Evidence for Dark Matter in Galaxies

Weigh a galaxy the way Newton weighed the planets — orbit by orbit — and it comes out far too heavy. A star or a hydrogen cloud on a circular path of radius $r$ obeys one line of mechanics: centripetal acceleration equals the gravity of the mass inside it, $v_c^2/r=GM(<r)/r^2$ . Rearranged, $M(<r)=v_c^2 r/G$ , the circular speed is a scale that reads out the mass interior to each radius. Point that scale at a spiral galaxy and something breaks. Its starlight fades exponentially, so if mass followed light the speed should crest near the edge of the bright stars and then fall off in the Keplerian way, $v_c\propto r^{-1/2}$ , just as the planets slow with distance from the Sun. Instead the curves stay flat — often for two or three optical scale lengths past the last luminous star, out where the 21-cm line of neutral hydrogen is the only tracer left. A flat curve is not a mild anomaly; it is a demand. Constant $v_c$ forces $M(<r)$ to rise in direct proportion to $r$ , which forces a density falling as $\rho\propto r^{-2}$ — the singular isothermal sphere — a distinct, extended, non-luminous component that dominates the outer galaxy. The same excess appears without a single circular orbit: ellipticals reveal it through the Jeans equation applied to stellar dispersions, through their hydrostatic hot X-ray gas, and through the way they bend background light. In every system the dynamical mass outruns the stellar mass, and the gap widens outward. This article assembles the galactic-scale case: the Newtonian mass estimator, the anatomy of rotation curves, the pressure-support and lensing arguments for ellipticals, and the mass-to-light ratios that put one number on the missing mass.

Key takeaways

The circular speed of stars and gas in spiral galaxies stays roughly constant, $v_c(r)\approx\text{const}$ , out to and beyond the visible edge, instead of falling as $v_c\propto r^{-1/2}$ the way it would if mass tracked light.
A flat rotation curve forces the enclosed mass to grow linearly, $M(<r)=v_c^2 r/G$ , and the density to fall as $\rho\propto r^{-2}$ — an extended, roughly isothermal dark halo.
Neutral-hydrogen 21-cm observations push the curve far past the optical disk, giving the cleanest evidence that the flatness persists where there is essentially no starlight.
Ellipticals show the same excess through stellar dispersions, X-ray hydrostatics, and gravitational lensing: dynamical masses beat stellar masses by factors of several to tens.
Mass-to-light ratios climb from $\sim1\text{–}5\,(M/L)_\odot$ in bright cores to $\sim10\text{–}100$ or more once the halo is counted in — the missing mass as one number.
Key idea: for any tracer on a circular orbit $M(<r)=v_c^2 r/G$ ; a curve that stays flat where the light has faded forces the enclosed mass to keep rising linearly with radius, demanding an extended, non-luminous, roughly $\rho\propto r^{-2}$ halo.

The core insight

For any tracer on a circular orbit, $M(<r)=v_c^2 r/G$ ; a rotation curve that stays flat where the light has faded forces the enclosed mass to keep rising linearly with radius, requiring an extended, non-luminous, roughly $\rho\propto r^{-2}$ halo.

Where this lives today

The galactic mass argument is now a precision industry. In the Milky Way, ESA's Gaia astrometry of over a billion stars — with stellar streams that drape through the halo like tracer dye — has tightened the Galactic rotation curve and the local dark-matter density to $\sim0.3\text{–}0.4\ \mathrm{GeV\,cm^{-3}}$ , the very number direct-detection experiments (LZ, XENONnT) must assume to turn a null result into a particle limit. Beyond the Local Group, the SPARC database of $\sim175$ disk galaxies with decomposed rotation curves is the modern battleground where cold-dark-matter halo profiles are tested against the radial-acceleration relation and against MOND. Weak-lensing surveys (DES, KiDS, the Vera C. Rubin Observatory's LSST) weigh halos statistically by stacking the shear of millions of background galaxies, while the "core-versus-cusp" and "missing-satellites" puzzles keep the halo's inner structure open. The flat curve Rubin measured by hand is now a benchmark every galaxy-formation simulation and every alternative to dark matter must reproduce.

