To jump straight to the chase, the main stellar classes are called O, B, A, F, G, K and M. These types correspond to a star’s surface temperature, with O stars being hottest, upwards of about 40,000 K, and M stars being coolest, down around 3000 K. The classifications are based upon patterns of spectral lines. If you are not familiar with the electromagnetic spectrum, or if you would like to brush up on it before continuing, use the two links provided to read some background material.
A Note About Temperature
Astronomers always talk about a star’s temperature in terms of the kelvin scale. If you are not familiar with the kelvin scale, the following brief description will serve to acquaint you with it.
To begin, note that the freezing temperature of water (0 deg C) is 273.16 kelvin, or 273.16 K. The boiling point of water (100 deg C) is 373.16 K. From this you can see that kelvin temperature units are the same size as Celsius degrees, but they have a 273.16 offset. Stars are so hot, even the coolest ones, that the 273 offset can be pretty much ignored if you like. You can think of kelvin temperatures as being the same as Celsius as a rule of thumb, at where stars are concerned, and you won’t be making a very larger error.
Why put in this offset in the first place? The Celsius scale is based on the freezing and boiling points of water at standard (Earth surface) temperature and pressure. This is a useful scale for chemists, who do much of their work between these two temperature points. However, in physics there is nothing special about water or its phase transition points. The kelvin scale is based upon a more physically meaningful concept: the kinetic energy content of a material (specifically, of an idealized gas). At zero kelvin, all motion of the molecules within the gas would stop. As the temperature increases from zero the kinetic energy of the gas molecules would increase apace. This is a useful concept in physics, though it is certainly an idealization. This classical notion of the energy of a system is not consistent with the quantum understanding of materials that superseded it, at least at low temperatures. It is nonetheless useful under many circumstances.
A final (minor) difference between Celsius and kelvin measures is how we talk about them. Kelvin temperature units are not degrees, but Celsius units are. This is merely semantics, of course, but it is how the scales are used. When we discuss kelvin temperatures, we just say kelvin, not degrees or degrees kelvin.
If you are more used to Fahrenheit temperatures, then you have to keep in mind that every degree Celsius (or every kelvin) is 1.8 degrees F. And of course, the freezing point of water is 32 F, and it boils at 212 F. So for Fahrenheit you have to take into account both an offset (32 degrees) and a difference in scale (1.8, or ~2). That is a pain. For very hot things like stars it is approximately true that the temperature in Fahrenheit is about twice the Celsius (or kelvin) temperature. About. We won’t worry about Fahrenheit or Celsius temperatures anymore. Hopefully this short discussion will help you put star temperatures into perspective. In short, even the coolest stars are really, really (REALLY) hot by our usual standards.
Early Development of Stellar Classification Schemes
Stellar classes happen to correspond to the surface temperatures of stars, but they were not (and are not) based upon the temperature. They were based on the spectra of the stars. The spectrum is the pattern of dark absorption lines seen when the light from the star is passed through a narrow aperture and then a prism. The earliest work to classify stars by these lines was done by Angelo Secchi in the 1860s. Secchi’s scheme used Roman numerals, I to V, to classify stars that were blue (class I), yellow (class II) and red (class III). The additional classes were used to subdivide the red stars based upon the strength of certain spectral features. Secchi’s work was taken up and greatly expanded upon at the Harvard College Observatory over the next several decades. This monumental task eventually culminated with the publishing of the first systematic catalogue of stellar spectra, The Draper Catalogue of Stellar Spectra, in 1890. The system of classification used in the Draper Catalogue is based upon that of Secchi, but his classification system was refined by observatory staff member Williamina Fleming, who took this system and greatly subdivided it, using capital letters in the place of Roman numerals. The letters ran, in order, from A to P. An additional letter, Q, was used for peculiar stars that did not seem to fit naturally into the main system. These capital letters were the classes originally listed in The Draper Catalogue.
The original Secchi classes are shown at right. Class I is at top, class II is next, followed by class III. An example of class IV is at the bottom. The letters used to designate the lines, shown at the bottom of each spectrum, have nothing to do with modern stellar classes. These letters were used to identify lines back before their origins were understood. We now know, for example, that the strong line designated as D in the yellow part of the spectrum is caused by sodium. The H line, on the other hand, seen in the violet part of the spectrum, is due to calcium.
