During the 1860s and 1870s, pioneering stellar spectroscopist Father Angelo Secchi created the Secchi classes in order to classify observed spectra. By 1866, he had developed three classes of stellar spectra:
In 1868, he discovered carbon stars, which he put into a distinct group:
In 1877, he added a fifth class:
In the late 1890s, this classification began to be superseded by the Harvard classification, which is discussed in the remainder of this article.
|Class||Temperature||Conventional color||Apparent color|| Mass|
|Luminosity||Hydrogen lines||% of all Main Sequence Stars|
|O||30,000–60,000 K||blue||blue||64 M☉||16 R☉||1,400,000 L☉||Weak||~0.00003%|
|B||10,000–30,000 K||blue to blue white||blue white||18 M☉||7 R☉||20,000 L☉||Medium||0.13%|
|A||7,500–10,000 K||white||white||3.1 M☉||2.1 R☉||40 L☉||Strong||0.6%|
|F||6,000–7,500 K||yellowish white||white||1.7 M☉||1.4 R☉||6 L☉||Medium||3%|
|G||5,000–6,000 K||yellow||yellowish white||1.1 M☉||1.1 R☉||1.2 L☉||Weak||7.6%|
|K||3,500–5,000 K||orange||yellow orange||0.8 M☉||0.9 R☉||0.4 L☉||Very weak||12.1%|
|M||2,000–3,500 K||red||orange red||0.4 M☉||0.5 R☉||0.04 L☉||Very weak||76.45%|
The mass, radius, and luminosity listed for each class are appropriate only for stars on the main sequence portion of their lives and so are not appropriate for red giants. A popular mnemonic for remembering the order is "Oh Be A Fine Girl/Guy, Kiss Me" (there are many variants of this mnemonic). The spectral classes O through M are subdivided by Arabic numerals (0–9). For example, A0 denotes the hottest stars in the A class and A9 denotes the coolest ones. The Sun is classified as G2.
|I||A, B, C, D||Hydrogen lines dominant.|
|II||E, F, G, H, I, K, L|
|IV||N||Did not appear in the catalogue.|
|O||Wolf-Rayet spectra with bright lines.|
In 1897, another worker at Harvard, Antonia Maury, placed the Orion subtype of Secchi class I ahead of the remainder of Secchi class I, thus placing the modern type B ahead of the modern type A. She was the first to do so, although she did not use lettered spectral types, but rather a series of 22 types numbered from I to XXII. In 1901, Annie Jump Cannon returned to the lettered types, but dropped all letters except O, B, A, F, G, K, and M, used in that order, as well as P for planetary nebulae and Q for some peculiar spectra. She also used types such as B5A for stars halfway between types B and A, F2G for stars one-fifth of the way from F to G, and so forth. Finally, by 1912, Cannon had changed the types B, A, B5A, F2G, etc. to B0, A0, B5, F2, etc. This is essentially the modern form of the Harvard classification system.
The fact that the Harvard classification of a star indicated its surface temperature was not fully understood until after its development. In the 1920s, the Indian physicist Megh Nad Saha derived a theory of ionization by extending well-known ideas in physical chemistry pertaining to the dissociation of molecules to the ionization of atoms. First applied to the solar chromosphere, he then applied it to stellar spectra. The Harvard astronomer Cecilia Helena Payne (later to become Cecilia Payne-Gaposchkin) then demonstrated that the OBAFGKM spectral sequence is actually a sequence in temperature. Because the classification sequence predates our understanding that it is a temperature sequence, the placement of a spectrum into a given subtype, such as B3 or A7, depends upon (largely subjective) estimates of the strengths of absorption features in stellar spectra. As a result, these subtypes are not evenly divided into any sort of mathematically representable intervals.
O, B, and A stars are sometimes misleadingly called "early type", while K and M stars are said to be "late type". This stems from a early 20th century model of stellar evaluation in which stars were powered by gravitational contraction via the Kelvin–Helmholtz mechanism in which stars start their lives as very hot "early type" stars, and then gradually cool down, thereby evolving into "late type" stars. This mechanism provided ages of the sun that were much smaller than what is observed, and was rendered obsolete by the discovery that stars are powered by nuclear fusion. However, brown dwarfs, whose energy comes from gravitational attraction alone, cool as they age and so progress to later spectral types. The highest mass brown dwarfs start their lives with M-type spectra and will cool through the L, T, and Y spectral classes.