The physical picture

Think of every circular orbit as a spring balance. Hang a star or a gas cloud on it, read its speed, and the reading is the mass enclosed: $M(<r)=v_c^2 r/G$ . In the Solar System the balance behaves — outside the Sun's mass the speeds drop as $r^{-1/2}$ , the Keplerian fingerprint of a centrally concentrated mass. Galaxies refuse: the balance keeps reading more mass shell after shell, even in the outer regions where the plate is black and there is essentially no light to weigh. The luminous galaxy is only the bright core of a much larger, invisible structure, and the first figure draws it to scale: a thin disk of stars adrift inside a vast dark halo.

Figure

A spiral galaxy's luminous disk sits at the center of a vastly larger dark halo whose density falls as ρ ∝ 1/r²; each circular orbit weighs the mass inside it through M(<r) = v_c² r/G, so the orbit at 10 kpc encloses ≈1.1×10¹¹ M⊙ while the one at 30 kpc encloses ≈3.4×10¹¹ M⊙, even though almost no starlight lies between them

Why does a flat curve mean $\rho\propto r^{-2}$ ? Each spherical shell holds a mass $4\pi r^2\rho\,\mathrm{d}r$ . If $\rho=v_c^2/(4\pi G r^2)$ the $r^2$ factors cancel and every shell adds the same mass per unit radius, so $M(<r)$ climbs linearly and $v_c$ never changes. This singular isothermal sphere is the crudest halo that fits; its formally diverging mass (truncated in realistic NFW models) makes the point that there is always more mass just outside where you last looked. Light, meanwhile, is a biased tracer: stars emit essentially all the luminosity but carry only a fraction of the gravitating mass, and the mismatch shows up as a mass-to-light ratio that climbs outward. The second figure is the observation that started it all — the rotation curve of a massive spiral, split into what the stars can supply and what the halo must add.

Figure

Rotation-curve decomposition for a massive spiral: the observed circular speed stays flat near 220 km/s well past the optical edge, while the luminous disk alone peaks near two scale lengths and falls off Keplerian-style as v ∝ 1/√r; the shortfall is exactly the dark halo, combining as v_c² = v_disk² + v_halo²

In ellipticals there are no orderly circular orbits, so the argument switches from centripetal balance to pressure. Random stellar motions and hot gas hold the system up against its gravity, and the Jeans equation and hydrostatic equilibrium play the role that $v_c^2/r=GM/r^2$ plays in a disk. They too return more mass than the stars provide, and adding gravitational lensing — which weighs mass regardless of how it moves — makes the independent probes agree across disks, spheroids, and light-bending. That concordance is what makes the halo interpretation robust.

The core equations

The whole edifice rests on one estimator: a circular orbit balances gravity against inertia, so its speed measures the enclosed mass — Newton's shell theorem plus centripetal acceleration, nothing more.

v_c^2(r)=\frac{G\,M(<r)}{r}\quad\Longleftrightarrow\quad M(<r)=\frac{v_c^2(r)\,r}{G} \qquad (1)

where: $v_c(r)$ — circular (rotation) speed at galactocentric radius $r$ ; $M(<r)$ — gravitating mass enclosed within $r$ ; $G$ — Newton's gravitational constant. Reading: the speed at each radius weighs everything inside it and is blind to spherically distributed mass outside, so a flat $v_c$ at large $r$ says mass keeps accumulating. Impose the observed flatness, $v_c=\text{const}$ , differentiate the enclosed mass by radius, and divide the shell mass by the shell area $4\pi r^2$ to get the density a flat curve demands.

