M–sigma Relation and Black Hole–Galaxy Coevolution

Weigh the supermassive black hole at the centre of a nearby galaxy, then measure how fast the stars of its bulge jostle about — their line-of-sight velocity dispersion $\sigma$ — and the two numbers turn out to be locked together. Across four decades of black hole mass, from the Milky Way's four-million-solar-mass Sgr A\* to the six-billion-solar-mass giant in M87, the points fall on a single power law, $M_\bullet \propto \sigma^{4\text{–}5}$ , with a scatter of only $\sim 0.3$ dex. This is the $M_\bullet$ – $\sigma$ relation, discovered around 2000, and it is the tightest of all the black hole–host scaling laws. What makes it a genuine puzzle rather than a curiosity is a clash of scales. A black hole's gravity rules only inside its sphere of influence, $r_\mathrm{h} = GM_\bullet/\sigma^2$ , a region tens of parsecs across. The dispersion $\sigma$ , by contrast, is set by the whole bulge, a thousand times larger. Newtonian gravity gives the black hole no way to reach out and organise stars it cannot feel. Some other channel must couple the nuclear engine to the galaxy, and the leading candidate is feedback: during its active (AGN) phase the black hole pours radiation, winds, and jets into the surrounding gas, and that energy and momentum can halt its own growth. Two idealised feedback models — momentum-driven, giving $M_\bullet \propto \sigma^4$ , and energy-driven, giving $M_\bullet \propto \sigma^5$ — bracket the observed slope and, remarkably, fix the normalization from fundamental constants alone. The relation therefore sits at the crossroads of accretion physics, galactic dynamics, and cosmology, encoding the idea that black holes and their host bulges grew up together.

Key takeaways

The $M_\bullet$ – $\sigma$ relation is a tight empirical power law linking a galaxy's central black hole mass to its bulge stellar velocity dispersion, $M_\bullet \propto \sigma^{4\text{–}5}$ , with $M_\bullet \approx 2\times10^8\,M_\odot$ at $\sigma = 200\ \mathrm{km\,s^{-1}}$ .
Its tightness ( $\sim 0.3$ dex scatter) is the real surprise: the black hole's parsec-scale sphere of influence is a thousand times smaller than the bulge whose $\sigma$ enters the relation.
The bridge across those scales is AGN feedback — energy or momentum from accretion self-regulates growth, shutting off both black hole and bulge assembly at a critical condition.
Momentum-driven feedback (radiation pressure balancing gravity) predicts $M_\bullet \propto \sigma^4$ ; energy-driven feedback (a shocked wind retaining its heat) predicts the steeper $M_\bullet \propto \sigma^5$ .
Black holes are weighed dynamically — stellar and gas kinematics, megamasers, reverberation mapping — in roughly a hundred galaxies, anchoring the local black hole census.
The slope, normalization, and scatter are a primary calibration target for every cosmological galaxy-formation simulation that includes AGN feedback.
Key idea: a black hole grows by accretion until the energy or momentum output of the AGN can unbind the surrounding gas reservoir; that self-regulation ties $M_\bullet$ to the depth of the bulge potential well, encoded in $\sigma$ , yielding $M_\bullet \propto \sigma^{4\text{–}5}$ .

Where this lives today

The $M_\bullet$ – $\sigma$ relation is now a working instrument rather than a curiosity. Because a direct dynamical mass demands resolving $r_\mathrm{h}$ — feasible only out to tens of megaparsecs — the relation is the standard ladder for estimating black hole masses in more distant systems from a single spectroscopic $\sigma$ , and for calibrating reverberation-mapping masses of active galaxies, where the light-travel time across the broad-line region weighs the hole indirectly (the SDSS Reverberation Mapping and OzDES campaigns are extending this to thousands of quasars). The relation is also the primary calibration target for cosmological galaxy-formation simulations: codes such as IllustrisTNG, EAGLE, and SIMBA cannot reproduce the massive end of the galaxy population without AGN feedback, and they tune that feedback so that their black holes land on the observed $M_\bullet$ – $\sigma$ line with the right slope and scatter. On the frontier, the Event Horizon Telescope has now imaged the shadows of M87\ and Sgr A\, pinning two anchor points of the relation by a method entirely independent of stellar dynamics; pulsar-timing arrays (NANOGrav and partners) have detected a nanohertz gravitational-wave background consistent with a cosmic population of merging supermassive black hole binaries, whose demographics feed back on the coevolution picture; and JWST has begun finding surprisingly massive black holes in the first billion years — the "little red dots" and overmassive holes that appear to lie above the local relation, sharpening the open question of whether black holes led or lagged their bulges in the early universe.

