The modern spectral classification system is so successful that it has hardly been changed since 1943. It is based on just two physical properties that imprint themselves on the spectrum of a star's light: the star's temperature and atmospheric pressure. These reveal an abundance of information that paints the star's portrait and tells its life story.

The temperature sets the star's color and tells its surface brightness, how much light it emits from each square meter of its surface. The pressure depends on the star's surface gravity and therefore, roughly, on its size -- telling whether it is a giant, dwarf, or something in between. The size and surface brightness in turn yield the star's luminosity (its total light output, or absolute magnitude) and often its evolutionary status as young, middle-aged, or nearing death. The luminosity also gives a good idea of the star's distance. Appended to the basic spectral type may be letters for chemical peculiarities, an extended atmosphere, unusual surface activity, fast rotation, or other special characteristics.

Every starwatcher needs to have a feel for spectral types. Here are the most important things to know.

Dissecting Starlight

The tale begins in 1802, when the English experimenter William Wollaston passed a beam of sunlight through a thin slit and then through a prism. The slit provided a sharp, high-resolution view of the familiar rainbow spectrum, with no colors overlapping each other. When seen this way, Wollaston noticed, the Sun's spectrum was marked by many narrow, black lines of various intensities. These dark lines stayed at exactly the same places in the colorful band from day to day and year to year. They were later measured and cataloged by Josef von Fraunhofer, for whom they are still called "Fraunhofer lines."

Similar spectral lines showed up in laboratory experiments. Using a slit and prism, physicists discovered that when a solid, a liquid, or a dense gas is heated to glow, it emits a smooth spectrum of light with no lines: a continuum. A rarefied hot gas, on the other hand, glows only in certain colors, or wavelengths: bright, narrow emission lines instead of a rainbow band. If a cooler sample of the same gas is placed in front of a glowing object, dark absorption lines appear at the wavelengths where the emission lines would be if the gas were hot.

By 1859 the situation was clear: we see the Sun's hot surface through a cooler solar atmosphere that imposes the dark lines.

Every element, every chemical compound, shows its own set of spectral lines. They are as unique as fingerprints. They reveal not only which atoms and molecules are present but also many other physical conditions, starting with temperature. Here, scientists realized, was a way to bring the Sun down into the laboratory. When they put slit-and-prism devices (spectroscopes) on telescopes, they could even see spectral lines in the light of stars.

It was the 19th century's greatest astronomical breakthrough. Philosophers had cited the makeup of stars as something beyond all possible human knowing. Now finding the composition of the Sun and stars was just a matter of comparing spectral lines seen in a telescope to those in a laboratory. This wasn't always simple, but it gave birth to modern astrophysics -- the treatment of stars as physical objects to be studied and understood, rather than as mere points of light on the sky to be measured.

Spectral Classes

The first great classifier of stellar spectra was Angelo Secchi in Rome. In the 1860s he examined the spectra of hundreds of stars visually in a telescope and classed them into five main types, mostly named for bright examples. Sirian stars, for instance, showed spectra like Sirius's: dominated by absorption lines of hydrogen atoms.

Today's classification scheme was born at Harvard College Observatory. Starting in 1886 under Edward C. Pickering, the observatory staff photographed and classified thousands of stellar spectra. They assigned them letters from A through Q, generally in alphabetical order from the simplest-looking to the most complex. But soon a more natural system became clear. By rearranging and merging classifications, Antonia C. Maury and Annie J. Cannon found they could fit nearly all stars' spectra into one smooth, continuous sequence. The sequence matched the stars' color temperatures, from the hottest, blue-white stars to cool, orange-red ones.

But it was too late to reassign the letters. When the dust cleared, the rearranged sequence ran O B A F G K M from hot to cool. Spectral types on the blue end were called "early" and those on the red end "late." These terms are still used today, though the incorrect idea of stellar evolution they embody -- that stars simply cool with age -- has been obsolete for generations.

The sequence could be cut even more finely. Cannon subdivided each letter with numbers from 0 to 9, so that a spectrum whose appearance placed it halfway between standard G0 and K0 stars was called G5. Using this scheme, Cannon led the classification at Harvard of 325,300 spectra recorded on wide-field photographs. The resulting Henry Draper Catalogue (HD) and Henry Draper Extension (HDE), published beginning in 1918, remain standard references today.

The time-honored mnemonic for remembering the spectral sequence, invented by Henry Norris Russell when astronomy's leadership was all male, is "Oh Be A Fine Girl Kiss Me." In 1995 Mercury magazine published a student's rejoinder: "Only Boys Accepting Feminism Get Kissed Meaningfully." Take your pick.

In the late 1990s two more types were added to the spectral sequence: L and T. Both indicate cool brown dwarfs, substellar objects not massive enough to sustain nuclear fusion in their cores. L and T stars are best observed at infrared wavelengths. In fact, substars of type T cannot be detected visually at all.

A few other spectral types don't fit the sequence but instead parallel it. Type W or Wolf-Rayet stars are as hot and blue as the hottest O stars but show strong emission lines, either of nitrogen (WN), carbon and oxygen (WC), or neither (WR). Emission lines indicate an especially large, thick shroud of hot gas surrounding a star.