Note also that examples of specific stars are given for each type. Examples of Secchi class I stars are given as Sirius, Vega, Altair and Regulus. In the modern classification scheme, these do not all have the same stellar classification. The same is true for the stars given for the other Secchi classes.
This image is taken from the Tools of Cosmology website of the American Institute of Physics. https://history.aip.org/history/exhibits/cosmology/tools/pic-spectroscopy-lithograph.htm
The classification work of Fleming was continued at Harvard by Antonia Maury, who began her work there in 1897. She discarded Fleming’s letters in favor of Roman numerals (again), but used 22 of them (I to XXII) instead of the four that had been used by Secchi. This system allowed her to separate stars into finer spectral subclasses, based on finer details, than had been done previously. She also rearranged the ordering of certain stars. So the letters of Fleming no longer ran in sequence when they were translated into the Maury system. While Maury’s classification system allowed for fine distinctions to be made based upon small differences in stellar spectra, it was quite complicated and difficult to use. Ms. Maury eventually left the observatory staff because of disagreements with the observatory director, Edward Pickering. She later returned, years later, after Pickering was replaced by Harlow Shapley. In her absence, the stellar classification work was continued by one of her colleagues, Annie J. Cannon.
Annie J. Cannon had come to work at the observatory the year before Antonia Maury. They shared the duties of stellar classification to some extent, with Maury classifying the northern stars and Cannon the southern ones. However, Cannon disliked the system of Roman numerals used by Maury. She discarded them in favor of capital letters, but, even here she retained only seven of the original sixteen. To take account of the small deviations within a class she used numbers 0 to 9. In this way she was able to classify all the stars in a simple system while still making distinctions between them, just as Maury’s more complicated system did. Cannon also rearranged the ordering of some of the stars in the sequence, as Maury had done. The changes, made in 1902 by Cannon, gave us essentially the stellar classification scheme we use today. But what do all these letters (or Roman numerals) mean?
Keep in mind that the various classification schemes of stellar spectra were based on two things: the color of the stars and the patterns of the lines seen in those stars. Furthermore, at the time that these classification systems were created, in the last few decades of the 19th century, neither the color nor the lines were understood. Not at first. Only slowly did scientists come to realize that the patterns of dark lines seen in the stars corresponded to the patterns of bright lines seen in the spectra of hot gases in the laboratory. And thus only slowly was it understood that the lines were due to the presence of particular types of atoms in the atmospheres of the stars. The existence of the relationship between the color of a star and its surface temperature had a similarly slow dawning. For the scientists working to classify stars, the job was like that of a biologist who classifies trees by the shape of their leaves or the type of flowers they produce. Basically, these Harvard Observatory astronomers were doing stellar taxonomy; there was no underlying physical framework on which to build a linkage between the star colors and their spectra.
Just as taxonomy in biology allows its practitioners to discover patterns in the variations among different species of organism, so stellar classification allowed astronomers to begin to understand systematic differences between different kinds of stars. This led, eventually, to an understanding of stellar temperatures, sizes, masses and, most vitally, composition. Indeed, many other observations and theoretical underpinnings had to be employed to arrive at this understanding. The work done at the Harvard College Observatory by its staff laid vital groundwork, and so when new discoveries in physics became available, their application to the stars was made much easier. And it should be noted, nearly all of observatory staff members working on these projects were women, though the observatory directors, first Edward Pickering and later Harlow Shapley, were men.
