An interactive journey through the kinds of stars

The sky is not a collection of points

It is a catalog of states. Every point of light up there is a particular mass in a particular phase of life — and its color tells you both. This journey shows you how to read it.

126 scenes 9 chapters 44 live graphics, 13 of them interactive Interpretive layer throughout
Begin the journey
Table of contents
Prologue
“Two things fill the mind with ever new and increasing admiration and awe, the more often and the more steadily one reflects on them: the starry heavens above me and the moral law within me.”
Immanuel Kant · Critique of Practical Reason, Conclusion · 1788
Looking up

For some 200,000 years the night sky was the one constant of human experience. It was calendar, compass, place of worship and rebuke all at once. Only for about 200 years have we known what these points are made of — and for roughly 100, why they shine.

This lecture follows two tracks side by side. One is astrophysical and testable. The other is cultural history: what people have read into the same points of light. The two stay apart — the second always appears in gold panels and never claims to be physically true.

Interpretive layer · Philosophy

“Le silence éternel de ces espaces infinis m'effraie.” — The eternal silence of these infinite spaces frightens me.

Blaise Pascal wrote that line around 1660. It is the counter-voice to Kant: the same sky that stirs awe can also stir dread. Modern astronomy has not resolved either reaction — it has only made precise what they are about.

Pensées, fragment 201 (Lafuma) / 206 (Brunschvicg)
Definition

A star is a balance, not an object.

Physically, a star is a ball of gas in which two opposing forces cancel exactly: gravity pulls all the mass inward. The pressure gradient of hot gas and radiation pushes outward. That state is called hydrostatic equilibrium.

The energy for the counter-pressure comes from nuclear fusion at the center. That is precisely the defining criterion: a star is a self-gravitating body that fuses hydrogen into helium in its core, or at least did so once.

A star is therefore not a thing but a process in balance — and every process ends. The entire variety of stellar kinds we are about to work through is nothing more than this: different masses, at different points of the same decay.

Gravity ⟶ inward Pressure ⟵ outward Balance = stability
Hydrostatic equilibrium
Move the slider below
Fusion rate in the core100 %
What is a star?
Interpretive layer · Philosophy

The idea that stability is not rest but balanced tension appears in Greek natural philosophy long before physics. Heraclitus calls it “palintropos harmoniē” — a back-turning attunement, illustrated by the bow and the lyre: the bow holds because string and wood pull against each other.

That the stars, for thousands of years the very emblem of the unchanging, should turn out to be the purest case of this principle is one of the finer ironies in the history of science.

Heraclitus, fragment DK 22 B 51
Scale

Before we sort: how much is there to sort?

These figures are estimates with substantial error bars, but the order of magnitude is robust. And it leads to the question that made all of modern astrophysics possible in the first place: if there are that many objects — are they billions of individual cases, or are there types?

The answer is surprisingly clear. There are types. Few of them, in fact. And the single most important parameter that fixes a star is exactly one: its mass at birth. Almost everything else follows from it. This is known as the Vogt-Russell theorem — though in its strict form it is only a rule of thumb, not a proven statement.

Logarithmic scale
From the Earth to the observable universe
≈ 5,000
Visible to the naked eye (whole celestial sphere, ideal conditions)
100–400 bn
Stars in the Milky Way (estimated range)
≈ 2 tn
Galaxies in the observable universe (Conselice et al. 2016)
1022–24
Stars in the observable universe (order of magnitude)
Uncertainty The number of stars in the Milky Way is not measured but extrapolated from the total mass and an assumed mass distribution function. Depending on the assumption about the share of low-mass stars, serious estimates vary by a factor of 4.
Chapter I — How we learned to read the stars

1814: an optician finds gaps in the light

Joseph von Fraunhofer, a glassmaker in Benediktbeuern, only wanted to build better lenses. For that he needed light of exactly known wavelength. When he sent sunlight through a prism and a fine slit, he saw hundreds of sharp dark lines in the band of color.

He cataloged 574 of them and labeled the strongest with letters — A, B, C, D, E, F, G, H. Those labels are still in use today: the Fraunhofer D line is the sodium doublet, the H and K lines come from singly ionized calcium.

Fraunhofer did not know what the lines meant. He only knew this: they are always there, always in the same place. With that, the sky became something you could read for the first time.

Solar spectrum · Fraunhofer lines
Hover over the lines
Fraunhofer's dark lines

The classic Fraunhofer lines

A, B
O₂ (Earth's atmosphere!) · 759 / 687 nm
C (Hα)
Hydrogen · 656.3 nm
D₁, D₂
Sodium · 589.6 / 589.0 nm
E
Iron · 527.0 nm
F (Hβ)
Hydrogen · 486.1 nm
G
Fe + Ca (CH band) · 430.8 nm
H, K
Ca II · 396.8 / 393.4 nm
Detail Lines A and B do not come from the Sun but from the oxygen of our own atmosphere. They are called telluric lines — the first known case of an observer contaminating their own measurement.
1835 / 1859

The most famous failed prediction in the history of science

“We shall never be able to study their chemical composition.”

Comte, the founder of positivism, wanted an example of knowledge that is unreachable in principle. The distance of the stars, he argued, makes any sampling impossible — and without a sample there is no chemistry.

Kirchhoff's three laws
Auguste Comte on the heavenly bodies · Cours de philosophie positive, 19th lecture · 1835
Comte's error

Twenty-four years later, Gustav Kirchhoff and Robert Bunsen in Heidelberg showed that every chemical element produces an unmistakable pattern of lines. Kirchhoff saw that Fraunhofer's dark lines sit exactly where the same elements show bright lines in the laboratory. The Sun had been analyzed chemically — without a sample.

The lesson is not about Comte being foolish. His argument was reasonable. The lesson is this: the limits of knowledge are almost always moved by new methods, not by more effort with the old ones.

Interpretive layer · Philosophy of science

Comte's error has become the standard argument against any form of “we will never know that”. The counter-position is worth noting, though: the philosopher of science Paul Feyerabend would have objected that refuting one positivist says nothing about the reach of knowledge as a whole — it only shows that predictions about future methods rarely succeed.

What Comte saw correctly: we will never take a sample from a star. What he did not see: that the star keeps sending us one — in the form of light.

1863–1868 · Rome

Angelo Secchi sorts 4,000 stars by hand

A Jesuit priest and director of the observatory of the Collegio Romano points an objective prism at the sky and becomes the first person to order stars not by brightness or position but by their physical constitution.

Class I

White-blue, broad strong hydrogen lines. Sirius, Vega Today: A and early F

Class II

Yellow, many metal lines, weaker hydrogen. Sun, Capella, Arcturus Today: G and K

Class III

Secchi: the first classes

Orange-red, broad bands with a sharp edge toward the blue. Betelgeuse, Antares Today: M (TiO bands)

Class IV

Deep red, bands bounded toward the red. R Leporis Today: carbon stars (C)

Secchi later added a Class V for stars with bright emission lines (today these include Be stars and Wolf-Rayet stars). What matters is the achievement behind it: he recognized that the seemingly chaotic variety of spectra collapses into a handful of recurring patterns. He simply did not yet know why.

Interpretive layer · Cultural history

That a Catholic priest produced the first physical classification of stars is no coincidence but has an institutional explanation: from the 16th century onward the Jesuits ran a worldwide network of observatories. Secchi's predecessors in the post had contributed to the Gregorian calendar reform.

The story of a fundamental opposition between “church and astronomy” is a crude historical simplification. The Galileo affair was real and consequential — but it was a conflict about who gets to interpret, not about observing the sky as such.

1886–1924 · Harvard College Observatory

The catalog that industrialized astronomy

After the death of the physician and amateur astronomer Henry Draper, his widow Mary Anna Draper funds a project to record every known star spectroscopically. The observatory's director, Edward Charles Pickering, hires a team of women for it — partly because they worked more carefully, partly because they could be paid considerably less. They entered history as the Harvard Computers.

1890
Williamina Fleming
Orders more than 10,000 stars by the strength of their hydrogen lines and assigns letters from A (strongest lines) to Q. A purely descriptive, alphabetical system — with no physical meaning.
1897
Antonia Maury
Recognizes that line width carries information of its own and introduces 22 Roman-numbered groups plus a suffix c for strikingly narrow, sharp lines. Pickering considered this a needless complication. It was the discovery of the luminosity class — 46 years too early.
The Harvard Computers

Why the order is so strange

1901–1924
Annie Jump Cannon
Drops most of Fleming's letters as redundant, reorders the rest along a continuous transition of the spectra and adds decimal subdivisions. The result: the sequence O B A F G K M. Over her lifetime Cannon classified about 350,000 stars — at times more than 200 an hour, peering through a magnifier at photographic plates.
The Harvard Computers

Fleming's A–Q was sorted alphabetically by hydrogen line strength. Cannon kept the established letters but put them into the physically meaningful order — and that order was not the alphabet. Out of consideration for catalogs already published, nothing was renamed.

That is why the hottest class is called O and not A. A piece of scientific history preserved in a sequence of letters that has been baffling first-year students for over 120 years.

Mnemonics

Oh Be A Fine Girl/Guy, Kiss Me

Only Bad Astronomers Forget Generally Known Mnemonics

Interpretive layer · Sociology of science

Antonia Maury's c suffix is a lesson in how knowledge can be lost. She had observed the effect correctly; Pickering pushed the system back to the simpler version. Only Ejnar Hertzsprung realized in 1905 that Maury's c stars are systematically very luminous — she had found the giant stars without being able to name them.

In 1925 Cannon became the first woman to receive an honorary doctorate from the University of Oxford. She never held a regular professorship at Harvard; until 1938 her official job title was simply Curator of Astronomical Photographs.

The most important doctoral thesis in astronomy

1925: stars are not made of what the Earth is made of.

Into the 1920s it was taken for granted that the Sun has essentially the same composition as the Earth — iron, silicon, oxygen. The spectra seemed to confirm it: they are full of metal lines.

Cecilia Payne, 25 years old, applies the newly formulated Saha equation from quantum physics to stellar spectra in her Harvard doctoral thesis. It describes how strongly an element is ionized at a given temperature — and therefore whether it can produce lines at all.

Her result: the strength of a spectral line says almost nothing about how abundant an element is, and almost everything about temperature. Once you account for that, the Sun turns out to be more than 90 % hydrogen by particle count, with almost all the rest helium. Everything else together: under 0.2 %.

Her reviewer Henry Norris Russell judged the result “clearly impossible” and had her write in the text that the values were “almost certainly not real”. Four years later Russell reached the same conclusion by a different route and published it — he cited Payne, but the discovery was long credited to him.

Cecilia Payne, 1925

The astronomer Otto Struve later called her work “the most brilliant PhD thesis ever written in astronomy”. In 1956 she became the first woman to hold a full professorship in Harvard's Faculty of Arts and Sciences.

Values follow the standard model of solar composition (Asplund et al.); depending on the reference model they vary by tenths of a percent. In astronomy every element heavier than helium is called a “metal” — including oxygen and carbon.

