Velocity Dispersion

What velocity dispersion really means

Take a snapshot of every star in an elliptical galaxy and tag each with its velocity. Some race toward you, some away, some sideways. The average of all those velocities is just the galaxy's bulk motion through space — uninteresting for measuring its mass. What matters is the scatter around that average: how far individual stars deviate from the mean. That scatter, expressed as a standard deviation, is the velocity dispersion, written σ.

Crucially, we almost never measure the full three-dimensional velocity of a distant star. Spectroscopy gives only the line-of-sight component — the part of the motion pointing toward or away from us — via the Doppler shift. So when astronomers quote σ for a galaxy, they nearly always mean the line-of-sight velocity dispersion, σ_los. For an isotropic system (orbits equally likely in any direction) the full 3D dispersion is √3 times the line-of-sight value.

Velocity dispersion captures random motion, in contrast to ordered motion like the smooth rotation of a spiral disk. The distinction sorts galaxies into two families. Spiral galaxies are rotation-supported: their stars orbit coherently in a disk, and the rotation speed v far exceeds the dispersion σ. Elliptical galaxies are pressure-supported: they barely rotate, and it is the random buzzing of stars on plunging, criss-crossing orbits — like molecules in a hot gas — that resists gravitational collapse. The ratio v/σ is a compact way to say which kind of support a galaxy uses.

Reading σ from broadened spectral lines

You cannot resolve individual stars in a galaxy a hundred million light-years away — they blur into a single smear of light. But each star's spectrum carries narrow absorption lines (calcium, magnesium, iron, the prominent Ca II triplet near 850 nm), and each star Doppler-shifts those lines by a different amount depending on its velocity. Stack millions of slightly shifted spectra and the lines no longer look narrow: they broaden into a smooth, bell-shaped trough. The width of that broadening is the velocity dispersion.

This is the elegant trick at the heart of the measurement. The integrated galaxy spectrum is the spectrum of one star convolved with the line-of-sight velocity distribution. To extract σ, astronomers cross-correlate the galaxy spectrum against a template — a sharp-lined nearby star of the right type — and recover the broadening function. A broader trough means a larger σ. The technique routinely measures dispersions down to a few km/s, far below the resolution of most spectrographs, because line broadening is a statistical width, not a single shift.

Modern instruments push this further. Integral-field spectrographs (such as MUSE on the VLT) map σ across an entire galaxy pixel by pixel, revealing how dispersion rises toward the dense center and how it varies with radius — exactly the kind of data the visualization above evokes, where a tight spectral line fans out into a broad band as the stars scatter in velocity.

Weighing galaxies with the virial theorem

Here is why σ is so prized: it weighs things you cannot see. For a self-gravitating system that has settled into equilibrium, the virial theorem states that twice the kinetic energy equals the magnitude of the gravitational potential energy (2K = |U|). Working that out gives a mass estimate:

M ≈ k · σ² R / G

where R is a characteristic radius (often the half-light or effective radius), G is Newton's gravitational constant, and k is a dimensionless factor of order a few that depends on the system's structure and orbital anisotropy. Plug in σ ≈ 200 km/s and R ≈ 3 kpc for a giant elliptical and you recover ~10¹¹–10¹² solar masses — and that total includes the dark matter, because σ responds to all the gravitating mass, not just the glowing stars.

This is exactly how Fritz Zwicky stumbled onto dark matter in 1933. Measuring the galaxy-to-galaxy velocity dispersion of the Coma cluster (~1000 km/s), he applied the virial theorem and found the cluster needed far more mass than its visible galaxies could supply — a factor of hundreds. He called the deficit dunkle Materie, dark matter. The same logic now weighs dwarf spheroidal galaxies, whose tiny σ ≈ 5–10 km/s is still far too large for their handful of visible stars, marking them as the most dark-matter-dominated objects known.

The M-sigma relation: σ and the central black hole

One of the most surprising results in modern astrophysics is that a galaxy's velocity dispersion predicts the mass of the supermassive black hole hiding in its core. The M-sigma relation, established around 2000 by Ferrarese & Merritt and Gebhardt and collaborators, is a remarkably tight power law:

M_BH ∝ σ⁴ to σ⁵

A bulge with σ ≈ 200 km/s reliably hosts a black hole of roughly 10⁸ solar masses, with surprisingly little scatter. What makes this strange is that the black hole's gravity is utterly negligible across most of the bulge — it is millionths of the galaxy's mass and dominates only the innermost parsecs. Yet σ, measured over kiloparsecs, "knows" the black hole's mass. The accepted explanation is co-evolution: as a galaxy assembles, energy and momentum injected by the accreting black hole (AGN feedback) heats and expels gas, self-regulating both the black hole's growth and the bulge's. The dispersion and the black hole grow in lockstep. The M-sigma relation is now a workhorse for estimating black hole masses in galaxies too distant for direct orbital measurement.

A ladder of velocity dispersions

The same quantity spans more than three orders of magnitude across the cosmic hierarchy. The larger and more massive the bound system, the deeper its potential well and the faster its members must move to stay in equilibrium:

System	Typical σ (line-of-sight)	What it tells you
Open cluster	~0.5–1 km/s	Loosely bound; many dissolve within ~10⁸ yr
Globular cluster	~7–15 km/s	Tightly bound; little to no dark matter
Dwarf spheroidal galaxy	~5–10 km/s	Tiny σ but huge mass-to-light → dark-matter dominated
Milky Way stellar disk	~20–60 km/s	Rises with stellar age (disk heating)
Giant elliptical galaxy	~200–300 km/s	Pressure-supported; tracks bulge & black-hole mass
Brightest cluster galaxy	~300–400 km/s	Among the most massive single galaxies
Rich galaxy cluster	~700–1200 km/s	Weighs the cluster's full dark-matter halo

Notice that σ also feeds two of the most useful scaling laws in extragalactic astronomy. The Faber-Jackson relation links an elliptical galaxy's luminosity to its dispersion, L ∝ σ⁴, and is the elliptical-galaxy analogue of the Tully-Fisher relation for spirals. Folding in galaxy size and surface brightness tightens it into the fundamental plane, a near-flat surface in the three-dimensional space of σ, radius, and brightness that lets σ serve as a distance indicator across the universe.

Subtleties and pitfalls

• Anisotropy degeneracy. The virial mass depends on whether orbits are mostly radial (plunging) or tangential (circular). The same σ can correspond to different masses, so the structure factor k is genuinely uncertain — the classic mass–anisotropy degeneracy.
• Equilibrium assumption. The virial theorem assumes a relaxed, settled system. A galaxy caught mid-merger, or a cluster still collapsing, can yield a badly inflated σ and a fictitious mass.
• Aperture effects. σ usually falls with radius, so the value you measure depends on how big a slit or fiber you used. Comparisons must be corrected to a standard aperture (e.g. within one effective radius).
• Rotation contamination. If even a slowly rotating galaxy is observed off-axis, ordered rotation leaks into the measured line width and must be modeled out to isolate the true random dispersion.