On August 22, Nature Astronomy published two papers in the same issue: the original research by Cheng et al. (first author from Leiden), DOI 10.1038/s41550-026-02932-4, plus a companion commentary in the same journal, DOI 10.1038/s41550-026-02947-x. The conclusion in one sentence: after a bottom-heavy initial mass function (IMF) correction, 9 early massive galaxies that have already "stopped forming stars" are actually 3 to 4 times more massive than previously estimated — one galaxy formed less than 1.5 billion years after the Big Bang was revised to 4 times its prior mass.
What JWST saw
The team combined high-quality JWST NIRSpec spectra with ground-based VLT LEGA-C survey data to, for the first time, reliably estimate the ratio of low-luminosity to high-luminosity stars in 9 mature, long-quiescent massive galaxies. Cheng (a Leiden PhD graduate) offers a "skyscraper-residents" analogy: bright galaxies in the night sky are like skyscrapers in a city — visually prominent, but hiding large numbers of low-rise residents between them. These residents had previously been impossible to identify in spectra. Penn State co-author Joel Leja put the conclusion more bluntly: 3 to 4 times.
Why a bottom-heavy IMF
The classical IMF assumes stars form in the same proportions throughout cosmic history; observations of the local universe over the past 20 years have supported this assumption (for Milky-Way-like smaller galaxies). Cheng's team's finding draws a boundary on this assumption: in early massive galaxies, the fraction of low-mass stars is significantly above average. This is not fine-tuning of a model parameter, but a structural change in the ratio.
The "impossibly early" problem deepens
Since JWST launched, astronomers have repeatedly found unusually large and unusually mature galaxies appearing relatively early after the Big Bang — a persistent tension with existing galaxy formation models, formally named the "impossibly early" problem by Steinhardt et al. in 2016. Cheng's team's paper means early galaxy masses must be revised upward by another factor of 4, tightening that tension further. Existing models invoking "too-fast star formation / too-efficient gas accretion" now also need an additional mechanism making the early IMF favor "many small stars." Leja's original words: if these galaxies truly have 4 times more stars, "we need more physical explanation for why so many small stars appear so early."
Three undercurrents accompanying this work
1. The temptation to extrapolate earlier in cosmic time. The paper explicitly says the next step is to apply the same technique to even earlier galaxies — if galaxies within 1.5 billion years get a 3-4x correction, could galaxies a few hundred million years old see 6-8x corrections? Classical descriptions of cosmology and structure formation will keep being pushed.
2. Planet formation. Leja and Kriek both emphasize: many planets orbit low-mass stars. If low-mass stars in the early universe were systematically undercounted, then the number of planets formed in the early universe was correspondingly underestimated. This pushes the scale of "cradles for life" to an earlier and broader timeline. Planet formation and habitability are separate judgments, but the numerical revision is explicit.
3. Another round between Penn State and the Yale/Leiden camp. Van Dokkum / Conroy proposed a "concordance IMF" in 2024, attempting to keep early galaxy masses from being inflated into conflict with contemporary physics. Cheng's team's work effectively rejects this middle path, pulling the IMF to the "bottom-heavy" extreme. If subsequent observations accumulate in the same direction, galaxy formation models may be forced to make "when and by what mechanism does the bottom-heavy IMF appear" the standard question of the next decade.
The limits — and power — of JWST itself
JWST gained no new mirrors and no new detectors; it was simply NIRSpec spectroscopy combined with ground-based high-precision radial velocity data from the VLT that upgraded "how many small stars are in early galaxies" from a guess to an empirical result. It suggests the biggest puzzle of the JWST era may not be "what are the objects we can't see," but "the things we thought we couldn't see were there all along." Whether the next JWST time allocation will more aggressively fund the "early IMF" question will be an early signal in the 2027 observing-resource contest.