The physical picture

Picture the black hole as a thermostat wired into its galaxy. Gas sinks toward the centre, feeds accretion, and lights up the nucleus; the AGN in turn heats and pushes on the surrounding gas. As long as the push is weak the gas keeps flowing in and the black hole keeps growing. But growth makes the push stronger — the Eddington luminosity is proportional to $M_\bullet$ — so at some critical mass the outflow finally overpowers the gravity holding the gas in, sweeps the reservoir out of the bulge, and starves the black hole. Growth stops. A deeper potential well (larger $\sigma$ ) sets a higher bar, and so demands a more massive black hole before the thermostat trips. That single sentence — bigger well, bigger black hole — is a $M_\bullet$ – $\sigma$ relation. The scale gap is what forces this feedback reading. The dispersion $\sigma$ is measured over kiloparsecs; the black hole's gravity reaches only parsecs. There is no direct gravitational handle by which the hole could tune itself to distant stars. The resolution is that $\sigma$ is not really about the stars at all — it is a proxy for the depth of the potential well, that is, for how tightly the gas the AGN must expel is bound. Feedback couples the nuclear engine to the galactic-scale gas reservoir, and the stars are simply the most convenient tracer of the same potential. The first figure below makes the scale gap and the force balance concrete for a fiducial black hole. The second is the payoff: the measured relation itself, three real galaxies spanning four decades of mass strung along one line.

Figure

A 2×10⁸ M⊙ black hole sits inside a sphere of influence r_h = GM•/σ² ≈ 20 pc — a thousand times smaller than the ~1 kpc bulge whose σ ≈ 200 km s⁻¹ it must somehow track. The bridge is feedback: the AGN push L_Edd/c must equal the bulge self-gravity 4 f_g σ⁴/G, and that balance forces M• ∝ σ⁴

The relation is not a loose trend with a best-fit line drawn through a cloud; it is a knife-edge correlation. That tightness is the single most important fact about it. A weak correlation could be explained away — many black holes and many bulges grew from the same gas supply, so of course the big ones go together. But scatter of only $0.3$ dex over four decades of mass means something is actively enforcing the relation, continually pulling outliers back onto the line. Feedback self-regulation is exactly such an enforcing mechanism. The second figure shows the relation with three galaxies whose black holes have been weighed by independent, gold-standard methods.

Figure

The M•–σ relation on log axes: black hole mass climbs as roughly the 4.4 power of bulge dispersion, log(M•/M⊙) = 8.3 + 4.4 log(σ/200 km s⁻¹). Three dynamically weighed anchors span the line — Sgr A* in the Milky Way (4×10⁶ M⊙, σ ≈ 105), the megamaser galaxy NGC 4258 (4×10⁷ M⊙, σ ≈ 115), and M87 (6×10⁹ M⊙, σ ≈ 324) — four decades of mass with only ~0.3 dex scatter

The core equations

The relation itself is stated as a straight line in log–log space: the logarithm of the black hole mass is a linear function of the logarithm of the dispersion, pivoted about a fiducial $\sigma = 200\ \mathrm{km\,s^{-1}}$ so that the intercept has a clean meaning. This is the form fitted to the data and quoted in catalogues.

\log_{10}\!\left(\frac{M_\bullet}{M_\odot}\right) = \alpha + \beta\,\log_{10}\!\left(\frac{\sigma}{200\ \mathrm{km\,s^{-1}}}\right) \qquad (1)

where: $M_\bullet$ — black hole mass; $\sigma$ — bulge line-of-sight stellar velocity dispersion; $\alpha \approx 8.3$ — intercept, so $M_\bullet \approx 2\times10^8\,M_\odot$ at $\sigma = 200\ \mathrm{km\,s^{-1}}$ ; $\beta \approx 4$ – $5$ — slope. Reading it: a $0.30$ dex change in $\log\sigma$ (a factor of two in $\sigma$ ) moves $M_\bullet$ by $\beta\times0.30 \approx 1.3$ – $1.5$ dex, more than a factor of twenty in mass. The slope is more transparently a power law: black hole mass goes as $\sigma$ raised to a power between four and five, and the observed scatter about that power law is small.

M_\bullet \propto \sigma^{\beta}, \qquad \beta \approx 4\text{–}5, \qquad \text{scatter} \approx 0.3\ \mathrm{dex} \qquad (2)

where: $\beta$ — the logarithmic slope that the two feedback models below aim to predict. Reading it: the two clean theoretical values are $\beta=4$ (momentum-driven) and $\beta=5$ (energy-driven), and the data sit between them. The scale of the puzzle is set by the sphere of influence — the radius inside which the black hole, rather than the bulge stars, dominates the gravitational potential. Equate the black hole's Keplerian speed at radius $r$ to the ambient dispersion and solve for $r$ .

r_\mathrm{h} = \frac{G M_\bullet}{\sigma^2} \qquad (3)

where: $r_\mathrm{h}$ — sphere-of-influence radius; $G$ — gravitational constant. Reading it: for $M_\bullet = 2\times10^8\,M_\odot$ and $\sigma = 200\ \mathrm{km\,s^{-1}}$ this is only $\approx 20\ \mathrm{pc}$ , which must be spatially resolved to weigh the hole directly — the reason direct masses exist only for nearby galaxies. The push in the feedback picture is set by the Eddington luminosity, the luminosity at which the outward force of radiation on ionised gas (via electron scattering) balances the inward pull of the black hole's gravity.

