Galactic Rotation Curves

A rotation curve is the run of circular speed $v(R)$ with cylindrical galactocentric radius $R$ in the plane of a disk galaxy, and it is the observational face of the galaxy's gravitational potential. A star or gas cloud on a circular orbit balances gravity against the centripetal term, $v^2/R=\partial\Phi/\partial R$ , so measuring $v(R)$ weighs the galaxy at every radius without any assumption about what the mass is made of. That indifference to composition is the whole power of the method: a bright star, a cold hydrogen cloud, and a dark rock at the same radius all orbit at the same speed, so the curve reports total gravitating mass, not luminous mass. The dynamical prediction from the visible matter alone is sharp and wrong. Interior to the bulk of the light $v(R)$ rises, and once most of the luminous mass is enclosed Newtonian gravity predicts a Keplerian decline $v\propto R^{-1/2}$ , as the planets do. What is actually seen — first convincingly by Vera Rubin and Kent Ford in optical emission lines, then extended to enormous radii with the neutral-hydrogen 21-cm line — is a flat plateau that persists far beyond the edge of the starlight. A constant $v$ forces the enclosed mass to keep climbing, $M(<R)=v^2R/G\propto R$ , and the local density to fall only as $\rho\propto R^{-2}$ : a vast dark halo outweighing the stars tenfold. This article shows, step by step, how a measured $v(R)$ becomes the foundational galactic-scale evidence for dark matter, how a real curve is decomposed into its parts, and how the flat plateau anchors the Tully–Fisher distance ladder.

Key takeaways

A rotation curve $v(R)$ gives the circular orbital speed of tracers — stars, HI gas — as a function of galactocentric radius $R$ , and it directly weighs the enclosed mass.
If light traced mass, dynamics would give a Keplerian fall $v\propto R^{-1/2}$ beyond the luminous disk; instead observed curves stay flat, $v\approx$ constant, out to many optical radii.
A flat curve forces the enclosed mass to grow linearly, $M(<R)\propto R$ , and the local density to fall as $\rho\propto R^{-2}$ — the singular isothermal signature of an extended dark halo.
For the Milky Way $v_0\approx220$ – $235\ \mathrm{km\,s^{-1}}$ at $R_0\approx8.2\ \mathrm{kpc}$ , giving $M(<R_0)\approx9\times10^{10}\,M_\odot$ ; the 21-cm line carries the curve far past the stars.
A real curve is decomposed as $v_\mathrm{tot}^2=v_\mathrm{bulge}^2+v_\mathrm{disk}^2+v_\mathrm{gas}^2+v_\mathrm{halo}^2$ ; the transition radius fixes both the disk mass-to-light ratio and the halo scale.
Key idea: circular velocity directly weighs the enclosed mass, $v^2(R)=GM(<R)/R$ . A flat $v(R)$ therefore forces $M(<R)\propto R$ and $\rho\propto R^{-2}$ — mass that keeps accumulating where there is essentially no light.

Where this lives today

Rotation curves remain a frontline tool. Gaia astrometry, combined with maser parallaxes and Cepheid distances, has refined the Milky Way curve enough to measure its gentle outer-disk decline beyond $\sim15\ \mathrm{kpc}$ and weigh the Galactic halo at $\sim10^{12}\,M_\odot$ . In external galaxies, deep HI surveys such as THINGS and SPARC assemble decomposed curves for hundreds of systems, powering the sharpest current debate in galactic dynamics: the core–cusp problem. Cold-dark-matter simulations predict cuspy NFW halos ( $\rho\propto R^{-1}$ at small radius), while the curves of many dwarf and low-surface-brightness galaxies favor a constant-density core — bearing on whether dark matter is truly cold and collisionless or self-interacting. The same flat velocity anchors the Tully–Fisher relation, $L\propto v_\mathrm{flat}^{4}$ , a distance indicator good to $\sim10\%$ ; and the tight radial-acceleration relation between observed and baryonic acceleration is what modified-gravity proposals like MOND were built to explain — keeping rotation curves central to whether the missing mass is a particle or a failure of Newtonian gravity.

