Double Star Astronomy

Part 4: Double Star Color

A characteristic and problematic aspect of visual double star astronomy is the fascination with star color. Two stars, narrowly separated, of contrasting magnitude and spectral type — this is often the essential description of a "showcase pair."

Unfortunately, this enjoyable aspect of astronomy has been by idiosyncratic color language and a poor understanding of how star color arises and how it is perceived. These three issues must be understood by the astronomer who wants to describe star color reliably.

Color Perception Is Idiosyncratic

The fundamental issue that emerges when two astronomers compare their color impressions is that the color terms used to describe star color do not agree. This is a difficulty recognized by William Herschel at the very beginning of double star astronomy:

Here I must remark, that different eyes may perhaps differ a little in their estimations [of star colors]. I have, for instance, found, that the little star which is near α Herculis, by some to whom I have shewn it has been called green, and by others blue. Nor will this appear extraordinary when we recollect that there are blues and greens which are very often, particularly by candle-light, mistaken for each other. (Preface to Catalogue of Double Stars, 1782)

Even astronomers who are aware of this problem often do not appreciate the extent to which star colors will vary with the observer. To illustrate this variation, we can turn to the testimony of one of the most ardent star color advocates, Sissy Haas, who helpfully quotes historical star color descriptions in her Double Stars for Small Telescopes (Sky Publishing, 2008). Here is a sample:

double starObserver 1Observer 2Observer 3

36/37 Herpale blue, blueyellowish white, yellowish peach
ρ Herazure white, azure whitebluish white, pale emeraldgreenish white, greenish
μ Herbrilliant orange, ghostly wisppale white, purple
μ Herboth pure goldlight apple green, cherry red
ν Draboth brilliant whiteboth pale grayboth yellow white
ψ Draboth lemony whiteboth pearly whitewhitish yellow, lilac
STF 2248 Drawhite, blueyellowish, ash
STF 2440 Dragreenish white, bluepale yellow, blue
ζ Lyrboth goldish whitetopaz, greenishgreenish white, yellow
36 Ophboth citrus orangeruddy, pale yellowboth golden yellow
61 Ophboth straw yellowboth silvery white
η Oristraw yellow, silvery yellowwhite, purplish
ε Booamber yellow, deep bluepale orange, sea green
39 Booboth whitish goldboth whitewhite, lilac
ξ Boobright white, vivid grayorange, purpleyellow, deep orange
44 Booboth grapefruit orangepale white, lucid gray
ψ Drawhite, blueyellowish, ash

These examples come from observers such as T. Webb, Smyth, Olcott and others, and can be multiplied to fill a page.

These examples illustrate two problems: a naive imprecision in the use of color names (such as vivid gray, yellowish peach or silvery green); and a difficulty in recognizing star colors accurately, for example in the confusion between yellow and orange, white and lilac, blue and ash, or gold and cherry red. We can begin to clarify this muddle by looking at the mechanism that creates star color.

The Blackbody Limits of Star Color

Temperature is kinetic energy in atoms — motion in a gas, vibration in a solid — but when we say that an object is at a certain temperature, we do not mean that every atom in the object has exactly the same kinetic energy. Instead there is a distribution in the energy of the individual atoms that creates an average temperature.

A simple way to evaluate the temperature of an object is to record the photons (quantum units of energy) that the object emits into its less energetic (lower temperature) surroundings. These photons can be very long wavelength, low energy radio waves, moderate energy heat, high energy light, or very short wavelength, very high energy ultraviolet and xrays. This distribution in the photons emitted by a body at a specific temperature is described in quantum theory as an idealized blackbody curve or blackbody flux profile.

This curve has the same basic shape across all temperatures: skewed so that the distribution extends farther from the peak into low energy than into high energy states. But as temperature increases the emitted flux changes in two ways: (1) the overall flux, including the peak of the curve, shifts from lower into higher energy photons, and (2) the total quantity of flux, expressed as the peak radiance in watts, increases in proportion (diagram, below).

The peak radiances of the curves in this diagram have been set to the same height, to show the differences in the shape of the curves. If the curves for 24000K and 1500K were drawn together at their actual peak values, the curve for the 1500K body would be a flat line across the bottom of the graph.