L_\mathrm{Edd} = \frac{4\pi G M_\bullet m_p c}{\sigma_T} \qquad (4)

where: $L_\mathrm{Edd}$ — Eddington luminosity; $m_p$ — proton mass; $c$ — speed of light; $\sigma_T$ — Thomson cross section. Reading it: numerically $L_\mathrm{Edd} \approx 1.26\times10^{38}\,(M_\bullet/M_\odot)\ \mathrm{erg\,s^{-1}}$ , and because $L_\mathrm{Edd}\propto M_\bullet$ the push grows in lockstep with the hole — the key to why growth is self-limiting. The bulge is modelled as a singular isothermal sphere, the simplest mass distribution with a flat rotation curve. Its enclosed mass grows linearly with radius and its circular speed is a fixed multiple of $\sigma$ .

M(r) = \frac{2\sigma^2 r}{G}, \qquad v_c^2 = 2\sigma^2 \qquad (5)

where: $M(r)$ — mass enclosed within radius $r$ ; $v_c$ — circular speed. Reading it: the linear $M(r)\propto r$ is what makes the gravitational force on a swept-up shell independent of radius, which is why the feedback balance closes into a clean power law. Setting the momentum-driven feedback force against the bulge self-gravity yields the critical black hole mass — the article's central quantitative result.

M_\sigma = \frac{f_g\,\sigma_T}{\pi G^2 m_p}\,\sigma^4 = \frac{f_g\,\kappa}{\pi G^2}\,\sigma^4 \qquad (6)

where: $f_g$ — gas fraction, $\approx \Omega_b/\Omega_m \approx 0.16$ ; $\kappa = \sigma_T/m_p$ — electron-scattering opacity. Reading it: the normalization is built entirely from constants and $f_g$ , and it evaluates to a few $\times10^8\,M_\odot$ at $\sigma = 200\ \mathrm{km\,s^{-1}}$ — no free amplitude was inserted by hand. Finally, empirically, the black hole carries a roughly fixed small fraction of the bulge stellar mass — the sibling scaling relation that the $M_\bullet$ – $\sigma$ law refines.

M_\bullet \approx 0.001\,M_\mathrm{bulge} \qquad (7)

where: $M_\mathrm{bulge}$ — bulge stellar mass. Reading it: the $\sim 0.1\%$ ratio is a useful mnemonic, but the $M_\bullet$ – $\sigma$ relation is tighter, because $\sigma$ tracks the potential depth that feedback actually regulates.

Derivation AThe isothermal bulge and its flat rotation curve

The goal of this first derivation is to justify the one piece of galactic dynamics that both feedback arguments lean on: the singular isothermal sphere, whose enclosed mass rises linearly with radius and whose circular speed is a constant multiple of the velocity dispersion. We want to show, starting from the isothermal density profile, that $M(r) = 2\sigma^2 r/G$ and $v_c^2 = 2\sigma^2$ , because it is precisely this linear mass growth that makes the gravitational weight on a driven shell independent of radius and lets the feedback balance collapse to a single power of $\sigma$ .

Step 1. Take the density profile of a singular isothermal sphere, $\rho(r) = \sigma^2 / (2\pi G r^2)$ , the self-consistent solution for an isothermal self-gravitating gas (or a stellar system with isotropic, radius-independent velocity dispersion $\sigma$ ). Why: this is the canonical bulge model in Binney and Tremaine — it reproduces the flat rotation curves galaxies actually show, so it is the right idealisation for the potential the AGN must fight.

Step 2. Integrate the density to get the enclosed mass, $M(r) = \int_0^r 4\pi r'^2\,\rho(r')\,\mathrm{d}r' = \int_0^r 4\pi r'^2 \cdot \dfrac{\sigma^2}{2\pi G r'^2}\,\mathrm{d}r' = \dfrac{2\sigma^2}{G}\int_0^r \mathrm{d}r' = \dfrac{2\sigma^2 r}{G}$ . Why: the $r'^2$ in the volume element cancels the $r'^{-2}$ of the density, leaving a constant integrand, so the mass piles up linearly — the feature that makes everything downstream tractable.

Step 3. Form the circular speed from $v_c^2 = GM(r)/r = G\cdot(2\sigma^2 r/G)/r = 2\sigma^2$ , a constant independent of radius. Why: a radius-independent circular speed is a flat rotation curve, confirming the model is realistic, and it fixes the numerical link $v_c = \sqrt{2}\,\sigma$ between the two speeds we measure.