Some giant stars at the cool end of the spectrum have an excess of carbon. These were originally called R and N but have been merged to form type C. "Carbon stars" can often be spotted at a glance in a telescope by their deep red color. A bright example in the autumn sky is 19 Piscium (TX Piscium) in the Circlet of Pisces, spectral type C5. Their distinctive absorption bands (masses of overlapping spectral lines) due to the carbon compounds C₂, CN, and CH darken or "blanket" the blue end of the spectrum. In other words, a carbon star's atmosphere is a red filter. When seen in emission instead of absorption, these same spectral bands glow blue; the same compounds that redden a carbon star give comets their blue-green tint.

The rare type-S stars are also red giants. They parallel type M but show strong bands of zirconium oxide and lanthanum oxide instead of an M star's titanium oxide. We can imagine that planets of S stars, bathed in chemically peculiar stellar winds, might be encrusted with gems of cubic zirconia.

Luminosity Classes

Even in stars of the same spectral type, the absorption lines don't always look alike. In some stars the lines are narrow and sharp; in others they are broadened by various effects. Chief among these is atmospheric pressure, which also changes the intensity ratios of certain pressure-sensitive lines.

Astronomers quickly realized that atmospheric pressure tells a star's surface gravity and therefore suggests its size. Narrow lines indicate an immense, bloated star with a weakly compressed atmosphere far from its center of gravity. In the Henry Draper Catalogue, spectral types were prefixed with d for dwarf, sg for subgiant, g for giant, and c for supergiant.

You'll still run across these letters from time to time, but beginning in 1941 they were replaced by a more detailed scheme first published by William W. Morgan and Philip C. Keenan. With only minor changes, this "MK" system of spectral classification remains the standard today. Stars are assigned to luminosity classes by Roman numerals: I for supergiants (often subdivided into classes Ia-0, Ia, Iab, and Ib in order of decreasing luminosity), II for bright giants, III for normal giants, IV for subgiants, V for dwarfs on the main sequence, and occasionally VI for subdwarfs.

Thus a designation such as G2V, the Sun's spectral type, tells temperature and luminosity. When these are plotted against each other on a graph, the result is called a Hertzsprung-Russell or H-R diagram. This has been a fundamental astrophysical tool ever since it was invented around 1911. Most stars gather in certain narrow regions of the H-R diagram according to their masses and ages.

Stars arrive on the main sequence soon after they are born, and this is where they spend most of their lives. Massive stars blaze brightly on the hot, blue end of the main sequence. They burn up their nuclear fuel in only millions or tens of millions of years. Stars with lower masses comprise the yellow, orange, and red dwarfs on the lower-right part of the main sequence, where they remain for billions of years.

As a star begins to exhaust the hydrogen fuel in its core, it evolves away from the main sequence toward the upper right and becomes a red giant or supergiant. Stars that began with more than six times the Sun's mass then evolve left and right through complicated loops on the H-R diagram as if in a frenzy to keep up their energy production, then finally explode as supernovae. Less massive giants evolve to the left and then down to become white dwarfs; this is the track the Sun will trace through the H-R diagram 8 billion years from now (S&T: May 1994, page 12).

Odds and Ends

Spectra can reveal many other things about stars. Accordingly, lowercase letters are sometimes added to the end of a spectral type to indicate peculiarities. A list appears in the following table.

Symbols can be added for elements showing abnormally strong lines. For example, Epsilon Ursae Majoris in the Big Dipper is type A0p IV:(CrEu), indicating strong chromium and europium lines. The colon means uncertainty in the IV luminosity class.

Certain spectral subtleties are not widely known among amateurs. Some visual observers pride themselves on being able to nail a star's type to the nearest letter by its color in the eyepiece. Color is indeed a close indicator of spectral type for stars earlier (hotter) than about K5, assuming no interstellar reddening is present. But the relationship often breaks down among the later K and M stars. Compare the tint of Betelgeuse, type M2 Iab, to that of Aldebaran, K5 III. Most people can't see a difference. At the same time, two red giants of the identical type may show different tints; compare Mu and Eta Geminorum, both cataloged as type M3 III.

In addition, dwarf G, K, and M stars are not as red as giants and supergiants of the same types. The color difference is equivalent to about one-half to one letter class.

Lastly, differences between spectra are far greater than differences in the actual compositions of stars. An A star might seem to be almost pure hydrogen, while a K star shows only trace evidence of hydrogen in a spectrum packed with lines of "metals" (the astronomer's term for all elements other than hydrogen and helium). But A and K stars are made of the same stuff. Different atoms and ions merely display their spectral lines at different temperatures. Even carbon stars are made mostly of hydrogen and helium. The true "abundances" of elements can indeed be measured in a star. But it's a tough job of comparing precise line strengths in a high-quality spectrum with those predicted by atomic theory or measured in the lab.

For much of the 20th century, the study of visible-light spectra practically was astronomy. In recent decades the opening of nonvisible wavelengths and other exciting advances have distracted attention from this field. Nevertheless it remains the bedrock on which modern astronomy rests.

Alan MacRobert is an associate editor of Sky & Telescope magazine and an avid backyard astronomer.