~73.8 % by mass
Hydrogen in the Sun
~24.9 % by mass
Helium
~1.3 % by mass
Everything else (“metals”)
Interpretive layer · Epistemology

Payne-Gaposchkin's case shows a pattern that recurs throughout the history of science: the obstacle was not the data but a background assumption nobody recognized as an assumption — that matter throughout the cosmos is uniform, on the model of Earth.

Thomas Kuhn would have called this an example of how tenacious paradigms are. The real break was not in the calculation but in the willingness to trust the calculation against intuition.

1943 · Yerkes Observatory

Two letters are not enough. The atlas of Morgan, Keenan and Kellman.

The Harvard system describes temperature and nothing else. But two stars of the same temperature can be entirely different objects: a compact main-sequence star, or a bloated supergiant a thousand times wider.

William Wilson Morgan, Philip Keenan and Edith Kellman publish the Atlas of Stellar Spectra in 1943 and solve the problem through line width — exactly the effect Antonia Maury had seen 46 years earlier.

Pressure broadening
Dwarf (below) vs. supergiant (above)
The MK system

The physical cause: pressure broadening. In the dense photosphere of a dwarf star, neighboring particles constantly disturb the energy levels of the atoms, and the lines become broad and smeared. In the extremely thin, far-extended envelope of a supergiant those disturbances are absent — the lines come out razor-sharp.

Line width reveals density. At a known temperature, density reveals the radius. The radius reveals the luminosity. And from luminosity plus apparent brightness follows the distance — the spectroscopic parallax.

The Sun's full MK code is G2 V: temperature class G, subclass 2, luminosity class V (main sequence). Two characters, and the star is largely determined physically.

Interpretive layer · Chapter close

The sky had been read for 3,000 years — just differently.

The idea that the sky contains information you can decode did not begin with Fraunhofer. It is one of the oldest intellectual practices there is. What changed was the kind of code.

Babylonian sky writing

The cuneiform series Enūma Anu Enlil — some 70 clay tablets with roughly 7,000 omens, compiled in the first millennium BC from considerably older material — is the first systematic attempt to treat celestial appearances as a sign system. The structure is strictly conditional: “If X happens in the sky, then Y will happen in the land.”

What matters is what these texts are not: individual horoscopes. Babylonian omen astrology dealt almost exclusively with king and state. Birth astrology for private persons only emerges from the 5th century BC onward — the oldest known individual horoscope dates to 410 BC.

The sky as a text
Babylonian sky writing

And these observers were excellent empiricists. Out of the need to predict omens came humanity's first precise planetary tables. The Babylonian values for the length of the synodic month were accurate to the second. Astronomy grew out of astrology, not against it.

Enūma Anu Enlil; oldest dated individual horoscope: 410 BC, Babylon

The break comes later, and can be dated more precisely than is often assumed. In the 2nd century AD Claudius Ptolemy writes both works: the Almagest as mathematical celestial mechanics and the Tetrabiblos as a doctrine of interpretation. To him these were two chapters of the same discipline. The separation only takes place in the 17th century — and even then hesitantly.

Chapter II — The temperature ladder · Interactive

Why color means temperature

Every hot body radiates across all wavelengths — but not evenly. The distribution follows Planck's law of radiation (Max Planck, 1900). The curve always has the same shape and merely shifts with temperature.

G2 V
Planck curve · star color · spectral lines
Effective temperature5,772 K
O5 B2 A0 · Vega F5 G2 · Sun K5 M5 · Proxima
The blackbody slider

Wien's displacement law

λmax · T = 2.898 · 10−3 m·K

Double the temperature and the wavelength of peak radiation halves. The Sun at 5,772 K peaks at about 502 nm — green. That it still looks white to us is because it radiates all visible colors strongly.

Stefan-Boltzmann law

L = 4π R² · σ T⁴

Luminosity rises with the fourth power of temperature. A star twice as hot at the same size shines 16 times brighter. That is why the rare hot stars dominate our view of the sky so completely.

A common error There are no green stars. A star whose peak lies in the green radiates so broadly that the eye sees white. Green can only arise if other colors are missing — for a continuum radiator that is impossible.
Overview

O B A F G K M A ladder from 45,000 down to 2,400 kelvin

Click a class
The seven letters
ClassTeff (K)Color (true)Mass (M☉)Radius (R☉)Luminosity (L☉)Marker linesShare*
O≥ 30,000Blue≥ 16≥ 6.6≥ 30,000He II, He I, N III~0.00003 %
B10,000–30,000Blue-white2.1–161.8–6.625–30,000He I, H (increasing)0.12 %
A7,500–10,000White1.4–2.11.4–1.85–25H (maximum), Ca II weak0.61 %
F6,000–7,500Yellow-white1.04–1.41.15–1.41.5–5H weaker, metals strong3.0 %
G5,200–6,000Yellowish-white0.8–1.040.96–1.150.6–1.5Ca II H+K strong, Fe I, G band7.6 %
K3,700–5,200Orange0.45–0.80.7–0.960.08–0.6Neutral metals, first TiO12 %
M2,400–3,700Orange-red0.08–0.45≤ 0.7≤ 0.08TiO bands dominant, VO76 %
The seven letters

* Share of the main-sequence stars in the solar neighborhood. Values follow the compilation customary in the literature (among others Habets & Heintze 1981, LeDrew 2001). The temperature boundaries are convention, not constants of nature — different sources shift them by a few hundred kelvin. Each class is further divided decimally from 0 (hot) to 9 (cool); the Sun is G2, so in the hot third of the G class.

Spectral class O · ≥ 30,000 K

The torches that barely exist

O stars are the most extreme main-sequence stars: hotter than 30,000 K, at least 16 solar masses, ten thousand to a million times more luminous than the Sun. Their radiation peaks deep in the ultraviolet — the visible blue part is only the tail end.

Their signature in the spectrum is lines of doubly ionized helium (He II). Tearing a second electron off helium takes 54.4 eV — only an extremely hot star produces photons like that.

O5 V · 42,000 K
Class O — the short-lived

Their lifetime is a few million years. For comparison: when the first O stars that still shine today were born, there were already humans on Earth. They are so rare that in the entire solar neighborhood about one in three million main-sequence stars is an O star — and yet they define the appearance of every star-forming region, because their ultraviolet light makes the surroundings glow (H II regions).

Zeta Puppis · O4 · ~1,100 ly Alnitak (ζ Ori) · O9.5 · ~1,260 ly Theta¹ Ori C · O7 · ~1,340 ly
Consequence Where there are O stars, star formation is under way now. They are cosmic time markers: a region with O stars is younger than about 10 million years.
Spectral class B · 10,000–30,000 K

Blue-white, fast-spinning, sociable

B stars are what most people picture when they think of a “blue star”. Their spectrum shows neutral helium (He I) as its marker line; the hydrogen lines grow steadily stronger from B0 toward B9.

They are strikingly often found in OB associations — loosely bound groups of young massive stars that formed together from the same molecular cloud and drift apart over ten to a hundred million years. The Pleiades cluster is one such group.

Many B stars rotate extremely fast — equatorial speeds above 200 km/s are normal. In some of them this produces a subclass of its own, the Be stars, which we will meet later.

B2 V · 20,000 K
Rigel · B8 Ia · ~860 ly Spica · B1 · ~250 ly Regulus · B8 · ~79 ly Achernar · B6 · ~139 ly
Class B — the builders
Interpretive layer · The Pleiades

The Pleiades are probably the most widely attested star group across cultures anywhere. Greece: the seven daughters of Atlas. Japan: Subaru — “to unite” (hence the car logo with six stars). Māori: Matariki, whose reappearance marks the new year and has been a public holiday in New Zealand since 2022. Babylon: MUL.MUL, “the star of stars”, the first line in the star catalog MUL.APIN.

One motif recurs strikingly: seven sisters, but only six are visible — a “lost Pleiad”. That story turns up in Greece, among Aboriginal Australians, in North America and in Siberia. In 2020 the astronomers Ray Norris and Barnaby Norris proposed the hypothesis that this could be the memory of an actual sight: some 100,000 years ago Atlas and Pleione stood farther apart and were separately visible. This is a seriously discussed but unproven hypothesis — the competing explanation is simply coincidence plus the cultural popularity of the number seven.

Interpretive layer · The Pleiades
Norris & Norris, “Why are there Seven Sisters?”, 2020 (arXiv:2101.09170)
Spectral class A · 7,500–10,000 K

Where the hydrogen lines reach their maximum

Historically the A class is the most important one, because it was the starting point of Fleming's alphabet: this is where the Balmer lines of hydrogen are strongest.

The reason is subtle, and it was exactly Cecilia Payne's point: not because there is the most hydrogen here — but because at around 9,500 K the largest fraction of hydrogen atoms sits in the first excited state, out of which the visible Balmer transitions take place. Cooler: too little excitation. Hotter: the hydrogen is already ionized and has no electron left to excite.

A stars appear white to the eye. For decades Vega was the zero point of the photometric system: by convention its brightness defined magnitude 0 in every filter, and with it the color index B−V = 0.

A0 V · 9,600 K
H (Balmer) line strength against temperature
Sirius A · A1 V · 8.6 ly Vega · A0 V · 25 ly Altair · A7 V · 16.7 ly Deneb · A2 Ia · ~2,600 ly
Class A — pure hydrogen
Interpretive layer · Sirius

Sirius, the brightest star in the sky, is the single star with the densest cultural record. In ancient Egypt it was Sopdet (Greek Sothis): its heliacal rising — the first visibility at dawn after weeks of absence — coincided with the beginning of the Nile flood and structured the Egyptian calendar. The Greek name Seirios means “the scorching” or “the glowing”; its assignment to the constellation Canis Major gives us the dog days.

On the Dogon claim: a widespread story holds that the Dogon people of Mali possessed traditional knowledge of the invisible companion star Sirius B. It goes back to Marcel Griaule and Germaine Dieterlen (1950). The anthropologist Walter van Beek restudied the Dogon in 1991 and found no confirmation whatsoever: no informant knew the material Griaule reported. The most likely explanation is cultural contamination — Sirius B had been known since 1862 and was a popular science topic in the 1920s, and European visitors were on site. In the scholarly literature the claim counts as refuted.

Interpretive layer · Sirius
Van Beek, “Dogon Restudied”, Current Anthropology 32(2), 1991
Spectral class F · 6,000–7,500 K

The tipping point: where metals take over the stage

F stars mark the transition. The hydrogen lines weaken while more and more lines of ionized metals emerge — above all Ca II, iron and chromium. The spectrum becomes denser, more wiry.

Something decisive happens here physically: at a surface temperature of about 7,000 K a pronounced outer convection zone appears for the first time — a churning mantle in which hot gas rises and cooled gas sinks. With it comes the capacity for a magnetic dynamo: starspots, flares, activity cycles. Everything we know from the Sun begins here.

F stars are considered interesting in astrobiology: brighter and longer-lived than they need to be unstable, but with markedly stronger ultraviolet radiation than the Sun.