Step 4. Record the consequence used later: the gravitational force on a thin shell of gas mass $M_g = f_g M(r)$ at radius $r$ is $F_\mathrm{grav} = GM(r)M_g(r)/r^2 = 4 f_g \sigma^4 / G$ , with the two factors of $r$ from the masses cancelling the $r^2$ below. Why: this constancy in $r$ is the mathematical heart of the feedback derivations — the outward push need only beat one fixed number, and that number scales as $\sigma^4$ . *The results $M(r)=2\sigma^2 r/G$ and $v_c^2=2\sigma^2$ are the standard singular-isothermal-sphere results; the shell weight $4f_g\sigma^4/G$ has dimensions of force and is radius-independent.*

Go deeper · Derivation B · 2 minMomentum-driven feedback and the sigma-to-the-fourth law

Derivation BMomentum-driven feedback and the sigma-to-the-fourth law

The goal here is to derive the shallower prediction, $M_\bullet \propto \sigma^4$ , by balancing the outward momentum flux of Eddington-limited radiation against the inward gravitational weight of a swept-up gas shell in the isothermal bulge. In the momentum-driven regime the shocked wind gas cools efficiently, so only its momentum — not its thermal energy — survives to push the ambient gas; setting the radiation momentum rate equal to the shell weight defines a critical black hole mass above which the gas is expelled and accretion shuts off. That critical mass is $M_\sigma$ , and its value emerges fixed by constants alone.

Step 1. Write the outward force as the momentum per unit time carried by the radiation, capped at the Eddington limit: $F_\mathrm{rad} = L/c \le L_\mathrm{Edd}/c = 4\pi G M_\bullet m_p / \sigma_T$ . Why: at and above the Eddington luminosity the black hole cannot push harder by shining brighter, so this is the maximum steady momentum the nucleus can deliver to the gas by single scattering.

Step 2. Write the inward force as the self-gravitational weight of the shell computed in the previous derivation, $F_\mathrm{grav} = GM(r)M_g(r)/r^2 = 4 f_g \sigma^4/G$ , independent of radius. Why: because the gas shell must be lifted out of the bulge's own potential, the relevant weight involves the total enclosed mass acting on the gas mass $M_g = f_g M(r)$ , and the isothermal profile makes it a pure function of $\sigma$ .

Step 3. Impose the critical condition — the outflow just overcomes gravity — by setting the two forces equal: $\dfrac{4\pi G M_\sigma m_p}{\sigma_T} = \dfrac{4 f_g \sigma^4}{G}$ . Why: below this the shell falls back and accretion continues; above it the shell is driven away, cutting off the fuel — so equality marks the mass at which the black hole regulates itself.

Step 4. Solve for the critical mass, $M_\sigma = \dfrac{f_g\,\sigma_T}{\pi G^2 m_p}\,\sigma^4$ , the fourth-power law with a normalization set only by $f_g$ and fundamental constants. Why: the black hole mass appears linearly on the left (through $L_\mathrm{Edd}\propto M_\bullet$ ) and $\sigma^4$ appears on the right, so isolating $M_\sigma$ hands back a clean $\sigma^4$ scaling with no adjustable amplitude.

Step 5. Insert cgs numbers — $\sigma_T = 6.65\times10^{-25}\,\mathrm{cm^2}$ , $m_p = 1.67\times10^{-24}\,\mathrm{g}$ , $G = 6.67\times10^{-8}$ , $f_g = 0.16$ , $\sigma = 2\times10^7\,\mathrm{cm\,s^{-1}}$ — to get $M_\sigma \approx 3.7\times10^8\,M_\odot$ , i.e. $\log_{10}(M_\sigma/M_\odot) \approx 8.6$ . Why: this lands within a factor of two of the observed intercept $\alpha \approx 8.3$ , so a two-line force balance reproduces both the slope and, astonishingly, the amplitude of a relation measured across a hundred galaxies. *Numerically, $M_\sigma = f_g\sigma_T\sigma^4/(\pi G^2 m_p) = 7.3\times10^{41}\,\mathrm{g} = 3.7\times10^8\,M_\odot$ ( $\log 8.56$ ) at $\sigma = 200\ \mathrm{km\,s^{-1}}$ , with slope $\beta = 4$ exact.*

Go deeper · Derivation C · 2 minEnergy-driven feedback and the steeper sigma-to-the-fifth law

Derivation CEnergy-driven feedback and the steeper sigma-to-the-fifth law

The goal of this derivation is the steeper prediction, $M_\bullet \propto \sigma^5$ , which applies when the fast nuclear wind shocks and the shocked gas cannot radiate its thermal energy efficiently. Now the retained heat, not just the momentum, does work sweeping up the interstellar medium, and balancing the wind's mechanical power against the rate at which the shell gains energy in the bulge potential adds one extra power of $\sigma$ . The physical switch between the two regimes is simply whether the shocked wind cools quickly (momentum-driven, $\sigma^4$ ) or holds its heat (energy-driven, $\sigma^5$ ).