The peak radiance of this generic curve, and therefore its overall shape, is explicitly related to the temperature of the blackbody by Wien's displacement law:

λmax = 2.898 x 10–3/T[K] meters

which shows that as temperature (the denominator) increases, the peak wavelength is displaced to shorter wavelengths (higher energies). The amount of energy emitted depends on the size of the object, but for comparison, at a temperature of 1500K (Kelvins) the peak flux of a standard object is at a wavelength of around 1930 nm with a radiance of about 30 thousand watts, but at a temperature of 24,000K the peak flux is around 120 nm with a radiance of around 30 billion watts. The invariant rule is, as the temperature of a body increases, its peak emittance is at shorter wavelengths and its total emittance is brighter.

An important detail is that the median of the blackbody curve is at a lower energy than the peak, which means more than half the emitted photons are at energies below the peak value. Infrared (heat) is always powerfully present in radiating bodies.

This blackbody curve describes the energy radiated by any body whose radiation arises from its temperature, which includes the radiation emitted by the photosphere temperature of stars. For example, the Sun, at a surface temperature of 5780K, has its peak flux at about 2.898/5780 x 10–3 = 0.000000501 meters or 501 nm (blue green).

The change in the peak emittance of a blackbody radiator is evaluated within the narrow range of our light sensitivity, from about 400 nm to 700 nm, indicated in the diagram by the band of spectral hues. At a temperature of 3000K the blackbody curve slopes strongly upward into the red wavelengths, indicating a reddish color. At a temperature of 12,000K the curve slopes upward into the blue wavelengths, indicating a blue color.

We can calculate the visual color associated with a blackbody by means of standard sensitivity curves for the human cones. These calculations reduce the flux profile to a specific hue for every color temperature, illustrated in the table (below).

color temperature
for common illuminants and light sources
 color mnemonic correlated illuminant or light source
1000S9 lower limit of blackbody curve
1850S5 candle flame
2000S3 sunlight at sunrise/sunset (clear sky)
2750M8 60W incandescent tungsten light bulb
2860M7  CIE A: 120W incandescent light bulb
3400M2 photoflood or reflector flood lamp
3500M2 direct sunlight one hour after sunrise
4100K7 CIE F11: triband fluorescent light
4300K6 morning or afternoon direct sunlight
5000K2 white flame carbon arc lamp (CIE D50: warm daylight illuminant)
5400G8 noon summer sunlight
6400F6 xenon arc lamp
6500F5 average summer daylight (CIE D65: cool daylight illuminant)
7100F1 light summer shade
7500A8 indirect northern skylight
8000A6 deep summer shade
9300A1 white point of a CRT (television screen)
10640B9 clear blue sky
30,000B0 mountain noon sky
Sources: Mitchell Charity, MIT; Kodak USA

Note: These color samples grossly exaggerate the chromatic contrast and drastically reduce relative luminance due to the gamut limits of digital displays.

The important point illustrated by this table is that the range of blackbody colors is actually quite limited: there are no yellow greens, greens, blue greens, or green blues; no blue violets, violets, magentas, bluish reds or reds. The color range is from red orange through yellow to white, then into pale blue.

In fact, three other factors limit the color range. The first is that, at very high blackbody temperatures, a very large change in the temperature produces a very small change in the apparent color. This is in contrast to the hues that appear at low temperatures, which produce a clear hue transition from red through yellow to white across small increases in temperature.

This limitation arises from the shape of the blackbody curve within the wavelength limits of our visual sensitivity. The eye judges color as the relative proportional luminance of the flux profile between about 400 to 700 nm. Within this range, the proportional difference in luminance between the peak and minimum luminance of a blackbody profile is very different across blackbody temperatures. At high temperatures, the upward slope in blue wavelengths changes relatively little, and these small differences are difficult for our eyes to perceive. This can be illustrated by plotting the radiance of the flux profiles on a logarithmic scale as a proportion of a constant peak radiance, which approximates the overall response of the eye to radiance differences (diagram, below).

In fact, the eye is not equally sensitive to all wavelengths within its range, but gives the most weight to wavelengths in the central 500 to 620 nm. Even so, the diagram illustrates that there is a visually large difference between the flux profiles at 1500K and 3000K, which will produce a readily visible color difference. In contrast, the flux profile at 6000K is essentially flat, and despite its peak in blue green wavelengths, will appear as a pure white. And the difference between the flux profiles at 12000K and 24000K will be visually subtle because the slopes of the profiles are nearly parallel.