\rho(r)=\frac{1}{4\pi r^2}\frac{\mathrm{d}M}{\mathrm{d}r}=\frac{v_c^2}{4\pi G r^2}\quad(v_c=\text{const}) \qquad (2)

where: $\rho(r)$ — local mass density; $\mathrm{d}M/\mathrm{d}r$ — mass per unit radius in a spherical shell. Reading: an inverse-square density is the singular isothermal sphere; stellar disks and bulges fall off far faster, so luminous matter cannot supply it far out. To confront a real curve you decompose it. Potentials add linearly, so squared circular speeds of independent components add, and the halo is the piece left over after photometry fixes the luminous contributions.

v_c^2(r)=v_{\rm disk}^2+v_{\rm bulge}^2+v_{\rm gas}^2+v_{\rm halo}^2 \qquad (3)

where: $v_{\rm disk},v_{\rm bulge},v_{\rm gas}$ — contributions of the stellar disk, bulge, and gas, fixed by photometry and an assumed mass-to-light ratio; $v_{\rm halo}$ — the dark-halo contribution inferred as the residual needed to fit the data. Reading: this is the operational definition of the halo — not a fitted invention but the shortfall between observed and luminous rotation. The single number that quantifies "missing mass" is the mass-to-light ratio: how many solar masses of gravitating matter accompany each solar luminosity within a given radius.

\Upsilon\equiv\frac{M}{L}\quad(\text{in }M_\odot/L_\odot) \qquad (4)

where: $M$ — dynamical mass within some radius; $L$ — luminosity within the same radius; $\Upsilon$ — mass-to-light ratio, $\sim1\text{–}5$ for stellar populations and $\gg10$ once the halo is enclosed. Reading: $\Upsilon$ rises steadily with radius, tracking the growing dominance of non-luminous mass. For pressure-supported ellipticals the estimator is the spherical Jeans equation — the first velocity moment of the collisionless Boltzmann equation for a static, non-rotating system, in which velocity dispersion acts as a pressure and orbital anisotropy enters through $\beta$ .

\frac{1}{\rho_*}\frac{\mathrm{d}(\rho_*\sigma_r^2)}{\mathrm{d}r}+\frac{2\beta\sigma_r^2}{r}=-\frac{\mathrm{d}\Phi}{\mathrm{d}r}=-\frac{GM(<r)}{r^2} \qquad (5)

where: $\rho_*(r)$ — stellar (tracer) number density; $\sigma_r$ — radial velocity dispersion; $\beta=1-\sigma_t^2/\sigma_r^2$ — orbital anisotropy ( $\beta=0$ isotropic); $\Phi$ — total potential. Reading: the same $M(<r)$ appears on the right, now entangled with $\beta$ — the mass–anisotropy degeneracy that makes ellipticals harder than disks. Hot gas gives an independent weigh-in: if the X-ray plasma is in hydrostatic equilibrium, its temperature and density gradients fix the enclosing mass directly.

M(<r)=-\frac{k_B T\,r}{\mu m_p G}\left(\frac{\mathrm{d}\ln\rho_{\rm gas}}{\mathrm{d}\ln r}+\frac{\mathrm{d}\ln T}{\mathrm{d}\ln r}\right) \qquad (6)

where: $T$ — temperature of the hot gas; $\rho_{\rm gas}$ — gas density from the X-ray surface brightness; $\mu$ — mean molecular weight; $m_p$ — proton mass; $k_B$ — Boltzmann constant. Reading: no stars need move — the gas pressure gradient alone reports a mass far above the stellar mass. Finally, lensing weighs mass by how it bends light, independent of dynamics. A source aligned behind a mass forms a ring of angular radius set by the enclosed projected mass.

\theta_E=\sqrt{\frac{4GM}{c^2}\frac{D_{LS}}{D_L D_S}} \qquad (7)

where: $\theta_E$ — angular Einstein radius; $M$ — projected mass within $\theta_E$ ; $D_L,D_S,D_{LS}$ — angular-diameter distances to lens, to source, and between them; $c$ — speed of light. Reading: because gravity deflects light regardless of the matter's motion, lensing masses cross-check the dynamical ones and break the anisotropy degeneracy.