The physical picture

Compare a merry-go-round with the Solar System. Solid-body rotation, as the inner galaxy or a bar turns, keeps everyone at one angular speed, so outer riders move faster: $v\propto R$ . Keplerian orbits, like the planets, have outer bodies moving slower: $v\propto R^{-1/2}$ . A spiral galaxy does neither in its outskirts: every tracer at large $R$ moves at nearly the same linear speed, as if there were always just enough extra mass outside your orbit to hold the speed fixed. Because circular speed is a pure statement about gravity — the orbiting object feels only $\Phi(R)$ — the mismatch between the mass the curve demands and the mass the light provides is unavoidable, not a modeling artifact. The first figure sets the Keplerian expectation against the flat reality; the second decomposes a real curve into its components.

Figure

Rotation curve of a spiral galaxy — expectation versus reality. The observed curve rises then holds a flat plateau at v_flat ≈ 220 km s⁻¹ far beyond the edge of the starlight near 12 kpc, while a Keplerian fall from the luminous mass, v ∝ R^−1/2, would drop toward 100 km s⁻¹ by 30 kpc. The widening gap is the dark halo.

In the flat region every logarithmic shell contributes the same mass: $\mathrm{d}M=(v^2/G)\,\mathrm{d}R$ is independent of $R$ , so the halo is a nested set of equal-mass shells reaching far past the visible disk — a singular isothermal sphere in disguise. But no single component makes a curve flat by itself. The stellar disk produces a bumped, declining term that peaks near two scale lengths and then falls faster than Keplerian; the gas adds a smaller contribution; the halo supplies a rising term that fills the deficit. Because accelerations (and hence $v^2$ ) add, the observed curve is the quadrature sum of the parts.

Figure

Mass-model decomposition of a Milky-Way-like curve. Velocities add in quadrature, v_tot² = v_disk² + v_gas² + v_halo², so the amber stellar-disk term peaks near R ≈ 2.2 R_d ≈ 6.5 kpc then declines, the green halo term rises monotonically, and their sum (blue) holds the observed points flat near 220 km s⁻¹.

The core equations

The starting point is the circular-orbit condition. A tracer on a circular orbit in an axisymmetric potential needs a net inward acceleration $v^2/R$ , and gravity supplies it through the radial gradient of the potential in the midplane.

\frac{v^2(R)}{R}=\left.\frac{\partial\Phi}{\partial R}\right|_{z=0} \qquad (1)

where: $v(R)$ — circular (rotational) speed at galactocentric radius $R$ ; $R$ — cylindrical distance from the center in the disk plane; $\Phi$ — gravitational potential of the whole galaxy; $z$ — height above the midplane, with the gradient taken at $z=0$ . Reading: the curve is a direct probe of the potential gradient — and hence the mass — at every radius, regardless of what that mass radiates. For a spherical (or spherically averaged) mass distribution the shell theorem collapses equation (1) into the most useful relation in the subject: the speed squared weighs the mass interior to the orbit.

v^2(R)=\frac{G\,M(<R)}{R}\qquad(\text{spherical symmetry}) \qquad (2)

where: $G$ — Newton's constant, $6.674\times10^{-11}\ \mathrm{m^3\,kg^{-1}\,s^{-2}}$ ; $M(<R)$ — total gravitating mass enclosed within $R$ . Reading: invert to $M(<R)=v^2R/G$ and a measured curve becomes a mass profile, no photometry required. Applied to a flat curve this inversion is dramatic. If $v$ is constant, the enclosed mass cannot level off — it must climb in direct proportion to radius, far past the light.

M(<R)=\frac{v^2(R)\,R}{G}\ \xrightarrow{\ v=\mathrm{const}\ }\ M(<R)\propto R \qquad (3)

where: $v(R)$ — measured circular speed, equal to a constant $v_\mathrm{flat}$ in the plateau. Reading: doubling the radius doubles the enclosed mass, so the mass-to-light ratio soars. Differentiating that linear mass profile fixes the density run of the halo — the singular isothermal sphere, the reference model against which every real halo is compared.

\rho(R)=\frac{1}{4\pi R^2}\frac{\mathrm{d}M}{\mathrm{d}R}=\frac{v_\mathrm{flat}^2}{4\pi G R^2} \qquad (4)

where: $\rho(R)$ — local mass density at radius $R$ ; $v_\mathrm{flat}$ — plateau speed. Reading: a flat curve is equivalent to $\rho\propto R^{-2}$ — a profile needing a central core and an outer truncation. The kinematics of nearly circular orbits are governed by two frequencies: the angular speed $\Omega$ and the epicyclic frequency $\kappa$ for small radial oscillations, whose ratio encodes the local shape of the curve.

