Why the Earliest Galaxies Look Like Swarms of Stellar Clumps

JWST has changed the picture of the early universe by showing that many galaxies in the first billion years are not smooth stellar systems. Instead, many are resolved into compact star-forming clumps — bright knots only a few tens to a few hundreds of parsecs across, packed into small galaxies. Gravitational lensing, the magnification by a foreground mass such as a galaxy cluster, has been crucial for resolving individual clumps in some intrinsically faint galaxies that would otherwise remain unresolved.

A new preprint by Andrea Ferrara and colleagues asks how those clumps form, how they evolve, and why we can see them. The paper, posted on arXiv, is an analytical study rather than a new telescope observation. It builds a mathematical bridge between the global properties of early galactic disks and the statistical population of clumps that JWST has begun to reveal.

From Toomre instability to a clump mass spectrum

The clumps are not minor blemishes. They can dominate a galaxy's star formation, enrich the surrounding gas with heavier elements, and inject energy and momentum through stellar winds, radiation, and supernovae. Some may migrate toward galactic centers and help build compact bulges or nuclear star clusters.

But their origin is debated. One possibility is that they form in mergers and tidal interactions, which should have been common in the rapidly assembling halos of the early universe. Another is that they form in place, when a gas-rich disk becomes gravitationally unstable and fragments.

The new paper focuses on that second route. It starts from the classical Toomre instability. In simple terms, a rotating gas disk becomes unstable when its own gravity can overcome the combined support from rotation and pressure and turbulent support. When that balance tips, the disk can break into dense, star-forming condensations.

Earlier theoretical work often asked which clump mass grows fastest. That gives one characteristic scale. But JWST sees a broad population, not a single clump mass. So the authors integrate over the entire range of unstable wavelengths in the disk. That lets them derive an intrinsic clump mass spectrum, a prediction for how many clumps of different masses gravitational instability should produce.

The fastest-growing mode predicts simple relationships: at a given turbulent velocity, more massive clumps should be larger, with clump mass proportional to effective radius, and their surface density should fall as radius increases. The authors find that observed clumps fit these trends if the parent disks have turbulent gas velocities of about 10 to 80 kilometers per second. That range overlaps with the turbulent dispersions, roughly 30 to 80 kilometers per second, inferred from resolved gas kinematics in gas-rich galaxies at z greater than about 4.

The model also introduces a compactness parameter, set by how much gas settles into the disk and how concentrated that disk is. At a fixed turbulent velocity, a more compact parent disk should produce smaller, less massive clumps. A less compact, more diffuse disk should produce fewer, larger, more massive condensations.

The intrinsic mass spectrum spans nearly two orders of magnitude, from about 300,000 to 20 million solar masses. That is a broad distribution, not a tidy single scale. It reflects the fact that a real unstable disk does not choose one wavelength; it hosts a range of growing perturbations.

The heaviest clumps sink before we see them

The spectrum that forms is not necessarily the spectrum we observe. Massive clumps moving through gas experience stronger dynamical friction, a drag-like force that removes orbital energy and drives them inward. The paper shows that the migration time scales inversely with clump mass, so the heaviest clumps spiral toward the galactic center much faster than lighter ones.

This selective removal suppresses the high-mass tail of the clump population. The resulting slope, a mass function proportional to M to the minus 1.89, agrees well with one recent observational estimate from the Claeyssens et al. sample. The authors are careful to note that this comparison is indicative. That sample combines clusters over a range of redshifts and environments and is restricted to relatively young, massive clusters, so it is not a complete census of clumps under identical conditions.

Another recent JWST survey, JADES, finds a shallower slope, about minus 1.50. The paper suggests that this disagreement may be natural. Individual galaxies in the model do not have perfect power-law mass functions, and the observed slope can depend on disk compactness, halo properties, and how long the clump population has to evolve. The authors argue that the clump mass function may not be universal.

One counterintuitive prediction follows: disks with more gas are expected to form smaller clumps, not larger ones. Higher gas surface density makes the fastest-growing instability occur on smaller scales. More compact, gas-rich disks should therefore host larger numbers of lower-mass clumps. That offers a testable link between galaxy compactness and clump multiplicity.

Radiation pressure exposes the embedded knots

Forming a gaseous clump is one thing. Seeing it is another. Young stellar clumps are born embedded in dense gas and dust. To appear in JWST images, the natal material must be cleared away.

The paper compares three clearing mechanisms. Supernova explosions are the intuitive candidate, but the authors find they are insufficient. The dense gas in compact Toomre clumps is bound too tightly for supernova momentum alone to expel it.

Radiation pressure does better. In dust-poor clumps, Lyman-alpha photons, produced by young massive stars, scatter many times and can impart a strong outward push. As clumps become enriched with dust, Lyman-alpha trapping weakens, but infrared radiation pressure grows stronger because dust absorbs and re-emits stellar light. The two mechanisms are complementary, so radiation pressure can clear the natal gas across a broad range of dust abundances.

That matters for observability. Feedback does not stop the clumps from forming; it ends their embedded phase and exposes the young stellar systems. The model therefore connects the same instability that makes the clumps to the process that makes them visible.

What the analytical model leaves open

This is an analytical framework, not a full simulation. It assumes that stellar systems roughly preserve the size of their parent gas fragments and that star formation in the clumps is efficient. It does not follow what happens after the gas is removed, when a stellar clump may expand, be stripped by tides, or dissolve.

The dynamical friction treatment also depends on an adopted migration time of about 70 million years, representing how long the disk keeps its clump-forming conditions before being replenished or rearranged. The authors note that a factor-of-two change in that time shifts the suppression scale by only about 0.3 in logarithmic mass, without changing the qualitative behavior. The infrared radiation-pressure calculation is an upper limit, because a turbulent, porous medium can let radiation escape through low-density channels.

The paper does not settle whether every JWST clump formed in place. Mergers and interactions remain plausible contributors, and distinguishing them may require resolved kinematics and chemistry. Genuine disk fragments should share the rotation and enrichment of their surroundings; captured or merger-born systems may not.

Still, the work gives a coherent account of why early galaxies look clumpy, why their clumps span a broad mass range, and why the most massive ones may be missing from the orbiting population. In that picture, the bright knots are not just chaotic debris. They are visible signatures of gravitational instability, major sites of early star formation, and possibly helping build compact bulges or nuclear star clusters in early galaxies.