Webb spectra found an ancient massive galaxy packed with far more small, faint stars than expected, implying some of the universe’s earliest giants may be roughly four times heavier than astronomers calculated — deepening the mystery of how they formed so quickly

Astronomers cannot place a distant galaxy on a scale. They estimate its stellar mass from the light that reaches a telescope, then use a model to infer the stars too faint to contribute much of that light.

That hidden part of the calculation may be much larger than expected in some ancient massive galaxies. Ultra-deep James Webb Space Telescope spectra of nine old, quiet systems indicate that several contain an excess of low-mass stars compared with the birth distribution usually assumed for the Milky Way.

The oldest galaxy in the sample produced the strongest result. Its stars appear to have formed beyond redshift 5, when the universe was less than about 1.2 billion years old. If that stellar population resembles those inside the still earlier massive galaxies Webb is now discovering, their stellar masses could be roughly four times higher than estimates based on a Milky Way-like population.

The distinction between measurement and implication is essential. Webb did not resolve trillions of faint stars in a galaxy from the first few hundred million years. It measured the combined spectrum of a later galaxy whose old stars may preserve the conditions under which an earlier giant formed.

The peer-reviewed study in Nature Astronomy, led by Chloe M. Cheng of Leiden Observatory and published on 18 August 2026, calls the result evidence for a bottom-heavy stellar initial mass function. This is one study, not settled consensus. Its sample contains nine selected galaxies, and the fourfold correction is an inference with an uncertainty, not a new direct weighing.

The galaxies are ancient, but they were not observed at cosmic dawn

Cheng and colleagues selected nine massive quiescent galaxies at redshift about 0.7. Quiescent means that the intense star formation that built them had largely ended. Removing the glare and short-lived features of active star formation makes the accumulated old stellar population easier to analyse.

Redshift 0.7 corresponds to looking back roughly six to seven billion years. That is a long view into the past, but not the first billion years after the Big Bang. The early-universe connection comes from stellar ages. The oldest galaxy in the set has an estimated formation redshift above 5, meaning much of its stellar population was assembled when the universe was younger than roughly 1.2 billion years.

The paper describes that object as a possible descendant of Webb’s very early massive galaxies. This evolutionary bridge matters. The team measured the descendant directly at redshift 0.7, then asked what its bottom-heavy population would imply if the same stars were already present in its high-redshift ancestor.

A galaxy’s brightest stars are a biased census

Stars do not contribute light in proportion to their mass. High-mass stars are enormously luminous but comparatively rare and short-lived. Low-mass stars are cool, faint and extraordinarily long-lived. A single red dwarf adds little to a galaxy’s integrated light, yet a vast population of them can hold a large share of its stellar mass.

Astronomers describe the distribution of stellar birth masses with the initial mass function, usually shortened to IMF. When the stars in another galaxy cannot be counted individually, models commonly adopt a well-tested Milky Way distribution, such as the Kroupa IMF. Observed luminosity is converted to total mass by extrapolating down into the unseen low-mass population.

A bottom-heavy IMF means that low-mass stars are more numerous relative to heavier stars than in that Milky Way reference. It does not mean the galaxy contains no massive stars, or that its individual stars somehow gained extra weight. It means the mix is shifted toward the faint end, raising the mass-to-light ratio.

Webb found spectral fingerprints, not separate dwarf stars

Even Webb cannot separate individual red dwarfs at these distances. Cheng’s team used the Near-Infrared Spectrograph, NIRSpec, to collect exceptionally deep spectra through a programme called JWST Initial Mass Function of Early Red NIRSpec Objects, or IMFERNO.

The Very Large Telescope’s Large Early Galaxy Astrophysics Census supplied complementary observations at bluer rest-frame wavelengths. The third LEGA-C data release contains high-quality spectra for thousands of galaxies beyond redshift 0.6, giving the new Webb observations a detailed optical foundation.

The method relies on absorption features that respond differently to surface gravity and temperature. Red dwarfs and red giants can have similar colours, but their atmospheres leave subtly different fingerprints in sodium, iron hydride and molecular bands. Webb’s infrared reach captured features sensitive to the relative number of dwarfs and giants, while the optical data constrained ages and chemical abundances.

Those effects overlap. A change in sodium abundance can imitate part of an IMF signal, and a mixture of stellar ages changes the spectrum too. The researchers therefore fitted the full wavelength range simultaneously. Their model allowed stellar age, 18 elemental abundances and two low-mass slopes of the IMF to vary rather than assigning every difference to dwarf stars.

The mass correction comes from a ratio, not a direct scale

The paper expresses the result through an IMF mismatch parameter. It compares the mass-to-light ratio allowed by the best-fitting variable IMF with the ratio obtained when the model is fixed to a Milky Way-like IMF. A value of one means no correction. A higher value means more mass is required to produce the observed amount of light.

Several of the nine galaxies sat significantly above one, particularly systems with higher stellar velocity dispersion, iron abundance and inferred virial mass. The oldest galaxy had the most bottom-heavy fit. For an early counterpart with the same IMF, the authors estimate that a conventional stellar mass could rise by a factor of about 4, with an uncertainty of roughly 1.

