Molecular Clouds and Giant Molecular Clouds

A molecular cloud is what the interstellar medium becomes when gas finally wins the fight to keep itself cold. Cosmic rays and starlight still pour heat in, but a cloud grown dense enough shields its interior behind a skin of dust, hydrogen pairs into $\mathrm{H_2}$ , and CO lines and thermal grain radiation carry the heat away until the gas settles near $10\text{–}20\,\mathrm{K}$ . These clouds fill a negligible fraction of the Galactic disk's volume yet hold much of its gas mass, and — decisively — they are the only place in a galaxy where stars are born. The physics is a contest between one attractor and three defenders. Self-gravity pulls the cloud toward collapse; against it stand thermal pressure (feeble at $10\,\mathrm{K}$ ), supersonic turbulence (dominant on large scales), and magnetic fields threading the gas. Which wins is decided by three touchstone quantities that recur through this article: the Jeans mass, the mass unstable at a given temperature and density; the free-fall time, how fast a pressureless collapse proceeds; and the virial theorem, the energy bookkeeping of gravity versus support. Together they explain the central, humbling fact of the field — only a few percent of a cloud's gas turns into stars per free-fall time, so star formation is a slow drip, not a runaway avalanche. And a catch shapes every observation: the $\mathrm{H_2}$ carrying the mass is nearly invisible at these temperatures, so almost all we know is read through surrogates — CO lines, dust, and the conversions back to $\mathrm{H_2}$ .

Key takeaways

Molecular clouds are the coldest ( $T\sim 10\text{–}20\,\mathrm{K}$ ) and densest ( $n\gtrsim 10^2\text{–}10^6\,\mathrm{cm^{-3}}$ ) phase of the interstellar medium, mostly $\mathrm{H_2}$ with dust and trace CO; giant molecular clouds (GMCs) reach $10^5\text{–}10^6\,M_\odot$ over tens of parsecs and host essentially all star formation.
Because $\mathrm{H_2}$ has no permanent dipole moment and stays dark, clouds are traced indirectly — through the CO(1–0) line at $115\,\mathrm{GHz}$ ( $2.6\,\mathrm{mm}$ ), dust extinction and emission, and the empirical CO-to- $\mathrm{H_2}$ conversion $X_{\rm CO}$ .
A cloud's fate is fixed by three quantities: the Jeans mass $M_J\propto T^{3/2}\rho^{-1/2}$ , the free-fall time $t_{\rm ff}\propto\rho^{-1/2}$ , and the virial parameter $\alpha_{\rm vir}=5\sigma_v^2 R/GM$ .
Observed GMCs cluster near $\alpha_{\rm vir}\sim 1\text{–}2$ : marginally bound, held up mostly by supersonic turbulence and magnetic fields, not thermal pressure.
Larson's relations tie velocity dispersion to size, $\sigma_v\propto L^{0.38\text{–}0.5}$ , exposing internal motions that are supersonic from parsec scales down to individual cores.
Star formation is strikingly inefficient — only $\epsilon_{\rm ff}\sim 1\%$ of the gas becomes stars per free-fall time — and explaining that number links cloud microphysics to galactic star-formation laws.
Key idea: A molecular cloud is a parcel of cold, dust-shielded, self-gravitating $\mathrm{H_2}$ whose fate is decided by the competition between self-gravity and the combined support of thermal pressure, supersonic turbulence, and magnetic fields — quantified by the virial theorem, the Jeans mass, and the free-fall time.

Where this lives today

Molecular-cloud physics is a hotbed of observation, driven by instruments that resolve the hierarchy in the first figure. The Atacama Large Millimeter/submillimeter Array (ALMA) images prestellar cores and protostellar disks at sub-arcsecond resolution across the Galaxy and nearby galaxies, testing fragmentation core by core. The Herschel far-infrared survey showed that clouds are laced with filaments of a near-universal $\sim0.1\,\mathrm{pc}$ width, along which cores form like beads on a string — reframing star formation as a filament-mediated process. Galaxy-scale CO surveys, from the Dame Milky Way map to the PHANGS-ALMA catalogues, test whether $\alpha_{\rm vir}\sim1$ and one $X_{\rm CO}$ hold everywhere, while Zeeman and polarization surveys (Planck, BLASTPol, ALMA) probe whether cores are sub- or supercritical. The central open problem endures: no first-principles theory yet predicts $\epsilon_{\rm ff}\sim1\%$ without tuning — closing that gap is the frontier linking clouds to the Kennicutt–Schmidt law of galaxies.