F5 V · 6,600 K
Class F — the in-between
Procyon A · F5 IV-V · 11.5 ly Polaris · F7 Ib · ~447 ly Canopus · F0 II · ~310 ly
Interpretive layer · Polaris and the world axis

Nobody knew that Polaris is an F supergiant and a Cepheid on top of that when it became the most important orientation mark of the northern hemisphere. All that mattered was its apparent immobility. In Finnish mythology it is called Pohjantähti and holds up the vault of heaven like a nail; in Norse tradition the sky turns around Veraldar nagli, the world nail. Similar images — post, axis, hinge — are found in Mongolia, Siberia and India.

Interpretive layer · Polaris and the world axis

The catch: Polaris is not an eternal pole star. Through the precession of the Earth's axis (a period of about 25,800 years) the celestial north pole wanders. Around 2700 BC, when the Great Pyramid was built, Thuban (α Draconis) stood at the pole. In about 12,000 years it will be Vega. The “world axis” is itself a process — an image whose philosophical irony is hard to beat.

Spectral class G · 5,200–6,000 K

The only star we know from close up

The Sun is a G2 V star: temperature class G, subclass 2, luminosity class V. That makes it brighter and more massive than about 90 % of all stars in its neighborhood — the term “average star” is misleading.

G2 V · 5,772 K
Granulation · spots · prominences
Class G — our Sun
Teff
5,772 K
Mass
1.989 · 10³⁰ kg (by definition: 1 M☉)
Radius
695,700 km (109 × Earth)
Luminosity
3.828 · 10²⁶ W
Core temperature
≈ 15.7 M K
Core density
≈ 150 g/cm³ (13 × lead) — and still a plasma
Age
≈ 4.57 bn years
Time left on the main sequence
≈ 5 bn years
Mass loss
≈ 4.3 M tonnes/s converted to energy by fusion

The most conspicuous features in the solar spectrum are the broad H and K lines of ionized calcium and the so-called G band, a tangle of CH molecular transitions at 430 nm. Both are characteristic of the whole G class.

Perspective According to current model calculations, a photon from the solar core needs between 10,000 and a few hundred thousand years for its energy to reach the surface — through countless absorptions and re-emissions. The remaining 150 million kilometers to Earth take it 8 minutes and 20 seconds.
Spectral class K · 3,700–5,200 K

Orange, frugal, tough

K stars are cooler and less massive than the Sun. Their spectrum is dominated by lines of neutral metals; in the late subclasses (K5 and cooler) the first faint titanium oxide bands appear — the first sign that molecules can already survive in their atmosphere.

Astrobiologically they are considered especially promising and are sometimes called “Goldilocks stars”: their main-sequence lifetime runs to 15 to 45 billion years. They are steadier than M dwarfs, which bombard their planets with violent flares, and far more common and longer-lived than G stars.

The nearest K dwarf is Alpha Centauri B at 4.37 light years — part of the closest star system to us. Together with the G star Alpha Centauri A and the M dwarf Proxima it forms a triple system that gathers three spectral classes into a single field of view.

K5 V · 4,300 K
Arcturus · K1.5 III · 36.7 ly Aldebaran · K5 III · ~65 ly α Cen B · K1 V · 4.37 ly Epsilon Eridani · K2 V · 10.5 ly
Class K — the enduring
Interpretive layer · Aldebaran and the “royal stars”

In esoteric literature Aldebaran, Regulus, Antares and Fomalhaut are often called “the four royal stars of Persia” or “watchers of the sky” — each assigned to a cardinal direction and a season.

Historical assessment: around 3000 BC the four stars did in fact mark the approximate positions of the equinoxes and solstices — that is astronomically calculable and the plausible core of the idea. The specific label “royal stars” with fixed archangels attached (Michael, Gabriel, Raphael, Uriel), however, cannot be documented in Old Persian sources in this form. It is largely a construction of 19th and 20th century Western esotericism, recombining older motifs.

Interpretive layer · Aldebaran and the “royal stars”

A clean example of how an invented tradition acquires venerability after the fact. Aldebaran's Arabic name al-dabarān, by contrast, is well attested: “the follower” — because it follows the Pleiades across the sky.

Spectral class M · 2,400–3,700 K

Three out of four stars are M dwarfs

And not one of them is visible to the naked eye. That may be the most consequential fact in stellar statistics: the sky we see is not representative. We see the rare exceptions, because they are bright.

M stars are so cool that molecules remain stable in their atmospheres. The dominant absorption bands come from titanium oxide (TiO), and in the coolest ones from vanadium oxide. Instead of sharp lines the spectrum shows broad, jagged troughs — it looks nothing like that of a hot star.

M5 V · 3,050 K
Class M — the silent majority

Low-mass M dwarfs are fully convective: the whole star mixes, instead of just an outer layer. That makes nearly its entire hydrogen supply available as fuel — not only the core region. The calculated lifetime of the smallest examples runs to more than a trillion years, a hundred times the present age of the universe. Not a single M dwarf anywhere in the universe has died yet.

4.25 ly
Proxima Centauri, M5.5 Ve — the nearest star of all, yet at magnitude 11.1 about 70 times too faint for the naked eye
~0.0017 L☉
Proxima's luminosity — it radiates less than 0.2 % of the Sun's output
An important distinction Not every M star is small. The class says something about temperature and nothing else. Betelgeuse and Antares are M stars too — but supergiants with a hundred million times the volume of an M dwarf. That is precisely why a second dimension was needed: the luminosity class.
Reality check

Two completely different skies

Left: the stars that actually exist. Right: the stars we see with the naked eye. It is almost the same list — only reversed.

The effect is called Malmquist bias: in any brightness-limited sample, luminous objects are massively overrepresented, because you can see them across greater distances. It is not a quirk of astronomy but a general trap of statistics — the same structure as survivorship bias.

Of the 20 brightest stars in the night sky, not one is an M dwarf, although M dwarfs make up three quarters of all stars. Conversely, of the 20 nearest stars the overwhelming majority are M dwarfs.

Share per spectral class · logarithmic scale on the left
Who is really out there
Interpretive layer · Epistemology

That makes the night sky a perfect model of a more general problem: what stands out is rare. What is common does not stand out.

All of human history built its cosmologies, myths and systems of orientation on a sample consisting of some 5,000 statistical outliers. That working calendars and navigation systems came out of it anyway says something for the practice — but the world pictures derived from it described a universe that does not exist in that form.

Extension downward · since 1999

The classes that only exist as of yesterday: L · T · Y

Into the 1990s the scale ended at M. Then infrared surveys found objects cooler than any known star — and a new alphabet became necessary.

Class L

≈ 1,300–2,400 K

Too cool for TiO — titanium and vanadium condense into dust clouds and rain out of the atmosphere. Metal hydrides (FeH, CrH) and neutral alkali metals dominate instead: sodium, potassium, rubidium, cesium. The color is a deep magenta-red.

Contains both genuine very low-mass stars and brown dwarfs — the class alone does not settle which.

Class T

≈ 500–1,300 K

The cool sequence: M → L → T → Y · colors are model calculations
L, T, Y — beyond M

The signature is strong absorption bands of methane (CH₄) in the near infrared — hence the nickname “methane dwarfs”. These objects are cooler than some planets and emit almost entirely in the infrared. To the human eye they would be faintly magenta to invisible.

Class Y

below ≈ 500 K

Only detected from 2011 onward, by the NASA mission WISE. Here ammonia (NH₃) appears in the spectrum. The record holder WISE 0855−0714, 7.3 light years away, has an effective temperature of about 285 K — roughly 12 °C. That is room temperature. JWST observations point to water clouds in its atmosphere.

State of research The temperature of WISE 0855−0714 was first estimated at 225–260 K in 2014; a JWST/NIRSpec analysis (2023/24) arrives at about 285 K. Shifts like this are normal for objects of this kind, because the values depend on models. The catchphrase “colder than water freezes” holds narrowly — or narrowly fails — depending on the model.
Parallel to the main ladder

Four classes that do not fit the temperature scale

OBAFGKMLTY is a temperature sequence. These four classes, by contrast, describe a deviant chemical composition or an entirely different type of object. They are side branches, not rungs.

W — Wolf-Rayet

Stars whose hydrogen envelope has been blown away completely and whose bare fusion core is visible. Temperatures from 30,000 to over 200,000 K. Instead of absorption lines they show broad emission lines.

WN — nitrogen dominant (CNO ash) WC — carbon dominant (helium ash) WO — oxygen dominant, extremely rare

C — carbon stars

Cool giants whose atmospheres contain more carbon than oxygen. All the oxygen is locked up in CO; the remaining carbon forms soot and molecules such as C₂ and CN.

The side branches: W, C, S, D

The result: the reddest objects in the sky. R Leporis is called “Hind's Crimson Star” for its color. Formerly divided into R and N, today listed as C-R and C-N.

S — zirconium oxide

The intermediate type between M and C: carbon and oxygen are present in a ratio of roughly 1:1. Neither TiO nor pure carbon dominates; instead bands of zirconium oxide (ZrO) appear.

Diagnostically valuable: S stars show technetium — an element with no stable isotopes and a half-life of at most 4.2 million years. Its presence proves that the star is producing it right now and carrying it outward.

D — white dwarfs

The side branches: W, C, S, D

Stellar corpses. No longer a star in the proper sense, since fusion has stopped. The classification describes only which element forms the atmosphere — usually a single-element layer, because the enormous gravity makes everything heavier sink.

DA hydrogen · DB helium · DQ carbon · DZ metals · DC no features

Suffixes in the MK code

e — emission lines present m — enhanced metal lines n — smeared (“nebulous”) lines, fast rotation s — strikingly sharp lines p — peculiar, not otherwise classifiable sh — shell star (gas envelope) var — variable spectrum f — He II and N III emission (O stars only)
Interpretive layer · Chapter close

Red and white, before there was a kelvin

That stars come in different colors is visible to the naked eye — and always was. All that changed was the explanation.

Ancient color readings

In the Almagest (2nd century) Ptolemy explicitly assigns a reddish hue to several stars — among them Aldebaran, Antares, Betelgeuse and Arcturus. From today's standpoint those assignments are correct: all of them are K and M giants.

Ancient color readings

One famous puzzle remains Sirius. Several ancient sources, including Ptolemy himself and Seneca, list it among red stars. Today Sirius is unambiguously blue-white (A1 V). The debate has been running since the 19th century. An astrophysical color change within 2,000 years is ruled out for an A star — the timescales are off by orders of magnitude. The most plausible explanations are errors of translation and transmission, confusion with the scintillation effect near the horizon (seen from the Mediterranean, Sirius stands low and twinkles strongly reddish), or a misattribution in the textual history. No consensus, but a clear exclusion.

Color as meaning
Alchemy · The color stages

Western alchemy knew a fixed sequence of color stages in the Magnum Opus: nigredo (blackening), albedo (whitening), citrinitas (yellowing) and rubedo (reddening). The order ran from decomposition to completion — red was the goal, the end product.

