For four years the complaint about JWST's earliest galaxies had a hopeful subtext.
They looked too massive, too early — impossibly bulky for a universe only a few hundred million years old — and the obvious escape was that we'd overcounted. Better data, cleaner analysis, and the paradox would shrink back to something the models could stomach.
A new measurement just delivered the opposite verdict. The galaxies aren't lighter. They're heavier — by as much as a factor of four — with most of that extra mass hiding in stars too faint to see.
The galaxies aren't lighter. They're heavier — by as much as a factor of four — with most of that extra mass hiding in stars too faint to see.
Last issue this same tension showed up in the little red dots, where the fix on the table was hidden black holes doing the shining. This week it's hidden dwarf stars doing the weighing. Same crack in the model, different corner of it.
The new work comes from Chloe Cheng, who recently finished her PhD at Leiden, and colleagues, published in Nature Astronomy. And the crucial thing to understand is that nobody ever counted the stars in these galaxies. They inferred them.
You cannot resolve individual stars in a galaxy seen 300 million years after the Big Bang. You get the combined spectrum of everything at once, and you work backward to a mass — using an assumed recipe for how many stars form at each size, the initial mass function.
That recipe has always been the Milky Way's. Astronomers took the IMF measured in our own galaxy and extrapolated it across cosmic time and distance, because they had nothing else to extrapolate from.
Here's why that matters. The bright, massive stars dominate the light, but they're rare and they die fast. The dim, low-mass stars are where most of the actual mass lives — and they're nearly invisible in the glare.
Cheng's metaphor is exact. "If a galaxy were a city, the brightest stars would be the skyscrapers that immediately catch your eye from afar," she said. "Our models demonstrate that a far more numerous population of low-mass stars is concealed by those rare, bright stars, like houses hidden between skyscrapers."
Count only the skyscrapers and multiply by the Milky Way's skyscraper-to-house ratio, and you get a mass estimate. Get the ratio wrong, and every mass built on it is wrong too.
So Cheng's team did what nobody had done before: measure the IMF beyond the local universe directly, rather than assuming it.
The trick is that even invisible low-mass dwarfs leave fingerprints. Certain absorption lines in a galaxy's combined spectrum are sensitive to how much light comes from faint dwarf stars versus luminous giants — the "fine print" of the spectrum.
Reading that fine print took extraordinary data. The team used ultra-deep JWST NIRSpec spectra of nine massive, quiescent galaxies at redshift roughly 0.7, from the JWST IMFERNO program, and stretched the wavelength coverage bluer with deep spectra from the Very Large Telescope's LEGA-C survey.
They chose quiescent galaxies deliberately — no messy young stars, no nebular emission clouding the signal — so the subtle dwarf-star lines could be read cleanly.
"Until recently, measurements like these were simply impossible," said co-author Martje Slob. "We needed not only a telescope capable of magnifying very distant galaxies, but also spectra of exceptional quality and new analysis techniques to reliably detect the subtle signatures of faint, low-mass stars hidden within these cosmic titans."
The result: the most massive galaxies really do carry an excess of low-mass stars. Their IMFs are bottom-heavy compared to the Milky Way's — more houses per skyscraper than the local recipe allows.
And the excess tracks with the galaxies' properties in exactly the way local studies have long hinted. The more bottom-heavy IMFs show up in galaxies with higher velocity dispersion and higher metallicity — deeper gravitational wells, more chemically processed stars.
Then comes the sentence that turns this from a technical correction into a problem.
"Remarkably, our oldest galaxy," the authors write — one that formed at redshift above 5 — "has the most bottom-heavy IMF." That galaxy is a likely descendant of JWST's "impossibly early" galaxies, which implies those galaxies had similarly bottom-heavy IMFs.
Apply that correction and their masses go up by a factor of about four, plus or minus one.
Now sit with why that stings. Bottom-heavy IMFs were not a random discovery. They were one of the proposed escape hatches.
If the early galaxies formed stars under a different recipe than the Milky Way, went the hope, then our mass estimates might be systematically off — and maybe, once corrected, the tension would ease.
The correction arrived. It points the wrong way. Confirming bottom-heavy IMFs doesn't dissolve the "too massive, too early" problem; it makes the galaxies more massive, and the strain worse.
"Our findings may thus amplify the tension with galaxy formation models," the authors write. The escape hatch was a trapdoor.
This is worth being precise about, because it's easy to overstate. None of this yet touches ΛCDM itself, the cold-dark-matter framework that predicts how dark-matter halos grow.
ΛCDM predicts halos. A separate layer of astrophysics — how gas falls in, cools, forms stars, makes dust, and what IMF it forms them under — translates those halos into the light JWST actually sees. It's that astrophysical layer taking the hits.
And that layer is genuinely elastic. Some groups reproduce the JWST luminosity functions with refined halo models and modest star-formation efficiencies, arguing the overabundance stems from oversimplified collapse physics, not new physics. Others showed years ago that Hubble's UV data leave little room to blame cosmology, pointing back at astrophysics.
The honest position, held by cosmologists like Mike Boylan-Kolchin, has been that most people would put their money on the astrophysical explanation. Cheng's result doesn't overturn that. But it does something quietly corrosive: it removes one of the astrophysical fixes people were counting on.
Every "crisis" in this saga has rested on an unmeasured assumption — the extrapolated Milky Way IMF. Now that the IMF has actually been measured beyond the local universe, the assumption didn't just fail to save the models. It made their problem bigger.
And the rewriting isn't happening only at the far edge of the census. It's happening at home, too.
The same week, a separate Nature Astronomy paper reported definitive evidence that a dwarf galaxy — named LKH, for Low-energy, Kraken, and Heracles — collided with the young Milky Way about 11.8 billion years ago, roughly two billion years after the Big Bang.
The team, led by Davide Massari in Bologna, pinned it down using a distinct third group of globular clusters, dated with Hubble imaging and Gaia measurements, that formed in neither the Milky Way nor the later Gaia–Sausage–Enceladus merger.
LKH had a total mass around 500 million solar masses — a bit heavier than the Sagittarius dwarf the galaxy is swallowing now. And the finding extends our own galaxy's known assembly history 1.8 billion years further back than before.
So galaxy assembly is being rewritten from both ends at once: the earliest cosmic census gets heavier, and the Milky Way's own biography gets a new opening chapter in its first two billion years.
Which leaves the framing question. Each of these results, taken alone, is absorbable — a reanalysis here, a merger there, all within the flexible astrophysical layer that sits above ΛCDM's halos.
But the fixes keep failing in the same direction. Last issue's black holes, this issue's hidden dwarfs — every escape hatch, on inspection, deepens the thing it was meant to resolve.
The astrophysical layer can absorb a great deal. The open question is how many reanalyses it can absorb before the load-bearing assumption underneath it — not the astrophysics, but the cosmology it's been protecting — is what finally gives.