Step 1. Estimate the wind's mechanical power. A wind launched near Eddington carries momentum $\dot M_w v_w \approx L_\mathrm{Edd}/c$ , so its kinetic luminosity is $\dot E_w = \tfrac12 \dot M_w v_w^2 \approx \tfrac12 (v_w/c)\,L_\mathrm{Edd} \equiv \eta\,L_\mathrm{Edd}$ with $\eta \sim 0.05$ . Why: fixing the wind by its Eddington momentum and then forming its energy is what carries the extra factor $v_w/c$ that ultimately steepens the slope.

Step 2. Estimate the rate at which the driven shell must gain energy to escape the isothermal potential: it needs energy of order $M_g\,\sigma^2$ delivered over a flow time $t_\mathrm{flow} \sim r/v_\mathrm{esc}$ with $v_\mathrm{esc}\sim\sigma$ , and since $M_g = f_g M(r) = 2 f_g \sigma^2 r/G$ the required power scales as $M_g\sigma^2/t_\mathrm{flow} \sim f_g\sigma^5/G$ . Why: the shell must be lifted out of a well of depth $\sim\sigma^2$ per unit mass on a timescale set by its own speed $\sim\sigma$ , and both the mass and the timescale bring in factors of $\sigma$ that combine to the fifth power.

Step 3. Balance supply against demand, $\eta\,L_\mathrm{Edd} \sim f_g\sigma^5/G$ , and substitute $L_\mathrm{Edd} = 4\pi G M_\bullet m_p c/\sigma_T$ to get $\dfrac{4\pi \eta G M_\bullet m_p c}{\sigma_T} \sim \dfrac{f_g\sigma^5}{G}$ . Why: this is the energy-conserving analogue of the momentum balance — the same self-regulation logic, but with power rather than force as the conserved currency.

Step 4. Solve for the black hole mass, $M_\bullet \sim \dfrac{f_g\,\sigma_T}{\eta\,G^2 m_p c}\,\sigma^5 \propto \sigma^5$ , one power of $\sigma$ steeper than the momentum-driven law. Why: because $L_\mathrm{Edd}\propto M_\bullet$ supplies one fewer power of $\sigma$ than the $\sigma^5$ energy demand — compared with the $\sigma^4$ force demand — the energy-driven balance is steeper by exactly the extra factor $\sigma/c$ relative to the momentum case. *The extra factor $\sigma/c$ relative to the momentum-driven balance is what converts $\sigma^4 \to \sigma^5$ ; the result is dimensionally consistent, and the reduced-radiative-loss premise underlies the $\sigma^5$ scaling.*

Limiting cases

Different feedback couplings and dynamical regimes predict distinct slopes and behaviours, and the same table doubles as a map of when each idealisation applies:

Case	Result	Meaning
Momentum-driven feedback	$M_\bullet \propto \sigma^4$	Radiation-pressure or single-scattering wind momentum balances bulge self-gravity; the shallower slope, normalization $M_\sigma = f_g\sigma_T\sigma^4/(\pi G^2 m_p)$ , applies when the shocked wind cools efficiently.
Energy-driven feedback	$M_\bullet \propto \sigma^5$	A fast wind shocks and retains its thermal energy; the steeper slope, favoured when the shock cannot radiate — common at high luminosity, and matching the high-mass end better.
Small scales, $r \ll r_\mathrm{h}$	$v_c(r) \to \sqrt{GM_\bullet/r}$	Inside the sphere of influence the black hole dominates the potential; stellar kinematics rise in a Keplerian cusp, enabling a direct dynamical mass.
Large scales, $r \gg r_\mathrm{h}$	$\sigma \approx \text{const}$	The bulge is nearly isothermal, so $\sigma$ is flat and insensitive to the black hole; this is the $\sigma$ that enters the relation.
No classical bulge (pure disk)	relation breaks down	Disk-dominated and bulgeless galaxies scatter off or fall below the relation — feedback needs a bulge-scale reservoir to regulate.

Common trapThe relation is with the bulge, not the whole galaxy — and the aperture matters

The single most common way to misuse the $M_\bullet$ – $\sigma$ relation is to feed it the wrong $\sigma$ . The correlation is with the classical bulge (or elliptical) velocity dispersion, not the disk, not the total galaxy, and not the dark halo. Pure disk galaxies, bulgeless spirals, and pseudobulges built by secular disk processes scatter off the relation or lie below it, and the correlation with total stellar mass or halo mass is much weaker with larger scatter. Even for a genuine bulge, $\sigma$ must be measured within a consistent aperture — conventionally the effective radius $R_e$ — because $\sigma$ falls with radius, so an inconsistent aperture silently shifts the normalization by tenths of a dex. Quoting a black hole mass from $\sigma$ without stating the aperture is quoting a number without its units.