The Conventional color descriptions are traditional in astronomy, and represent colors relative to Vega, a star that is perceived as white under naked eye observational conditions, but which magnified appears as blue. The Apparent color descriptions is what the observer would see if trying to describe the stars under a dark sky without aid to the eye, or with binoculars. The table colors used, are D65 standard colors, which are what you would see if the star light would be magnified to be filling non-dazzlingly bright areas. Most stars in the sky, except the brightest ones, appear white or bluish white to the unaided eye because they are too dim for color vision to work.
Our Sun itself is white. It is sometimes called a yellow star (spectroscopically, relative to Vega), and may appear yellow or red (viewed through the atmosphere), or appear white (viewed when too bright for the eye to see any color). Astronomy images often use a variety of exaggerated colors (partially founded in faint light conditions observations, partially in conventions). But the Sun's own intrinsic color is white (aside from sunspots), with no trace of color, and closely approximates a black body of 5780 K (see color temperature). This is a natural consequence of the evolution of our optical senses: the response curve that maximizes the overall efficiency against solar illumination will by definition perceive the Sun as white. The sun is known as a G type star.
The Yerkes spectral classification, also called the MKK system from the authors' initials, is a system of stellar spectral classification introduced in 1943 by William Wilson Morgan, Phillip C. Keenan and Edith Kellman from Yerkes Observatory. This classification is based on spectral lines sensitive to stellar surface gravity which is related to luminosity, as opposed to the Harvard classification which is based on surface temperature. Later, in 1953, after some revisions of list of standard stars and classification criteria, the scheme was named MK (by William Wilson Morgan and Phillip C. Keenan initials).
Since the radius of a giant star is much larger than a dwarf star while their masses are roughly comparable, the gravity and thus the gas density and pressure on the surface of a giant star are much lower than for a dwarf. These differences manifest themselves in the form of luminosity effects which affect both the width and the intensity of spectral lines which can then be measured. Denser stars with higher surface gravity will exhibit greater pressure broadening of spectral lines.
A number of different luminosity classes are distinguished:
Marginal cases are allowed; for instance a star classified as Ia0-Ia would be a very luminous supergiant, verging on hypergiant. Examples are below. The spectral type of the star is not a factor.
|-||G2 I-II||The star is between super giant and bright giant.|
|+||O9.5 Ia+||The star is a hypergiant star.|
|/||M2 IV/V||The star is either a subgiant or a dwarf star.|
The following illustration represents star classes with the colors very close to those actually perceived by the human eye. The relative sizes are for main sequence or "dwarf" stars.
Class B stars are extremely luminous and blue. Their spectra have neutral helium, which are most prominent at the B2 subclass, and moderate hydrogen lines. Ionized metal lines include Mg II and Si II. As O and B stars are so powerful, they only live for a very short time, and thus they do not stray far from the area in which they were formed. These stars tend to cluster together in what are called OB associations, which are associated with giant molecular clouds. The Orion OB1 association occupies a large portion of a spiral arm of our galaxy and contains many of the brighter stars of the constellation Orion. They constitute about 1 in 800 main sequence stars in the solar neighborhood —rare, but much more common than those of class O.
Class G stars are probably the best known, if only for the reason that our Sun is of this class. Most notable are the H and K lines of Ca II, which are most prominent at G2. They have even weaker hydrogen lines than F, but along with the ionized metals, they have neutral metals. There is a prominent spike in the G band of CH molecules. G is host to the "Yellow Evolutionary Void". Supergiant stars often swing between O or B (blue) and K or M (red). While they do this, they do not stay for long in the G classification as this is an extremely unstable place for a supergiant to be. G stars represent about 1 in 13 of the main sequence stars in the solar neighborhood.
Class M is by far the most common class. About 76% of the main sequence stars in the solar neighborhood are red dwarfs (78.6% if we include all stars: see the note under Class O), such as Proxima Centauri. M is also host to most giants and some supergiants such as Antares and Betelgeuse, as well as Mira variables. The late-M group holds hotter brown dwarfs that are above the L spectrum. This is usually in the range of M6.5 to M9.5. The spectrum of an M star shows lines belonging to molecules and all neutral metals but hydrogen lines are usually absent. Titanium oxide can be strong in M stars, usually dominating by about M5. Vanadium oxide bands become present by late M.
A number of new spectral types have been taken into use from newly discovered types of stars.
