Jeans Instability

A one-percent ripple that builds a star

Take a cold, vast cloud of molecular hydrogen — the kind that drifts through the spiral arms of the Milky Way, hundreds of light-years across, a few hundred atoms per cubic centimetre. It is, very nearly, in balance. Gravity pulls every atom toward every other atom, trying to draw the whole thing inward. Gas pressure pushes back out, resisting compression. For most clouds, most of the time, this is a stalemate that can last tens of millions of years.

Now squeeze one spot just one percent denser. Whether that tiny ripple fades back into the cloud or grows without limit into a collapsing core decides whether a star is born. The answer comes from a single comparison — the size of the perturbation against one critical length. If the ripple is smaller than that length, pressure smooths it away and the cloud bounces. If it is larger, gravity wins, the region falls inward, and the collapse runs away. That critical length is the Jeans length, and the whole phenomenon is the Jeans instability — the foundational trigger for star formation and, in its cosmological form, for the growth of every galaxy and cluster in the universe.

What makes it powerful is that the criterion is local and quantitative. You do not need to simulate the whole cloud; you only need to compare a perturbation's wavelength to λ_J = c_s √(π / Gρ), or equivalently its mass to the Jeans mass. Everything else — fragmentation, the stellar initial mass function, the timescale of collapse — follows from how this single threshold behaves as the gas evolves.

How it works: pressure waves versus free fall

The cleanest way to understand the Jeans criterion is a race between two clocks.

The pressure clock. If you compress a patch of gas, the overdensity launches a pressure (sound) wave outward. That wave is how a fluid "communicates" support — it redistributes pressure to push back against the compression. The time for the wave to cross a region of size λ is

t_sound ≈ λ / c_s

where c_s is the sound speed in the gas (for cold molecular gas, c_s ≈ 0.2–0.3 km/s).

The gravity clock. Independently, gravity is pulling the region inward. The time for a self-gravitating blob to collapse — the free-fall time — depends only on its density:

t_ff = √(3π / 32 Gρ)  ≈  1 / √(Gρ)

Now race them. If t_sound < t_ff, the pressure wave crosses the region and stiffens it against collapse before gravity can act — pressure wins, the perturbation oscillates and dies away as an ordinary sound wave. If t_sound > t_ff, gravity pulls the region together faster than any pressure wave can react — gravity wins, and the region collapses. The boundary, t_sound = t_ff, is exactly the Jeans length:

λ / c_s = 1 / √(Gρ)   ⟹   λ_J ≈ c_s / √(Gρ)
Full linear result:    λ_J = c_s √(π / Gρ)

The order-of-magnitude race gets the scaling right; the exact √π prefactor comes out of solving the linearised fluid equations properly, which we do below.

Derivation: the dispersion relation

Start with an infinite, uniform, self-gravitating gas at rest with density ρ₀, pressure P₀, and sound speed c_s² = ∂P/∂ρ. Write the three governing equations — continuity, Euler (momentum), and Poisson (gravity):

∂ρ/∂t + ∇·(ρ v) = 0                 (continuity)
∂v/∂t + (v·∇)v = −(1/ρ)∇P − ∇Φ      (Euler)
∇²Φ = 4πGρ                          (Poisson)

Perturb each quantity by a small plane wave ∝ exp[i(k·x − ωt)] — density ρ = ρ₀ + δρ, velocity v = δv, potential Φ = Φ₀ + δΦ — and keep only first-order terms. (Here is where the famous "Jeans swindle" lives: the uniform background Φ₀ is quietly dropped, which is not strictly consistent in static Newtonian gravity but is rigorously recovered in an expanding universe.) Combining the linearised equations gives the dispersion relation:

ω² = c_s² k² − 4πGρ₀

Read this carefully. For large k (small wavelength), the c_s²k² term dominates: ω² > 0, ω is real, the perturbation oscillates — these are stable sound waves. For small k (large wavelength), the −4πGρ₀ term dominates: ω² < 0, ω is imaginary, and the perturbation grows exponentially as exp(|ω|t) — collapse. The crossover, ω² = 0, defines the Jeans wavenumber and Jeans length:

k_J² = 4πGρ₀ / c_s²
λ_J = 2π / k_J = c_s √(π / Gρ₀)

So perturbations with λ > λ_J (k < k_J) collapse; those with λ < λ_J are stable. This is the Jeans instability in its exact form. The maximum growth rate as k → 0 is |ω| → √(4πGρ₀) ≈ 1/t_ff, confirming that collapse, once underway, proceeds on the free-fall time.

Worked example: a molecular cloud core

Take a dense core inside a giant molecular cloud — the kind that becomes a single star or small multiple. Adopt typical numbers: temperature T = 10 K, mean molecular weight μ = 2.3 (mostly H₂ with some helium), and number density n = 10⁴ molecules/cm³.