Understanding Spectral Classification Systems
The work of the Harvard Observatory staff was impressive, but probably most impressive was the work of Cecelia Payne (later Cecelia Payne-Gaposchkin). Payne was a British astronomer who worked at the observatory after receiving her undergraduate degree in physics at Cambridge University in 1923. Her passion was to understand the relationships of the strengths of the various lines to the abundances of the elements producing them. Payne made careful calculations of the strengths using the nascent quantum theory of atoms (the Saha equation, developed in 1920 by the Indian scientist Meghnad Saha) to deduce the chemical abundances of the stars. Payne was the first person to understand that these strengths are not simply related to abundances at all. They are mostly the result of the temperature, not the composition. In fact, she found the foundational result, completely astounding at the time, that all stars have essentially the same composition. They are all made almost entirely of hydrogen gas, about 90%, with an additional ~10% that is helium. Only a tiny fraction of the atoms in a star is made of other elements like oxygen, calcium, sodium, magnesium, etc., with each typically comprising only a few parts per million of the star. Or less. Cecelia Payne found that the reason the lines of these other elements are so prominent is due almost entirely to the stars’s surface temperature and the arrangement of the allowed energy levels in each atom. Payne thus found that the pattern of lines in stars of different spectral types is essentially the result of the temperature differences between those types. For her effort, Cecelia Payne was awarded a doctorate in astronomy from Harvard in 1925. Hers was the first doctorate in astronomy awarded to a woman by Harvard or any other university in North America. In fact, Cecelia Payne received the first doctorate in astronomy ever awarded by Harvard to anyone, man or woman. This is because Harlow Shapley was granted permission to create the Department of Astronomy expressly for the purpose of awarding Payne her degree when the Chair of the Department of Physics refused to award a doctorate to a woman.
Of course, any of the other Harvard College Observatory astronomers (they were called “computers” because they did many computations of various types) could have been awarded a doctorate for their work as well. None of them ever was, aside from honorary degrees from various other colleges and universities. It is a real injustice that women were not awarded PhD degrees at this time. Even though several of them, especially Annie Cannon, were respected internationally and attended and spoke at conferences the world over, none were ever given the academic credit they deserved. None until Cecelia Payne, that is. She went on to become a professor in the Harvard Astronomy Department, and its Chair as well.
The figure below shows the modern stellar classes arranged in order. Hot stars at the top and cool stars at the bottom. The figure is like the one above. It shows the dark spectral lines superimposed on the rainbow continuum of the stars. Note how the pattern of the lines changes as the temperature of the stars decreases from top to bottom. So a B0 star is the hottest type of B star, and a B9 would be the coolest. A B5 star sits right in the middle of the B range, midway between the hottest A star and the coolest O star. Similarly for the other spectral types. Note that even decimal numbers are allowed: the hottest star shown here is an O6.5.
The figure at the right shows examples of these same (or nearly the same) stellar types, but now the intensity of the light in the spectra have been plotted on a graph. Intensity is on the vertical axis, wavelength on the horizontal. Shorter wavelengths are blue, whereas longer wavelengths are red. Our eyes are sensitive to wavelengths from around 380 nm (deep violet) to 680 nm, which is a deep red, though these limits can vary slightly among different individuals. Beyond these visible wavelengths lie the ultraviolet at the short wavelength end, and the infrared longward of the red wavelengths. And of course, additional wavelengths of light, like x-rays, gamma rays, microwaves and radio all lie outside the visible window as well.
Can you spot the dark lines in the figure above that correspond to the dips of the intensity in the graphs at right?
The hot stars, spectral types A, B and O, all show a regular pattern of deep absorption (dark) lines at the blue part of the spectrum. These are seen at wavelengths between 400 and 500 nm, so from blue to green. These lines (called hydrogen Balmer) are due to hydrogen gas that has its electron excited to the first excited state instead of sitting in the ground state - the lowest state. In the hottest stars, O, B and A types, the temperatures are high enough to keep hydrogen’s electron constantly excited up into the first excited state. In cool stars, like G, K and M class stars, the lines are not present. That is because the stars are too cool to excite hydrogen.
The hydrogen atoms in these cooler stars are typically in the ground state, meaning that their electrons sit in the lowest possible energy level of the atom. As a result, they cannot create the Balmer transitions. However, around F stars the lines begin to be prominent. This is not because the stars have more hydrogen than cooler stars, it is because the electrons in these hotter stars are populating the first excited state more and more. In classes A and B the Balmer lines are the most dominant. In fact, they are pretty much the only features seen in the spectrum. However, by the time we get to O stars, the lines are getting weaker again. What is going on there?
The O stars are the hottest stars. They are so hot that their hydrogen gas is excited above the first excited state. In fact, the hydrogen in these stars is pretty much completely ionized, meaning that most electrons in the hydrogen of these stars are not even attached to an atom. As a result these “atoms” (essentially just bare protons in a sea of free electrons) cannot absorb at all in the visible. Only the relatively small fraction of hydrogen that still retains an electron in the first excited state is able to absorb and create the weak lines seen.
