. As you look closely at one of the flowers the question suggests itself:

Why is this buttercup yellow?

This seems to be a perfectly reasonable question to ask concerning the nature of colour. It can be answered in many different ways, however, according to what you are after - i.e. what kind of answer you are prepared to accept as a satisfactory explanation. Whether it is a physical, a chemical, a biological, a psychological or a philosophical elucidation you would like to get. If the one you ask is a physicist, you may get the answer:

"It looks yellow because the petals reflect the yellow wavelengths of light and absorb the rest".

This explanation conforms with the colour theory which Isaak Newton presented in his famous Optics (1704) where he proposed that a coloured object gets its colour because it has "a Disposition to reflect this or that sort of Rays more copiously than the rest".

Strictly speaking the rays are not coloured, he said, but each sort of ray has the power to stir up a sensation of this or that colour in the observers sensorium. Thus a yellow surface looks yellow because it reflects the "yellow-making" sort of light more copiously than the rest.

On closer scrutiny this answer to our question is found untenable and even misleading in certain respects. Let me show you why.

THE SPECTRAL REFLECTANCE OF A YELLOW SURFACE

Let us study the reflectance property of a yellow surface by the method Newton himself invented.

I project this dia-slide onto a white screen and by help of a glass prism (in front of the projector's lens) turn the narrow slit into a colourful spectrum.

The white light in this case comes -- not from the sun -- but from a halogen 100 watt bulb, and is essentially thermal radiation.

Scrutinizing this spectrum you see that the yellow part of it is quite narrow, which mean that if the colour of a yellow object were the result of absorption of all light, except this narrow region, then yellow would be a very dark colour; dark brown, maybe, but definitely not yellow, which is by its very nature a bright colour.

Next, I mount a yellow patch on a white screen, that can be moved horisontally. I project the light spectrum onto the screen and move it so as to make the spectrum fall onto the yellow patch.

As you see, it reflects the red and green with almost full strength but darkens the blue-violet part of the spectrum – in other words, the short wavelengths of the light falling onto it.

So the correct explanation of the yellow colour of the buttercup seems rather to be: Yellow results when the short wavelengths in the illumination are absorbed, and all the rest reflected.

Suppose you could measure the degree of darkening at the various parts of the spectrum and then plot the resulting values in a diagram, it would look like this:

This diagram represents the so called reflectance of the yellow sample. So, this is what the surface colour "yellow" is like, from the point of view of physics. In the same way you can measure the reflectance of surfaces of other colours (and I will return to this.) Likewise, you can measure the so called transmittance of a transparent sheet of coloured glass or plastic.

But what is the correlation between such reflectance and transmittances curves and the colour the material is seen to have? This is a question that needs to be looked deeper into and I will return to it in a while.

PERCEPTUAL CONSEQUENCES

Our observation of the optical nature of yellow has some interesting consequences. One is that yellow is a bright colour and much like white. The essential difference concerns the blue-violet, dark part of the spectrum. Hence, yellow should be more distinct in daylight (as in any cool, blueish light) than in a warm light. We can simulate such illumination with the help of an orange transparency.

The difference between the yellow and white areas almost disappears (look at the borders). This is because the white areas get a yellow tint in this light, but more important in practice: the yellow areas shift towards white, due to chromatic adaptation. In many situations yellow light can take on the rule of white, i.e. colourless, illumination.

This optical affinity of white and yellow is the reason why yellow is the brightest among hues. Goethe, in his Farbenlehre, speaks about yellow as the colour most akin to white:

In its highest purity it always carries with it the nature of brightness, and has a serene, gay, softly exciting character. (Die Farbenlehre §766)

So, whereas it is true that yellow merges out of white … it is still a bit uneasy on a white ground …. On grey it gets its real colourfulness, being without competition. T o speak of "dark yellow" is a contradiction in terms – it looses its yellow quality, turning into olive green, as seen on this scale of succesively darker yellow stripes.

In fact, pure yellow is so dependent on brightness that even a moderate shadowing of a yellow surface may give it a greenish tint.