The astrophysical color scale runs structurally similar and is entirely independent of it: blue-white means young and hot, red means either very low in mass or very far advanced. A red giant really is an end stage. That both systems put red at the end is coincidence — but a coincidence that shows why analogies are so seductive. Structural similarity is not evidence.

Chapter III · Interactive — the most important diagram in astronomy

Two axes, and the order falls out by itself

Around 1910 Ejnar Hertzsprung in Denmark and Henry Norris Russell in the United States independently plot the luminosity of stars against their spectral class. Both expect a scatter of points.

Hertzsprung-Russell diagram
Hover the points for details · filters below
Main sequence Giants Super-/hypergiants White dwarfs Evolutionary tracks Instability strip
The HR diagram

What appears instead is a pattern: the great majority of stars lie on a narrow diagonal band running from upper left to lower right — the main sequence. Above it sits a separate group of bright, cool objects: the giants. Below it, a handful of hot but extremely faint points: the white dwarfs.

The interpretation will take decades. But the diagram shows immediately that stars do not fill the parameter space evenly. There are permitted and forbidden zones.

How to read it

x axis — temperature, for historical reasons running backwards: hot on the left, cool on the right. Originally the spectral class O→M stood here.

y axis — luminosity in solar luminosities, logarithmic across about ten orders of magnitude.

Diagonals — lines of constant radius. They follow from L = 4πR²σT⁴ and run from lower left to upper right.

Position — reveals mass and evolutionary stage in a single point.

Not a timeline The HR diagram is not a time axis. A star does move through it, but neighboring points belong to entirely different stars. It is a snapshot of a population, not a biography.
The Roman numerals

Same color, a thousandfold difference

Size comparison at identical surface temperature (class M)

The classes are conventions along a continuum, not sharp categories. That is why intermediate designations exist, such as IV-V (Procyon A) or Iab-Ib. Remarkably, a red supergiant of class Iab has roughly the same surface temperature as an M dwarf of class V — but a hundred million times the volume. Classifying by temperature alone would therefore be not merely incomplete but misleading.

Luminosity classes I–VII
ClassNameCharacterExampleRadius (R☉)
0 / Ia⁺HypergiantThe most extreme luminosities known, at the stability limit; heavy mass lossRho Cassiopeiae, Eta Carinaeup to ~1,500
IaBright supergiantVery luminous, very extendedDeneb, Rigel~70–200
IabSupergiant (intermediate)Middle rungBetelgeuse~200–900
IbLess luminous supergiantLower end of the supergiantsPolaris, Antares (Iab/Ib)~40–150
IIBright giantBetween giants and supergiantsCanopus, Mira~10–100
IIIGiantThe most common post-main-sequence stage; core hydrogen exhaustedArcturus, Aldebaran, Pollux~10–100
IVSubgiantJust leaving the main sequenceProcyon A, Alnair~2–8
VDwarf (main sequence)Core hydrogen fusion — a star spends ~90 % of its life hereSun, Sirius A, Proxima~0.1–10
VI / sdSubdwarfMetal-poor, therefore below the main sequence; mostly Population IIKapteyn's Star~0.1–1
VII / DWhite dwarfBurnt-out core, degenerate matter, no fusionSirius B, Procyon B~0.008–0.02
Interactive · Applying it

Build yourself a star and read it back

G2 V
Yellowish-white main-sequence star
Spectral classG
Subclass2
Luminosity classV
Scale relative to the Sun (logarithmic)
Teff
≈ 5,800 K
Radius
≈ 1 R☉
Luminosity
≈ 1 L☉
Example
Sun

The values are rounded typical figures from standard calibration tables, not exact model calculations. Not every combination exists in nature — an O star of class VII, for instance, is physically impossible.

A third dimension: chemistry

Stars come in generations

In 1944, during the wartime blackout of Los Angeles — which gave Mount Wilson Observatory an exceptionally dark sky — Walter Baade establishes that the stars of the Andromeda galaxy fall into two clearly separate groups.

Population I

Metal-rich ([Fe/H] ≈ 0 to +0.5). Young stars in the galactic disk and in spiral arms, on nearly circular orbits. Formed out of gas already enriched by earlier generations of stars.

The Sun belongs here. Rocky planetary systems are practically confined to Population I — you need heavy elements to build them.

Population II

Populations & metallicity

Metal-poor ([Fe/H] from −1 to below −4). Old stars in the galactic halo and in globular clusters, on strongly eccentric, inclined orbits. Remnants of the Milky Way's first construction phase.

The most metal-poor star known, SMSS J031300.36−670839.3, contains less than one ten-millionth of the solar iron fraction.

Population III

Metal-free. The very first stars, formed from the pure hydrogen-helium gas of the early universe. Without metals a gas cloud cools less efficiently, which is why models predict very large masses — in some cases several hundred solar masses.

Not yet detected directly. Such stars would have been so short-lived that none can have survived. There are candidate signatures in high-redshift JWST observations, but no confirmation.

Metallicity — the notation

Populations & metallicity

[Fe/H] = log₁₀(NFe/NH)star − log₁₀(NFe/NH)Sun A logarithmic measure relative to the Sun. [Fe/H] = 0 means “like the Sun”, −1 means a tenth of the solar iron fraction, −3 a thousandth. Metallicity is the best available indicator of the age of a stellar population, because the interstellar medium is enriched with every generation of stars.

Interpretive layer · Genealogy

The division into populations introduces an idea into astronomy that had previously belonged to biology and historiography: descent. Stars did not all form at once and are not interchangeable; they stand in a relationship of kinship that can be read off their composition — the way the age of a rock is read from sediment layers.

With that the cosmos acquires a history in the strong sense: a directed, irreversible sequence in which every state is the condition of the next. The Aristotelian notion of an eternal, unchanging heaven — European doctrine for more than 1,800 years — is thereby not merely refuted but turned into its exact opposite.

Chapter IV — Birth

Everything begins in an almost perfect vacuum

Stars form in giant molecular clouds — accumulations of molecular hydrogen (H₂) and dust, 15 to 600 light years across and with masses up to a few million solar masses.

Their density is typically 100 to 1,000 particles per cubic centimeter. For comparison: the best vacuum achievable in a laboratory is about a hundred times denser. These clouds are emptier than anything we can produce on Earth — and still the densest places in interstellar space.

Molecular cloud · collapse
Click inside the cloud
Molecular clouds

They are extremely cold: 10 to 20 kelvin. That is exactly what makes them unstable, because cold gas has little internal pressure. Once a region exceeds the Jeans mass, gravity wins against gas pressure and the collapse begins.

Jeans criterion (James Jeans, 1902)

MJ ∝ T3/2 · ρ−1/2

The colder and the denser a cloud is, the less mass it takes to collapse. That is why stars form only in the coldest places in the galaxy.

Interpretive layer · Cosmogonies

Practically every creation story on record begins with an unordered, unstructured initial state out of which order arises through separation. Greek: Chaos in Hesiod (Theogony 116) — not a “muddle” but literally the yawning abyss, an empty gap. Babylonian: the fresh water Apsu and the salt water Tiamat in the Enūma Eliš. Hebrew: tohu wa-bohu and the darkness over the primeval waters (Genesis 1:2).

The physical description shares the structure — diffuse initial matter, then condensation and separation — but not the mechanism. No will, no intention, no separation by an act of speech; only gravity against pressure, decided by an inequality. The narrative shape is similar, the explanation is something else entirely. Taking one as evidence for the other is the classic fallacy of such comparisons.

The first 100,000 years

From cloud core to protostar

The collapse is not an even falling-together. Because the cloud rotates and angular momentum is conserved, a disk necessarily forms — and perpendicular to it two jets. The same geometry turns up in young stars, active galactic nuclei and black holes.

Class 0

The embedded core

The central condensation is still entirely embedded in its dust envelope and invisible in visible light. It can be seen only in the submillimeter and far infrared. The envelope still holds more mass than the core.

Duration: ~10,000–100,000 years

Class I

Disk and jets

Collapse → disk formation → jets → ignition
Protostars

The envelope thins out and an accretion disk is clearly developed. Bipolar jets blow material out at hundreds of km/s and clear channels. Where they strike surrounding gas, glowing Herbig-Haro objects appear.

Duration: ~100,000–500,000 years

Class II

T Tauri stage

The envelope is largely gone, the star becomes optically visible and is surrounded by a protoplanetary disk. This is where planets form. The star itself is not yet fusing — it shines purely on the gravitational energy released by the collapse.

Duration: ~1–10 M years

Class III

Pre-main-sequence

Protostars

The disk has largely dispersed or been built into planets. The star keeps contracting until its core temperature reaches the ignition threshold. Then hydrogen fusion sets in, radiation pressure halts the collapse — and the star arrives on the zero-age main sequence (ZAMS).

Sun-like star: ~50 M years in total

Two names for the same youth

Young stars are restless

T Tauri stars — named after the prototype in the constellation Taurus — are pre-main-sequence stars of less than about two solar masses. They are irregularly variable, show strong emission lines (Hα above all), enormous starspots that can cover a quarter of the surface, and X-ray flares orders of magnitude stronger than the present Sun's.

In the HR diagram they lie above the main sequence: they are larger and brighter than they will later be, because they are still contracting. Their path there first follows the nearly vertical Hayashi track (cooling at almost constant temperature, fully convective), then, for more massive stars, the leftward Henyey track.

T Tauri · disk + jets
T Tauri & Herbig Ae/Be

Herbig Ae/Be stars are the more massive counterpart: 2 to 8 solar masses, spectral class A or B, likewise not yet on the main sequence. Because massive stars run through the pre-main-sequence phase very quickly, they are considerably rarer to observe. Above about 8 solar masses this phase practically no longer exists — such stars ignite while they are still accreting.

FU Orionis outbursts Some young stars increase their brightness a hundredfold within a few months and then stay bright for decades. The cause is presumably a sudden instability in the accretion disk that lets a surge of material fall onto the star. The young Sun probably went through such episodes as well.
The boundary

At 0.08 solar masses everything is decided

Whether a collapsing ball of gas becomes a star hangs on a single number. It is not a convention but follows directly from quantum mechanics.

Drag the slider through the three regimes
Mass of the object1.0 M☉
Ignition
≈ 0.08 M☉
Hydrogen burning limit — about 80 Jupiter masses (at solar metallicity; metal-poor objects need ~0.09 M☉)
≈ 3 M K
Core temperature at which the pp chain ignites efficiently

Below 0.08 M☉ a quantum-mechanical effect intervenes: the electrons in the core become degenerate. By the Pauli principle no two electrons may occupy the same quantum state — and out of that comes a pressure that does not depend on temperature. This degeneracy pressure halts the contraction before the core gets hot enough. The object cools forever, having never ignited.

≈ 13 MJup
Lower limit for deuterium fusion — the conventional boundary between planet and brown dwarf
Ignition
Interpretive layer · The threshold

Nature offers few such clear examples of a genuine threshold: at 0.079 solar masses an object cools quietly for trillions of years. At 0.081 it ignites and shines. No transition, no intermediate form, no gradual approach — the difference comes from an inequality between two pressures.