The second factor limiting our perception of the blackbody color range is that our eyes accept a slightly bluish light, equivalent to the daylight mixture of sunlight and skylight, as a pure white at all luminance levels. As a result, our eyes are predisposed to perceive faint and bluish starlight as indistinguishable from faint white light.

The third factor limiting the color range in stars is the filtering effect of the Earth's atmosphere and of the lens and macular pigment in the human eye. As shown in the diagram above, the atmospheric "windows" that permit photons to pass do not extend below around 300 nm, the peak radiance of very high temperature bodies. The eye lens also yellows with age, and this effectively filters much of the "blue" light that enters the eye.

All three factors combine to produce a significant and easily perceived shift in blackbody hue with star temperatures below about 6000K, in contrast to bodies that appear similarly white or bluish white at temperatures above that.

Despite the precision and convenience we obtain by equating star color with blackbody color, the actual circumstances are more complicated. Interstellar dust absorbs starlight primarily in the very short wavelengths, making stars appear redder and dimmer than they actually are. Stellar spectra are selectively filtered by the star's own photosphere, which produces absorption bands that selectively darken wavelengths within the blackbody flux. Stars can also be surrounded by clouds of dust or gas that alter the spectrum through absorption and excitation emittance.

While it might seem that color is also affected by relative motion — bodies rapidly receding from us would appear more red, and bodies approaching us would appear more blue — it turns out that this Doppler effect is perceptible only in the extremely high velocity differences between galaxies, or in the slow moving wavelengths of sound.

Thus, the blackbody spectrum is not necessarily an accurate representation of star color in every situation. But it is a useful way to understand the physical range of star colors, including the hues we can exclude — yellow green, green, blue green, violet, magenta and red — as having a physical origin in a star's flux profile, and the relative color sensitivity of our eye to different blackbody temperatures.

Color Perception

The principal difficulty in the observation of star color is the complex behavior of color perception. Our visual system interprets color into the color appearance of stars based on the star's apparent brightness (its contrast with a dark background), its contrast with the brightness and color of nearby stars, and the existing chromatic adaptation or "white point" of the eye. This interpretative behavior of color vision has its own patterns and rules, which astronomers need to understand in order to report star color.

Understanding is helped by a consistent and unambiguous set of terms for color attributes. The color of all uniform lights can be described using just three terms: brightness for the apparent luminance or radiance of the light; hue for the specific "color name" of the light (red, yellow, blue, etc.); and saturation for the hue purity of the light. Brightness varies from bright to dim, and saturation varies from intense [hue] to white.

Chromatic Adaptation. Astronomers can experience the importance of chromatic adaptation by observing a distinctly colored K or M star, and a "white" G or F star under three conditions: (1) normal dark adaptation, (2) after staring for several seconds at a book page illuminated by a strong red light, and (3) staring at a page illuminated with a white light. The red star will appear whitish or strongly red under conditions (2) and (3), while the white star will appear faintly green under (2).

In addition, the eye tends to adapt to any light so that its color shifts toward white, reducing perceived color intensity. For this reason, star colors will appear most vivid on first glance, and the color will become less prominent across prolonged observation.

Saturation. The hue intensity or saturation in lights is perceptually reduced by an increased proportion of "white" light in the color. All star colors are low in saturation, meaning that they appear strongly diluted with white light — the hues appear as tints. In addition, stars with an effective temperature above ~6000K appear much less saturated (closer to white) than lights below ~3000K. This means that chromatic adaptation has a much larger effect on star color than it would on a light of "pure" color, and the effect is stronger on the apparent color of type O, B and A stars.

Saturation is dependent on brightness: as lights become dimmer, their saturation can appear to increase. Saturation is also dependent on the visual width of a color area, with larger color areas appearing more saturated. For these reasons star color is often more visible or striking if the stars are viewed slightly out of focus, so that the star image appears as a disk of reduced brightness.

Hue. Unlike brightness or saturation, which vary as a linear dimension of intensity or strength from a zero point, hue is described by a hue circle in which one hue blends imperceptibly into another. Spectral hues wrap around the circle in their natural order from red to violet, and the circle is closed by joining red and violet through their mixture, which creates a red violet or magenta hue (diagram, below).