Colour and wavelength

Before concluding this session, let me take us one step further towards a proper definition of optical colour. If you think about it, the eye couldn't be expected to make a detailed spectral analysis of the detected light at each spot of the retinal image. (Actually, measuring a spectral distribution is not a single measurement, but implies a series of measurements, one at a small interval around each wavelength.) Instead it registers some genereal feature of the distribution. Call it the "colour valency". Infinitely many different spectral distributions can have the same valency, that is, be equivalent as stimuli to the eye.

This equivalence has been surveyed by the method of "colour matching" -- pioneering work done by James Clerk Maxwell around 1860 -- with the conclusion that the gamut of colour valencies is essentially three-dimensional. This is the so called "trichromacy" of normal human colour vision.

Here is an example of how different equivalent spectral distributions can be:

Diagrams to the right: (1) the colour of a lemon-peel in daylight, (2) an imitation of the same colour by a narrow peak at suitable wavelength together with some uniform background, and (3) how the spectrum would look if the lemon were illuminated with light from a three-band fluorescent lamp.

In the following I will use a computer program, displaying the relation between spectral distribution and colour, based on the trichromatic equivalencies, conventionally settled for a "standard observer" and valid for 2 degree visual extension of the sample.

In the big window you recognize the diagram representing the reflectance of a yellow sample, here plotted against a wavelength scale, with thirty steps from 400 to 700 nanometers. (The colours of the bins are for illustrative purpose only.)

The square shows the corresponding coloured light, as produced on the computer screen by help of its RGB system of pixels. (To be perfectly correct this presumes a calibration of the screen/program interface, but let us accept standard data, if other data are not available.)

Looking at this presentation you may argue: All right, a broad distribution my be typical of yellow, but the yellow wavelengths are still there, and determine the yellow of the petals, whereas the other wavelengths, on the sides, contribute to its brightness.

Not even this is a necessary condition for yellow. Consider the following.

We start from the typical yellow spectral dsitribution, and take away the yellow wavelengths, at 570 and 580 nm, transferring them to another distribution, shown at the right, building up a light, consisting of only yellow wavelengths (in addition to a small amount of white, which is necessary to make it possible to show the resulting colour on a computer screen.) Eventually we arrive at a colour matching the colour on the left side. Let be that we will have to increase its intensity to make the two identical:

Anyway, here we have: Two light spots with identical colour, one of them with a spectrum dominated by wavelengths at 570 and 580 nm; the other one missing exactly these wavelengths.

Here is one more example (compare the lemon peel, above) :

The reflected light from a yellow sample is chromatically equivalent with a mixture of three spectral lights at 460nm, 540 nm and 620 nm.

The general lesson we learn is: The colour of a light flux is not dependent on the absence or presence of any particular wavelength.

You can get the program and play around with it - visit: visualspd.htm

IN CONCLUSION

It is true that a yellow hue is seen in a narrow region of a spectrum, at a position corresponding to wavelengths around 580 nm. And it is true that a light flux strongly dominated by photons of this wavelength looks yellow to a dark-adapted eye. But exactly the same yellow colour is seen when the eye is stimulated with light of such a composition that there are no photons of wavelength at and around 580 nm. The point is that the sensation of yellow is not dependent on the presence of some kind of "yellowmaking" photons, but is equally dependent on the simultaneous absence of certain other sorts of photons. It is first and foremost the fact that photons of short wavelengths are missing in the light, impinging upon the eye, which signals "yellow". The answer to our initial question would thus be: "The buttercup looks yellow in ordinary daylight because it is a property of its petals to absorb photons of wavelengths below 500 nm with much greater probability than above. "

Evidently, very different spectral compositions can represent exactly the same colour to the eye, under given conditions. What is common to all such equivalent spectral compositions?

We have noted that it is the spectral composition as a whole that counts -- not specific wavelength components. Wavelengths that are absent in the distribution is as important as the ones that are present. It is -- in some way -- the general shape of the distribution that determines its "colour valency".