The physical picture

Picture a cold, dark cloud wrapped in a dusty skin that soaks up ultraviolet starlight. Outside the skin the gas is warm and atomic; inside, shielded from the photons that tear molecules apart, hydrogen pairs into $\mathrm{H_2}$ and cools to near $10\,\mathrm{K}$ . What holds the cloud up against gravity is not thermal pressure — at $10\,\mathrm{K}$ the sound speed is a feeble $0.2\,\mathrm{km\,s^{-1}}$ — but supersonic turbulence and the stiffness of magnetic field lines. The structure is hierarchical: a complex of tens of parsecs contains denser clumps, and inside those, sub-parsec prestellar cores where gravity finally wins and a star is made. The first figure gives the anatomy of that hierarchy with real numbers attached.

Figure

Anatomy of a giant molecular cloud: a UV-shielded interior at T ≈ 10–20 K and n ≳ 100 cm⁻³ spans ≈ 50 pc and 10⁵–10⁶ solar masses, clumps reach n ≈ 10⁴ cm⁻³, and a prestellar core at n ≳ 10⁶ cm⁻³ collapses toward a single star while magnetic field lines thread the complex and a CO photon escapes at 2.6 mm

Where turbulence sweeps gas into a dense enough knot, gravity wins locally and a core is born; elsewhere the same turbulence disperses the gas again. That is why only a small fraction of a cloud collapses at any moment, and why star formation is slow and regulated rather than a free-fall stampede. CO is the flashlight we shine into this darkness — abundant and easily excited, it lights up in the millimetre even though the $\mathrm{H_2}$ stays dark — but it is imperfect, saturating in its core and freezing onto grains in the coldest interiors. The deepest handle on cloud structure is Larson's relations: bigger clouds have bigger internal velocities. The second figure plots that linewidth–size law, whose punchline is that the motions are supersonic everywhere, ten to forty times the thermal sound speed.

Figure

Larson's linewidth–size relation σ_v ≈ 1.1 (L/pc)^0.38 km/s across three decades: a 0.3 pc core near 0.7 km/s, a 3 pc clump near 1.7 km/s, a 30 pc GMC near 4 km/s — all far above the 0.2 km/s thermal sound speed at 10 K, so the motions are supersonic and turbulent

The core equations

The stability of a cloud begins with the Jeans mass — the smallest mass that gas of temperature $T$ and density $\rho$ cannot hold up against its own gravity. It balances thermal against gravitational energy for a uniform sphere and scales as $T^{3/2}\rho^{-1/2}$ : cold, dense gas collapses easily; warm, rarefied gas does not.

M_J = \left(\frac{5 k_B T}{G \mu m_H}\right)^{3/2}\left(\frac{3}{4\pi\rho}\right)^{1/2} \qquad (1)

where: $M_J$ — minimum mass unstable to collapse; $T$ — gas temperature; $\rho$ — mass density; $\mu$ — mean molecular weight per particle ( $\approx 2.3$ for molecular gas with helium); $m_H$ — hydrogen mass; $k_B,\,G$ — Boltzmann's and Newton's constants. A region more massive than $M_J$ at its local $T$ and $\rho$ is doomed to collapse. Once a region is unstable, its collapse clock is the free-fall time — the time for a pressureless uniform sphere to fall to a point from rest. It depends only on density, so it sets one timescale for a whole cloud.

t_{\rm ff} = \sqrt{\frac{3\pi}{32 G \rho}} \qquad (2)

where: $t_{\rm ff}$ — collapse time of a pressureless uniform sphere; $\rho$ — initial mean mass density; $G$ — gravitational constant. Numerically $t_{\rm ff}\approx 3.4\times10^{6}\,\mathrm{yr}\,(n/10^2\,\mathrm{cm^{-3}})^{-1/2}$ for molecular gas — denser gas collapses faster. The decision between collapse and support is made by the scalar virial theorem, the steady-state energy budget of a bound cloud: neglecting surface and time-derivative terms, its kinetic, thermal, magnetic, and gravitational energies sum to zero.