This circle is created by two basic dimensions of chromatic sensitivity in the eye: yellow/blue and red/green, although "red" here means a violet red or crimson, and "green" is actually a cyan or turquoise. Chromatic adaptation, the adjustment of the eye's white point, is achieved by changes in the relative sensitivity of these two dimensions, and all hues are produced by their combination.

The hue of star color can be described as a dominant wavelength, which is the single of wavelength of spectral light whose hue, adjusted for brightness and saturation, exactly matches the apparent hue of the total blackbody emittance. The diagram shows the hues associated with specific spectral wavelengths (in nanometers, as indicated in italic type), and the corresponding blackbody temperatures. The colored circles provide hue samples of the different colors, but note that these are all much more saturated than the corresponding star colors, especially across the blue blackbody series.

All red hues in a blackbody spectrum appear to have some yellow content — the most extreme (coolest) color is a red orange — and at the effective temperature of M type stars (~3000K) the hue is actually an orange yellow. In the same way, none of the blackbody "blue" colors come close to purple, and these hues are all on the "green" side of the hue circle. As illustrated in the previous section, effective temperatures of around 6000K, though technically green hues, are so flat across the visible range that they will appear indistinguishable from white.

Chromatic contrast. A major consequence of the opponent dimension organization of hues is that hues located on opposite sides of the hue circle will appear to have enhanced saturation, and an intensely chromatic light will cause a white light next to it to appear tinted with the hue opposite on the hue circle. Thus an orange yellow star at a dominant wavelength of around 600 nm will shift the hue of a white or faint companion toward a blue hue at around 470 nm, and enhance the apparent saturation of an already blue companion in this direction.

Alternately, stars that are of similar or analogous hues can appear more contrasted. Thus, a pair of stars in which one star is orange and the other star is yellow orange can appear more contrasted, so that the pair appears red orange and yellow.

Both types of effect are called chromatic contrast, and they can significantly alter the apparent color of paired stars.

It is important to understand that chromatic contrast can create the appearance of color where none exists. The classic 18th century demonstration of this effect is quite pretty, and easy to view if you have never seen it. Illuminate a white wall or large sheet of white paper with a "white" halogen light and the light of a candle or yellow light, and adjust their positions so that they cast separate shadows of equal darkness. Then place an object such as your hand or an upright book so that it casts a shadow from both light sources onto the wall. The shadow cast by the white light will appear colorless or slightly yellow, while the shadow cast by the candle will appear to be a beautiful dark blue. A similar arrangement using a white and red light will produce a luminous green shadow (image, right).

The same effects can appear in double stars in which one star is colored but the other star is very pale, white or dim: a red star will push its fainter companion star toward a green hue, and a yellow star will push its companion toward blue, and this effect is more pronounced when the companion star is significantly fainter than the primary.

Star color is extremely bright, although far away, so the proportion of chromaticity to brightness is very small. As a result most stars (excepting carbon stars and late type giant stars) have a very attenuated color, so that the blackbody colors appear as a delicate tint; and as already indicated, this is accentuated by the small visual extent or "pinpoint" appearance of the Airy disk.

The diagram below summarizes the most important visual attributes of star color:

(1) All star hues, with the exception of cool and strongly filtered emittance from carbon stars (type C), are weakly saturated: the color appears mixed with a significant amount of white light.

(2) The "blue series" of star colors appear much less differentiated than the "red series" colors, largely because the saturation of the blue hues is lower.

(3) Hue differences are reduced when the color area is very small, as illustrated by the upper row of dots below the square color swatches.

(4) Changing the brightness of a light can produce a shift in the hue. The blue series of hues will appear to shift toward white (apparent gray) as their brightness declines; the extremely hot blue hues may appear to shift toward blue violet; the yellow to red hues will appear to shift toward red. These brightness related hue shifts are illustrated in the lower row of dots.

The brightness effects on hue are more significant on the red side of the black body series. Thus, Betelgeuse (alp Ori) and Herschel's Garnet Star (mu Cep) have the same spectral type (M2 Ia), but because Betelgeuse is much brighter (visual magnitude 0.4 vs. 4.1) its color appears orange yellow while the fainter mu Cep appears distinctly orange red.

