Nine massive galaxies, frozen for billions of years, have just given up a manufacturing secret: they hide crowds of small stars nobody could see. A team led by Leiden Observatory (Mariska Kriek), using the James Webb Space Telescope and the Very Large Telescope in Chile, shows that these galaxies weigh up to 4 times more than previous estimates. The puzzle of the first "impossible" galaxies just got tougher.

Nine galaxies under the microscope, one clear verdict

Published on 18 August 2026 in Nature Astronomy, the study led by Chloe Cheng and supervised by Mariska Kriek targets nine "quiescent" galaxies: giants that stopped forming stars billions of years ago. Their light traveled nearly 6 billion years, showing them as they were 6.5 billion years after the Big Bang (redshift z ≈ 0.7).

Why this choice? Because a dead galaxy holds an old, calm stellar population, ideal for counting stars. The JWST-IMFERNO program collected 31.2 hours of ultra-deep spectra with the NIRSpec spectrograph, complemented by archives from the VLT's LEGA-C survey. The result: for the first time, astronomers could estimate the fraction of small stars in galaxies this distant.

Spectra of the nine massive galaxies studied by the JWST-IMFERNO program, alongside their images taken by the James Webb Space Telescope.
The nine galaxies observed by JWST, ordered by increasing redshift. Their spectra reveal signatures of a hidden population of small stars. Credit: Cheng et al. / Leiden Observatory, published in Nature Astronomy.

The IMF problem, the astronomers' star counter

To weigh a galaxy, you add up its stars. The catch: you only see the bright ones. Red dwarfs, which dominate the stellar headcount the way the Milky Way holds roughly 200 billion stars, are too faint to be detected individually, even by Webb. So astronomers extrapolate their numbers from a statistical recipe: the initial mass function (IMF), which describes how many stars are born small and how many are born giants.

The trouble is that this recipe was assumed to be universal. Everyone applied the Milky Way version everywhere. The Leiden study shows that assumption fails for the giants of the young universe: the recipe severely undercounts small stars there.

How to flush out the invisible

The method relies on spectroscopy: splitting each galaxy's light into its constituent colors. Red giants and red dwarfs leave different fingerprints in the spectrum, because their surfaces have different temperatures and gravities. Three near-infrared signatures do the job: the Na I bands (8,180 Å), the CaT (8,475 to 8,725 Å) and the Wing-Ford band (9,905 to 9,945 Å), the latter being an exclusive fingerprint of red dwarfs, first identified in the spectrum of the dwarf star Wolf 359 in 1969.

By fitting stellar population models to these spectra, the researchers measured a parameter α for each galaxy, the ratio between the true mass and the mass computed with the Milky Way IMF. The result: α ranges from 1.1 to 4.0 across the sample, with a median of 1.4. Two galaxies clearly exceed the statistical threshold: they contain a significant excess of small stars.

Combined VLT and JWST spectrum of galaxy 1134272, showing the Na I, CaT and Wing-Ford absorption bands, sensitive to the red dwarf population.
The spectrum of galaxy 1134272, the oldest and most 'bottom-heavy' in the sample. The insets zoom in on the three signatures (Na I, CaT, Wing-Ford band) that betray the presence of red dwarfs. Credit: Cheng et al., Nature Astronomy (arXiv:2601.20864).

The oldest galaxy: up to 4 times more massive

One galaxy in the sample, catalogued as 1134272, steals the show. Its estimated stellar age of 5.86 billion years implies formation before redshift 5, less than 1.5 billion years after the Big Bang. It would already have gone quiet by z ≈ 4.6: a giant, all grown up while the universe was still a child.

For this galaxy, α reaches 4.01 with an uncertainty of ± 0.7. Its mass, computed until now with the Milky Way IMF, should therefore be multiplied by about 4 (± 1). And that is no curiosity: the authors suggest this galaxy is a descendant of the "impossibly early" galaxies found by JWST, the ones that seemed too massive, too mature, too soon. If the descendants share this stellar profile, the ancestors had it too. Their true masses could be 4 times higher than published estimates (factor 4 ± 1).

What it changes for cosmology

Every time JWST spots a massive galaxy very early in cosmic history, galaxy formation models break a sweat. By quadrupling the estimated masses of their descendants, the study turns a tension into a major enigma: how could astronomical numbers of small stars form so fast, in a universe still young?

There is also a surprising lead: small stars live very long and readily host planets. A larger-than-expected population of red dwarfs in the young universe could mean more planets formed within the first billions of years. The team plans to push the method toward ever more distant galaxies, at the gates of the first generations of stars.

Comparison of initial mass function shapes: the IMFERNO galaxies (thin lines) against the Kroupa and Salpeter IMFs, references calibrated on the Milky Way.
IMF shapes measured in the nine galaxies (thin lines) compared with the Milky Way recipes (Kroupa, dashed) and Salpeter (thick). The steeper the slope at low mass, the more small stars the galaxy contains. Credit: Cheng et al., Nature Astronomy (arXiv:2601.20864).

Going further