Aristotle would have called that a change of form, not a mere more or less. Modern physics knows many such phase transitions, but rarely one whose consequence is so total: shining or not shining, for the object's entire existence. The difference between a star and a failed star is about two percent of mass.

The failed stars

Neither star nor planet: brown dwarfs

In 1962 Shiv Kumar predicts theoretically that there should be objects too low in mass to ignite. In 1975 Jill Tarter coins the name brown dwarf — although they are in truth neither brown nor dwarfs in the classical sense. The first unambiguous detection only comes in 1995, with Gliese 229 B and Teide 1.

Brown dwarfs do not fuse ordinary hydrogen, but above about 13 Jupiter masses the core temperature suffices briefly to fuse deuterium (heavy hydrogen), and above about 65 Jupiter masses to burn lithium as well. Both are tiny reservoirs, exhausted within a few million years.

That yields an elegant test: the lithium test. An object that still shows lithium in its atmosphere has never exceeded about 65 Jupiter masses — a genuine star would long since have destroyed its lithium and burned it evenly through convection.

Size comparison
Jupiter · brown dwarf · M dwarf · Sun
Luhman 16 · 6.5 ly · nearest BD pair WISE 0855 · 7.3 ly · ~285 K Gliese 229 B · first detection 1995
Brown dwarfs
A curious detail Brown dwarfs are all roughly Jupiter-sized, almost regardless of their mass. More mass does not mean more volume, only higher density — because degeneracy pressure scales differently from gas pressure. A brown dwarf of 70 Jupiter masses is even slightly smaller than Jupiter.
Chapter V — The main sequence

Two ways to turn hydrogen into helium

Proton-proton chain

dominant below ≈ 1.3 M☉ · core temperature ≲ 17 M K

Four protons fuse into a helium nucleus through several intermediate steps. The first step is the bottleneck: two protons have to fuse and, at the same moment, one of them must decay into a neutron via the weak interaction. The probability of that is so absurdly small that a single proton in the solar core waits billions of years on average.

pp chain
Click to switch process
pp chain and CNO cycle

That, precisely, is why the Sun burns slowly instead of exploding — and why there was time for evolution at all.

CNO cycle

dominant above ≈ 1.3 M☉ · core temperature ≳ 17 M K

Carbon, nitrogen and oxygen act as catalysts: they take up four protons one after another, eject a helium nucleus, and end up unchanged. The net result is the same as in the pp chain, but the rate depends extremely steeply on temperature — roughly as T¹⁵ to T²⁰.

That steepness has consequences: the energy output concentrates in a tiny core region, which therefore becomes convective. Massive stars consequently have a convective core and a radiative envelope — exactly the reverse of the Sun.

Predicted in 1938/39 independently by Hans Bethe and Carl Friedrich von Weizsäcker. Bethe received the Nobel Prize for it in 1967.

Mass defect When four hydrogen nuclei fuse into one helium nucleus, about 0.7 % of the mass is converted into energy (E = mc²). The Sun turns some 4.3 million tonnes of mass into radiation every second — against a total mass of 2 · 10²⁷ tonnes, that amounts to less than a tenth of a percent over its entire life so far.
Interactive · The most important relation in stellar physics

A single parameter decides almost everything

Mass-luminosity relation

L ≈ L☉ · (M/M☉)3.5

1.0 L☉
Luminosity
10 bn yr
Time on the main sequence
5,800 K
Surface temperature
G2 V
Spectral type
Luminosity and lifetime against mass · log-log
Initial mass1.0 M☉
Mass is destiny

A star of twice the Sun's mass does not shine twice as brightly but about eleven times as brightly. At ten times the mass, roughly 3,000 times.

Lifetime

t ≈ 10 bn yr · (M/M☉)−2.5

More mass means more fuel — but consumption rises far faster than the supply. A 20-solar-mass star has 20 times the fuel and burns it 36,000 times faster.

An approximation The exponent 3.5 holds only roughly in the range 0.5–20 M☉. For very low-mass stars it is closer to 2.3; for very massive ones it drops toward 1, because radiation pressure dominates there (the Eddington limit).
Interpretive layer · The measure of things

“What burns fast and bright burns briefly” is a platitude that applies to stars literally — and with an exponent, at that. The temptation to draw a lesson for life from it is an old one: in the Iliad, Achilles chooses between a long, unremarkable life and a short one with undying fame.

The analogy does not hold, though. The star chooses nothing; its initial mass is the accidental product of turbulent fragmentation in a gas cloud. The fascination lies rather in the opposite: that one of the oldest human narrative figures finds an exact quantitative counterpart in the sky which has nothing to do with meaning. The metaphor is accurate. The conclusion drawn from it is not.

The quiet majority

The stars that own the future

Red dwarfs — M dwarfs on the main sequence — make up about three quarters of all stars. They are so frugal that their lifetime makes every other timescale in the universe look ridiculous.

The reason is their complete convection: below about 0.35 M☉ the whole star mixes. Fresh hydrogen is continually delivered into the core and helium carried back out. A Sun-like star can use only about 10 % of its hydrogen; a red dwarf nearly 100 %.

M dwarf · flare activity
Light curve below
≈ 1012 years
Calculated lifetime of the smallest M dwarfs — 70 times the present age of the universe
Red dwarfs

Their great drawback: flares. Their strong, twisted magnetic fields discharge in outbursts that can double or tenfold the star's total brightness within minutes. Such stars are called UV Ceti variables or flare stars. For planets in the narrow habitable zone — which around such faint stars lies only a few million kilometers out — this is a serious problem: atmospheric erosion and hard ultraviolet radiation.

0
Red dwarfs that have died anywhere in the universe so far
Proxima Centauri · M5.5 Ve · 4.25 ly Barnard's Star · M4 V · 5.96 ly TRAPPIST-1 · M8 V · 40.7 ly · 7 planets
Red dwarfs
Interpretive layer · Deep time

There is a strange asymmetry here: the universe is 13.8 billion years old, but the era of the red dwarfs has barely begun. On our present understanding we live in an extremely early epoch — comparable to the first seconds of a human life.

In 1979, in “Time Without End”, the physicist Freeman Dyson tried to take these spans seriously and ask what life and consciousness might mean in an ever-cooling universe. It is one of the few philosophical works that treats timescales of 10¹⁰⁰ years not as rhetoric but as something to calculate. Most of the stars that will ever exist do not exist yet.

Freeman Dyson, “Time Without End: Physics and Biology in an Open Universe”, Reviews of Modern Physics 51, 1979
At the upper end

Is there an upper limit to stellar mass?

Yes — and it has a name. The Eddington limit describes the luminosity at which outward radiation pressure exactly balances inward gravity. Exceed it and the star blows its own outer envelope away. It cannot grow heavier, because it pushes away the very matter it would have to grow from.

For a long time about 150 solar masses was taken as the practical upper limit. That figure was revised by observations of the cluster R136 in the Large Magellanic Cloud.

R136a1 — the most massive star known

Eddington limit
Radiation pressure against gravity
The giants of the main sequence
Mass
≈ 196 M☉ (range 169–230)
Temperature
≈ 46,000 K
Luminosity
≈ 4.7 M L☉
Type
WNh (Wolf-Rayet with residual hydrogen)
Distance
≈ 163,000 ly (Tarantula Nebula, LMC)
An important correction from 2022 Earlier work quoted up to 315 solar masses for R136a1. A study using the Zorro speckle instrument at the Gemini South telescope (Kalari et al. 2022) resolved the object more sharply and arrived at about 196 M☉. The reason for the earlier overestimate: what looked like one star was partly the blended light of several neighboring objects. A lesson in the limits of resolving power.

Stars like this lose up to 10⁻⁵ solar masses per year to their stellar wind — ten whole solar masses in a million years. They die with considerably less mass than they had. Their lifetime is only a few million years.

The special cases of the main sequence

Stars that break ranks

Not every star fits neatly into a scheme. A whole series of subtypes arises from rotation, magnetic fields or diffusion processes in the atmosphere — effects that distort the surface chemistry without the star being unusual inside.

Be stars

B stars with emission lines (hence the “e”). They rotate so fast — often above 80 % of breakup speed — that material escapes at the equator and forms a decretion disk. That disk produces the emission lines.

The disk can disperse and re-form over years — the same star then alternates between “B” and “Be”. Achernar is so strongly flattened that its equatorial diameter is about half again its polar diameter.

Ap / Bp stars

Chemically peculiar, with extremely strong, ordered magnetic fields — over 3 tesla in some cases, ten thousand times the Sun's average field. These fields suppress convection, so elements can separate out by diffusion.

The result is absurd enrichments in patches: strontium, chromium, europium — in places by a factor of 10,000 above normal. Since the star rotates, the spectrum changes periodically (the oblique rotator model).

Am stars

Chemically peculiar stars

Metal-rich A stars. They rotate strikingly slowly — usually because tides in a close binary have braked them. Without rotation and without convection, heavy material settles out in the quiet atmosphere while radiation pressure drives other elements upward.

Notably: calcium and scandium are underabundant, rare earths overabundant. A pure surface effect — the stellar interior is perfectly normal.

Blue stragglers

In old clusters, where every massive star should long since have died, one finds hot blue main-sequence stars — sitting above the turnoff point, where nobody ought to be left.

The explanation: they are not original but arose from the merger of two stars or from mass transfer off a companion. Cosmic rejuvenation by cannibalism.

Subdwarfs (sdB, sdO)

Hot, compact stars of about half a solar mass that are already fusing helium in the core but have lost almost all of their hydrogen envelope — usually to a close companion.

They are the main reason for the UV excess of old elliptical galaxies: although only old, red stars should be there, these galaxies radiate in the ultraviolet.

Lambda Boötis stars

The opposite of the Am stars: A stars with strikingly few heavy elements at the surface, while carbon, nitrogen, oxygen and sulfur stay normal.

The common explanation: the star accretes gas from a surrounding disk out of which the dust has already been filtered — the heavy elements are locked in dust grains that do not come along. Only about 2 % of all A stars.

Chapter VI — Variable stars

Stars that breathe

In the HR diagram there is a narrow, almost vertical band in which stars cannot be stable: the instability strip. Anything that wanders through it on leaving the main sequence starts to pulsate — expanding, contracting, again and again, with a period of hours to years.

The engine is called the kappa mechanism (κ stands for opacity, the opaqueness of the stellar material). At a certain depth there is partly ionized helium. Compress that layer and the energy supplied goes not into heating but into further ionization. Opacity rises with it — the layer becomes less transparent, dams up radiation and is pushed outward.

Kappa mechanism
Radius, temperature and light curve
The instability strip

As it expands it cools, the ions capture electrons, opacity falls, the dammed energy escapes — and the layer drops back. A self-sustaining cycle. The star becomes a heat engine driving a mechanical oscillation.

δ Cephei stars

Why a narrow band? The helium ionization zone has to lie at exactly the right depth. If the star is too hot, the zone sits too far out and holds too little mass to accomplish anything. If it is too cool, convection takes over and carries the energy away before pressure can build. The strip is the narrow zone in between.

