Dwarf galaxies keep only about 11% of the oxygen their stars made, Hubble census finds

Across 37 nearby dwarf galaxies, the share of stellar oxygen still in the galaxy runs from 4% to 25% (mean about 11%), well below what three cosmological simulations predict, though no one has yet measured where the rest went.

Peer reviewed · Observational census · Retention figures are estimates

Astronomers have added up the oxygen in 37 nearby dwarf galaxies and found most of it missing. Kristen McQuinn of Rutgers University and co-authors, in The Astrophysical Journal on 2 November 2026, estimate that each galaxy still holds 4% to 25% of the oxygen its stars have made, about 11% on average. Simulations predict more. The catch: the team did not find the lost oxygen, and the supernova-driven winds they favour as the cause are inferred from the numbers, not observed directly.

Evidence at a glance

Is this a big deal?

A solid, careful measurement rather than a surprise discovery. It puts a number on a long-suspected leak and shows that leading simulations leak too little. The open question of where the oxygen went is the part to watch.

What scientists found

Think of a bank statement. Total earnings are known, but the balance is small. For the 37 galaxies, the “earnings” are the oxygen their stars forged and the “balance” is the oxygen now sitting in stars, gas and dust. The balance averages 10.7% of the earnings (standard deviation 5.1 percentage points), with galaxies ranging from about 3.8% to 24.8%. The analogy breaks at one point: a bank knows where money went, and here the destination is unknown.

Nearly all the oxygen that remains is in gas, not in stars. Galaxies more than 1 megaparsec (about 3.3 million light-years) from a large neighbour of more than a billion Suns in stellar mass all kept less than 15%. Closer in, the numbers scatter widely, but the authors call that environment trend tentative because the sample is small and mostly isolated galaxies.

How they found it

The 37 galaxies are within about 6 megaparsecs (roughly 20 million light-years), close enough for Hubble to pick out individual stars, which most dwarf galaxies are too far away to allow. Brightness and colour of those stars let the team reconstruct when each galaxy formed its stars and how its chemistry changed.

Three more measurements complete the budget. Infrared light from the Spitzer Space Telescope gives each galaxy’s stellar mass. Radio telescopes (the Very Large Array and the Australia Telescope Compact Array) map the hydrogen gas. Oxygen’s emission lines in glowing gas clouds give its abundance in the gas, and a dust model accounts for oxygen locked in grains.

Dividing what is there by what stars should have made gives “retention”. For the denominator, the team uses an oxygen yield of 0.010 and a return fraction of about 0.43. Together these mean each solar mass of stars alive today implies about 0.018 solar masses of oxygen made in total.

Worked example: one hypothetical galaxy

SciQuest calculation, not a galaxy from the paper. Take a galaxy with 10⁸ Suns of stars. Oxygen made is 0.01 ÷ (1 − 0.433) × 10⁸ ≈ 1.8 million solar masses. At the 11% average, about 190,000 solar masses remain and about 1.6 million are missing. The missing oxygen is about eight times what the galaxy still holds (89 ÷ 11).

Why it matters to you

Nothing here changes daily life. The oxygen in air and water was made in stars, and this work tracks how much of it small galaxies hold onto and how much they send into intergalactic space. Rutgers’ release quotes McQuinn on why oxygen: “Oxygen is one of the primary elements formed by stars.” For the star side of the story, see our guide to the life cycle of a star.

What the evidence does not show

The study does not locate the missing oxygen. Earlier measurements of ionised oxygen in the gas halos around dwarf galaxies, quoted in the paper, cover roughly 1% to 7% of the oxygen produced. After a correction for ionisation, the paper puts the halo’s possible share at roughly 5% to 35%. Most oxygen is therefore still unaccounted for, and whether it has left for the space between galaxies is open.

The yield matters too. A higher published yield of 0.015 would push the retention figures lower still, and the choice of stellar-evolution model shifts values modestly.