\Omega(R)=\frac{v(R)}{R},\qquad \kappa^2(R)=R\frac{\mathrm{d}\Omega^2}{\mathrm{d}R}+4\Omega^2 \qquad (5)

where: $\Omega$ — angular orbital frequency; $\kappa$ — epicyclic frequency. Reading: solid-body rotation gives $\kappa=2\Omega$ , Keplerian gives $\kappa=\Omega$ , and a flat curve gives $\kappa=\sqrt{2}\,\Omega$ — so $\kappa$ alone diagnoses the mass distribution. Near the Sun, the differential rotation is summarized by the two Oort constants, the shear $A$ and the vorticity $B$ , defined from the value and slope of the curve at the solar radius.

A=\tfrac12\left(\frac{v_0}{R_0}-\left.\frac{\mathrm{d}v}{\mathrm{d}R}\right|_{R_0}\right),\qquad B=-\tfrac12\left(\frac{v_0}{R_0}+\left.\frac{\mathrm{d}v}{\mathrm{d}R}\right|_{R_0}\right) \qquad (6)

where: $A$ — Oort shear constant; $B$ — Oort vorticity constant, with $A-B=\Omega_0=v_0/R_0$ ; $v_0,R_0$ — circular speed and radius at the Sun. Reading: $A$ and $B$ from nearby stellar velocities fix both $\Omega_0$ and the local slope $\mathrm{d}v/\mathrm{d}R$ without the whole curve. For the inner Milky Way, seen from inside, the curve is built by the tangent-point method: along a sightline at longitude $\ell$ the peak line-of-sight velocity comes from gas at the tangent radius $R=R_0\sin\ell$ .

v_r^{\max}(\ell)=R_0\big[\Omega(R)-\Omega_0\big]\sin\ell,\qquad R=R_0\sin\ell \qquad (7)

where: $v_r^{\max}$ — peak line-of-sight velocity at galactic longitude $\ell$ ; $\ell$ — galactic longitude, method valid for $0<\ell<90^\circ$ . Reading: the 21-cm profile's terminal velocity gives $\Omega(R)$ , and hence $v(R)$ , point by point interior to the Sun. Finally, a real curve is the quadrature sum of its components, because it is the accelerations — proportional to $v^2/R$ — that add, not the velocities.

v_\mathrm{tot}^2(R)=v_\mathrm{bulge}^2+v_\mathrm{disk}^2+v_\mathrm{gas}^2+v_\mathrm{halo}^2 \qquad (8)

where: $v_\mathrm{tot}$ — observed total circular speed; $v_i$ — the contribution of component $i$ to $v^2$ . Reading: because the terms add in quadrature, the halo is whatever is left after the luminous components are subtracted — the deficit is measured, not assumed.

Derivation AWeighing a galaxy from its circular speed

The goal is the full logical chain that turns a measured plateau into a dark halo: from Newton's shell theorem to the mass-from-velocity relation, then feeding a flat curve into it to force a linearly growing enclosed mass, and differentiating that mass to find the density profile. Every step is elementary Newtonian mechanics, yet the endpoint — mass growing without bound where no light is emitted — is the most consequential inference in galactic astronomy.

Step 1. For a spherically symmetric mass, the shell theorem says a test particle at radius $R$ feels gravity only from the interior mass, as though it were a point at the center: the inward acceleration is $g(R)=GM(<R)/R^2$ . Why: shells outside $R$ exert zero net force on an interior point, so the exterior halo — however massive — drops out of the local orbit equation.

Step 2. A circular orbit requires that this gravitational acceleration exactly supply the centripetal term, $v^2/R=g(R)$ . Solving gives $v^2(R)=GM(<R)/R$ , equation (2): the measured speed squared, times radius over $G$ , is the enclosed mass. Why: nothing here refers to light — the balance is between gravity and inertia, so the mass that appears is the total gravitating mass.