2K + 2U_{\rm th} + \mathcal{M} + W = 0 \qquad (3)

where: $K$ — bulk (turbulent) kinetic energy; $U_{\rm th}$ — internal thermal energy; $\mathcal{M}$ — magnetic energy; $W<0$ — gravitational potential energy. The dimensionless virial parameter $\alpha_{\rm vir}\equiv 2K/|W| = 5\sigma_v^2 R/(GM)$ measures the balance: $\alpha_{\rm vir}<1$ is bound, $\alpha_{\rm vir}\sim 1$ marginal, $\alpha_{\rm vir}\gg 1$ unbound. Turning the theorem around gives a mass measurable without the CO calibration. Setting $2K = |W|$ for a uniform sphere and solving for $M$ yields the virial mass, built from a cloud's linewidth and size.

M_{\rm vir} = \frac{5\,\sigma_v^2 R}{G} \qquad (4)

where: $M_{\rm vir}$ — mass implied by virial balance; $\sigma_v$ — one-dimensional velocity dispersion from the line profile; $R$ — cloud radius from the map. Using only kinematics and geometry, $M_{\rm vir}$ is an independent cross-check on the CO mass. The linewidth in $M_{\rm vir}$ is not random: it obeys Larson's linewidth–size relation, an empirical power law in which larger clouds have larger internal motions — the signature of a turbulent cascade.

\sigma_v \approx 1.10\,\mathrm{km\,s^{-1}}\left(\frac{L}{\mathrm{pc}}\right)^{0.38} \qquad (5)

where: $\sigma_v$ — one-dimensional velocity dispersion; $L$ — cloud size scale. Larson's (1979) exponent is $0.38$ ; later surveys of virialized clouds find $\approx 0.5$ , the value expected when $\alpha_{\rm vir}\sim 1$ at constant surface density. Because the $\mathrm{H_2}$ is invisible, the mass is read from CO through the empirical X-factor, converting velocity-integrated CO(1–0) intensity into an $\mathrm{H_2}$ column density.

N_{\mathrm{H_2}} = X_{\rm CO}\, W_{\rm CO} \qquad (6)

where: $N_{\mathrm{H_2}}$ — $\mathrm{H_2}$ column density; $W_{\rm CO}$ — velocity-integrated CO(1–0) intensity in $\mathrm{K\,km\,s^{-1}}$ ; $X_{\rm CO}\approx 2\times10^{20}\,\mathrm{cm^{-2}\,(K\,km\,s^{-1})^{-1}}$ for the Milky Way. Finally, the payoff quantity for galaxies: the star-formation rate per free-fall time, the fraction of gas made into stars each $t_{\rm ff}$ .

\mathrm{SFR} = \epsilon_{\rm ff}\,\frac{M_{\rm gas}}{t_{\rm ff}},\qquad \epsilon_{\rm ff}\sim 0.01 \qquad (7)

where: $\mathrm{SFR}$ — star-formation rate; $M_{\rm gas}$ — molecular gas mass; $\epsilon_{\rm ff}$ — efficiency per free-fall time, observed to be about one percent — the number every theory of the star-forming ISM must reproduce.

Derivation AThe Jeans mass from the energy balance of a sphere

The goal is to obtain the Jeans mass, equation (1), from a single physical statement — that a cloud collapses when its self-gravity overpowers the thermal motions of its particles — and to see why the threshold scales as $T^{3/2}\rho^{-1/2}$ , so cold dense gas fragments while warm rarefied gas does not. We build the two competing energies for a uniform sphere, impose the collapse condition, and eliminate the radius to read off the critical mass.

Step 1. Write the total thermal (kinetic) energy of the $N=M/(\mu m_H)$ particles as $E_{\rm th}=\tfrac{3}{2}N k_B T=\tfrac{3}{2}\dfrac{M}{\mu m_H}k_B T$ . Why: this is the outward-pushing budget — the random motions resisting compression — growing linearly with mass and temperature.

Step 2. Write the gravitational binding energy of a uniform sphere of radius $R$ as $|E_{\rm gr}|=\tfrac{3}{5}\dfrac{GM^2}{R}$ . Why: this is the inward-pulling budget; its $M^2/R$ scaling lets gravity dominate as a region is squeezed to smaller $R$ , since it grows faster than the thermal term.