Spectra of some very hot and bluish stars exhibit marked emission lines from carbon or nitrogen, or sometimes oxygen.
Class W or WR represents the superluminous Wolf-Rayet stars, notably unusual since they have mostly helium in their atmospheres instead of hydrogen. They are thought to be dying supergiants with their hydrogen layer blown away by hot stellar winds caused by their high temperatures, thereby directly exposing their hot helium shell. Class W is subdivided into subclasses WC (WCE early-type, WCL late-type), WN (WNE early-type, WNL late-type), and WO according to the dominance of carbon, nitrogen, or oxygen emission in their spectra (and outer layers).
Intermediary between the genuine Wolf-Rayets and ordinary hot stars of classes O and early B, there are OC, ON, BC and BN stars. They seem to constitute a short continuum from the Wolf-Rayets into the ordinary OBs.
In lists of spectra, the "spectrum OB" may occur. This is in fact not a spectrum, but a marker which means that "the spectrum of this star is unknown, but it belongs to an OB association, so probably either a class O or class B star, or perhaps a fairly hot class A star."
The novel spectral types L and T were created to classify infrared spectra of cool stars. This included both red dwarfs and brown dwarfs which are very faint in the visual spectrum. The hypothetical spectral type Y has been reserved for objects cooler than T dwarfs which have spectra that are qualitatively distinct from T dwarfs.
Class L dwarfs get their designation because they are cooler than M stars and L is the remaining letter alphabetically closest to M. L does not mean lithium dwarf; a large fraction of these stars do not have lithium in their spectra. Some of these objects have mass large enough to support hydrogen fusion, but some are of substellar mass and do not, so collectively these objects should be referred to as L dwarfs, not L stars. They are a very dark red in color and brightest in infrared. Their atmosphere is cool enough to allow metal hydrides and alkali metals to be prominent in their spectra. Due to low gravities in giant stars, TiO- and VO-bearing condensates never form. Thus, larger L-type stars can never form in an isolated environment. It may be possible for these L-type supergiants to form through stellar collisions, however, an example of which is V838 Monocerotis.
Class T and L could be more common than all the other classes combined if recent research is accurate. From studying the number of proplyds (protoplanetary discs, clumps of gas in nebulae from which stars and solar systems are formed) then the number of stars in the galaxy should be several orders of magnitude higher than what we know about. It is theorized that these proplyds are in a race with each other. The first one to form will become a proto-star, which are very violent objects and will disrupt other proplyds in the vicinity, stripping them of their gas. The victim proplyds will then probably go on to become main sequence stars or brown dwarf stars of the L and T classes, but quite invisible to us. Since they live so long, these smaller stars will accumulate over time.
Carbon related stars are stars whose spectra indicate production of carbon by helium triple-alpha fusion. With increased carbon abundance, and some parallel s-process heavy element production, the spectra of these stars are becoming increasingly deviant from the usual late spectral classes G, K and M. The giants among those stars are presumed to produce this carbon themselves, but not too few of this class of stars are believed to be double stars whose odd atmosphere once was transferred from a former carbon star companion that is now a white dwarf.
Originally classified as R and N stars, these are also known as 'carbon stars'. These are red giants, near the end of their lives, in which there is an excess of carbon in the atmosphere. The old R and N classes ran parallel to the normal classification system from roughly mid G to late M. These have more recently been remapped into a unified carbon classifier C, with N0 starting at roughly C6. Another subset of cool carbon stars are the J-type stars, which are characterized by the strong presence of molecules of 13CN in addition to those of 12CN. A few dwarf (that is, main sequence) carbon stars are known, but the overwhelming majority of known carbon stars are giants or supergiants.
Class S stars have zirconium oxide lines in addition to (or, rarely, instead of) those of titanium oxide, and are in between the Class M stars and the carbon stars. S stars have excess amounts of zirconium and other elements produced by the s-process, and have their carbon and oxygen abundances closer to equal than is the case for M stars. The latter condition results in both carbon and oxygen being locked up almost entirely in carbon monoxide molecules. For stars cool enough for carbon monoxide to form that molecule tends to "eat up" all of whichever element is less abundant, resulting in "leftover oxygen" (which becomes available to form titanium oxide) in stars of normal composition, "leftover carbon" (which becomes available to form the diatomic carbon molecules) in carbon stars, and "leftover nothing" in the S stars. The relation between these stars and the ordinary M stars indicates a continuum of carbon abundance. Like carbon stars, nearly all known S stars are giants or supergiants.