Worked examples

1 · The sphere of influence, and why direct masses are rare (fully worked)

Estimate the sphere-of-influence radius of a $M_\bullet = 2\times10^8\,M_\odot$ black hole in a bulge with $\sigma = 200\ \mathrm{km\,s^{-1}}$ , and say how far away such a galaxy could be and still have its black hole weighed by resolving $r_\mathrm{h}$ with an 8 m telescope at $0.05''$ resolution.

Step 1. Convert to cgs: $M_\bullet = 2\times10^8 \times 1.989\times10^{33}\,\mathrm{g} = 3.98\times10^{41}\,\mathrm{g}$ , and $\sigma = 200\ \mathrm{km\,s^{-1}} = 2\times10^7\ \mathrm{cm\,s^{-1}}$ , so $\sigma^2 = 4\times10^{14}\ \mathrm{cm^2\,s^{-2}}$ .

Step 2. Apply equation (3): $r_\mathrm{h} = GM_\bullet/\sigma^2 = (6.67\times10^{-8})(3.98\times10^{41})/(4\times10^{14}) = 6.6\times10^{19}\ \mathrm{cm}$ . In parsecs ( $1\ \mathrm{pc} = 3.086\times10^{18}\ \mathrm{cm}$ ), $r_\mathrm{h} \approx 21\ \mathrm{pc}$ .

Step 3. For $r_\mathrm{h} = 21\ \mathrm{pc}$ to subtend the angular resolution $\theta = 0.05'' = 2.4\times10^{-7}\ \mathrm{rad}$ , the maximum distance is $d = r_\mathrm{h}/\theta = (6.6\times10^{19}\,\mathrm{cm})/(2.4\times10^{-7}) = 2.7\times10^{26}\ \mathrm{cm} \approx 90\ \mathrm{Mpc}$ . So this black hole can be dynamically weighed out to roughly 90 Mpc — the nearby universe only, which is exactly why direct masses number in the hundreds, not millions, and why the $M_\bullet$ – $\sigma$ relation is so valuable as a mass proxy for everything farther.

2 · Reproducing the normalization from constants (fully worked)

Use the momentum-driven critical mass, equation (6), to predict $M_\bullet$ at $\sigma = 200\ \mathrm{km\,s^{-1}}$ with $f_g = 0.16$ , and compare to the observed intercept.

Step 1. Assemble the pieces in cgs: numerator $f_g\,\sigma_T\,\sigma^4 = 0.16 \times (6.65\times10^{-25}) \times (2\times10^7)^4$ . Since $(2\times10^7)^4 = 1.6\times10^{29}$ , the numerator is $0.16 \times 6.65\times10^{-25} \times 1.6\times10^{29} = 1.70\times10^{4}$ .

Step 2. Denominator $\pi G^2 m_p = 3.1416 \times (6.67\times10^{-8})^2 \times (1.67\times10^{-24}) = 3.1416 \times 4.45\times10^{-15} \times 1.67\times10^{-24} = 2.34\times10^{-38}$ .

Step 3. Divide: $M_\sigma = 1.70\times10^{4}/2.34\times10^{-38} = 7.3\times10^{41}\ \mathrm{g}$ . In solar masses, $M_\sigma = 7.3\times10^{41}/1.989\times10^{33} = \mathbf{3.7\times10^{8}\,M_\odot}$ , i.e. $\log_{10}(M_\sigma/M_\odot) = \mathbf{8.56}$ .

Step 4. Compare: the observed intercept is $\alpha \approx 8.3$ ( $M_\bullet \approx 2\times10^8\,M_\odot$ ). The first-principles estimate agrees to within a factor of about 1.8 — extraordinary for a calculation with no fitted amplitude, and the strongest single piece of evidence that feedback, not coincidence, sets the relation.

3 · Predicting the Milky Way's black hole (a scatter check)

The Milky Way's bulge has $\sigma \approx 105\ \mathrm{km\,s^{-1}}$ . Predict $M_\bullet$ from the fiducial fit $\log(M_\bullet/M_\odot) = 8.3 + 4.4\log(\sigma/200)$ and compare to the measured mass of Sgr A\, $4.1\times10^6\,M_\odot$ .*

Step 1. Form the log ratio: $\log_{10}(105/200) = \log_{10}(0.525) = -0.280$ .

Step 2. Predicted mass: $\log(M_\bullet/M_\odot) = 8.3 + 4.4\times(-0.280) = 8.3 - 1.23 = 7.07$ , so $M_\bullet^\mathrm{pred} \approx 1.2\times10^7\,M_\odot$ .