Step 3. Impose the collapse condition. The virial theorem makes a cloud stable when $2E_{\rm th}\ge|E_{\rm gr}|$ and unstable when $|E_{\rm gr}|>2E_{\rm th}$ ; setting the two equal marks the critical mass. Why: the factor of two is the virial requirement that in equilibrium the kinetic energy is half the magnitude of the potential, so the stability line sits at $|W|=2K$ , not $|W|=K$ .

Step 4. Eliminate the radius using $R=\left(\dfrac{3M}{4\pi\rho}\right)^{1/3}$ and solve for $M$ , giving $M_J=\left(\dfrac{5k_BT}{G\mu m_H}\right)^{3/2}\left(\dfrac{3}{4\pi\rho}\right)^{1/2}$ . Why: writing $R$ in terms of $\rho$ turns the criterion into a statement purely about $T$ and $\rho$ — the observables — and exposes the $M_J\propto T^{3/2}\rho^{-1/2}$ scaling that governs which cores fragment.

Go deeper · Derivation B · 1 minThe free-fall collapse time of a pressureless sphere

Derivation BThe free-fall collapse time of a pressureless sphere

The aim is to derive the free-fall time, equation (2), by following a single mass shell falling under the gravity interior to it, and to show that the collapse time depends only on the initial density $\rho$ — not on the shell's radius or the cloud's total mass. This is why $t_{\rm ff}$ is one number for an entire cloud, and the yardstick against which the slowness of real star formation is measured.

Step 1. For a uniform sphere with pressure neglected, a shell at radius $r_0$ feels only the interior mass $M=\tfrac{4}{3}\pi r_0^3\rho$ , conserved as it falls, so its equation of motion is $\dfrac{\mathrm{d}^2 r}{\mathrm{d}t^2}=-\dfrac{GM}{r^2}$ . Why: with no pressure and no shell crossing, this is a one-body Kepler-like fall, the enclosed mass acting as a fixed point source at the centre.

Step 2. Integrate once from rest at $r=r_0$ to get the energy integral $\tfrac{1}{2}\left(\dfrac{\mathrm{d}r}{\mathrm{d}t}\right)^2=GM\left(\dfrac{1}{r}-\dfrac{1}{r_0}\right)$ . Why: this is energy conservation for the falling shell, converting potential into kinetic energy and giving the infall speed at every radius.

Step 3. Separate variables and integrate $r$ from $r_0$ to $0$ ; the substitution $r=r_0\cos^2\xi$ gives a standard cycloid integral, and the time comes out as $t_{\rm ff}=\sqrt{\dfrac{3\pi}{32G\rho}}$ , with $r_0$ cancelling. Why: the disappearance of $r_0$ is exactly what makes $t_{\rm ff}$ a property of the density alone — every shell reaches the centre at the same instant, a clock the whole cloud shares.

Go deeper · Derivation C · 1 minThe virial mass a cloud reveals through its linewidth

Derivation CThe virial mass a cloud reveals through its linewidth

The purpose is to derive the observational virial mass, equation (4), that weighs a molecular cloud from nothing but its measured linewidth and size — independent of the uncertain CO-to- $\mathrm{H_2}$ conversion. We specialize the virial theorem to a turbulence-supported cloud, drop the thermal, magnetic, and surface terms subdominant on GMC scales, and solve the balance for the mass.

Step 1. For a cloud supported by supersonic turbulence ( $\sigma_v\gg c_s$ ), fold all the internal motion into the kinetic term $K=\tfrac{3}{2}M\sigma_v^2$ , where $\sigma_v$ is the one-dimensional velocity dispersion and the factor three counts the three spatial directions. Why: on GMC scales the linewidth is bulk turbulent motion, not thermal broadening, so $U_{\rm th}$ is negligible and folds into $K$ .

Step 2. Take the gravitational energy of a uniform sphere, $W=-\tfrac{3}{5}\dfrac{GM^2}{R}$ , and neglect the magnetic and surface terms. Why: for a marginally bound cloud these are order-unity corrections, so dropping them changes only the prefactor — as does a centrally concentrated profile, which is why some texts carry a geometry factor.