Because chromatic contrast and brightness changes enhance the saturation of opposing hues and increase the apparent difference between similar hues, project contrast hues onto faint or white stars, and shift hue toward the spectrum extremes, the appearance of star color is very sensitive to context and therefore fundamentally deceptive.

Viewing and Describing Star Color

When we combine the interpretive behavior of color percerception with the deceptive effects of visual context and the wide variation in color perception that appears across "normal" color observers, the result is that star color is a highly idiosyncratic and unstable aspect of visual astronomy.

This is certainly no reason to ignore the gorgeous visual impact of several famous close binaries, and it is equally no reason not to report apparent star color. It is sufficient reason to be wary of one's own color perceptions, to use methods that can make color perception more accurate, and to describe color in language that is consistent and unambiguous.

Improving perceptual accuracy. A few simple methods can be used to make star color perception more accurate. These serve primarily to counteract the effects of brightness, chromatic adaptation and chromatic contrast on color appearance.

• Use a consistent optical presentation – Because aperture, type of objective and magnification can all significantly affect apparent color, it is important to use the same type of telescope and the same eyepiece when surveying star colors. The eyepiece should deliver a magnification for the aperture sufficient to make the Airy disk readily visible.

• Be aware of chromatic adaptation – Accurate color perception requires a consistent adaptation to white. A simple way to retain chromatic adaptation is to glace at and then away from the star, each time in a different direction, so that a different part of the retina exposed.

However a bright, colored star will shift this adaptation, as will prior exposure to red light from observatory lights or flashlights. In situations where chromatic adaptation has been distorted in these ways, it is acceptable to observe a bright, white (type late F or early G) star, or a book page illuminated with a dim white light, or the moon with a high power eyepiece or with the naked eye, as a way to consistently reset the white point of the eye. Some of these techniques may impair dark adaptation, but dark adaptation is not especially useful to observe star color.

• Emphasize first glance color – Chromatic adaptation will cause star color to change across an observing interval, especially if one or both stars are bright or very bright. It is therefore most important to capture the color of stars at first glance, especially if they are bright or very bright. Note also how color appears to change across an observing interval. Then rest or reset the eye, and observe again.

• Observe defocused star images – Especially with bright or very bright stars, the defocused image will produce a colored disk that is larger and fainter than the focused star image. This will somewhat reduce the effects of chromatic adaptation (because the light is fainter) and improve color perception (because the area is larger).

• Consider the effects of chromatic contrast – Refer to the color circle (above) to evaluate the possible effects of chromatic contrast.

• Hide the primary star – If possible, hide the light of the primary star by moving it past the edge of the eyepiece field of view to assess the color of a fainter secondary star. This eliminates chromatic contrast and the suppressing effect of glare from the brighter star.

Reducing language ambiguity. A major source of the unreliability in star color descriptions is in the color language. Not only are imprecise or exaggerated color descriptions difficult for other people to interpret, they encourage the observer to make inaccurate visual identifications.

Sissy Haas prefers a vocabulary that seems inspired by the farmer's market and commercial paint store — she reports star colors as grapefruit orange, straw yellow, gloss white, amber yellow, peach white, pearly white, citrus orange, yellowish peach, sun yellow, silvery blue, bright orange, bluish turquoise, tangerine orange, whitish powder blue, azure white, silvery sapphire, dusty gold, pumpkin orange, lemony white, ash white, aquamarine, plum red, deep white, vivid gray, banana yellow and so on. What for example is the difference between a citrus orange, bright orange, tangerine orange and a grapefruit orange? And how closely does a pumpkin orange star actually match the daylight color of a pumpkin orange pumpkin? The problem with this kind of verbal clutter is that it is impossible for another person to identify the color behind the poetic exaggeration.

In contrast, William Herschel lists as his standard color terms "garnet, red, pale red, pale rose-colour, white inclining to red, white, white inclining to blue, blueish white, blue, greenish, green, dusky," and these terms, excepting the greens, match very well the blackbody series. And his approach exemplifies the first principle of star color description: keep color description simple and consistent.

The simplest method of color description, used for all scientific and professional color description, relies on the three colormaking attributes. In describing lights (stars), these are the brightness of the color as a source of light; the hue or "color" of the color; and the saturation of the color, or its mixture with white.