Derivation AEnclosed mass and density for a flat rotation curve

The goal is to show, with no free parameters, that a strictly flat rotation curve is mathematically equivalent to an enclosed mass rising in direct proportion to radius and a density falling as the inverse square of radius — the singular isothermal sphere — and to make explicit why ordinary luminous matter cannot reproduce that profile at large radius. We start from the circular-speed estimator, impose $v_c=\text{const}$ , differentiate for the shell mass, and divide by the shell area for the density.

Step 1. Begin with force balance for a circular orbit, $v_c^2/r=GM(<r)/r^2$ , which rearranges to $M(<r)=v_c^2 r/G$ . With $v_c$ independent of $r$ , the right side is a constant times $r$ , so the enclosed mass grows linearly: doubling the radius doubles the mass inside. Why: this is the entire dark-matter argument in one line — a curve that does not turn over is a mass that does not stop growing.

Step 2. Differentiate with respect to radius: $\mathrm{d}M/\mathrm{d}r=v_c^2/G$ , a constant. Each spherical shell, of whatever radius, contributes the same mass per unit thickness. Why: constant shell mass per unit radius is the tell-tale of an isothermal structure and is exactly what a Keplerian, centrally concentrated mass does not do.

Step 3. Convert shell mass to local density by dividing by the shell area, $\rho(r)=(1/4\pi r^2)\,\mathrm{d}M/\mathrm{d}r=v_c^2/(4\pi G r^2)$ . The density falls as $r^{-2}$ — the singular isothermal sphere. Why: this closes the loop, turning a kinematic observation into a statement about how matter is distributed in space.

Step 4. Interpret. An exponential stellar disk and a de Vaucouleurs bulge both fall off far faster than $r^{-2}$ at large radius, so luminous matter cannot maintain this density where the curve is still flat. The required mass formally diverges as $r\to\infty$ and must be truncated in realistic models (the NFW profile does this), but the message survives: a distinct, extended, non-luminous component dominates the outer galaxy. Why: the derivation does not merely fit the data — it disqualifies the light-emitting matter as the source, which is what "dark matter" means here.

Go deeper · Derivation B · 1 minDynamical mass of an elliptical from its velocity dispersion

Derivation BDynamical mass of an elliptical from its velocity dispersion

The goal is to extract an order-of-magnitude dynamical mass for a pressure-supported elliptical directly from its measured stellar velocity dispersion, using the spherical Jeans equation, and to expose the mass–anisotropy degeneracy that keeps such masses from being pinned down by dispersions alone. We take the isotropic limit, approximate the logarithmic slopes as order unity, read off the virial-scale mass, then restore the caveat.

Step 1. Set $\beta=0$ in the Jeans equation, leaving $(1/\rho_*)\,\mathrm{d}(\rho_*\sigma_r^2)/\mathrm{d}r=-GM(<r)/r^2$ . Here $\rho_*$ is the luminous tracer density and $\sigma_r$ the radial dispersion; isotropy means the velocity ellipsoid is a sphere. Why: the isotropic case is the honest zeroth-order estimate — it keeps the physics (pressure support against gravity) while dropping the one parameter we cannot measure directly.

Step 2. Approximate the logarithmic derivative $\mathrm{d}\ln(\rho_*\sigma_r^2)/\mathrm{d}\ln r\sim-1$ and identify $\sigma_r$ with the observed line-of-sight $\sigma$ up to an order-unity constant, turning the equation into the scaling $\sigma^2/r\sim GM/r^2$ . Why: real dispersion profiles decline gently over a decade in radius, so a logarithmic slope near $-1$ is a defensible average — all an order-of-magnitude estimate needs.