Step 3. Impose virial balance $2K+W=0$ , i.e. $2\left(\tfrac{3}{2}M\sigma_v^2\right)=\tfrac{3}{5}\dfrac{GM^2}{R}$ , and solve to get $M_{\rm vir}=\dfrac{5\sigma_v^2 R}{G}$ . Why: one power of $M$ cancels, leaving the mass fixed by $\sigma_v$ and $R$ alone — an X-factor-free cross-check, whose agreement with CO masses shows GMCs are virialized.

Limiting cases

Which support term dominates, and how close the cloud is to the virial boundary, decides its fate:

Case	Result	Meaning
Thermally dominated (small cores)	$\sigma \approx c_s=\sqrt{\dfrac{k_B T}{\mu m_H}}\approx 0.2\,\mathrm{km\,s^{-1}}$	In quiescent cores the linewidth approaches the isothermal sound speed at $10\,\mathrm{K}$ ; the thermal Jeans mass sets the fragmentation scale.
Turbulence dominated (GMC scale)	$\sigma_v\gg c_s,\quad \sigma_v\propto L^{1/2}$	On GMC scales the dispersion greatly exceeds $c_s$ and follows Larson's law; support is non-thermal and the effective Jeans mass is far larger.
Magnetically subcritical	$\dfrac{M}{\Phi_B}<\left(\dfrac{M}{\Phi_B}\right)_{\rm crit}=\dfrac{c_\Phi}{\sqrt{G}},\ c_\Phi\approx0.13$	Below the critical mass-to-flux ratio, the frozen-in field alone holds the cloud up; contraction proceeds only as neutrals slip past ions (ambipolar diffusion).
Virial boundary	$\alpha_{\rm vir}=\dfrac{5\sigma_v^2 R}{GM}\sim 1\text{–}2$	Observed GMCs cluster here: marginally bound, neither in free-fall collapse nor safely stable — consistent with slow, regulated star formation.

Common trapCO luminosity is not the same as H₂ mass, and linewidth is not thermal

Two traps ensnare newcomers. First, do not read CO brightness as a direct measure of $\mathrm{H_2}$ : the CO(1–0) line is optically thick in its core and freezes onto grains in the coldest gas, so it is only a proxy calibrated through $X_{\rm CO}$ , uncertain by a factor of two and worse in low-metallicity or intensely irradiated gas — which is why "CO-dark" envelopes exist where $\mathrm{H_2}$ is present but CO has been photodissociated. Second, do not treat a GMC as a warm, pressure-supported ball: its linewidth is supersonic turbulence, not thermal motion, so plugging the $10\,\mathrm{K}$ sound speed into the Jeans mass badly underestimates the stability threshold — the mass scale for a whole cloud uses the turbulent dispersion, and the thermal Jeans mass applies only inside quiescent cores.

Worked examples

1 · Weighing a GMC from its linewidth (fully worked)

A GMC is mapped with velocity dispersion $\sigma_v=3\,\mathrm{km\,s^{-1}}$ and radius $R=20\,\mathrm{pc}$ . Estimate its virial mass and comment on the cross-check with CO.

Step 1. Convert to CGS: $\sigma_v^2=9\times10^{10}\,\mathrm{cm^2\,s^{-2}}$ and $R=20\times3.086\times10^{18}=6.17\times10^{19}\,\mathrm{cm}$ .

Step 2. Apply equation (4): $M_{\rm vir}=\dfrac{5\sigma_v^2 R}{G}=\dfrac{5\times(9\times10^{10})\times(6.17\times10^{19})}{6.674\times10^{-8}}=4.16\times10^{38}\,\mathrm{g}$ .

Step 3. Convert to solar masses: $M_{\rm vir}=4.16\times10^{38}/1.989\times10^{33}=\mathbf{2.1\times10^{5}\,M_\odot}$ — a textbook GMC mass. An independent CO-luminosity estimate typically agrees to within a factor of a few, which — repeated across the Galaxy — is the standard argument that GMCs are roughly virialized and one $X_{\rm CO}$ describes them all. $M_{\rm vir}\approx 2\times10^{5}\,M_\odot$ , independent of the CO calibration.