• Brightness is characterized by the contrast between the luminance adjectives bright and faint.

For consistency, avoid using any other terms such as brilliant, dim, shining, glaring, etc. Instead, use the terms in a standard visual scale that does not exceed five or six steps:

Very bright
Very faint

Why not just report the star's magnitude? After all, stars are seen against a common dark background and at approximately the same level of light adaptation, so relative brightness will correlate closely with a star's visual magnitude. The complication is that the same star will appear brighter in a larger aperture telescope, so the perceived brightness is also aperture dependent.

The solution is to create a perceptual brightness scale based on visual appearance, using visual magnitude to define intervals on the scale. For example, a very bright star will be any star that presents diffraction spikes from a secondary spider, or displays more than a specific number of diffraction rings when viewed at high power; a very faint star will be just above the visual limit of foveal detection (that is, the star is faint but is clearly visible when looked at directly). Note that color perception will fail for stars that are very faint, but the category is necessary to describe faint companions.

Once these two perceptual limits are established in a given instrument, the intermediate brightness steps can be determined by dividing the visual magnitude between the extremes into equal intervals. Thus, if observations indicate that diffraction spikes appear at magnitude 1.5, and foveal detection fails at magnitude 9.5, then:

Very bright Max. to 1.5
Very faint ? to 9.5

and four perceptual intervals must be divided across a magnitude range of (9.5 – 1.5) = 8.0, or intervals of 8.0/4 = 2 magnitudes. This gives:

Very bright – Max. to 1.5
Bright – 1.5 to 3.5
Average – 3.5 to 5.5
Faint – 5.5 to 7.5
Very faint – 7.5 to 9.5

Once the scale is determined in this way, stars of known magnitude can be observed to determine the visual appearance characteristic of stars within each perceptual category. Then the rating categories can be applied no matter what aperture of telescope is used, because the perceptual differences will appear in all situations.

One complication is that a very bright primary will make a nearby companion star appear fainter than it would appear in isolation — in part by direct contrast, and in part by brightening the background through glare or diffusion. But the scale can still be used to record this visual fact.

The remaining two attributes, hue and saturation, are actually two aspects of the single attribute of chromaticity or color content. One cannot appear without the other; hue becomes more ambiguous as saturation goes down, and both are affected by differences in brightness.

• Hue is the "color" of the color, and here it is important to use the standard hue terms red, orange, yellow, green, blue, violet, and to denote hue mixtures by placing the secondary or tinting hue first: yellow orange is an orange mixed with a smaller proportion of yellow, orange yellow is a yellow mixed with a smaller proportion of orange. Alternately, a hue combination can denote a color that appears midway between the two anchor hues: then yellow orange and orange yellow mean the same thing, and denote a hue that is equal proportions of yellow and orange. Obviously, whichever approach you choose, you should use it exclusively.

• Saturation is the concentration or purity of the hue, characterized by the adjectives intense to pale. Again the anchor terms can be expanded into a scale:

Very intense
Very pale

where very intense can mean the most intense color visible in a star (e.g., the color of a carbon star), and very pale can mean a color that contains so much white that a distinct hue is barely perceptible. (Obviously the term bright should not be used to describe intense saturation ["bright orange"], as this will confuse saturation and brightness.

Finally, the three attributes should always be listed in the order (1) brightness, (2) saturation and (3) hue (or hue mixture).

If the star does not appear to have a recognizable and stable color, then applying a specific hue description disguises this perceptual fact. Brightness will always be easy to describe, but when saturation is low then hue can become elusive. In these situations the description bright [or dim] and ambiguous is accurate. Don't reach for silvery green, gloss white or whitish powder blue instead.

The result is a color language that is easy to learn and remember, reliable to use and simple to interpret:

Very bright, pale yellow orange
Faint, ambiguous
Bright, medium orange
Faint, very pale blue green
Average, medium yellow

The difficulties in perceiving faint or very pale chromaticity lead to the second principle of star color description: the more that a star color appears subtle, elusive or difficult to describe, the more likely that it is illusory. This is especially true when the star color appears green or violet, colors outside the blackbody range. In these situations it may be useful to note any context features that might affect the color perception, for example a bright primary of a contrasting hue.


Last revised 11/26/13 • ©2013 Bruce MacEvoy