Mass-Luminosity Relation

The relation in one line

For main-sequence stars — those still fusing hydrogen in their cores — bolometric luminosity rises as a power of mass:

L / L☉ = (M / M☉)^α

with the exponent α depending on the mass regime. In the solar-mass range α is close to 4. A star with twice the Sun's mass is therefore about 2⁴ = 16 times as luminous; a star with ten times the Sun's mass is roughly 10⁴ = 10,000 times as luminous. The relation is one of the cleanest, most predictive scalings in all of stellar astrophysics, and it follows directly from how stars hold themselves up against gravity.

Eddington's 1924 derivation

The relation was given a structural explanation by Arthur Eddington in a 1924 Monthly Notices paper, a decade before the discovery of nuclear fusion. Eddington combined three equations for a steady, fully-radiative, gaseous star:

dP/dr   = − G M(r) ρ / r²              (hydrostatic equilibrium)
dT/dr   = − 3 κ ρ L(r) / (16π a c r² T³)  (radiative diffusion)
P       = (ρ k_B T) / (μ m_H)           (ideal gas)

Dimensional manipulation of these together with an opacity law κ = κ(ρ, T) gives, to leading order, L ∝ M³ / κ for a constant-opacity star. Eddington realised that the central temperature of a star scales with mass; combined with a Kramers-style opacity κ ∝ ρ T^(−3.5), the result steepens to roughly L ∝ M⁴. The exponent he calculated was 3.5; the modern value of ~4 differs because of mean-molecular-weight, mixing-length, and non-grey-opacity corrections that Eddington could not have computed without electronic computers.

What is remarkable about the derivation is its independence from nuclear physics. Eddington did not yet know that the energy source was fusion. The relation comes from the requirement that a self-gravitating gas ball be in equilibrium and that whatever heat it generates escape by radiation. The nuclear furnace simply has to keep up — it does not set the exponent.

The modern piecewise relation

A single power law does not fit the whole main sequence. Different opacity sources, energy-transport mechanisms, and pressure regimes dominate at different masses, so the exponent shifts. The standard modern partitioning (after Salaris & Cassisi, Demircan & Kahraman, Eker et al. and many others) reads:

Mass range (M☉)	Spectral type	α	Dominant physics
< 0.43	M dwarf	≈ 2.3	Fully convective, cool molecular opacity
0.43 – 2	K–F	≈ 4.0	Radiative core, Kramers opacity
2 – 20	A–B	≈ 3.5	Convective core, radiative envelope
> 20	O	≈ 3.0	Electron-scattering opacity, radiation-pressure significant
> 100	Wolf-Rayet, LBV	→ 1	Near-Eddington limit, mass loss dominates

The slopes reflect a real change in the underlying physics, not curve-fitting. Below 0.43 M☉ stars are fully convective and their opacities are dominated by molecules and H⁻ — a cooler, more strongly temperature-dependent opacity than Kramers'. Above 2 M☉ a convective core sets in and the energy-transport bookkeeping changes; above 20 M☉ the photon contribution to pressure cannot be ignored. At the very top of the main sequence the star runs up against its own Eddington luminosity, radiation drives mass loss, and the relation begins to saturate.

Worked numbers across the main sequence

Plugging numbers into the appropriate exponent for each regime:

Mass (M☉)	Spectral type	L (L☉)	T_eff (K)	t_MS
0.10	M8	0.0008	~2700	~10 Tyr
0.20	M5	0.005	~3200	~1 Tyr
0.50	K7	0.06	~4000	~120 Gyr
1.00	G2 (Sun)	1.0	5778	10 Gyr
1.50	F2	5	~7100	~3 Gyr
3.00	A0	54	~9500	~370 Myr
10.0	B2	10,000	~22,000	~30 Myr
50.0	O5	800,000	~42,000	~4 Myr
100	O3	~2 × 10⁶	~50,000	~3 Myr

The dynamic range spans nine orders of magnitude in luminosity for three orders of magnitude in mass — exactly what an exponent of three to four buys you. Above the Sun's mass the table shows what is sometimes called the unfair bargain: each factor of two in mass roughly trebles the lifetime cost, so a ten-solar-mass star burns through its supply 300 times faster per unit mass than the Sun does. Bright lives are short ones.