Step 3. Compare: the measured Sgr A\* mass corresponds to $\log(M_\bullet/M_\odot) = 6.61$ , about 0.45 dex below the prediction. The Milky Way sits at the low edge of the scatter — a reminder that $0.3$ dex is a population dispersion, not a guarantee for any one galaxy, and that our Galaxy's modest, secularly-influenced bulge is a known slight under-performer. Predicted $\sim10^7\,M_\odot$ ; actual $4\times10^6\,M_\odot$ — consistent within roughly $1.5\sigma$ of the scatter.

Physics of the portrait

The $M_\bullet$ – $\sigma$ relation is a rare case where the observation and its explanation arrived almost together, and where the deepest idea — that a black hole regulates its own galaxy — was written down before the relation it explains was even measured.

Origins · 1 minSilk and Rees and how this was discovered

Silk and Rees

Joseph Silk b. 1942 and Martin Rees b. 1942 · AGN feedback and black hole–galaxy coevolution

By the late 1990s several groups had established that essentially every massive galaxy hosts a central dark object of $\sim 0.1\%$ the bulge mass — the Magorrian relation of 1998. But the tightest link was still hidden. In 1998 Joseph Silk and Martin Rees, reasoning from quasar energetics rather than from data, argued that the enormous energy released as a black hole grows must eventually blow the surrounding gas away, and that this feedback should imprint a relation between black hole mass and the depth of the host potential — the energy-driven $M_\bullet \propto \sigma^5$ argument in embryo. Two years later, in 2000, Laura Ferrarese and David Merritt, and independently Karl Gebhardt and collaborators, measured it: black hole mass correlated with bulge velocity dispersion with startlingly little scatter, far tighter than the correlation with luminosity. Andrew King in 2003 sharpened the theory, showing that momentum-driven feedback gives the cleaner $M_\bullet \propto \sigma^4$ with a normalization fixed by fundamental constants, and that energy-driven feedback gives the steeper $\sigma^5$ — the two idealisations that still bracket the data. John Kormendy and Luis Ho, in their 2013 review, curated the modern sample and codified black hole–galaxy coevolution as a distinct subject. The through-line from Silk and Rees to today is a single physical claim: the black hole and its bulge finish growing at the same moment, when accretion energy finally unbinds the shared gas — coevolution, not coincidence.

History

1998

— Magorrian and collaborators find black hole mass correlates with bulge mass across nearby galaxies

1998

— Silk and Rees propose AGN feedback sets the black hole–bulge link, predicting a σ⁵ scaling

2000

— Ferrarese and Merritt, and Gebhardt and collaborators, announce the tight M•–σ relation

2003

— King derives the momentum-driven σ⁴ critical mass with a constants-only normalization

2013

— Kormendy and Ho review black hole–galaxy coevolution and calibrate the modern relation

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 28 (galactic nuclei and supermassive black holes) · Binney and Tremaine, Galactic Dynamics, 2nd ed., Ch. 4 (velocity dispersion, isothermal sphere, Jeans equations) · Mo, van den Bosch and White, Galaxy Formation and Evolution, Ch. 13–14 (AGN and feedback) · Ryden, Introduction to Cosmology, 2nd ed. (structure and galaxy-formation context) · Dodelson, Modern Cosmology (large-scale structure and galaxy formation). The slope and scatter ( $\beta \approx 4$ – $5$ , $\sim 0.3$ dex) and intercept ( $\alpha \approx 8.3$ ) are standard values. The momentum-driven critical mass $M_\sigma = f_g\sigma_T\sigma^4/(\pi G^2 m_p)$ evaluates in cgs ( $\sigma_T = 6.65\times10^{-25}\ \mathrm{cm^2}$ , $m_p = 1.67\times10^{-24}\ \mathrm{g}$ , $G = 6.67\times10^{-8}$ , $f_g = 0.16$ , $\sigma = 2\times10^7\ \mathrm{cm\,s^{-1}}$ ) to $\approx 3.7\times10^8\,M_\odot$ ( $\log 8.56$ ). The isothermal enclosed mass is $M(r) = 2\sigma^2 r/G$ ; the sphere of influence is $r_\mathrm{h} = GM_\bullet/\sigma^2 \approx 21\ \mathrm{pc}$ ; and the Eddington luminosity is $L_\mathrm{Edd} \approx 1.26\times10^{38}(M_\bullet/M_\odot)\ \mathrm{erg\,s^{-1}}$ . Anchor galaxy data: Sgr A\* $4.1\times10^6\,M_\odot$ , $\sigma\approx105\ \mathrm{km\,s^{-1}}$ ; NGC 4258 $3.9\times10^7\,M_\odot$ , $\sigma\approx115\ \mathrm{km\,s^{-1}}$ ; M87 $6.5\times10^9\,M_\odot$ , $\sigma\approx324\ \mathrm{km\,s^{-1}}$ .

