Dispelling Myths about Color

by Rich Pasco

This is my personal narrative of the learning process which brought me to my current understanding of how human beings perceive, discuss, display, and print color.

“All the Colors of the Rainbow”

continuous spectrum Isaac Newton (1642-1726) is generally credited with discovering that a prism can separate white light into a spectrum of its component colors. I see a spectrum as a continuous spread of color, but Newton (and everybody since) tried to assign them distinct names. Just how many names Newton chose is open to argument, but one source says that he chose six(6):
Our modern understanding of light and color begins with Isaac Newton (1642-1726) and a series of experiments that he published in 1672. He is the first to understand the rainbow – he refracts white light with a prism, resolving it into its component colors: red, orange, yellow, green, blue and violet.
The count of six (6) is also stated here, albeit with different colors:
At the age of 23, Isaac Newton reinvestigated this same dispersion of white sunlight into a rainbow of colors.... Newton organized his findings in a color wheel showing the three “primary colors” — red, green, and blue — separated by the three “secondary colors” — yellow, cyan, and magenta.

The “Three Primary Colors” Myth

3 crayons When I was a kid, my elementary school teacher drilled into us that “There are three primary colors; Red, Yellow, and Blue.” To prove it, she gave us sets of three crayons and told us that we could draw any color with them. For example, to make green, we could use Yellow and Blue. I swallowed that nonsense hook, line and sinker, until years later when as a high-school student, I studied color television technology.

RGB I learned that color television sets display their images by lighting up clusters of tiny dots of three primary colors: Red, Green and Blue. I wondered what happened to the Yellow my elementary school teacher had called “primary,” and quickly learned that color TVs display yellow as a combination of red and green. Maybe yellow wasn't quite so primary after all! My TV work convinced me that my elementary school teacher had lied, probably not deliberately but she had passed on false folklore.

cartridge Then I got a personal computer and eventually a color printer. Its color ink cartridge also had three primary colors of ink: Yellow, Cyan, and Magenta. Now I was really confused! Will the real three primary colors please stand up?

Delving deeper, I eventually learned that the spectrum is really a continuum of color, and that any division into names is arbitrary. However, the human eye has three kinds of cones—the cells on the retina that are sensitive to color—and that our perception of color is due to the relative stimulation of the cones most sensitive to red, green, and blue light. Thus, TV set makers and printers can fool the eye into perceiving any color by stimulating the three kinds of cones in the proper ratios.

additive color
Additive Color

subtractive color
Subtractive Color
Further, I learned, that because the color dots on a TV screen are beside one another, each adds its color to the mix of light going to the eye. For example, if the screen lights up a red dot and an adjacent green dot, from a distance a human eye will perceive a yellow dot.

Conversely, a printer lays its inks on top of one another onto white paper which by default reflects all colors of light. Each ink subtracts (absorbs) one of the primary colors of light going through it. For example, yellow ink absorbs the blue light, allowing only the red and green to pass through, stimulating their respective cones and creating the illusion we call “yellow.” For another example: laying yellow ink (which absorbs blue light) on top of cyan ink (which absorbs red) will allow only the green light to pass, stimulating only the green cones and creating the illusion we call “green.”

So the three primary additive colors of light are Red, Green, and Blue, while the three primary subtractive colors of ink are Yellow, Cyan, and Magenta. Notice that neither of these triads is the set my elementary school teacher preached; she was simply wrong.

The “Color Temperature” Myth

hob Another myth that I came across was that any color can be described by a single number, called its color temperature. Color temperature is defined as the temperature (usually in degrees Kelvin, °K) at which a heated ideal black body would glow (incandesce) at the stated color. I grew up with a wood stove which would glow a deep red when there was roaring fire inside, an electric range whose burners glowed orange when hot, and of course incandescent light bulbs that heated a filament to glow with a yellowish kind of white. And I witnessed arc welders whose work glowed brilliant bluish white, so the concept was plausible: a single number took me from deep red through orange and yellow to white and eventually blue. Brilliant!

The myth was reinforced by light bulb manufacturers who label their bulbs with the color temperature they deliver, and by film makers who labeled the film with the color temperature for which the film was designed. In theory, at least, using a film color temperature matching the light bulb would ensure that white objects in the scene appeared white in the photos.

fluorescent green This worked until I tried taking photos under fluorescent light tubes, most of which produce more green light (in proportion to their red and blue) than a black body at any temperature. No matter what color temperature of film I chose, my pictures always had a greenish cast! I realized had been betrayed by the “color temperature” myth.

I remembered what I had learned in college linear algebra and my work with television: Three numbers define a point in a three-dimensional space, and so it takes three numbers to address any point in that space, or to span the space. With light and color, if one number defines its brightness, or luminance, then it takes two more to define its color, or chroma. Thus it takes two numbers to span the two-dimensional space of color (color plane), and there is no way a single number like color temperature can do it. The best a single number can do is to locate a point along a particular linear (or curved) path on the color plane. For more on this, see the discussion of Planckian Locus in the last paragraph under The CIE Chromaticity Chart below.

Power Spectral Density

power spectral density
Power spectral densities of blackbodies at various temperatures
I was finally able to sort it all out in electrical engineering graduate school at Stanford, where I learned the concept of power spectral density. Let me try to explain.

Any electromagnetic wave, whether radio, light or X-rays, is characterized by its frequency (in cycles per second) or its wavelength (in meters per cycle). These vary inversely; their product is always the speed of light in meters per second.

Monochromatic light (light of a single color) comprises light waves all having the same frequency (and hence the same wavelength). In the real world, few light sources are actually monochromatic. The deep red of a laser beam and the pure yellow of a low-pressure sodium vapor street light are a few examples.

Most visible light contains a mix of light of all frequencies in some proportion. For any light source, one could draw a graph showing how much power there is at each frequency along the entire spectrum, ranging from infrared at the low frequency (long wavelength) end up to ultraviolet at the high-frequency (short wavelength) end. This graph shows a mathematical function of frequency (or wavelength) called power spectral density.

From a purely physical standpoint, it takes a power spectral density curve to fully characterize any light source. This is much more information than a single word (such as “red,” “green,” or “blue”) or even just three numbers stating how much red, green and blue.

The CIE Chromaticity Chart

CIE chart Fortunately, we seldom need to examine the complete power spectral density of a light source to “know” its color. For most practical purposes, it is sufficient to know the relative degrees to which it stimulates the red, green, and blue cones in the eye of a normal (not color-blind) human being. The mathematically inclined will note that each of these three numbers may be obtained by integrating, with respect to wavelength, the product of the power spectral density of the light source (as a function of wavelength) times the sensitivity of the respective cone of the human eye (also as a function of wavelength).

Thus, once we know the total brightness of a light source, it takes only two more numbers—rather than its detailed power spectral density—to specify the color as perceived by a human observer. In 1931 the International Commission on Illumination (CIE) defined a system for mapping all perceptual colors onto a plane defined by two numbers (arbitrarily called x and y).

The outer curve shows the monochromatic colors (comprising a single wavelength). The blue numbers designate their wavelength in nanometers. Everything in the interior is a mix of colors.

The black curve running through the middle of the chart shown here is called the Planckian Locus. It is the set of colors which can be defined by various values of the single number known as “color temperature.”


About the author

Contact the author