Since even before the discovery of spectra, scientists had tried to find ways to categorize stars. By observing spectra, astronomers realized that large numbers of stars exhibit a small number of distinct patterns in their spectral lines. Classification by spectral features quickly proved to be a powerful tool for understanding stars.
The current spectral classification scheme was developed at Harvard Observatory in the early 20th century. Work was begun by Henry Draper who photographed the first spectrum of Vega in 1872. After his death, his wife donated the equipment and a sum of money to the Observatory to continue his work. The bulk of the classification work was done by Annie Jump Cannon from 1918 to 1924. The original scheme used capital letters running alphabetically, but subsequent revisions have reduced this as stellar evolution and typing has become better understood. The work was published in the Henry Draper Catalogue (HD) and Henry Draper Extension (HDE) which contained spectra of 225,000 stars down to ninth magnitude.
The scheme is based on lines which are mainly sensitive to stellar surface temperatures rather than actual compositional differences, gravity, or luminosity. Important lines are the hydrogen Balmer lines, lines of neutral and singly ionized helium, iron lines, the H and K doublet of ionized calcium at 396.8 and 393.3 nm, the G band due to the CH molecule, the 422.7 nm neutral calcium line, several metal lines around 431 nm, and the lines of titanium oxide.
Ordered from highest temperature to lowest, the seven main stellar types are O, B, A, F, G, K, and M. Astronomers use one of several mnemonics to remember the order of the classification scheme. O, B, and A type stars are often referred to as early spectral types, while cool stars (G, K, and M) are known as late type stars. The nomenclature is rooted in long-obsolete ideas about stellar evolution, but the terminology remains. The spectral characteristics of these types are summarized below:
|Type||Color||Approximate Surface Temperature||Main Characteristics||Examples|
|O||Blue||> 25,000 K||Singly ionized helium lines either in emission or absorption. Strong ultraviolet continuum.||10 Lacertra|
|B||Blue||11,000 - 25,000||Neutral helium lines in absorption.||Rigel
|A||Blue||7,500 - 11,000||Hydrogen lines at maximum strength for A0 stars, decreasing thereafter.||Sirius
|F||Blue to White||6,000 - 7,500||Metallic lines become noticeable.||Canopus
|G||White to Yellow||5,000 - 6,000||Solar-type spectra. Absorption lines of neutral metallic atoms and ions (e.g. once-ionized calcium) grow in strength.||Sun
|K||Orange to Red||3,500 - 5,000||Metallic lines dominate. Weak blue continuum.||Arcturus
|M||Red||< 3,500||Molecular bands of titanium oxide noticeable.||Betelgeuse
The Yerkes scheme uses six luminosity classes:
|Ia||Most luminous supergiants|
|Ib||Less luminous supergiants|
|V||Main sequence stars (dwarfs)|
Thus the Sun would be more fully specified as a G2V type star.
|Some Spectral Peculiarity Codes|
|comp||Composite spectrum; two spectral types are blended, indicating that the star is an unresolved binary.|
|e||Emission lines are present (usually hydrogen).|
|m||Abnormally strong "metals" (elements other than hydrogen and helium) for a star of a given spectral type; usually applied to A stars.|
|n||Broad ("nebulous") absorption lines due to fast rotation.|
|nn||Very broad lines due to very fast rotation.|
|neb||A nebula's spectrum is mixed with the star's.|
|p||Unspecified peculiarity, except when used with type A, where it denotes abnormally strong lines of "metals" (related to Am stars).|
|s||Very narrow ("sharp") lines.|
|sh||Shell star (B to F main sequence star with emission lines from a shell of gas).|
|var||Varying spectral type.|
|wl||Weak lines (suggesting an ancient, "metal"-poor star)|
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.
Stars in the asymptotic giant branch are short-lived. The degenerate core of the star is more massive that it was in the single-shell burning phase, and due to the peculiar nature of degenerate matter, the more massive core is physically smaller. The gravity experienced by overlying layers is hence stronger, requiring higher luminosities to maintain the balance between pressure and gravity. Thus the star expends energy at a very high rate and may well become a red supergiant. Stars in this phase of stellar evolution have proven to be challenging to model. One problem is that the helium shell burning is not stable. The layer of helium fusion is thin. Slight positive perturbations in the nuclear energy generate extra pressure and the region is enlarged slightly. But because the layer is thin, the change in height is slight and hence the change in pressure on the hotter region is changed very little. The higher temperature will likely increase the rate of nuclear reactions (many reactions processes are very temperature sensitive, such as the triple-alpha process which will most likely be dominate in the helium shell). Thus local reaction rates will pick up, generating more heat before it can diffuse. Thus large runaway reaction spots can start from small local condition changes. The runaway is only checked after considerable expansion and the creation of a convective cycle to carry away the excess energy. Yet even once the runaway is checked and the layer resettles, the same underlying physical problem remains. There is no real stable helium shell burning mode. Thus the star will experience spasms of energy generation with convective cells which may carry material all the way up to the hydrogen burning shell, followed by longer periods of relaxation back to the thin shell.
If the convective cells created during this helium fusion runaways reach all the way to the hydrogen fusion layer, this would potentially provide a mechanism for material deep within the star to be dredged up to its surface. This may nicely explain several stellar types which seem analogous with K and M stars temperature-wise, but show some other spectral features as if their outer atmospheres had been enriched with heavier element. These types are the R, N, and S types.
Instead of (or in addition to) the usual lines of titanium, scandium, and vanadium oxides characteristic of M type giants, S type stars show heavier elements such as zirconium, yttrium, and barium. A significant fraction of all S type stars are variable.