Periods of 1–100 days. Massive supergiants, Population I. The cosmic distance gauges.

RR Lyrae stars

Periods of 0.2–1 day. Old, metal-poor stars of the horizontal branch. Nearly uniform luminosity ⇒ ideal standard candles in globular clusters.

δ Scuti stars

Periods from minutes to a few hours, small amplitudes. They sit where the instability strip crosses the main sequence.

ZZ Ceti stars

Pulsating white dwarfs, with periods of seconds to minutes. Through asteroseismology they allow a look inside stellar corpses.

1908 / 1912 · Measuring the universe

How Henrietta Swan Leavitt made the universe measurable

Leavitt, another of the Harvard Computers, studies variables in the Small Magellanic Cloud. Because all these stars are practically the same distance away, a difference in apparent brightness corresponds to a real difference in luminosity.

In 1908 she finds, and in 1912 refines, a strictly linear relationship: the longer a Cepheid's pulsation period, the brighter it is. From one easily measured quantity — the period — the absolute luminosity follows.

With that the decisive step is taken: if you know luminosity and apparent brightness, you know the distance. Cepheids become the first standard candle across galactic distances.

Period-luminosity relation
Hover over the points
Leavitt's ladder

What followed: in 1924 Edwin Hubble finds Cepheids in the Andromeda nebula and shows that it lies far outside the Milky Way — the end of the idea that our galaxy is the universe. In 1929 he derives the expansion of the universe from Cepheid distances and redshifts. Both results rest entirely on Leavitt's relation.

Type I and Type II In 1952 Walter Baade discovers that there are two Cepheid populations with different period-luminosity relations. Classical Cepheids (Pop I) are about four times brighter than W Virginis stars (Pop II) of the same period. The confusion had made every extragalactic distance too small by roughly a factor of 2 — overnight, the universe doubled in size.
Interpretive layer · Recognition

Leavitt died of cancer in 1921, aged 53. In 1925 the Swedish mathematician Gösta Mittag-Leffler wanted to nominate her for the Nobel Prize and had to learn that she had died four years earlier — the prize is not awarded posthumously. For a time her job title at Harvard was “assistant”; her hourly wage was 30 cents.

It is hard to overstate the reach of her discovery: without the period-luminosity relation there would be no measured size of the universe, no Hubble constant and no Big Bang cosmology. She supplied the yardstick in which everything else has been measured ever since.

When two stars orbit each other

The demon star and a paradox resolved

More than half of all Sun-like stars belong to a binary or multiple system. Among massive stars the fraction is higher still. The single star is rather the exception.

Algol (β Persei) is the best-known eclipsing binary: every 2 days, 20 hours and 49 minutes the fainter companion moves in front of the brighter star and the brightness drops within a few hours from magnitude 2.1 to 3.4 — a difference an attentive observer can see with the naked eye. John Goodricke explained the mechanism correctly in 1783.

The Algol paradox

Eclipsing binary
Orbit and light curve in real time
Algol & binary stars

In Algol the more massive star is still on the main sequence, while the less massive companion is already a subgiant. That contradicts stellar evolution head-on — the heavier star should evolve first.

Resolution: the star that is lighter today was originally the heavier one. As it swelled up it crossed its Roche limit — the surface within which material is still gravitationally bound to it. Over millions of years the companion siphoned off most of its mass. Direct evidence of mass transfer in close binaries.

Visual — separable in a telescope Spectroscopic — detectable only from line shifts Eclipsing — detectable only from the light curve Astrometric — detectable only from the orbital wobble
Algol & binary stars
Interpretive layer · Algol, the ghoul

The name comes from the Arabic raʾs al-ghūl — “head of the demon”. In the Greek tradition the star marks the severed head of Medusa, held in Perseus's hand. In Hebrew tradition it is associated with Lilith. A striking number of cultures attach something uncanny, changeable or malevolent to this particular star.

The open question: did they know it blinks? In 2013 a Finnish research group led by Lauri Jetsu argued that the ancient Egyptian Cairo Calendar (c. 1200 BC), which marks days as favorable or unfavorable, contains a statistically significant periodicity of 2.85 days — close to Algol's period of 2.867 days. If that were correct it would be the oldest known record of a variable star, some 3,000 years before Goodricke.

Interpretive layer · Algol, the ghoul

The thesis is contested. Criticism targets the statistical methodology and the reading of the passages. It is plausible enough to be discussed seriously, but far from established. The honest state of it: fascinating, unresolved.

Jetsu et al., Astrophysical Journal 773, 2013; critical replies followed
Chapter VII — When the fuel runs out

The core shrinks. The envelope explodes outward.

When the hydrogen in the core is used up, the energy supply fails — and with it the counter-pressure. The helium core contracts and heats up as it does. That heat ignites a hydrogen shell source: a thin burning layer around the dead core.

This shell releases more energy than the core did before. The envelope above it can no longer carry the energy flow away and swells enormously — to ten or a hundred times its size. Because the same luminosity is now spread over a far larger surface, the surface temperature drops. The star becomes large, cool and red.

Main sequence → red giant
Sequence starts automatically · click to restart
The red giant

In the HR diagram it travels up and to the right, climbing the red giant branch (RGB). It is brighter than ever before — and at the same time its end has already been set in motion.

The helium flash

In stars below about two solar masses the helium core becomes degenerate before it is hot enough to ignite. Degenerate matter has a treacherous property: its pressure does not depend on temperature. So when helium burning sets in at around 100 million kelvin, the core cannot expand and cool itself.

The consequence is a thermal runaway: the temperature rises, the fusion rate rises, the temperature rises further. Within minutes an amount of energy is released that briefly matches the luminosity of an entire galaxy.

None of this is visible from outside. All the energy is consumed in lifting the degeneracy of the core. Afterwards the core behaves like ordinary gas again, and the star settles down on the horizontal branch.

The future of the Earth In about 5 billion years the Sun will become a red giant with a radius of roughly one astronomical unit — about as far out as the Earth's present orbit. Whether the Earth will be swallowed is not certain: by then the Sun will have lost so much mass that the Earth's orbit moves outward. Tidal forces could more than cancel that effect, though. Either way the Earth will have stopped being habitable roughly a billion years earlier.
The asymptotic giant branch

The chemical factory of the universe

After helium burning the pattern repeats one level up: the core, now made of carbon and oxygen, contracts, and the star has two shell sources — an inner one for helium, an outer one for hydrogen. It ascends a second time, up the asymptotic giant branch (AGB).

This configuration is unstable. The helium shell ignites explosively at irregular intervals — thermal pulses, every 10,000 to 100,000 years. Each pulse stirs the star and carries fusion products from the depths to the surface. The process is called dredge-up.

Thermal pulse
Double shell burning and dredge-up
AGB, s-process, C and S stars

Between the pulses the s-process runs (slow neutron capture): atomic nuclei slowly capture single neutrons and convert step by step into heavier elements. This is how about half of all elements beyond iron come about — strontium, yttrium, zirconium, barium, lead.

The pulses bring these elements to the surface and the strong stellar wind releases them into space. AGB stars are the most important supplier of cosmic dust — and hence of the material for future planets.

M → S → C

With every dredge-up the carbon content of the atmosphere rises. The star runs through a sequence: first M (oxygen in excess, TiO bands), then S (C/O ≈ 1, ZrO bands), finally C (carbon in excess, soot formation).

AGB, s-process, C and S stars

An observable chemical evolution in one and the same star.

The technetium proof

In 1952 Paul Merrill finds technetium in S stars. Technetium has no stable isotope; the longest-lived decays within 4.2 million years.

Since these stars are billions of years old, the technetium cannot be original. It must be produced in the star right now and carried upward. The first direct proof that nucleosynthesis really does take place inside stars.

Mira variables

AGB stars pulsate with huge amplitudes and periods of 80 to over 1,000 days. The prototype Mira (ο Ceti) swings between magnitude 2 and 10 — a brightness ratio of 1:1,500.

David Fabricius discovered its variability in 1596. The name Mira (“the wonderful”) comes from Hevelius. Mira moves through the interstellar medium so fast that it trails a tail of material 13 light years long.

The largest stars

Betelgeuse — the most closely watched star in the world

Stars above about 8 solar masses do not end as giants but as red supergiants: objects with radii of several hundred to over a thousand solar radii. Put Betelgeuse where our Sun is and its surface would reach beyond the orbit of Mars.

From October 2019 to February 2020 Betelgeuse lost about two thirds of its brightness — the Great Dimming. There was worldwide speculation that it was about to go supernova. The explanation accepted today is more prosaic and is supported by VLT images: the star ejected an enormous bubble of gas which cooled, condensed into dust and blocked part of our view.

Red supergiant
Convection cells · mass loss
Class
M1–M2 Ia–Iab
Distance
≈ 500–650 ly (poorly determined)
Radius
≈ 640–900 R☉
Mass
≈ 14–19 M☉
Teff
≈ 3,600 K
Age
≈ 8–10 M years
Red supergiants: Betelgeuse
New: the companion star (2024/2025) Alongside its main period, Betelgeuse shows a puzzling cycle of about six years. In 2024 it was proposed that a close companion is responsible — jokingly dubbed “Betelbuddy”. In December 2024 the Alopeke instrument at the Gemini North telescope apparently achieved the first direct image; the results were published in 2025 (proposed name: Siwarha). On this account it is a low-mass companion of about 1.5 solar masses. That would make Betelgeuse the first red supergiant with a confirmed close stellar companion. Very recent research — confirmation by independent observations is still outstanding.
On the question “when will it explode?” Betelgeuse will end as a Type II supernova. The timing is narrowed to a window of about 100,000 years — “soon” in astrophysical terms, entirely indeterminate in human terms. It could have exploded 500 years ago with the light still on its way; it could explode in 100,000 years. At its distance it would be visible as bright as daylight for months, but completely harmless to Earth.
Red supergiants: Betelgeuse
Interpretive layer · Orion

Orion is probably the most universal constellation of all. Egypt: Sah, identified with Osiris, the god of rebirth. Babylon: SIPA.ZI.AN.NA, “the faithful shepherd of Anu”. Betelgeuse's Arabic name goes back to yad al-jauzāʾ — “hand of al-Jauza”; today's form arose from a misreading in medieval Latin.

On the Giza correlation: in 1989 Robert Bauval proposed that the three Giza pyramids map Orion's belt. The thesis is popular but largely rejected in Egyptology and archaeoastronomy — the angles do not match precisely, the brightness ratios do not fit the pyramid sizes, and there are no ancient Egyptian texts attesting such an intention. What is well documented, by contrast: the alignment of the pyramids on the celestial north pole, and the importance of Orion in the Pyramid Texts.

At the absolute limit

Three ways to become unstable

Yellow hypergiants

Extremely rare, extremely luminous stars in the temperature range between blue and red. Only about a dozen are known in the Milky Way.