Step 3. Solve for the mass: $M(<r)\sim\sigma^2 r/G$ , structurally identical to the disk estimator $v_c^2 r/G$ with $\sigma$ replacing $v_c$ . For a giant elliptical with $\sigma\approx250\ \mathrm{km\,s^{-1}}$ at $r\approx10\ \mathrm{kpc}$ this gives $M\sim1.5\times10^{11}\,M_\odot$ ; the virial coefficient pushes the careful value toward $\sim10^{11.5}\,M_\odot$ , already several times the stellar mass, with the ratio climbing at larger $r$ . Why: recovering the same $\sigma^2 r/G$ form shows that pressure support and rotation support are two dialects of one virial statement about self-gravitating systems.

Step 4. Restore the caveat. A single dispersion profile does not fix $M(<r)$ uniquely: radial orbits ( $\beta>0$ ) and tangential orbits ( $\beta<0$ ) reshape the projected dispersion in ways that mimic a change in mass. Breaking this mass–anisotropy degeneracy needs higher velocity moments, independent tracers such as globular clusters or planetary nebulae, or lensing. Why: naming the degeneracy keeps the estimate honest — the mass is real and large, but its precise value in an elliptical is harder to state than in a disk.

Go deeper · Derivation C · 1 minWhy light alone predicts a declining curve

Derivation CWhy light alone predicts a declining curve

The goal is to establish the null hypothesis the data refute: if a galaxy's mass followed its exponential light, the enclosed mass would saturate a few scale lengths out and the rotation curve would turn over into a Keplerian decline. Building that prediction explicitly is what makes the observed flatness surprising and defines the halo as the residual. We integrate the exponential surface density to a finite total mass and feed it through the circular-speed relation.

Step 1. Model the stellar disk with the Freeman exponential surface density, $\Sigma(R)=\Sigma_0 e^{-R/R_d}$ , and integrate the enclosed mass, $M(<R)=2\pi\Sigma_0\int_0^R e^{-R'/R_d}R'\,\mathrm{d}R'$ . Beyond several scale lengths the exponential kills the integrand and the mass approaches the constant $2\pi\Sigma_0 R_d^2$ . Why: the finiteness is the crux — an exponential disk has nearly all its mass within a few $R_d$ , so nothing is left to keep the curve up further out.

Step 2. Insert this saturating mass into $v_c^2=GM(<R)/R$ . With $M$ effectively fixed at large $R$ , $v_c^2\to2\pi G\Sigma_0 R_d^2/R$ , so $v_c\propto R^{-1/2}$ — a Keplerian falloff should set in just past the bright disk, as the planets slow beyond the Sun. Why: this is the sharp, testable prediction of the mass-follows-light hypothesis — unambiguous, a decline not a plateau.

Step 3. Confront the data. HI 21-cm curves stay flat to two or three optical radii with no turnover, so the luminous prediction fails by a growing margin. The residual $v_{\rm halo}^2=v_c^2-v_{\rm lum}^2$ increases outward — precisely the contribution an extended halo supplies, the green curve in the second figure. Why: subtracting a well-constrained luminous model from a well-measured total is the cleanest detection of dark matter — a difference of measured quantities, not a fit to a preferred model.

Limiting cases

Case	Result	Meaning
Keplerian, mass all interior	$v_c(r)=\sqrt{GM/r}\propto r^{-1/2}$	Expected outside a point-like or centrally concentrated mass; NOT seen in galaxy outskirts.
Solid-body inner rise	$v_c(r)\propto r,\ \rho\approx\text{const}$	A uniform-density core (inner bulge/disk) gives a linearly rising curve near the center.
Flat curve, isothermal halo	$v_c=\text{const}\ \Rightarrow\ \rho(r)=\dfrac{v_c^2}{4\pi G r^2}$	The observed regime; $M(<r)=v_c^2 r/G$ grows linearly with radius.
Pressure-supported elliptical	$M(<r)\sim\dfrac{\sigma^2 r}{G}$	Order-of-magnitude dynamical mass from the dispersion $\sigma$ ; the coefficient depends on anisotropy via Jeans.