Step 1 — sound speed. The isothermal sound speed is c_s = √(k_B T / μ m_H):

c_s = √[(1.38×10⁻²³ × 10) / (2.3 × 1.67×10⁻²⁷)]
    = √(3.6×10⁴) m²/s²
    ≈ 190 m/s  ≈ 0.19 km/s

Step 2 — mass density. ρ = n × μ m_H = 10⁴ cm⁻³ × (2.3 × 1.67×10⁻²⁷ kg) per cm³. Converting to SI (1 cm⁻³ = 10⁶ m⁻³):

ρ = (10⁴ × 10⁶ m⁻³) × (2.3 × 1.67×10⁻²⁷ kg)
  = 10¹⁰ × 3.84×10⁻²⁷
  ≈ 3.8×10⁻¹⁷ kg/m³

Step 3 — free-fall time. t_ff = √(3π / 32 Gρ):

t_ff = √[ (3π) / (32 × 6.67×10⁻¹¹ × 3.8×10⁻¹⁷) ]
     = √(9.42 / 8.1×10⁻²⁶)
     = √(1.16×10²⁶) s
     ≈ 1.08×10¹³ s
     ≈ 3.4×10⁵ years

A few hundred thousand years — and crucially, that number depends only on ρ, not on how big the core is.

Step 4 — Jeans length and Jeans mass. λ_J = c_s √(π / Gρ):

λ_J = 190 × √[ π / (6.67×10⁻¹¹ × 3.8×10⁻¹⁷) ]
    = 190 × √(π / 2.53×10⁻²⁷)
    = 190 × √(1.24×10²⁷)
    = 190 × 3.5×10¹³
    ≈ 6.7×10¹⁵ m  ≈ 0.22 light-years  ≈ 0.07 pc

M_J ≈ (4/3)πρ (λ_J/2)³
    = 4.19 × 3.8×10⁻¹⁷ × (3.35×10¹⁵)³
    ≈ 1.59×10⁻¹⁶ × 3.76×10⁴⁶
    ≈ 6×10³⁰ kg  ≈ 3 M☉

So a 10 K core at 10⁴ cm⁻³ has a Jeans mass of a few solar masses and a Jeans length of a few hundredths of a parsec. Any clump in this core more massive than a few suns must collapse — and it does so within ~10⁵–10⁶ years. This is precisely why molecular clouds make stars of roughly solar mass, on timescales short compared to the age of the cloud.

The surprise: collapse breeds more collapse

The most counter-intuitive — and most consequential — feature of the Jeans instability is what happens once collapse begins. As the gas contracts, its density ρ rises. But for a cloud that stays roughly isothermal (it efficiently radiates away the heat of compression, which cold molecular gas does well), the temperature barely changes. Look again at the Jeans mass scaling:

M_J ∝ c_s³ / √(G³ ρ) ∝ T^(3/2) / √ρ

With T fixed and ρ climbing, M_J falls. Regions that were individually too small to collapse at the start — sub-Jeans, stable, just bouncing as sound waves — find that as the parent cloud densifies, the threshold drops below them. They become Jeans-unstable and begin their own independent collapse. The cloud does not fall into a single point; it shatters into a hierarchy of ever-smaller collapsing fragments. This is hierarchical fragmentation, and it is why a single molecular cloud forms not one star but a whole cluster of hundreds or thousands.

The fragmentation cascade does not continue forever. Eventually a contracting fragment becomes dense enough to be opaque to its own infrared radiation. It can no longer dump the compression heat, so it begins to heat up; now T rises, M_J stops falling and turns around, and fragmentation halts. This opacity limit sets the smallest possible fragment mass at roughly 0.01 solar masses — comfortably below the hydrogen-burning limit, and a key reason the stellar initial mass function turns over near a few tenths of a solar mass rather than continuing to ever-smaller stars.

Variants and regimes

The textbook Jeans criterion assumes a static, non-rotating, unmagnetised, non-turbulent gas. Each of those idealisations corresponds to a real effect that modifies the threshold:

• Bonnor–Ebert sphere. A pressure-confined isothermal sphere — the realistic version of a cloud core sitting inside a hotter ambient medium. It has a maximum stable mass (the Bonnor–Ebert mass, ~1.18 c_s⁴ / √(G³ P_ext)) closely related to M_J; exceed it and the sphere collapses. Observed prestellar cores like B68 are well fit by Bonnor–Ebert profiles near the critical point.
• Magnetic support. Magnetic fields add pressure and tension. A cloud is supported if its mass-to-magnetic-flux ratio is below a critical value (subcritical); above it (supercritical), gravity wins. Subcritical cores collapse only slowly, via ambipolar diffusion that lets neutral gas slip past the field.
• Rotation. Angular momentum provides centrifugal support, raising the effective collapse threshold and flattening the collapse into a disk — which is how protoplanetary disks arise.
• Turbulence. Supersonic turbulence adds non-thermal pressure that supports clouds globally, but its shocks also create the dense filaments and cores where local Jeans collapse is triggered. Modern theory replaces c_s with an effective velocity dispersion in the Jeans formula.
• Expanding-universe (cosmological) Jeans. In an expanding background the growth is power-law rather than exponential, because expansion fights collapse. The comoving Jeans length divides oscillating acoustic modes from growing modes — the basis of cosmological structure formation.
• Non-isothermal / adiabatic gas. If the gas cannot radiate and behaves adiabatically (γ = 5/3), compression heats it, c_s rises, and the Jeans mass increases with density — fragmentation is suppressed. The isothermal-vs-adiabatic distinction is what decides whether a cloud fragments or collapses monolithically.

