Roughly 215,000 light-years (66,000 parsecs) from the Milky Way's disk sits a smudge of a galaxy so faint it went undetected for decades. Boötes I has barely a ten-thousandth of the Milky Way's stellar mass, and the stars it does have are almost devoid of anything heavier than hydrogen and helium. That makes it a rare kind of time capsule — a fossil of the environment the universe's very first galaxies grew up in, sitting close enough for a modern telescope to actually resolve individual stars inside it.
A team led by Keyi Ding of the University of Maryland, working with Mario Gennaro and Roberto J. Avila of the Space Telescope Science Institute, Massimo Ricotti, and 19 other co-authors, pointed JWST's NIRCam instrument at that fossil and asked a basic but hard-to-answer question: when a cloud of near-primordial gas collapses into stars, does it make the same mix of stellar masses that gas clouds make today? The paper, submitted in May 2026 and accepted to The Astrophysical Journal after an August revision, says yes — at least within the precision JWST can currently deliver.
Why It Matters
The stellar initial mass function, or IMF, is the statistical recipe that describes how many low-mass stars, mid-sized stars, and massive stars come out of a given episode of star formation. It's one of the load-bearing assumptions underneath huge swaths of astrophysics: models of galaxy evolution, estimates of how much dark matter a galaxy must contain to match its visible mass, and calculations of how quickly heavy elements get seeded into a young universe all lean on the IMF holding roughly steady from galaxy to galaxy. For decades, astronomers have had good reason to worry it doesn't. Metal-poor, low-density gas — the kind that dominated the early universe — behaves differently as it cools and fragments than the metal-enriched gas forming stars in the Milky Way today. If the recipe changes with those conditions, then a lot of the scaffolding used to interpret both nearby dwarf galaxies and the first galaxies JWST is now photographing at extreme redshift would need rethinking. Ding et al.'s result is a data point saying the recipe may be sturdier than that — evidence, drawn from an actual near-primordial environment rather than a simulation, that star formation converges on similar outcomes even when the ingredients look very different.
Boötes I is an ideal place to test this because it is, in effect, frozen. Ultra-faint dwarf galaxies like it stopped forming new stars billions of years ago, so the stellar population visible today is essentially the same population that formed in the early universe — no younger generations of stars muddying the sample. Its measured metallicity, [Fe/H] ≈ -2.4, means its stars contain roughly 1/250th the fraction of heavy elements found in the Sun, a chemical fingerprint closer to the universe's first star-forming clouds than to anything nearby.
Measuring an IMF this precisely requires counting individual stars down to low masses, which is where JWST's resolving power comes in. The observations trace back to STScI proposal 3849, led by Gennaro with Ding and Ricotti as co-investigators, which requested deep NIRCam imaging in the F150W and F322W2 filters across a 2-by-3 mosaic with 20 dithered exposures per tile. That strategy paid off: the team resolved more than 10,000 individual stars in Boötes I, reaching down to roughly 0.15 solar masses — a regime that even Hubble struggles to reach in a galaxy this distant and this faint.
With that star count in hand, the researchers tested competing mathematical shapes for the IMF. A simple single power law — essentially a straight line on a log-log plot of star mass versus how many stars of that mass exist — did not fit the data well and was ruled out. But the standard alternatives used to describe star formation in the Milky Way and other well-studied galaxies, a broken power law and a lognormal distribution, both fit the Boötes I data. More strikingly, when the team compared the shape of that fit to the canonical Milky Way IMF, the two matched within 68% confidence — a statistical way of saying the two galaxies' star-formation outputs are indistinguishable at the precision this dataset allows, despite one forming stars in the metal-rich present and the other in near-primordial conditions more than 13 billion years ago.
An AAS Nova highlight published September 22, framing the result for a broader astronomy audience, described ultra-faint dwarfs like Boötes I as "fossils" of the galaxies that populated the early universe, noting their stellar mass is roughly 10,000 times smaller than the Milky Way's. The summary underscored the same headline finding: the broken power-law and lognormal models that describe Boötes I's stars line up with Milky Way values, adding one more environment — arguably the most extreme yet directly measured — to the growing list of places where the IMF looks the same.
That doesn't settle the debate. "Within 68% confidence" is a real caveat, not a rounding error — it means the current data can't rule out a more subtly different IMF, only a dramatically different one like a single power law. And Boötes I is one galaxy, observed at one moment in a JWST proposal cycle. Ding and her colleagues frame the result as evidence for universality, not proof of it, in an environment where a startlingly different outcome would have been at least as scientifically interesting as a matching one. But the fact that JWST can now resolve individual stars, one by one, in a galaxy this ancient and this faint is itself the more durable news: the tools finally exist to keep asking this question with better precision, in more fossils like Boötes I, as more data comes in.