Worked examples

1 · The mass a flat curve demands (fully worked)

A massive spiral has a flat rotation curve at $v_c=220\ \mathrm{km\,s^{-1}}$ . Find the enclosed mass at $r=10\ \mathrm{kpc}$ and at $r=30\ \mathrm{kpc}$ , and compare with the $\sim2\times10^{10}\,L_\odot$ of starlight lying almost entirely inside $10\ \mathrm{kpc}$ .

Step 1. Use $M(<r)=v_c^2 r/G$ . In convenient units it becomes $M/M_\odot=2.325\times10^5\,(v_c/\mathrm{km\,s^{-1}})^2\,(r/\mathrm{kpc})$ , absorbing $G$ and the kpc-to-metre conversion. Squaring once, $v_c^2=(220)^2=4.84\times10^4$ in $(\mathrm{km\,s^{-1}})^2$ .

Step 2. At $r=10\ \mathrm{kpc}$ : $M=2.325\times10^5\times4.84\times10^4\times10=\mathbf{1.13\times10^{11}\,M_\odot}$ . This mass inside the bright disk already exceeds the stellar mass ( $\Upsilon_*\approx3$ on $2\times10^{10}\,L_\odot$ gives $\sim6\times10^{10}\,M_\odot$ ) by roughly a factor of two.

Step 3. At $r=30\ \mathrm{kpc}$ : because $M\propto r$ at fixed $v_c$ , the mass simply triples, $M=\mathbf{3.4\times10^{11}\,M_\odot}$ — yet essentially no new starlight lives between 10 and 30 kpc.

Step 4. Form the mass-to-light ratio out to 30 kpc: $\Upsilon=3.4\times10^{11}/2\times10^{10}\approx\mathbf{17\,(M/L)_\odot}$ , versus $\sim3$ for the stars alone. The enclosed mass grows by a factor of 3 across a region with no extra light, and $\Upsilon$ climbs from a stellar $\sim3$ to $\sim17$ — the dark halo made quantitative.

2 · Weighing a giant elliptical from its velocity dispersion

A luminous elliptical has a velocity dispersion $\sigma\approx250\ \mathrm{km\,s^{-1}}$ , roughly constant out to $r\approx10\ \mathrm{kpc}$ , and luminosity $\sim5\times10^{10}\,L_\odot$ . Estimate its dynamical mass within 10 kpc and its mass-to-light ratio.

Step 1. With no circular orbits, use the isotropic Jeans estimate $M(<r)\sim\sigma^2 r/G$ — the same convenient form with $\sigma$ for $v_c$ : $M/M_\odot\sim2.325\times10^5\,(\sigma/\mathrm{km\,s^{-1}})^2\,(r/\mathrm{kpc})$ .

Step 2. Insert the numbers: $\sigma^2=(250)^2=6.25\times10^4$ , so $M\sim2.325\times10^5\times6.25\times10^4\times10=\mathbf{1.5\times10^{11}\,M_\odot}$ . A careful virial coefficient of order a few raises this toward $\sim10^{11.5}\,M_\odot$ .

Step 3. The mass-to-light ratio within 10 kpc is $\Upsilon\sim1.5\times10^{11}/5\times10^{10}\approx\mathbf{3\text{–}6\,(M/L)_\odot}$ , already at the upper edge of a plausible old stellar population and rising outward where X-ray gas and lensing reach. A pressure-supported galaxy, with not one measured circular orbit, delivers the same verdict as a spiral: a dynamical mass of a few $\times10^{11}\,M_\odot$ and a $\Upsilon$ that outgrows its stars — subject to the mass–anisotropy caveat on the precise value.

Physics of the portrait

The flat rotation curve arrived not as a theory but as a stubborn measurement, made by an observer who trusted her spectra over the prejudice that galaxies needed no unseen mass.