From recombination to the first stars

The Jeans criterion is not only about star formation today — it explains why structure exists at all. After the Big Bang, baryons were tightly coupled to photons. The radiation pressure made the baryon–photon fluid extraordinarily stiff: the sound speed was c_s ≈ c/√3, an appreciable fraction of the speed of light. With such a huge c_s, the baryonic Jeans mass exceeded 10¹⁷ solar masses, far larger than anything that existed. Baryon perturbations could not collapse; they could only oscillate as the acoustic waves we now see frozen into the cosmic microwave background and the baryon acoustic oscillation scale.

Then came recombination at redshift z ≈ 1100, when the universe cooled enough for protons and electrons to combine into neutral hydrogen. Photons decoupled, the radiation pressure vanished almost overnight, and the baryonic sound speed crashed by orders of magnitude. The Jeans mass plummeted from 10¹⁷ M☉ to roughly 10⁵–10⁶ M☉. For the first time baryons could fall into the dark matter potential wells that had been growing unimpeded since matter–radiation equality. The result was the collapse of the first gas clouds and the formation of the first (Population III) stars at z ≈ 20–30. Every galaxy, every star, every planet traces back to that moment when the Jeans mass dropped and gas could finally fall.

Jeans length and mass across cosmic environments

Environment	T (K)	n (cm⁻³)	c_s (km/s)	λ_J	M_J	t_ff
Diffuse ISM (warm)	8000	0.3	~8	~5 kpc	~10⁶ M☉	~80 Myr
Cold neutral medium	80	40	~0.8	~30 pc	~10³ M☉	~7 Myr
Molecular cloud	20	100	~0.3	~6 pc	~10³ M☉	~4 Myr
Cloud core	10	10⁴	~0.19	~0.07 pc	~1–3 M☉	~0.3 Myr
Prestellar core	10	10⁶	~0.19	~0.007 pc	~0.1 M☉	~30 kyr
Opacity limit	~10	~10¹¹	~0.2	tiny	~0.01 M☉	~30 yr
Post-recombination gas	~3000	~200 (then)	~5	cosmological	~10⁵–10⁶ M☉	—

The trend down each column tells the whole story: as density rises (down the table), the Jeans mass and Jeans length collapse and the free-fall time shrinks. That monotonic shrinking is the engine of hierarchical fragmentation, carrying the gas from kiloparsec-scale diffuse clouds all the way down to hundredth-of-a-solar-mass protostellar seeds.

Where the Jeans instability shows up

• Star formation. The direct application: it sets which molecular cloud cores collapse, on what timescale, and into roughly what mass — the starting point for the entire theory of star formation and the stellar initial mass function.
• Cosmological structure formation. The expanding-universe Jeans analysis underlies the growth of the density perturbations seen in the CMB into the cosmic web of galaxies and clusters.
• Galaxy disk stability. The Toomre Q criterion is the rotating, disk-geometry analogue of the Jeans criterion, governing whether a galactic disk fragments into spiral arms and giant molecular clouds.
• Planet formation by disk instability. In massive protoplanetary disks, a Jeans/Toomre-type instability can directly fragment gas into gas-giant planets, an alternative to core accretion.
• Globular cluster and first-object formation. The post-recombination Jeans mass of ~10⁵–10⁶ M☉ matches the mass scale of the first collapsed gas clouds and is invoked in models of globular cluster and Population III star formation.

Common pitfalls and misconceptions

• Thinking the cloud needs a trigger to collapse. The Jeans instability is spontaneous: any perturbation larger than λ_J grows on its own. External triggers (a passing shock, a nearby supernova) help by raising ρ and lowering λ_J, but they are not required.
• Assuming t_ff depends on cloud size. It does not — t_ff = √(3π/32Gρ) depends only on density. A small dense core and a large cloud of the same density collapse in the same time. Size enters only through whether the cloud exceeds λ_J in the first place.
• Forgetting that the Jeans mass falls during collapse. Treating M_J as fixed misses the entire fragmentation cascade. Because M_J ∝ T^(3/2)/√ρ and ρ rises while T stays low, M_J drops and the cloud shatters.
• Confusing the isothermal and adiabatic sound speeds. Star-forming gas is nearly isothermal (it radiates efficiently), so use c_s = √(k_BT/μm_H). Using the adiabatic c_s overestimates support by a factor of √γ.
• Taking the "Jeans swindle" as a fatal flaw. Dropping the background potential is mathematically inconsistent in a static medium, but the resulting dispersion relation is rigorously reproduced in an expanding universe, so the criterion is physically sound.
• Ignoring magnetic and turbulent support. Pure thermal Jeans gives a lower bound. Real cores are often near the magnetic-critical and virial boundaries; the thermal criterion is the right zeroth-order picture, not the full story.