2 · The Jeans mass and free-fall time of a prestellar core (fully worked)

A dense core has $T=10\,\mathrm{K}$ , $n_{\mathrm{H_2}}=10^{4}\,\mathrm{cm^{-3}}$ , $\mu=2.3$ . Find its sound speed, Jeans mass, and free-fall time.

Step 1. Mass density: $\rho=\mu m_H n=2.3\times(1.67\times10^{-24})\times10^{4}=\mathbf{3.85\times10^{-20}\,g\,cm^{-3}}$ .

Step 2. Isothermal sound speed: $c_s=\sqrt{\dfrac{k_B T}{\mu m_H}}=\sqrt{\dfrac{(1.38\times10^{-16})(10)}{2.3\times1.67\times10^{-24}}}=1.9\times10^{4}\,\mathrm{cm\,s^{-1}}=\mathbf{0.19\,km\,s^{-1}}$ — the feeble thermal speed turbulence overwhelms on larger scales.

Step 3. Jeans mass from equation (1). With $5k_BT/(G\mu m_H)=2.69\times10^{16}\,\mathrm{cm}$ and $3/(4\pi\rho)=6.20\times10^{18}\,\mathrm{cm^3\,g^{-1}}$ , $M_J=(2.69\times10^{16})^{3/2}(6.20\times10^{18})^{1/2}=1.10\times10^{34}\,\mathrm{g}=\mathbf{5.5\,M_\odot}$ — a few solar masses, the scale of a single star. That is the whole point: cores, not whole GMCs, fragment at stellar masses. As a core contracts and $n$ climbs toward $10^{5}\text{–}10^{6}\,\mathrm{cm^{-3}}$ , $M_J\propto\rho^{-1/2}$ drops below $1\,M_\odot$ .

Step 4. Free-fall time from equation (2): $t_{\rm ff}=\sqrt{\dfrac{3\pi}{32G\rho}}=1.07\times10^{13}\,\mathrm{s}=\mathbf{3.4\times10^{5}\,yr}$ — a dense core collapses in a few hundred thousand years. $c_s=0.19\,\mathrm{km\,s^{-1}}$ , $M_J\approx 5\,M_\odot$ , $t_{\rm ff}\approx 3\times10^{5}\,\mathrm{yr}$ .

3 · Why star formation is so slow (fully worked)

If every molecular cloud in the Galaxy collapsed on its free-fall time, how fast would stars form, and how does that compare with reality?

Step 1. GMCs have mean densities $n\sim100\,\mathrm{cm^{-3}}$ , giving $t_{\rm ff}\approx 3.4\times10^{6}\,\mathrm{yr}$ .

Step 2. The Milky Way holds $M_{\rm mol}\sim10^{9}\,M_\odot$ of molecular gas; collapsing it in one free-fall time gives $\mathrm{SFR}_{\rm naive}\sim10^{9}/(3.4\times10^{6})\approx\mathbf{300\,M_\odot\,yr^{-1}}$ .

Step 3. The observed Galactic rate is only $\approx1\text{–}2\,M_\odot\,\mathrm{yr^{-1}}$ — over two orders of magnitude smaller. The efficiency per free-fall time is therefore $\epsilon_{\rm ff}=\dfrac{\mathrm{SFR}_{\rm obs}\,t_{\rm ff}}{M_{\rm mol}}\approx\dfrac{(1.5)(3.4\times10^{6})}{10^{9}}\approx\mathbf{0.5\%}$ , of order one percent. Turbulence, magnetic fields, and stellar feedback disrupt and delay collapse. This factor-of-a-hundred gap is the defining puzzle connecting cloud physics to galactic star-formation laws. $\mathrm{SFR}_{\rm naive}\sim300$ vs $\mathrm{SFR}_{\rm obs}\sim1.5\,M_\odot\,\mathrm{yr^{-1}}$ , so $\epsilon_{\rm ff}\sim1\%$ .

Physics of the portrait

The modern picture is less one person's discovery than a century of assembled ideas — Jeans's stability criterion, the detection of interstellar CO, and the scaling laws that made clouds quantitative. If one figure anchors it, it is the man who found those laws and first argued that star formation must be slow.