In between the M class and the S class, border cases are named MS stars. In a similar way border cases between the S class and the C-N class are named SC or CS. The sequence M → MS → S → SC → C-N is believed to be a sequence of increased carbon abundance with age for carbon stars in the asymptotic giant branch.
The class D is the modern classification used for white dwarfs, low-mass stars that are no longer undergoing nuclear fusion and have shrunk to planetary size, slowly cooling down. Class D is further divided into spectral types DA, DB, DC, DO, DQ, DX, and DZ. The letters are not related to the letters used in the classification of other stars, but instead indicate the composition of the white dwarf's visible outer layer or atmosphere.
The white dwarf types are as follows:
The type is followed by a number giving the white dwarf's surface temperature. This number is a rounded form of 50400/Teff, where Teff is the effective surface temperature, measured in kelvins. Originally, this number was rounded to one of the digits 1 through 9, but more recently fractional values have started to be used, as well as values below 1 and above 9.
Two or more of the type letters may be used to indicate a white dwarf which displays more than one of the spectral features above. Also, the letter V is used to indicate a variable white dwarf.
Extended white dwarf spectral types:
Variable star designations:
Additional nomenclature, in the form of lower-case letters, can follow the spectral type to indicate peculiar features of the spectrum.
|Code||Spectral peculiarities for stars|
|:||Blending and/or uncertain spectral value|
|…||Undescribed spectral peculiarities exist|
|e||Emission lines present|
|[e]||"Forbidden" emission lines present|
|er||"Reversed" center of emission lines weaker than edges|
|ep||Emission lines with peculiarity|
|eq||Emission lines with P Cygni profile|
|ev||Spectral emission that exhibits variability|
|f||NIII and HeII emission|
|f+||Si IV emission additional to HeII and NIII emission|
|f*||NIV emission stronger than NIII emission|
|(f)||Weak emission lines of He|
|((f))||Displays strong HeII absorption accompanied by weak NIII emissions|
|He wk||Weak He lines|
|k||Spectra with interstellar absorption features|
|m||Enhanced metal features|
|n||Broad ("nebulous") absorption due to spinning|
|nn||Very broad absorption features due to spinning very fast|
|neb||A nebula's spectrum mixed in|
|p||Unspecified peculiarity, peculiar star.|
|pq||Peculiar spectrum, similar to the spectra of novae|
|q||Red & blue shifts line present|
|s||Narrowly "sharp" absorption lines|
|ss||Very narrow lines|
|v||Variable spectral feature (also "var")|
|w||Weak lines (also "wl" & "wk")|
|d Del||Type A and F giants with weak calcium H and K lines, as in prototype Delta Delphini|
|d Sct||Type A and F stars with spectra similar to that of short-period variable Delta Scuti|
|Code||If spectrum shows enhanced metal features|
|Ba||Abnormally strong Barium|
|Ca||Abnormally strong Calcium|
|Cr||Abnormally strong Chromium|
|Eu||Abnormally strong Europium|
|He||Abnormally strong Helium|
|Hg||Abnormally strong Mercury|
|Mn||Abnormally strong Manganese|
|Si||Abnormally strong Silicon|
|Sr||Abnormally strong Strontium|
|Code||Spectral peculiarities for white dwarfs|
|:||Uncertain assigned classification|
|P||Magnetic white dwarf with detectable polarization|
|E||Emission lines present|
|H||Magnetic white dwarf without detectable polarization|
|PEC||Spectral peculiarities exist|
For example, Epsilon Ursae Majoris is listed as spectral type A0pCr, indicating general classification A0 with a strong emission lines of the element chromium. There are several common classes of chemically peculiar stars, where the spectral lines of a number of elements appear abnormally strong.
Stars can also be classified using photometric data from any photometric system. For example, we can calibrate color index diagrams of U−B and B−V in the UBV system according to spectral and luminosity classes. Nevertheless, this calibration is not straightforward, because many effects are superimposed in such diagrams: interstellar reddening, color changes due to metallicity, and the blending of light from binary and multiple stars.
Photometric systems with more colors and narrower passbands allow a star's class, and hence physical parameters, to be determined more precisely. The most accurate determination comes of course from spectral measurements, but there is not always enough time to get qualitative spectra with high signal-to-noise ratio.