Origins · 1 minVera Rubin and how this was discovered

Vera Rubin

1928–2016 · optical rotation curves and galactic dark matter

Vera Cooper Rubin turned dark matter from a theorist's puzzle into an observer's fact. Working at the Carnegie Institution with instrument-builder Kent Ford and his image-tube spectrograph, she measured the rotation of the Andromeda galaxy in the late 1960s by tracking the Doppler shifts of glowing HII regions to unprecedented radius, publishing the optical rotation curve of M31 with Ford in 1970. The curves did not fall the way a Keplerian, mass-follows-light galaxy demanded — they stayed stubbornly flat. Through the 1970s Rubin, Ford and Norbert Thonnard extended the work to a large sample of spirals of every luminosity, and in 1980 reported that flatness was generic: every spiral enclosed far more mass than its light could account for, the discrepancy worsening outward. The result meshed with Fritz Zwicky's 1933 virial analysis of the Coma cluster, with the 21-cm curves tracing hydrogen further out, and with Ostriker and Peebles' argument that cold disks need a massive halo to survive. Rubin, temperamentally cautious, presented the kinematics and let the mass speak; the community's shift toward dark-matter halos followed from data no one could wave away, and she is now credited with the decisive galactic evidence for the missing mass.

History

1933

— Zwicky's virial analysis of the Coma cluster reveals unseen mass

1939

— Babcock measures the rotation of Andromeda out to large radius

1970

— Rubin and Ford publish the optical rotation curve of Andromeda

1978

— Bosma's HI survey shows rotation curves staying flat past the stellar disk

1980

— Rubin, Ford and Thonnard establish flat curves as generic in spirals

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 24–25 (galactic dynamics and dark matter) · Binney & Tremaine, Galactic Dynamics, 2nd ed., §2.2 (isothermal sphere) and ch. 4 (Jeans equations) · Mo, van den Bosch & White, Galaxy Formation and Evolution, ch. 11 (galaxy dynamics) · Ryden, Introduction to Cosmology, 2nd ed., ch. 8 · Dodelson, Modern Cosmology, ch. 1 · Rubin, Ford & Thonnard (1980), ApJ 238, 471. Key results used throughout: $M(<r)=v_c^2 r/G$ with $\rho\propto r^{-2}$ by differentiation; the convenient form $M/M_\odot=2.325\times10^5\,(v/\mathrm{km\,s^{-1}})^2(r/\mathrm{kpc})$ , giving $1.13\times10^{11}\,M_\odot$ at 10 kpc for $v_c=220\ \mathrm{km\,s^{-1}}$ ; the exponential-disk enclosed-mass integral yielding $v_c\propto R^{-1/2}$ ; the Jeans reduction to $M\sim\sigma^2 r/G$ and $\sim1.5\times10^{11}\,M_\odot$ for $\sigma=250\ \mathrm{km\,s^{-1}}$ at 10 kpc; and the Einstein-radius mass formula. Quoted amplitudes ( $v_c\sim200\text{–}300\ \mathrm{km\,s^{-1}}$ , $R_0\approx8\ \mathrm{kpc}$ , local DM density $\sim0.3\text{–}0.4\ \mathrm{GeV\,cm^{-3}}$ , Tully–Fisher $L\propto V^4$ ) are standard values.

Prerequisites. Galactic Rotation Curves · Neutral Hydrogen 21-cm Line and Galactic Kinematics · Structure of the Milky Way Galaxy · Elliptical Galaxies and the de Vaucouleurs Profile. Leads to. Galactic Dark Matter Halos and Density Profiles · Galactic Halo and Globular Cluster System · Spiral Galaxies and Density Wave Theory.

Ask the tutor: Ask me to walk through how a measured 21-cm rotation curve gets decomposed into disk, gas, and halo — or how the Bullet Cluster separates dark matter from baryons.

Does a flat rotation curve prove dark matter, or could it be modified gravity?