Step 3. Now impose the observation. If $v(R)=v_\mathrm{flat}$ is constant over some range, then $M(<R)=v_\mathrm{flat}^2R/G\propto R$ : the enclosed mass grows linearly with radius across the entire plateau. Why: a constant speed cannot be sustained by a fixed interior mass, which would give the falling Keplerian $v\propto R^{-1/2}$ ; holding $v$ up demands ever more mass just inside each larger orbit.

Step 4. Differentiate: $\mathrm{d}M/\mathrm{d}R=v_\mathrm{flat}^2/G$ is constant, and equating it to the shell mass $4\pi R^2\rho(R)$ gives $\rho(R)=v_\mathrm{flat}^2/(4\pi G R^2)$ , equation (4). Why: equal mass per unit radius spread over a shell whose area grows as $R^2$ forces the density to fall as $R^{-2}$ — the singular isothermal sphere, formally requiring a central core and an outer truncation. *Verified: $v^2=GM/R$ from the shell theorem; flat $v$ gives $M\propto R$ and $\rho\propto R^{-2}$ ; the epicyclic ratio $\kappa=\sqrt{2}\,\Omega$ follows from equation (5) for constant $v$ .*

Go deeper · Derivation B · 1 minWhy luminous matter predicts a Keplerian fall

Derivation BWhy luminous matter predicts a Keplerian fall

The purpose here is to make the dark-matter argument quantitative rather than rhetorical. We model the visible galaxy as a mass $M_\mathrm{vis}$ effectively enclosed within an optical radius $R_\mathrm{opt}$ , ask what curve that predicts beyond the light, and compare with 21-cm observations. The point is not that the light-only prediction is roughly wrong but that it fails by a definite, large, radius-growing factor that no stellar mass-to-light ratio can absorb.

Step 1. Outside the starlight, if there were no other mass, the enclosed mass would stop growing: $M(<R)\approx M_\mathrm{vis}=\mathrm{const}$ for $R>R_\mathrm{opt}$ . Why: with the luminous mass exhausted, each larger orbit encloses nothing new, exactly the Solar-System situation where the Sun holds essentially all the mass.

Step 2. Equation (2) then gives $v(R)=\sqrt{GM_\mathrm{vis}/R}\propto R^{-1/2}$ , a Keplerian decline. Between $R_\mathrm{opt}$ and $2R_\mathrm{opt}$ the speed should fall by $\sqrt{2}\approx1.41$ , a drop of about 30%, and by $2R_\mathrm{opt}$ to $4R_\mathrm{opt}$ it should halve. Why: this is the same $R^{-1/2}$ law the planets obey, and it is a firm prediction of Newtonian gravity acting on the visible mass alone.

Step 3. Observation contradicts it flatly. HI curves stay flat out to two-to-five optical radii with variations of order 10% or less — no hint of the predicted fall. Why: the neutral-hydrogen disk extends far past the stars, so the 21-cm line probes precisely the regime where the Keplerian decline should appear, and does not find it.

Step 4. The required mass follows from inverting the flat curve: $M(<R)=v_\mathrm{flat}^2R/G$ , which at large $R$ exceeds $M_\mathrm{vis}$ by factors of five to ten. Why: to keep $v$ flat instead of falling, the true enclosed mass must keep rising linearly, so a dark component must dominate the mass budget wherever the light has faded. *Verified: a curve flat at $v_\mathrm{flat}=220\ \mathrm{km\,s^{-1}}$ demands $M(<R)\propto R$ ; the predicted halving of $v$ by $4R_\mathrm{opt}$ is excluded by the observed flatness (Rubin, Ford and Thonnard 1980; Carroll and Ostlie Ch. 24).*

Go deeper · Derivation C · 2 minThe exponential stellar disk cannot stay flat

Derivation CThe exponential stellar disk cannot stay flat

This derivation shows that the flatness cannot be blamed on the stars themselves. A real stellar disk is not a point mass; it is a razor-thin exponential sheet of surface density $\Sigma(R)=\Sigma_0e^{-R/R_d}$ , and its flattened geometry gives a curve quite different from a spherical mass of the same profile. Working it out through thin-disk potential theory yields a term that rises, peaks, and then declines faster than Keplerian — so a bare disk of any mass-to-light ratio leaves a growing deficit at large radius that only a separate halo can fill.