How this fits what we already knew

Oxygen loss from galaxies is not a new idea. The paper compares its numbers with Leo P, a tiny galaxy of about 200,000 Suns that keeps roughly 5%, and with the Andromeda galaxy (M31), which keeps about 38%, mostly in its stars. A 2014 study of Milky Way-like galaxies, cited in Rutgers’ release, found much of their oxygen had also left the main body. Retention rises with galaxy mass, and the GLOW galaxies sit below an earlier estimate of roughly 20% near 1 billion Suns.

Against simulations the picture is clear. The FIRE-2 simulations keep two to four times more oxygen than observed across this mass range. A second set, by Christensen and colleagues, comes closest, mostly within a factor of two but still high, and a third, by Piacitelli and colleagues, predicts a few times more. All three reproduce the observed link between galaxy mass and metal content, so the mismatch lies in how much oxygen they keep.

Outside views and the authors’ interpretation

SciQuest has not interviewed an independent researcher and found no on-record outside comment yet. The authors’ own view, from the Rutgers release, is that galactic winds do the removing. Grace Telford of the University of Utah, a co-investigator, said “you need energy from massive stars and supernova explosions to drive material out of the host galaxies.” The paper also fits a simple galaxy-regulator model, which favours winds that weaken as galaxies grow and carry gas about twice as metal-rich as the galaxy’s own, together with roughly 40% suppression of fresh gas accretion. That is a model fit, not a direct measurement.

How other headlines framed it

• NASA framed it as a case of missing oxygen, led by Hubble.
• Rutgers put the emphasis on the cycle: stars make oxygen, small galaxies lose most of it.
• Sci.News gave the figure, about a tenth retained, in its headline.
• Earth.com stressed that something strange is missing, a more mysterious tone than the paper’s measured one.

One small inconsistency: Rutgers puts the galaxies within 22 million light-years, while the paper’s cut is 6 megaparsecs, about 20 million. SciQuest uses the paper.

What happens next

McQuinn plans to use NASA’s Roman Space Telescope, under an approved program, to find smaller, fainter dwarf galaxies over a large area of sky, then study them with Hubble and the James Webb Space Telescope. Telford plans to use Hubble to study individual massive stars in some GLOW galaxies. The authors also call for a larger sample to test the environment trend. For a sense of how astronomers read galaxy history from starlight, see our report on Gaia star ages and the Milky Way’s disk.

In one sentence

Nearby dwarf galaxies appear to hold only about a tenth of the oxygen their stars forged, simulations expect more, and where the rest went remains to be measured.

Read next

• Life cycle of a star: where oxygen is made.
• What is the universe made of? The wider cosmic inventory.
• JWST finds a barred spiral galaxy candidate: more on how galaxies grow.

Sources and disclosure

1. McQuinn K.B.W., Telford O.G., Kravtsov A.V., Burchett J., Cohen R., Hunter L., Skillman E.D., Roman-Duval J., Berg D.A., Cannon J.M., Chisholm J., Dalcanton J.J., Dolphin A.E., van Zee L., Williams B. “GLOW. II. A Census of Oxygen in Low-mass Galaxies.” The Astrophysical Journal 1011, 158 (2026). doi:10.3847/1538-4357/ae93a6. Preprint: arXiv:2608.07665.
2. NASA, “NASA’s Hubble Uncovers Case of Missing Oxygen in Nearby Dwarf Galaxies,” 2 November 2026: science.nasa.gov.
3. Rutgers University news, 2 November 2026: Stars make oxygen; small galaxies lose most of it.
4. Coverage read for framing: Sci.News, Earth.com (2 and 3 November 2026).

Disclosure: This article was drafted with AI assistance and checked against the sources above, including the paper’s open-access full text. Calculations by SciQuest: 1 megaparsec ≈ 3.26 million light-years, so 6 Mpc ≈ 20 million light-years; oxygen made per unit of present stellar mass = 0.01 ÷ (1 − 0.433) ≈ 0.0176; hypothetical 10⁸-Sun galaxy: 1.76 million solar masses made, 11% kept ≈ 190,000, missing ≈ 1.57 million. The graphic is a SciQuest original built from the paper’s reported values. Spotted an error? See our corrections policy.