Origins · 1 minRichard Larson and how this was discovered

Richard Larson

1941–2021 · turbulence, scaling laws, and the slowness of star formation

Richard B. Larson spent his career at Yale asking why galaxies turn gas into stars so grudgingly, and his answers still frame the subject. In a landmark 1979 paper he mined the CO surveys of Galactic clouds and found two regularities — a velocity dispersion that grows with size, $\sigma_v\propto L^{0.38}$ , and a mean density that falls with size — the relations now called Larson's laws. Read with the virial theorem, they revealed clouds threaded by supersonic turbulence and sitting close to marginal binding, overturning the earlier picture of quiescent, thermally supported gas. Larson had already argued, from the mismatch between the short free-fall time and the observed Galactic star-formation rate, that only a few percent of a cloud's gas can be converting to stars at any time — the low-efficiency problem that drives the field to this day. His later work tied the temperature of collapsing gas to the characteristic stellar mass, linking cloud microphysics to the initial mass function. Through it all ran his insistence that star formation is regulated and inefficient — vindicated by every modern survey of turbulent, magnetized, marginally bound clouds.

History

1902

— Jeans derives the mass above which a self-gravitating gas cloud is unstable to collapse

1947

— Bok and Reilly propose dark globules as sites of star formation

1970

— Wilson, Jefferts and Penzias detect the 2.6 mm CO line, opening molecular-cloud astronomy

1979

— Larson finds the linewidth–size and density scaling relations for molecular clouds

1987

— Solomon and collaborators establish virial masses and the CO calibration for GMCs

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 12 (The Interstellar Medium and Star Formation) · Draine, Physics of the Interstellar and Intergalactic Medium, Chs. 32–42 (molecular clouds, CO, gravitational collapse) · Binney & Tremaine, Galactic Dynamics, 2nd ed. (scalar virial theorem, gravitational stability) · Ryden, Introduction to Cosmology (Jeans analysis and gravitational instability) · Stahler & Palla, The Formation of Stars (cloud structure and collapse).

Provenance. The equations — Jeans mass and length, free-fall time, scalar virial theorem, virial mass, and critical mass-to-flux ratio — are standard and are re-derived above; verify against Carroll & Ostlie Ch. 12 and Draine. Numbers re-checked here: $c_s=0.19\,\mathrm{km\,s^{-1}}$ at $10\,\mathrm{K}$ with $\mu=2.3$ ; the Jeans mass with the stated prefactor is $M_J\approx5\,M_\odot$ at $T=10\,\mathrm{K}$ , $n=10^{4}\,\mathrm{cm^{-3}}$ (order a few solar masses — the older " $\sim1\,M_\odot$ " is the same result quoted loosely, and $M_J$ falls below $1\,M_\odot$ only for $n\gtrsim10^{5}\,\mathrm{cm^{-3}}$ ); $t_{\rm ff}\approx3.4\times10^{6}\,\mathrm{yr}$ at the GMC mean density $n\sim10^{2}\,\mathrm{cm^{-3}}$ (and $\approx3\times10^{5}\,\mathrm{yr}$ at $n\sim10^{4}$ ); $M_{\rm vir}\approx2\times10^{5}\,M_\odot$ for $\sigma_v=3\,\mathrm{km\,s^{-1}}$ , $R=20\,\mathrm{pc}$ ; naive vs observed Galactic SFR of $\sim300$ vs $\sim1.5\,M_\odot\,\mathrm{yr^{-1}}$ giving $\epsilon_{\rm ff}\sim1\%$ ; $X_{\rm CO}\approx2\times10^{20}\,\mathrm{cm^{-2}(K\,km\,s^{-1})^{-1}}$ ; CO(1–0) at $115\,\mathrm{GHz}$ / $2.6\,\mathrm{mm}$ ; $c_\Phi\approx0.13$ ; Larson exponent $0.38\text{–}0.5$ . All values are textbook-standard; none fabricated.

Prerequisites. Phases of the Interstellar Medium · Interstellar Gas Heating and Cooling Processes · CO as a Molecular Gas Tracer and the X-Factor · The Jeans Instability and Jeans Mass. Leads to. Virial Theorem for Molecular Clouds · Turbulence in the Interstellar Medium · Interstellar Polarization and Magnetic Fields · Thermal Dust Emission and Infrared Radiation.