Prerequisites. Supermassive Black Holes · Eddington Luminosity and Accretion Limits · Active Galactic Nuclei: Unified Model · AGN Feedback. Leads to. Quasars and Quasar Evolution · AGN Feedback · Relativistic Jets and Launching Mechanisms.

Ask the tutor: Ask me to walk through why the momentum-driven balance gives $M_\bullet\propto\sigma^4$ from first principles, or to work out the black hole mass implied by any galaxy's measured velocity dispersion.

Why does the relation use velocity dispersion instead of galaxy luminosity or halo mass?

Because feedback regulates the black hole against the depth of the gas-binding potential, and $\sigma$ measures that depth directly and locally. The correlations with total luminosity or halo mass exist but are looser and noisier, because those quantities depend on the whole galaxy's assembly history, star-formation efficiency, and dark-matter content — extra variables the black hole never "feels." Velocity dispersion, set within the bulge where the AGN outflow does its work, is the cleanest observable proxy for the potential the black hole must overcome, which is why the $M_\bullet$ – $\sigma$ relation is tighter than any of its cousins.

How can a black hole whose gravity reaches only parsecs correlate with kiloparsec-scale dynamics?

Not through gravity — through feedback during the active phase. Direct gravity from the black hole is utterly negligible a kiloparsec out; the sphere of influence is tens of parsecs at most. What couples the two scales is the energy and momentum the AGN injects while it is accreting: radiation, winds, and jets push on the galactic gas reservoir, and when the output crosses a critical threshold the gas is expelled and both the black hole's fuel and the bulge's star-forming supply are removed at once. So the black hole ends up "knowing" about the large-scale potential because it had to fight that potential to shut itself off. The correlation is a fossil of that self-regulation, not a live gravitational link.

Why do the two feedback models give σ⁴ and σ⁵, and which is right?

The difference is bookkeeping on what the wind conserves. Momentum-driven feedback assumes the shocked wind cools fast, so only its momentum ( $L/c$ ) pushes the gas; balancing that against the bulge weight $\propto\sigma^4$ gives $M_\bullet\propto\sigma^4$ . Energy-driven feedback assumes the shock cannot radiate, so the retained thermal energy does $P\,\mathrm{d}V$ work; balancing power against the energy the shell needs to escape brings in an extra factor of $\sigma$ , giving $\sigma^5$ . Reality is probably both, in different phases and luminosities: dense nuclear regions favour momentum-driving, while fast winds in diffuse gas become energy-conserving. Observed AGN outflows often carry momentum flux well above $L/c$ , which points to energy-driven boosting and helps explain why the measured slope leans toward the steep end at high mass.

What is the sphere of influence, and why does it make black holes hard to weigh?

The sphere of influence, radius $r_\mathrm{h}=GM_\bullet/\sigma^2$ , is the region where the black hole rather than the surrounding stars dominates the gravitational potential. Only inside it do stellar or gas orbits carry a clean imprint of the black hole's mass, so a direct dynamical measurement requires resolving $r_\mathrm{h}$ with the telescope. For a typical $10^8\,M_\odot$ hole that radius is only about 10–20 pc, which subtends a resolvable angle out to at most tens of megaparsecs even with adaptive optics or space telescopes. Beyond that the black hole's kinematic signature blurs into the bulge, and one must fall back on indirect methods — reverberation mapping, or the $M_\bullet$ – $\sigma$ relation itself — to estimate the mass.

Do pseudobulges and disk galaxies follow the same relation?

No, and that is an important diagnostic. Classical bulges and elliptical galaxies — systems built by mergers and violent relaxation — follow the tight relation. Pseudobulges, which form slowly out of disk material by secular processes, and pure disk or bulgeless galaxies scatter off it, usually lying below, with black holes undermassive for their $\sigma$ . This is expected in the feedback picture: the relation is forged during the intense, merger-driven accretion episodes that build classical bulges, whereas secular disk growth never drives the black hole through the self-regulation that sets the relation. So a galaxy's location relative to the line encodes how it assembled, which is part of why the relation is a probe of coevolution and not just a lookup table.

Does the relation evolve with cosmic time, and what would that tell us?

This is one of the field's live questions. If the normalization stayed fixed with redshift, black holes and bulges would have grown in lockstep, exactly tracking each other — pure coevolution. If it shifted, one component led the other: black holes overmassive at early times would mean the hole grew first and later feedback let the bulge catch up, while undermassive holes would mean the reverse. Measuring $\sigma$ at high redshift is hard, so evidence is still mixed, but JWST is now finding black holes in the first billion years that appear overmassive relative to their hosts — hinting that early black holes may have outpaced their bulges before feedback brought the population onto the local relation. Settling this pins down the causal order of black hole and galaxy growth.