Step 1. Solve Poisson's equation for the razor-thin exponential disk using a Hankel (Bessel) transform of $\Sigma(R)$ ; the exponential profile has an analytic transform, so the midplane potential $\Phi_\mathrm{disk}(R)$ is obtained in closed form in terms of modified Bessel functions. Why: a flattened mass cannot be replaced by an interior point mass — the shell theorem fails for a disk — so the full potential theory is unavoidable.

Step 2. The circular speed is $v_\mathrm{disk}^2=R\,\partial\Phi_\mathrm{disk}/\partial R|_{z=0}$ , which evaluates to

v_\mathrm{disk}^2(R)=4\pi G\Sigma_0 R_d\,y^2\big[I_0(y)K_0(y)-I_1(y)K_1(y)\big],\qquad y=\frac{R}{2R_d}

the Freeman (1970) result, with $I_n,K_n$ the modified Bessel functions. Why: the Bessel functions are the fingerprint of a flattened, extended mass, and they replace the simple $\sqrt{GM/R}$ that a spherical mass would give.

Step 3. This term rises from the center, reaches a maximum near $R_\mathrm{peak}\approx2.15\,R_d$ (about two-and-a-fifth scale lengths), and then declines. Why: interior to the peak the enclosed disk mass grows fast enough to raise $v$ , but once most of the exponential mass is passed the curve turns over.

Step 4. Beyond the peak $v_\mathrm{disk}$ falls off faster than the Keplerian $R^{-1/2}$ , because the disk's mass is exhausted even more sharply than a point mass would be. A flat total therefore requires a rising halo term, well fitted by a cored form $v_\mathrm{halo}^2\propto R^2/(R^2+r_c^2)$ , added in quadrature per equation (8). Why: no scaling of $\Sigma_0$ can turn a bumped, declining disk term into a flat curve — the shape is wrong, not merely the amplitude, so a distinct dark component is forced. *Verified: standard potential-theory result — Binney and Tremaine Ch. 2 and Freeman (1970), ApJ 160, 811; the peak at $R\approx2.15\,R_d$ and the faster-than-Keplerian outer decline are exact for the razor-thin exponential disk.*

Limiting cases

Different mass distributions leave distinct, diagnostic fingerprints on $v(R)$ , so the local shape of the curve alone tells you what kind of matter dominates there.

Case	Result	Meaning
Solid-body (inner disk / bar)	$v(R)\propto R$	Constant angular speed $\Omega=v/R$ ; arises where density is nearly uniform, as inside a homogeneous core; epicyclic frequency $\kappa=2\Omega$ .
Keplerian (point mass / outside all mass)	$v(R)=\sqrt{GM/R}\propto R^{-1/2}$	Expected beyond the luminous edge if there were no halo; the observed absence of this fall is the dark-matter signature; $\kappa=\Omega$ .
Flat (singular isothermal halo)	$v=v_\mathrm{flat}\Rightarrow M\propto R,\ \rho\propto R^{-2}$	Enclosed mass grows linearly; the singular isothermal sphere reproduces it exactly; $\kappa=\sqrt{2}\,\Omega$ .
Exponential stellar disk alone	$v_\mathrm{disk}^2=4\pi G\Sigma_0R_d\,y^2[I_0K_0-I_1K_1],\ y=R/2R_d$	Peaks near $R\approx2.2\,R_d$ , then declines faster than Keplerian, so a bare disk cannot hold a plateau.
Cored halo, small radius	$v\propto R$ , $\rho\to\rho_0$ finite	A constant-density core removes the central divergence of the singular isothermal sphere; the curve rises linearly from the center.

Common trapa flat curve does not mean the mass has stopped growing

The most common misreading is that a flat rotation curve implies a fixed enclosed mass. It implies the opposite: constant $v$ forces $M(<R)=v_\mathrm{flat}^2R/G$ to grow linearly — the mass keeps piling up precisely where the light is gone. A second trap is applying $v^2=GM/R$ as if it were exact for any galaxy. It is exact only for spherical symmetry; a flattened disk contributes a different term (Freeman's Bessel-function expression) that peaks and then falls faster than a point mass would. That is why serious work never naively inverts $v(R)$ into a single spherical $M(<R)$ , but decomposes the curve via equation (8) — the halo is the measured residual after the luminous, non-spherical components are subtracted.