They sit in a region called the Yellow Evolutionary Void, where stars can hardly be stable — they cross it quickly or swing back and forth. Rho Cassiopeiae ejects several percent of a solar mass in an outburst roughly every 50 years, its temperature dropping by more than 1,000 K within months.

Luminous blue variables

LBVs are the most unstable stars known. They lie so close to the Eddington limit that they shed large parts of their envelope at irregular intervals.

Stripping of the envelope: O star → LBV → WN → WC → WO
Hypergiants, LBVs, Wolf-Rayet

Eta Carinae went through the Great Eruption between 1837 and 1856: for a time the star became the second-brightest object in the night sky, hurling some 10 to 45 solar masses into space — more energy than some supernovae release, without dying. This is called a supernova impostor. The ejected material now forms the two-lobed Homunculus Nebula.

Wolf-Rayet stars

Discovered in 1867 by Charles Wolf and Georges Rayet in Paris. Stars that have lost their entire hydrogen envelope — what we see is the exposed fusion core.

Their spectra show broad emission lines instead of absorption lines: the stellar wind is so dense and fast (up to 3,000 km/s) that the extended envelope itself glows. Mass loss rates reach 10⁻⁵ M☉ per year.

WR 102 (type WO2) is one of the hottest stars known at about 210,000 K — while being only about half a solar radius across. Such stars are shortly before collapse; for WR 102 the estimate is less than 1,500 years.

On the “largest stars” Popular lists circulate record holders such as UY Scuti or Stephenson 2-18 with supposedly more than 2,000 solar radii. Those values are highly uncertain: they depend critically on the distance determination, and the stars have no sharp surface but blend gradually into their stellar wind. Several record claims were revised sharply downward by better Gaia parallaxes. “The largest star” is not a robust category but a snapshot of the current measurement error.
Chapter VIII — Final states · Interactive

How a star ends is decided by its mass alone

Drag the slider to choose the initial mass
Initial mass on the main sequence1.0 M☉

White dwarf

The mass boundaries are approximations and depend considerably on metallicity, rotation and the possible presence of a companion. Metal-poor stars lose less mass to their stellar wind and are therefore more likely to collapse into black holes. The transitions are gradual, and individual ranges — above 40 M☉, for instance — are the subject of active research.

The end for 97 % of all stars

A misleading name for a precise ending

At the end of the AGB phase the star ejects its entire outer envelope. The exposed core, at about 100,000 K, ionizes the outflowing gas with its ultraviolet light and makes it glow: a planetary nebula.

The name is a historical misunderstanding. William Herschel coined it in the 18th century because in the telescopes of the day these objects looked like round greenish disks — like planets. They have nothing to do with planets. The phase lasts only about 10,000 to 20,000 years — a blink in cosmic terms, which is why only some 3,000 are known in our galaxy.

Planetary nebula
The envelope is ejected and ionized
Planetary nebula & white dwarf

What remains is the white dwarf: the bare ash, a core of carbon and oxygen about the size of the Earth but with half to a full solar mass. A cubic centimeter of it weighs about a tonne. It fuses nothing any more; everything it radiates is purely stored residual heat.

The Chandrasekhar limit

The only thing holding out against gravity is electron degeneracy pressure — a pure consequence of the Pauli principle. In 1930, aged 19 and on the voyage to England, Subrahmanyan Chandrasekhar calculated that this pressure suffices only up to a mass of about 1.4 solar masses. Above that, the object collapses.

Planetary nebula & white dwarf

Arthur Eddington, then the most influential astrophysicist in Britain, dismissed the result publicly and condescendingly: “I think there should be a law of Nature to prevent a star from behaving in this absurd way.” Chandrasekhar was right and received the Nobel Prize in 1983 — 53 years later.

Sirius B — the first white dwarf known

In 1844 Friedrich Bessel infers an invisible companion from the wobbling proper motion of Sirius. In 1862 Alvan Graham Clark finds it while testing a new objective lens. Sirius B has about 1.0 solar masses at roughly 0.8 Earth radii and a surface temperature of about 25,000 K. It is hotter than Sirius A — and still some 10,000 times fainter.

Crystallization — and the black dwarf

Once a white dwarf cools below about 10,000 K, the carbon and oxygen nuclei arrange themselves into a crystal lattice. In 2019, Gaia data (Tremblay et al. 2019) showed for the first time a pile-up in the color-magnitude diagram matching exactly the latent heat released on solidification — the first confirmation of this phase, predicted back in 1960.

At the very end stands, in theory, the black dwarf: fully cooled, dark. The time needed for that exceeds the present age of the universe by many orders of magnitude — none exists yet.

The typology

Supernovae are classified by their spectrum, not by their mechanism

This is one of the most confusing classifications in astronomy — historically grown and lying crosswise to the actual cause. The basic question is simply: does the spectrum show hydrogen lines?

Type Ia — thermonuclear

no hydrogen · silicon line at 615 nm

A white dwarf in a binary system collects mass from its companion until it approaches the Chandrasekhar limit. Then the carbon in the degenerate core ignites abruptly throughout the entire volume — the star is torn apart completely and no remnant is left.

Classification tree by spectral features
Supernovae

Because the triggering mass is always nearly the same, the energy released is nearly the same too. Type Ia are therefore standard candles for cosmological distances. It was with them that the accelerating expansion of the universe was discovered in 1998 — Nobel Prize 2011.

Type II, Ib, Ic — core collapse

II: with H · Ib: no H, with He · Ic: neither

A massive star has fused all the way to iron. Iron is the end: fusing it consumes energy instead of supplying it. The core collapses within fractions of a second, the envelope falls in after it, rebounds off the newly forming neutron star and is flung outward.

The differences between II, Ib and Ic say nothing about the mechanism, only about how much envelope the star had already lost. Ib and Ic are exploded Wolf-Rayet stars. Subtypes of II (IIP, IIL, IIn, IIb) describe the shape of the light curve and narrow emission lines.

SN 1987A and the neutrinos In a core-collapse supernova about 99 % of the energy escapes as neutrinos — the spectacular display of light is only the fraction left over. For SN 1987A in the Large Magellanic Cloud, three detectors registered a total of 24 neutrinos on 23 February 1987, roughly three hours before the visible flare-up. It was the birth of neutrino astronomy and the first direct confirmation of the core-collapse model.
Matter at the limit

An object 20 kilometers across, heavier than the Sun

In core collapse, electrons and protons are merged into neutrons. What remains is an object about 20 km across with 1.4 to 2.3 solar masses. Its density matches that of an atomic nucleus: around 10¹⁷ kg/m³. A teaspoon of it would weigh about a billion tonnes on Earth.

Conservation of angular momentum makes them rotate extremely fast — the collapse acts like a figure skater pulling in their arms. And the magnetic field lines are compressed along with it: field strengths of 10⁸ tesla are normal.

Pulsar · lighthouse model
Rotation, beam cone and signal
Neutron stars & pulsars

If the magnetic axis is tilted against the rotation axis, the star acts like a lighthouse: a focused cone of radiation sweeps across the Earth once per turn. We receive a pulse of absurd regularity — that is a pulsar.

LGM-1

In November 1967 the doctoral student Jocelyn Bell in Cambridge finds a signal that recurs exactly every 1.3373 seconds. Nothing known was that precise. Half in jest the source was labeled LGM-1 — “Little Green Men”. Once a second source appeared elsewhere in the sky it was clear: a natural phenomenon. Today the object is called PSR B1919+21.

The 1974 Nobel Prize went to her supervisor Antony Hewish and to Martin Ryle. Bell Burnell was not included — to this day one of the most discussed cases in the prize's history. She has always commented on it diplomatically, and in 2018 she donated the entire 3 million dollars of her Breakthrough Prize to scholarships for under-represented physics students.

Millisecond pulsars

“Recycled” by taking up matter from a companion and spun up to as much as 716 rotations per second (PSR J1748−2446ad). The equator then moves at a substantial fraction of the speed of light. Their timekeeping rivals atomic clocks.

Neutron stars & pulsars

Magnetars

Neutron stars with magnetic fields up to 10¹¹ tesla — the strongest fields known in the universe. A magnetar at the distance of the Moon would wipe credit cards on Earth and deform atoms. They show irregular X-ray and gamma-ray bursts.

The TOV limit

By analogy with Chandrasekhar there is an upper mass limit for neutron stars too — the Tolman-Oppenheimer-Volkoff limit, on current estimates at about 2.2 to 2.3 M☉. It is not known exactly, because the equation of state of matter at these densities is unknown. Above it: collapse into a black hole.

Where classification ends

No longer a star — only geometry

If the core remnant exceeds the TOV limit, there is no known pressure left that could halt the collapse. What remains is not an object made of matter but a region of spacetime from which nothing can escape — bounded by the event horizon.

Its radius follows from a remarkably simple formula which Karl Schwarzschild derived as early as 1916 — as a soldier on the Eastern Front, a few months after Einstein's field equations and shortly before his own death:

rs = 2GM / c²

For the Sun that would be 2.95 km, for the Earth 8.9 millimeters. A stellar black hole of 10 solar masses has a horizon radius of about 30 km.

Event horizon · photon sphere
Gravitational lensing of the accretion disk
Black holes

A black hole can be described completely by three quantities: mass, angular momentum and electric charge. Everything else — the composition, the history, the structure of the original star — can no longer be determined from outside. This is called the no-hair theorem. The classification this lecture has unfolded ends here completely.

M87* · first image 2019 · 6.5 bn M☉ Sgr A* · image 2022 · 4.3 M M☉ Gaia BH3 · 2024 · 33 M☉ · only 2,000 ly
Interpretive layer · The limit of description

The no-hair theorem has a peculiar consequence: a black hole formed from a star is fundamentally indistinguishable from an equally heavy one formed from antimatter or from encyclopedias. Individuality is not lost because we measure badly, but because it physically ceases to exist.

This is exactly what ignites the information paradox: quantum mechanics demands that information never disappear. Stephen Hawking initially claimed the opposite, revised his position in 2004 and lost a famous bet over it. The problem is still not conclusively solved.

For classification this means: at the end of every typology stands an object that undercuts every typology. Three numbers, and the entire history of a star is gone.

17 August 2017

Where the gold comes from

The elements up to iron form in stars. Half of the heavier ones form in the s-process on the AGB. For the other half — gold, platinum, uranium, rare earths — you need a neutron flux no ordinary star can supply: the r-process (rapid neutron capture).

On 17 August 2017 the LIGO and Virgo detectors registered a gravitational wave signal lasting 100 seconds — GW170817, the merger of two neutron stars about 130 million light years away.

1.7 seconds later the Fermi space telescope caught a short gamma-ray burst from the same direction. Within eleven hours telescopes around the world had identified the source optically. Some 70 observatories on every continent and in orbit observed the same event.

The afterglow — the kilonova — showed exactly the spectral signature that models predict for freshly created r-process elements. Estimates suggest several Earth masses of gold and platinum were made in that single event.

It was the birth of multi-messenger astronomy: for the first time the same cosmic event was observed through gravitational waves and electromagnetic radiation. And it was the answer to a question that had been open since the 1950s.