That is why “four times heavier” should be read as a model-dependent stellar-mass revision. The observed photons have not changed. The accounting beneath them has. More faint stars require more mass for the same luminosity.

The paper’s preprint described two old systems with corrections of roughly three to four. The final journal abstract gives the most direct claim for the oldest galaxy and frames the factor as an implication for its possible earlier counterparts. Neither version establishes a universal multiplier for every distant galaxy.

The result reverses one possible solution to Webb’s early-galaxy tension

SpaceDaily recently examined how Webb’s first four years produced more bright early galaxies than many pre-launch models forecast. One uncertainty in that debate is the IMF. If early galaxies formed proportionally more massive, ultraviolet-bright stars, they could look exceptionally luminous without containing as much total stellar mass. That top-heavy possibility can soften the formation problem.

This study points in the opposite direction for the ancient population it measured. A bottom-heavy IMF puts more mass into faint, long-lived stars. Rather than explaining an early galaxy’s brightness with fewer stars, it says considerably more stellar material may be hidden beneath the same observed light.

The result does not prove that the most distant galaxies share this distribution. The famous record-holders are being observed much earlier and often while they are still forming stars vigorously. SpaceDaily’s report on MoM-z14 at only 280 million years after the Big Bang shows how different that direct evidence is. The Cheng sample instead asks what an old stellar fossil can reveal about a later stage in the ancestry of early massive systems.

More low-mass stars would make rapid formation harder to explain

Galaxy-formation models already have to move gas into early dark-matter haloes, cool it, convert it into stars and then suppress further star formation quickly enough to produce massive quiescent galaxies. Raising stellar mass by three or four times increases the amount of ordinary matter that must complete that process within the available time.

It also changes what “efficient” formation means. Low-mass stars lock material away for much longer than massive stars. They produce less ultraviolet light per unit mass and return less material rapidly through winds and supernovae. A bottom-heavy population would therefore alter chemical enrichment, feedback and the recycling of gas as well as the headline mass.

This deepens tension with galaxy-formation recipes, not automatically with the entire cosmological framework. The standard cosmological model governs how dark-matter structure develops and how much ordinary matter is available. Connecting that supply to a specific stellar population requires uncertain physics involving gas pressure, cooling, turbulence, radiation and feedback.

Stellar motions offer a partial reality check

A spectral model could infer so many faint stars that their mass exceeds what the galaxy’s gravity allows. To test that possibility, the team compared IMF-corrected stellar masses with virial masses estimated from galaxy size and the speed of stellar motions.

For most targets, adding the bottom-heavy correction moved stellar and virial masses into better agreement. Within the quoted uncertainties, the stellar mass did not exceed the virial mass. That supports the physical feasibility of the inferred dwarf-rich populations.

The check is not independent in every respect, and it revealed an outlier. One galaxy’s corrected stellar mass entered an apparently unphysical range. It is almost perfectly round and may be a face-on disc, in which case line-of-sight velocity dispersion would miss much of its rotation and underestimate the gravitational potential. Alternatively, its abundance or IMF fit could be wrong.

The authors also note that simple virial estimates concentrate on the compact, star-dominated central region and can undercount dark matter, gas or rotation. Detailed dynamical modelling will be needed before spectra and gravity become a clean cross-check.

Nine selected galaxies cannot define the universal IMF

The sample is small by design. Its galaxies are massive, compact and quiescent, and they were selected because very deep spectra could be obtained. The study does not show that ordinary galaxies at the same epoch, much less every galaxy at cosmic dawn, formed stars in the same proportions.

Galaxy evolution adds another complication. Massive systems can build their outer regions through later mergers. The compact core may preserve the earliest star formation while imported stars dilute the galaxy-wide signal. An IMF measured in a descendant is therefore evidence about its history, not a time machine with every evolutionary step removed.

Past work has already found bottom-heavy populations in the centres of massive nearby elliptical galaxies. A 2020 review of IMF evidence in early-type galaxies described broad agreement among spectral, dynamical and lensing approaches while also stressing their assumptions. The new result pushes an individual-galaxy full-spectrum measurement much farther from the local universe.

The next spectra need to close the evolutionary gap

The strongest test is to observe more massive quiescent galaxies at higher redshift, closer to the time when their star formation ended. Larger samples can show whether low-mass excess tracks stellar age, density, metal content, velocity dispersion or some combination of them.

Researchers will also need to compare different stellar-population codes, mask or isolate abundance-sensitive features, and combine spectra with resolved dynamics or gravitational lensing. Agreement across those methods would make it harder for one modelling choice to manufacture a bottom-heavy result.

For now, the study changes the direction of one uncertainty. Faint stars were often treated as a hidden quantity filled in with the Milky Way’s proportions. Webb has made their collective fingerprints measurable in galaxies billions of light-years away. If the oldest massive systems truly formed with many more small stars, then the early universe did not merely assemble bright galaxies quickly. It may have locked several times more matter into long-lived darkness than their light alone suggests.