The lifetime corollary

Main-sequence lifetime is the most consequential corollary of the mass-luminosity relation. A star burns a roughly fixed fraction f ≈ 0.1 of its hydrogen as core fuel; the energy released per unit mass is the fusion efficiency η_nuc ≈ 0.007 c². The lifetime is the fuel divided by the burn rate:

t_MS = (f η_nuc M c²) / L
     ∝ M / L
     ∝ M / M^α
     = M^(1 − α)
     ≈ M^(-3)    (for α ≈ 4)

Anchoring at the Sun's 10 Gyr lifetime gives a rough rule of thumb:

t_MS ≈ 10 Gyr × (M/M☉)^(−2.5)   (general use)
t_MS ≈ 10 Gyr × (M/M☉)^(−3)     (α = 4 limit)

The exponent of −3 holds in the regime where α ≈ 4. For low-mass red dwarfs with α ≈ 2.3 the slope is shallower, t ∝ M^(−1.3), but combined with very low absolute luminosities the absolute lifetimes still reach the trillions of years. No M-dwarf that has ever been born has had time to leave the main sequence in the entire history of the universe.

Why exactly four, not three, in the solar range

The cleanest "physics-only" derivation gives L ∝ M^3. The extra factor of M comes from the temperature dependence of opacity. In the solar-mass radiative-envelope regime, Rosseland-mean opacity follows Kramers' law,

κ ∝ ρ T^(−3.5)

This opacity falls steeply with rising temperature: hotter plasma is more transparent to photons. As mass increases, central temperature rises, opacity drops, and radiation flows out more easily — boosting L for a given M beyond the bare M³ scaling. The exponent gets dragged from 3 toward 4. In the high-mass regime where electron scattering takes over, opacity becomes essentially constant (Thomson scattering does not care about T), the temperature kick disappears, and the slope returns to about 3. The mass-luminosity exponent is therefore a fingerprint of which opacity source dominates.

Opacity sources by mass regime

Regime	Dominant opacity	κ dependence	Effective α
M dwarfs	Molecules, H⁻, dust	complex, strongly T-dependent	2.3
Sun-like	Bound-free, free-free (Kramers)	κ ∝ ρ T^(−3.5)	4.0
Intermediate-mass	Kramers + electron scattering	transitional	3.5
High-mass	Electron (Thomson) scattering	κ ≈ const	3.0
Eddington-limited	Radiation-pressure-dominated	—	→ 1

Eclipsing binaries: how we measure it

The mass-luminosity relation is calibrated against detached double-lined eclipsing binaries — systems in which we can measure both the mass and the luminosity of each component without modelling assumptions. The recipe is straightforward in principle and exquisitely demanding in practice:

1. Orbital period and radial-velocity amplitudes. From spectroscopy we get the period P and the two RV semi-amplitudes K₁, K₂. The ratio K₁/K₂ gives the mass ratio q = M₂/M₁ exactly.
2. Inclination. If the system eclipses, the orbital plane is essentially edge-on (i ≈ 90°), so sin i ≈ 1 and the unknown projection factor drops out.
3. Total mass. Kepler's third law gives M₁ + M₂ from P, K₁, K₂, and sin i. Combined with q, the individual masses are determined to typically 1–3% precision.
4. Individual luminosities. From the depths of the primary and secondary eclipses one extracts the surface-brightness ratio. Combined with the distance (parallax from Gaia, or asteroseismic distance) and reddening, the individual bolometric luminosities follow.

The 1991 Andersen review listed about 45 such systems with high-precision data. The Henry & McCarthy 1993 calibration extended the relation to M-dwarf masses below 0.5 M☉ — the regime where α drops to 2.3 — using a combination of speckle interferometry and orbital astrometry. Gaia DR3 has since pushed the calibration sample into the thousands and reduced systematic errors below 1%.

Where the relation does not apply

• Pre-main-sequence stars. A protostar is contracting gravitationally and radiates the released potential energy. It is over-luminous relative to its main-sequence mass — sometimes by orders of magnitude — and lies above the relation on the Hayashi track and Henyey track.
• Giants and supergiants. Once a star exhausts core hydrogen, luminosity is driven by the core mass and the shell-burning rate, not the total mass. Red giants and AGB stars sit dramatically above the relation, by factors of 100 to 10,000.
• White dwarfs. Supported by electron degeneracy pressure rather than ideal-gas pressure, white dwarfs have luminosity set by residual cooling. The mass-luminosity relation does not apply at all — there is in fact an inverse mass-radius relation peculiar to degenerate objects.
• Brown dwarfs (M < 0.075 M☉). Below this mass, a contracting object never ignites stable hydrogen fusion. It cools forever, with a slowly falling luminosity. Brown dwarfs are not main-sequence stars and the L ∝ M^α scaling fails; their luminosity at a given age depends mostly on mass and age in a complicated, non-power-law way.
• Very massive stars (> 100 M☉). Approaching their own Eddington luminosity, these objects shed mass through radiation-driven winds. The relation flattens toward α ≈ 1 — luminosity becomes nearly linear in mass at the top end.