Worked examples

1 · The mass of the Milky Way inside the Sun's orbit (fully worked)

Take $v_0=220\ \mathrm{km\,s^{-1}}$ at the solar radius $R_0=8.2\ \mathrm{kpc}$ , treating the interior mass as spherical. How much mass lies inside the Sun's orbit?

Step 1. Convert to SI. With $1\ \mathrm{kpc}=3.086\times10^{19}\ \mathrm{m}$ , $R_0=8.2\times3.086\times10^{19}=2.53\times10^{20}\ \mathrm{m}$ , and $v_0=2.20\times10^{5}\ \mathrm{m\,s^{-1}}$ .

Step 2. Apply equation (2): $M(<R_0)=v_0^2R_0/G$ . The numerator is $(2.20\times10^{5})^2\times2.53\times10^{20}=4.84\times10^{10}\times2.53\times10^{20}=1.22\times10^{31}$ , and dividing by $G=6.674\times10^{-11}$ gives $M(<R_0)=1.84\times10^{41}\ \mathrm{kg}$ .

Step 3. In solar masses, $M/M_\odot=1.84\times10^{41}/1.989\times10^{30}=\mathbf{9.2\times10^{10}\,M_\odot}$ — about $10^{11}$ solar masses inside the Sun's orbit, matching the Carroll and Ostlie textbook estimate.

Step 4. Extrapolate outward: because the curve stays flat, $M(<R)\propto R$ , so by $R\approx50\ \mathrm{kpc}$ the enclosed mass climbs toward $\sim5\times10^{11}\,M_\odot$ — now overwhelmingly dark. Inside the Sun's orbit the Galaxy holds $\approx9\times10^{10}\,M_\odot$ , and the total grows in step with radius far beyond.

2 · Enclosed-mass ratio and the local halo density (fully worked)

A spiral has a flat curve with $v_\mathrm{flat}=200\ \mathrm{km\,s^{-1}}$ . (a) By what factor does the enclosed mass grow between $R=5\ \mathrm{kpc}$ and $R=25\ \mathrm{kpc}$ ? (b) What density does a singular isothermal halo of the Milky Way ( $v_\mathrm{flat}=220\ \mathrm{km\,s^{-1}}$ ) imply at the solar radius?

Step 1. For part (a), $M(<R)=v_\mathrm{flat}^2R/G\propto R$ at fixed $v_\mathrm{flat}$ , so the mass ratio is simply the radius ratio: $M(25)/M(5)=25/5=\mathbf{5}$ . Five times more mass is enclosed at $25\ \mathrm{kpc}$ than at $5\ \mathrm{kpc}$ , even though the starlight has all but vanished — the increase is almost entirely dark. (For scale, this curve gives $M(<10\ \mathrm{kpc})\approx9\times10^{10}\,M_\odot$ , the Milky Way's order of magnitude.)

Step 2. For part (b), use equation (4): $\rho(R_0)=v_\mathrm{flat}^2/(4\pi GR_0^2)$ . With $v_\mathrm{flat}=2.20\times10^5\ \mathrm{m\,s^{-1}}$ and $R_0=2.53\times10^{20}\ \mathrm{m}$ , the denominator is $4\pi(6.674\times10^{-11})(2.53\times10^{20})^2=5.37\times10^{31}$ , giving $\rho(R_0)=4.84\times10^{10}/5.37\times10^{31}=\mathbf{9.0\times10^{-22}\ \mathrm{kg\,m^{-3}}}$ .

Step 3. Convert to astronomers' units: $9.0\times10^{-22}\ \mathrm{kg\,m^{-3}}\approx0.013\,M_\odot\,\mathrm{pc^{-3}}\approx0.5\ \mathrm{GeV\,cm^{-3}}$ . This slightly overestimates the true local dark-matter density ( $\approx0.3$ – $0.4\ \mathrm{GeV\,cm^{-3}}$ ) because at $R_0$ the disk shares the rotational support, so not all of $v_\mathrm{flat}^2$ belongs to the halo. The isothermal estimate lands within a factor of two of the measured value — a striking consistency for such a crude model.