A single flat curve is consistent with either a dark halo or a modification of gravity such as MOND, which alters the force law at low accelerations rather than adding mass. What tilts the balance is the breadth of the evidence: the same missing mass shows up in ellipticals via the Jeans equation, in hot X-ray gas via hydrostatics, in lensing that weighs mass independent of dynamics, and — decisively at larger scales — in galaxy clusters and the cosmic microwave background, where the required dark component is non-baryonic. Modified gravity struggles to fit all of these at once, especially the Bullet Cluster, where the lensing mass is offset from the gas. The concordance reading is cold dark matter halos, though the tight Tully–Fisher and radial-acceleration relations remain a genuine constraint any theory must reproduce.

Why is the neutral-hydrogen 21-cm line so important for this argument?

Because it lets you measure $v_c(r)$ where there are almost no stars. The optical rotation curve, traced by starlight and HII regions, typically runs out near the edge of the bright disk — precisely where the Keplerian-versus-flat question is hardest to settle. Neutral atomic hydrogen, radiating the 21-cm hyperfine line, extends in a thin disk far beyond the stars, often to two or three optical radii, so any circular speed measured there is essentially all halo — the luminous contribution $v_{\rm lum}$ has already fallen off. The persistent flatness of the 21-cm curve is therefore the cleanest single detection of dark matter in an individual galaxy, and it anchors the halo-profile modelling that follows.

What exactly is a singular isothermal sphere, and is it realistic?

It is the density profile $\rho(r)=v_c^2/(4\pi G r^2)$ that a perfectly flat rotation curve implies — "isothermal" because it is the density of a self-gravitating gas at constant velocity dispersion, and "singular" because it diverges both at the center and, in total mass, at infinity. Both divergences are unphysical, so it is a limiting idealization, not a literal halo. Real halos are better described by the NFW profile from cold-dark-matter simulations, with a gentler $r^{-1}$ inner cusp and a steeper $r^{-3}$ outer falloff, giving finite mass and a curve only approximately flat over the observed range. The isothermal sphere survives as the simplest model that captures the essential fact — mass rising linearly, density falling as $r^{-2}$ — over the radii where curves are measured.

How can mass-to-light ratio alone signal dark matter?

A stellar population of a given age and metallicity has a fairly well-understood mass-to-light ratio: old stars run a few solar masses per solar luminosity, young ones less. When the dynamical $\Upsilon=M/L$ — the mass a rotation curve or dispersion demands, divided by the light actually seen — climbs to tens or hundreds as you integrate outward, that excess cannot be hidden in ordinary stars, whose $\Upsilon$ is bounded. The rise directly measures the growing fraction of mass that emits no light — the single-number version of the whole argument: each unit of light drags along far more mass than any stellar population can supply.

Could the missing mass just be faint baryons — cold gas, dim stars, black holes?

This was taken seriously and largely ruled out. Faint stars and stellar remnants (the MACHO hypothesis) were searched for by microlensing toward the Magellanic Clouds and found far too few to make up the halo, and cold molecular gas lacks the required mass. Most decisively, the total baryon budget is fixed independently by Big Bang nucleosynthesis and the cosmic microwave background at about 5% of the critical density, while the total matter density is about 30% — the gap is non-baryonic by construction, so it cannot be ordinary atoms, however dark. A single galaxy's halo may hold some faint baryons, but the dominant component must be a new kind of matter.

Why do ellipticals need a different method than spirals if the conclusion is the same?

Because ellipticals are held up by pressure, not rotation. A spiral's stars and gas move on nearly circular orbits, so a single speed at each radius cleanly weighs the enclosed mass. An elliptical's stars swarm on randomly oriented orbits with little net rotation; there is no $v_c$ , only a velocity dispersion $\sigma$ . The Jeans equation converts that dispersion into a mass, $M\sim\sigma^2 r/G$ , but it carries an extra unknown — the orbital anisotropy $\beta$ — and different $\beta$ can mimic different masses. So ellipticals give the same answer (dynamical mass well above stellar mass, rising outward) with a larger honest uncertainty, which is why hot X-ray gas and lensing are enlisted to pin them down.