Consequences across astronomy

• Initial mass function and galactic light. Because L ∝ M^4 in the solar regime, even a modest tail of massive stars dominates a galaxy's bolometric output. A single O5 star outshines a million Sun-like stars, so a young cluster's light traces only its most massive members.
• Stellar lifetimes set chemical evolution. Heavy elements are released back into the interstellar medium when stars die. The mass-dependent lifetime determines what enrichment timescale each chemical element follows: α elements from O stars (Myr), iron from type Ia (Gyr), s-process elements from AGB stars (Gyr).
• Habitable-zone duration. A planet around a 1.5 M☉ F-star enjoys ~3 Gyr of habitability; the same planet around a 0.3 M☉ M-dwarf has ~10 Tyr. M-dwarfs dominate the universe's lifetime budget for complex life — if you ignore the radiation flare problem.
• Cosmic clock. The lifetime of a 1 M☉ star happens to be roughly the age of the universe. The Sun is therefore on the cusp: stars a little heavier than ours have already finished; stars a little lighter have barely budged.
• Star formation history of galaxies. O and B stars die quickly enough that their presence in a galaxy is a real-time indicator of ongoing star formation. UV-bright galaxies trace stars younger than ~50 Myr; the optical light traces stars over ~1 Gyr. This is the basis of every age-dating method in extragalactic astronomy.

Worked example: how long does Sirius live?

Sirius A is a 2.0 M☉ A1 main-sequence star with measured L ≈ 25 L☉. Using the relation with α = 4 we would predict L = 2⁴ = 16 L☉ — close but slightly under. Sirius A actually sits a bit above the α = 4 curve because the intermediate-mass regime (2–20 M☉) has α ≈ 3.5 plus mild metallicity dependence; with α = 3.6 the predicted L = 2^3.6 ≈ 12 L☉, again under but for a different reason — at A-star masses, age and metallicity matter, and Sirius A is fairly metal-rich and partway through its life.

For a back-of-envelope lifetime,

t_MS(Sirius A) ≈ 10 Gyr × (M/M☉)^(−2.5)
              = 10 Gyr × 2^(−2.5)
              = 10 Gyr × 0.177
              ≈ 1.77 Gyr

The actual estimated age of Sirius from isochrone fitting is ~240 Myr, and Sirius A is expected to leave the main sequence in another ~1 Gyr — consistent with our quick estimate. The 800 Myr companion Sirius B is already a white dwarf, having lived its main-sequence life as a more massive (~5 M☉) star and finished it long ago. The two stars beautifully illustrate the relation: the heavier sibling died first.

The brown-dwarf cutoff at 0.075 M☉

Below 0.075 M☉, the contracting protostellar core never reaches the central temperature (~3 × 10⁶ K) required to fuse hydrogen-1 into helium via the pp-chain stably. Such an object — a brown dwarf — fuses deuterium briefly (down to ~0.013 M☉) but then cools at a roughly constant radius, set by electron degeneracy. A brown dwarf's luminosity is therefore set by its current internal temperature and falls monotonically over Gyr, never reaching a steady main-sequence value.

The hydrogen-burning minimum mass is a genuine threshold, not a smooth crossover. Above 0.075 M☉ a star settles onto the main sequence and obeys L ∝ M^2.3; below it, no stable fusion equilibrium exists. The mass-luminosity relation has, in this sense, a sharp lower edge — a structural fact of the relevant nuclear physics.

Common pitfalls

• Quoting a single exponent. "L ∝ M^3.5" is true on average but wrong in every specific regime. For real predictions one must pick the regime and use the right α.
• Forgetting the lifetime corollary. The relation is sometimes presented as a curiosity about stellar brightness. Its main practical use is the lifetime scaling t ∝ M^(1−α) ≈ M^(−3), which sets every clock in stellar evolution and galactic chemical evolution.
• Applying it to giants or WDs. The relation is for main-sequence stars only. A red giant of 1 M☉ has L ≈ 100 L☉, not 1 L☉ — using the relation here gives nonsense.
• Ignoring metallicity. At fixed mass, a metal-poor star is hotter and somewhat more luminous than a metal-rich one. The 5% precision needed for high-resolution stellar work requires a metallicity-dependent calibration.
• Confusing bolometric with visible luminosity. A hot O star emits most of its light in the UV; a cool M dwarf emits most in the IR. The relation is for bolometric luminosity — the total radiated power — not for visible V-band or any single bandpass.
• Eddington's exponent was 3.5, not 4. The exact value Eddington calculated and the modern value differ by detailed opacity and mean-molecular-weight corrections he could not have computed. The relation is empirical-and-theoretical, and the modern α ≈ 4 in the solar regime came from numerical stellar models in the 1960s–70s.