Physics of the portrait

The rotation curve is where a measured constant became a cosmic mystery. The circular-orbit condition is old, but the observation that the outer curve does not fall was a twentieth-century shock, and it took a patient observer with new detectors to make it undeniable.

Origins · 1 minVera Rubin and how this was discovered

Vera Rubin

1928–2016 · galactic rotation curves and the case for dark matter

Vera Rubin spent her career measuring how galaxies turn, and in doing so she converted a fringe suspicion into a pillar of modern cosmology. In the late 1960s and 1970s, using the sensitive image-tube spectrograph she developed with instrument-builder Kent Ford, she measured optical emission-line velocities across the disk of the Andromeda galaxy (M31) and then dozens of other spirals. The curves were the same story every time: after an inner rise, the circular speed flattened and stayed flat as far as the ionized gas could be traced, with no sign of the Keplerian decline that the concentrated starlight demanded. Radio astronomers, following the 21-cm line far beyond the optical edge, found the plateau continuing further still. The interpretation was inescapable from $v^2=GM/R$ : the galaxies were embedded in massive, invisible halos outweighing their stars several times over. Rubin's curves, together with Fritz Zwicky's earlier cluster dynamics and later gravitational lensing, made dark matter a fact to be explained rather than a hypothesis to be entertained. Characteristically cautious, she preferred the secure observations to any theory of the missing mass — a stance that has aged well as its identity remains open.

History

1932

— Oort infers extra mass in the Galactic disk from the vertical motions of stars

1933

— Zwicky finds a large missing mass in the Coma cluster and names it dunkle Materie

1939

— Babcock measures a surprisingly non-declining rotation curve for M31

1970

— Rubin and Ford publish the flat optical rotation curve of Andromeda

1980

— Rubin, Ford and Thonnard show that flat curves are generic across spiral galaxies

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 24 (The Milky Way) and Ch. 27 (The Nature of Galaxies) · Binney and Tremaine, Galactic Dynamics, 2nd ed., Ch. 2 (Potential Theory), Ch. 3 (orbits, epicyclic and Oort constants) and Ch. 6 (Disk Dynamics) · Mo, van den Bosch and White, Galaxy Formation and Evolution, Ch. 11 (disk dynamics and rotation curves) · Ryden, Introduction to Cosmology, 2nd ed., Ch. 8 (Dark Matter) · Dodelson, Modern Cosmology, Ch. 1 (evidence for dark matter).

Provenance. The circular-orbit condition, $v^2=GM/R$ , and the flat-curve consequences follow from Newtonian dynamics — verify the shell theorem and the isothermal integral $\rho=v_\mathrm{flat}^2/(4\pi GR^2)$ . Milky Way mass: recompute $M=v_0^2R_0/G$ with $v_0=220\ \mathrm{km\,s^{-1}}$ , $R_0=8.2\ \mathrm{kpc}$ to get $\approx9\times10^{10}\,M_\odot$ (Carroll and Ostlie Ch. 24). The exponential-disk Bessel result is Freeman (1970), ApJ 160, 811; see Binney and Tremaine Ch. 2. Oort constants and epicyclic frequency: Binney and Tremaine Ch. 3, with $A-B=\Omega_0=v_0/R_0\approx27\ \mathrm{km\,s^{-1}\,kpc^{-1}}$ . Observational flatness: Rubin and Ford (1970), ApJ 159, 379; Rubin, Ford and Thonnard (1980), ApJ 238, 471. Constants: $G=6.674\times10^{-11}\ \mathrm{m^3\,kg^{-1}\,s^{-2}}$ , $M_\odot=1.989\times10^{30}\ \mathrm{kg}$ , $1\ \mathrm{kpc}=3.086\times10^{19}\ \mathrm{m}$ .

Builds on. Structure of the Milky Way Galaxy · Galactic Coordinate System and the Solar Neighborhood · Neutral Hydrogen 21-cm Line and Galactic Kinematics. Leads to. Evidence for Dark Matter in Galaxies · Galactic Dark Matter Halos and Density Profiles · Spiral Galaxies and Density Wave Theory.

Ask the tutor: Ask me to decompose a real HI rotation curve into disk, gas, and halo, or to show how the tangent-point method turns a 21-cm spectrum into the inner Milky Way's $v(R)$ .