Now take a look at the cooler stars at the bottom of the diagrams. These stars don’t show hydrogen, but they do show a lot of other lines. These lines are created by absorption into various kinds of atoms. For example, there are two very strong lines just shortward of 400 nm that are seen in K stars and continue to be strong up into the A-star region. These are caused by absorption into singly ionized calcium, or calcium that has lost one electron. There is another line seen in G stars right around 500 nm. It looks a little bit like one of the hydrogen lines, and in fact, there might be some hydrogen there, but it is mainly due to magnesium. Careful inspection would show that it is not quite at the same wavelength as the hydrogen line, and it has the wrong shape.
Finally, have a look at the M stars. They begin to show absorption, and very broad absorption at that, far out in the red part of the spectrum. These broad absorption troughs are very pronounced in the M5 stars shown. These broad troughs are not due to atoms at all, they are caused by molecular absorption. M stars have atmospheres so cool that molecules can form in them. Of course, these are not molecules that are familiar to us, like oxygen or hydrogen or water. Those molecules would quickly break apart in a gas as hot as even the coolest stars. M star atmospheres contain molecules of TiO (titanium oxide). Like all molecules, TiO shows very broad molecular band absorption created by its quantized modes of rotational motion.
So why don’t we see calcium and magnesium in stars of types A, B and O? Why don’t we see TiO? For the same reason that we don’t see hydrogen, or not much, in O stars. The hotter stars ionize away the atomic states seen in cooler stars. Other absorption features take their places as the atomic gas adjusts itself to the new, hotter temperature regime. This is exactly what Cecelia Payne was able to deduce using the Saha equation, and it is why her PhD thesis is often considered to be the most important ever done in astrophysics.
Stellar spectra, and thus spectral types, allow us to take the temperature of the stars. They also allow us to determine stellar chemical composition. They thus capture the most fundamental aspects of stellar astrophysics in a very simple scheme. It was Cecelia Payne, building upon the work of Annie Cannon and Antonia Maury (and Meghnad Saha, of course) who figured out how these spectra work.
If you would like to learn more about the early history of stellar classification and the other work done at the Harvard College Observatory (stellar spectral classification was only part of what they accomplished), Dava Sobel has written an excellent book on the subject. It is called The Glass Universe.
The MK Spectral Classification System for Stars
Since the original work on stellar classification, the system has been expanded to take into account the size of stars. The MK (for Morgan-Keenan, after the astronomers who helped develop it) spectral classes are essentially the same as Annie Cannon’s system, but with the addition of a Roman numeral (those again!!!) to designate a luminosity class. These luminosity classes are related to the size of a star. They take into account the fact that stars of the same temperature can have radically different sizes, and thus can have enormous differences in brightness. Stars are born small and expand slowly over their lives until the end, when they run out of hydrogen to burn in their core. At that point they expand rapidly into a giant or supergiant. The expansion, whether fast or slow, is related to their nuclear burning, and at the giant stage they can be truly colossal. Their radii can exceed the size of the orbit of Mars. Stars at the end of their lives also cool a lot, at least at their surface (not in their cores!), and so they tend to end up as either K or M type stars, regardless of how they started out.
So, for example, a G5V star is a relatively young G star that is halfway from an M star to a K star, right in the middle of the G class. However, a G5II star, while still in the middle of the G range, is not a young star at all. It is an evolved (and likely rapidly evolving) star. An M0V is a youngish (relatively) star at the extreme hot end of the temperature range for M stars, but an M0I, while having the same temperature as the first star, is immensely larger. This star would be one of those that would envelope the entire inner solar system within itself if was placed where the Sun is.
The Sun is currently a G2V star. Someday, in about 4 billion years, it will exhaust its hydrogen and expand into a K or M star of some sort. Until then it will slowly expand while remaining at approximately the same temperature, climbing the luminosity class ladder as it brightens from V to IV to III, etc, but remaining close to G2 in terms of temperature. That’s about 5800K.
From this brief discussion you can see that with MK spectral classes we can tell a lot about a star. We can designate its temperature and its current state of evolution, and we can use the spectral class to describe the entire lifecycle of a star. The reasons why spectral class gives such a complete description of a star has to do with the fairly simple physics of giant self-gravitating balls of gas. Perhaps that could be the topic of another blog post.