Ask the tutor: Ask me to walk through any of these — from why a $10\,\mathrm{K}$ cloud is still supersonic, to computing a GMC's virial mass or its one-percent efficiency step by step.

Why can't we just observe the H₂ that makes up most of a molecular cloud?

Because $\mathrm{H_2}$ is a symmetric, homonuclear molecule with no permanent electric dipole moment. Its lowest allowed rotational transitions are weak quadrupole lines whose upper levels sit hundreds of kelvin above the ground state — far out of reach at $10\text{–}20\,\mathrm{K}$ , where nothing is collisionally excited into them. The dominant mass reservoir is therefore spectroscopically silent. Instead we use surrogates: the bright CO(1–0) line at $2.6\,\mathrm{mm}$ ; dust extinction of background starlight; and thermal dust emission in the far infrared. Each has its own biases, which is why cross-checking CO against dust and against the virial mass is standard.

What is the difference between a molecular cloud, a giant molecular cloud, and a core?

They are levels of one nested hierarchy, set by scale and density. A giant molecular cloud (GMC) is the largest unit — tens of parsecs, $10^{5}\text{–}10^{6}\,M_\odot$ , mean density $\sim100\,\mathrm{cm^{-3}}$ . Within it, gravity and turbulence collect gas into clumps of a parsec and $\sim10^{4}\,\mathrm{cm^{-3}}$ , the birthplaces of star clusters. Inside clumps sit prestellar cores: sub-parsec, $\gtrsim10^{6}\,\mathrm{cm^{-3}}$ condensations of roughly a solar mass, each collapsing toward one star or small multiple. "Molecular cloud" is the umbrella term for the whole family, and Larson's relations quantify how dispersion and density change along the ladder.

Why is star formation so inefficient — only about one percent per free-fall time?

If clouds simply collapsed under gravity, the Galaxy's $\sim10^{9}\,M_\odot$ of molecular gas would form stars at hundreds of solar masses per year; the real rate is one to two. The resolution is that gravity is opposed on every scale: supersonic turbulence only occasionally compresses a region past its Jeans mass; magnetic fields, where clouds are subcritical, delay collapse until neutrals diffuse across field lines; and once stars form, their winds, radiation, and supernovae blow the gas apart. Each free-fall time thus converts only about one percent of the gas, and predicting that number from first principles remains open.

If clouds are near virial balance, why don't they just sit there forever?

A virial parameter near one means the cloud is marginally bound — kinetic and gravitational energies comparable — not that it is in stable, static equilibrium. Marginal binding is transient: turbulence decays on roughly a crossing time, so without replenishment the support weakens and parts collapse, while stellar feedback and Galactic motion disperse others. GMCs are best pictured as living a few tens of millions of years — assembled, forming stars at low efficiency, then torn apart — not as permanent structures in equilibrium.

What is the CO X-factor and why is it uncertain?

The X-factor $X_{\rm CO}$ converts the observed CO(1–0) brightness into an $\mathrm{H_2}$ column density (equation 6), with a Milky Way value near $2\times10^{20}\,\mathrm{cm^{-2}(K\,km\,s^{-1})^{-1}}$ . It works despite the CO line being optically thick, because a virialized cloud's CO brightness tracks its mass through the linewidth. But it is a calibration, not a law, and it drifts: low-metallicity gas makes less CO, strongly irradiated regions photodissociate CO while $\mathrm{H_2}$ survives (the "CO-dark" gas), and cold cores freeze CO onto grains. So $X_{\rm CO}$ is uncertain by a factor of two or more — which is exactly why independent virial and dust-based masses matter.

Do magnetic fields or turbulence hold clouds up — which one actually wins?

Both contribute, and the balance shifts with scale. On GMC scales the dominant support against collapse is supersonic turbulence: the linewidths imply kinetic energies comparable to the binding energy, giving $\alpha_{\rm vir}\sim1$ . Magnetic fields become decisive inside cores, where Zeeman and polarization measurements find many regions near the critical mass-to-flux ratio; a subcritical core cannot collapse until neutrals slip past the ions tied to the field (ambipolar diffusion). The modern view is not "turbulence versus fields" but a magnetized, turbulent medium in which both, plus stellar feedback, keep the efficiency low.