Does a flat rotation curve really prove dark matter, or could Newton's law just be wrong?

The curve proves that the observed dynamics require more gravity at large radius than the visible matter supplies — model-independent, straight from $v^2=GM/R$ . What it does not prove by itself is the cause. The mainstream response adds unseen mass, a dark halo whose $\rho\propto R^{-2}$ profile reproduces the plateau. The alternative, MOND and its relativistic cousins, modifies the gravitational law below a characteristic acceleration so that flat curves emerge without extra mass. Rotation curves alone cannot cleanly separate these; the case for particle dark matter rests on combining them with cluster lensing, the cosmic microwave background, and structure formation, where the modified-gravity accounts strain hardest. So the curve is decisive evidence for a discrepancy, and strong but not sole evidence for particle dark matter.

Why does the curve rise near the center instead of starting flat?

Deep in the inner galaxy the density is high and fairly uniform, so the enclosed mass grows roughly as $R^3$ and equation (2) gives $v=\sqrt{GM/R}\propto R$ — solid-body rotation, everyone sharing one angular speed. Only after you pass through the bulk of the central mass does the growth of $M(<R)$ slow enough for the curve to turn over and flatten. That transition radius is physically informative: it marks where the concentrated luminous mass gives way to the extended halo, and fitting it fixes the disk mass-to-light ratio and the halo scale together.

If the halo density falls as R^−2, isn't its total mass infinite?

Formally, yes — that is a known defect of the singular isothermal sphere, not of the data. Integrating $\rho\propto R^{-2}$ gives $M(<R)\propto R$ , which diverges as $R\to\infty$ , and it also diverges as $R\to0$ because the central density blows up. Real halos cure both ends: a finite-density core (or milder cusp) inside, and a truncation at large radius where cosmological infall sets an outer edge (the virial radius). The flat-curve $\rho\propto R^{-2}$ region is only the intermediate zone that happens to coincide with where we can measure gas. The NFW profile from simulations, $\rho\propto R^{-1}(R+R_s)^{-2}$ , is cuspy inside, isothermal-like in the middle, steep outside — finite in total mass.

How is the Milky Way's own curve measured when we sit inside the disk?

Sitting inside the disk forbids the face-on view we get of external galaxies, so inner and outer Galaxy need different tricks. Interior to the Sun ( $R<R_0$ ), the tangent-point method uses the 21-cm line: along a sightline at longitude $\ell$ , the largest line-of-sight velocity comes from gas at the tangent radius $R=R_0\sin\ell$ , where the orbit points straight along the line of sight, so the terminal velocity of the HI profile yields $v(R)$ directly. Beyond the Sun there is no tangent point, and one instead needs tracers with independently known distances — masers with VLBI parallaxes, Cepheids, and Gaia stars — to place each velocity at a radius, giving the full Galactic curve.

What sets the difference between a cored and a cusped halo, and why does it matter?

It is the small-radius behavior of $\rho(R)$ . A cored halo levels off to a constant central density (so the inner curve rises linearly, $v\propto R$ ), while a cusped NFW halo keeps rising as $\rho\propto R^{-1}$ into the center. Decomposed curves of dwarf and low-surface-brightness galaxies — where dark matter dominates even at the center — tend to favor cores, while cold-dark-matter simulations generically make cusps. This core–cusp mismatch is a small-scale challenge to the standard cosmology; its resolution may lie in baryonic feedback flattening the cusp, or in dark matter that is warm or self-interacting rather than cold and collisionless.

Why do the component velocities add as v² rather than linearly?

Because gravity is linear in the source but circular speed is not linear in gravity. The potentials of the bulge, disk, gas, and halo superpose additively, so their radial accelerations add. But the circular-orbit condition is $v^2/R=g$ , so $v^2=gR$ for each component, and summing the accelerations means summing the $v^2$ terms: $v_\mathrm{tot}^2=\sum_iv_i^2$ . Adding the velocities themselves would double-count in the wrong power. This is why a decomposition combines $v_i(R)$ in quadrature, and why the halo is extracted as the measured residual $v_\mathrm{halo}^2=v_\mathrm{tot}^2-\sum_\mathrm{lum}v_i^2$ .