Kilonova & the gold
An open question Whether neutron star mergers are the sole source of the r-process is unsettled. They are rare, and r-process elements are already found in very old, metal-poor stars — possibly too early for enough mergers to have taken place. Collapsars and magnetorotational supernovae are discussed as additional channels. The contribution of the individual channels is the subject of current research.
Interpretive layer · The metal of kings

In practically every high culture gold was the material of the sacred and of the ruler — among other reasons because it does not corrode and therefore counted as imperishable. In alchemy its production was the declared goal; the seven classical metals were firmly assigned to the seven planets, gold to the Sun.

The pairing gold ↔ Sun was astrophysically wrong: gold does not form in ordinary stars. But the underlying intuition — that this metal has a cosmic origin and is connected with the heavenly bodies — has turned out to be more accurate than any alchemist could have guessed. Every gold atom on Earth is older than the Sun and comes from a catastrophe for which there is no terrestrial analogue.

Kilonova & the gold
Interpretive layer · The metal of kings

One should not conclude from this that alchemy was “basically right”. It had an analogy, not an explanation. But it is remarkable how much less likely the actual answer is than any mythology anyone thought up.

Chapter IX — The cycle · Interactive

The periodic table, sorted by origin

Every element has a birthplace. Hover over the cells — the color shows which astrophysical process mainly produced that element.

Big Bang nucleosynthesis Cosmic rays (spallation) Dying low-mass stars (AGB, s-process) Exploding massive stars Exploding white dwarfs (SN Ia) Merging neutron stars (r-process) Made artificially
The origin of the elements

Presentation follows the scheme of Jennifer Johnson (Ohio State University), based on nucleosynthesis model calculations. Each assignment names the dominant channel; most elements form by several routes at once, and the shares are in places substantially model-dependent. The contribution of neutron star mergers relative to other r-process sources in particular is under active investigation.

The cycle closes

The elements in your body have an address

The cycle of matter
Cloud → star → nucleosynthesis → return → cloud
≈ 9.5 % of your mass
Hydrogen
Formed in the first three minutes after the Big Bang. These atoms are 13.8 billion years old and older than any star that has ever existed.
≈ 65 % of your mass
Oxygen
Produced in the helium burning of massive stars, flung into space by core-collapse supernovae.
Stardust
≈ 18.5 % of your mass
Carbon
From the triple-alpha process in red giants; distributed mostly by the stellar winds of dying AGB stars.
≈ 1.5 % of your mass
Calcium
From the oxygen and silicon burning of massive stars, shortly before they collapse.
Traces
Iron · iodine · zinc
Iron comes predominantly from Type Ia supernovae — exploding white dwarfs. The iodine your thyroid needs was probably produced in the merger of neutron stars.
Stardust
Interpretive layer · Carl Sagan

“The nitrogen in our DNA, the calcium in our teeth, the iron in our blood, the carbon in our apple pies were made in the interiors of collapsing stars. We are made of starstuff.

Carl Sagan, Cosmos, episode 9: “The Lives of the Stars”, 1980

The line has become a platitude and is often dismissed for that reason. That is premature — it is literally true and can be checked element by element. It is not a metaphor.

Interpretive layer · Carl Sagan

What is remarkable is what he does not claim. He says nothing about meaning, purpose or any special standing for human beings. He states a relation of origin — the same one that holds for a pebble, a bacterium and a car battery. That radical unspectacularness is precisely the point.

Two counter-voices are worth setting beside it. Steven Weinberg: “The more the universe seems comprehensible, the more it also seems pointless.” And Hans Blumenberg, who argued that the Copernican insult did not make humanity smaller but turned its position into a question for the first time. Both positions are compatible with the same facts.

Weinberg, “The First Three Minutes”, 1977 · Blumenberg, “The Genesis of the Copernican World”, 1975
The future

We live in the early days of the universe

The universe is 13.8 billion years old. That sounds like a lot. Measured against the timescales of stellar evolution, it is the beginning.

In 10¹² years

Star formation comes to a halt — the gas is used up. Only red dwarfs still shine. The universe turns red, then infrared. The era of degenerate matter begins.

In 10¹⁴–10¹⁵ years

Even the last and longest-lived M dwarfs go out. There are no shining stars left — only white dwarfs, neutron stars and black holes. The stellar age is over.

Logarithmic time axis · from today to 10¹⁰⁰ years
Deep time

In 10³⁴–10⁴⁰ years

If protons decay — which has not been demonstrated to this day — all white dwarfs and neutron stars dissolve. Experiments have so far only put a lower bound on the lifetime.

In 10⁶⁷–10¹⁰⁰ years

Even black holes evaporate through Hawking radiation. A stellar one in about 10⁶⁷ years, a supermassive one in up to 10¹⁰⁰. After that: an extremely dilute gas of photons and particles near absolute zero.

Degree of speculation The first two stages follow directly from well-tested stellar evolution theory and are solid. The third hangs on an unconfirmed prediction (proton decay). The fourth assumes that Hawking radiation is real — theoretically well founded, but never measured. And all of it holds only on the assumption that dark energy does not change. The further into the future, the less physics and the more extrapolation.
Epilogue
“I tell you: one must still have chaos in oneself to be able to give birth to a dancing star.”

Nietzsche meant it metaphorically and knew nothing of Jeans masses. That a star really does arise out of turbulent, disordered gas — and that without that turbulence there would be no fragmentation and hence no star — is a punchline he cannot have intended.

This journey has passed through 126 stations and close to a dozen systems of classification. Perhaps the real yield is this: the sky that humanity took for millennia to be the image of the eternal and unchanging is in truth a place of permanent destruction and renewal. Nothing up there is stable. Everything is transition.

And that is exactly why we exist. We are not a product of cosmic constancy but of cosmic unrest.

Friedrich Nietzsche · Thus Spoke Zarathustra, Prologue 5 · 1883
Epilogue
Interpretive layer · Two closing words

“Observe the movements of the stars as if you were running their courses with them, and let your mind dwell constantly on the changes of the elements into one another. Such imaginings wash away the filth of life on the ground.”

Marcus Aurelius, Meditations VII, 47 (c. 175 AD)

By “change of the elements” the Roman emperor meant the Stoic doctrine of four elements, not nucleosynthesis. His image has nonetheless become more precise, not less true: the elements really do change into one another. It just takes billions of years and happens at a hundred million kelvin.

Interpretive layer · Two closing words

“From this distant vantage point, the Earth might not seem of any particular interest. But for us, it's different. Consider again that dot. That's here. That's home. That's us.”

Carl Sagan on the “Pale Blue Dot” image, Voyager 1, 14 February 1990
Interactive · To close

Eight questions

0 / 8
Transparency

Sources, uncertainties, method

How facts were handled

All physical values come from standard astronomical references. Where sources disagree — which is the rule for temperature boundaries, stellar radii and distances — the customary range is given instead of a spurious single figure. Contested or very recent results are marked as such in orange notice panels.

The gold interpretive panels contain cultural-historical, mythological and philosophical material. They are meant as historical reporting and make no claim to scientific truth. Where popular claims have been refuted (Dogon/Sirius, the Giza-Orion correlation, the “Persian royal stars”), that is stated explicitly.

Known weak points

Sources & method

The classification tables map a continuum into boxes — the boundaries are conventions. The evolutionary tracks in the HR diagram are schematic and not computed from stellar models. The frequency figures apply to the solar neighborhood and cannot be transferred to the galaxy as a whole. The times given for the distant future are extrapolations with a growing degree of uncertainty.

Technical notes

This presentation is a single HTML file with no external dependencies. Every graphic and animation is generated at runtime via Canvas 2D and SVG — no image files are embedded. That lets it run offline, keeps it clear of licensing concerns and makes it embeddable in web projects. Star colors are computed from the Planck curve via CIE color-matching functions, not set by hand.

References used

Sources & method
  • 01 Morgan, Keenan & Kellman, “An Atlas of Stellar Spectra”, 1943
  • 02 Payne, “Stellar Atmospheres”, Harvard doctoral thesis, 1925
  • 03 Habets & Heintze, Astron. Astrophys. Suppl. 46, 1981 (calibration of spectral type ↔ Teff)
  • 04 Pecaut & Mamajek, ApJS 208, 2013, together with the continually maintained table “A Modern Mean Dwarf Stellar Color and Effective Temperature Sequence”
  • 05 Asplund et al., “The Chemical Composition of the Sun”, ARA&A 47, 2009
  • 06 Kalari et al., ApJ 935, 2022 (R136a1, revised mass ≈ 196 M☉)
  • 07 Kirkpatrick et al., ApJS 197, 2011 (introduction of spectral class Y)
  • 08 Luhman, ApJ 786, 2014 (WISE 0855−0714); Beiler et al., AJ 2023/24 (JWST temperature ≈ 285 K)
  • 09 Leavitt & Pickering, Harvard Circular 173, 1912 (period-luminosity relation)
  • 10 Chandrasekhar, MNRAS 95, 1935 (limiting mass of white dwarfs)
  • 11 Hewish, Bell et al., Nature 217, 1968 (discovery of pulsars)
  • 12 Abbott et al. (LIGO/Virgo), PRL 119, 2017 (GW170817)
  • 13 Tremblay et al., Nature 565, 2019 (crystallization of white dwarfs, Gaia)
Sources & method
  • 14 Montargès et al., Nature 594, 2021 (Betelgeuse, Great Dimming caused by dust)
  • 15 Howell et al., ApJL, 2025 (direct imaging of the Betelgeuse companion)
  • 16 Merrill, ApJ 116, 1952 (technetium in S stars)
  • 17 Baade, ApJ 100, 1944 (stellar populations I and II)
  • 18 Johnson, “Origin of the Elements in the Solar System”, Science 363, 2019
  • 19 Dyson, Rev. Mod. Phys. 51, 1979 (“Time Without End”)
  • 20 Van Beek, Current Anthropology 32(2), 1991 (refutation of the Dogon-Sirius thesis)
  • 21 Norris & Norris, “Why are there Seven Sisters?”, 2020 (arXiv:2101.09170)
  • 22 Jetsu et al., ApJ 773, 2013 (Cairo Calendar and Algol — contested)
  • 23 Carlson, Nature 318, 1985 (“A double-blind test of astrology”)
  • 24 Rochester & Koch, “Enūma Anu Enlil”, survey literature on Babylonian celestial science
  • 25 NASA/ESA/ESO mission documentation for Hubble, JWST, Gaia, WISE, Chandra, Event Horizon Telescope
Back to the beginning
A note on use If you want to add real photographs: NASA imagery is largely in the public domain, material from ESO and ESA/Hubble is under CC BY 4.0 (attribution required). Both can be included with legal certainty as long as the respective credit line travels with them. No external image is embedded in this file, deliberately, so that it works offline without restriction.
VAULT OF ASTROLOGY
Index
Prologue Title 1 / 60
to navigate I Index D Interpretive layer F Fullscreen
Close ✕

Contents