Small, bright galaxies offer a closer look at early star-forming systems

In the early universe, radiation from stars and other sources transformed much of the surrounding hydrogen gas from neutral to ionized. This period, called cosmic reionization, is thought to have been driven largely by star-forming galaxies, but their faintness makes their properties hard to measure. Could smaller, closer galaxies with similarly intense emission provide a more accessible way to investigate the kinds of systems involved?

M. Delgado Mancheño and colleagues searched for such galaxies in the Extended Groth Strip, a region observed by the Hubble Space Telescope’s CANDELS survey. They looked for extreme emission-line galaxies, or EELGs: galaxies whose gas produces unusually strong light at particular wavelengths. Their aim was to assemble a substantial sample around “cosmic noon,” when the universe’s overall rate of star formation was near its peak, and assess whether the galaxies resemble extreme emitters seen at much earlier times.

Finding galaxies by their colors

The team used CANDELS measurements in three broad wavelength bands. A strong emission line can make a galaxy look unusually bright in one band relative to neighboring bands, like a bump in an otherwise smoother spectrum. The researchers used this color excess to select candidates for two kinds of emitters: blended hydrogen-beta and oxygen lines at redshifts around 1.7, and hydrogen-alpha at redshifts around 1.0. Redshift measures how much the universe’s expansion has stretched a galaxy’s light, and thus helps indicate how far back in time we see it.

The selection yielded 102 candidates in the first group and 127 in the second. The researchers estimated galaxy properties using existing CANDELS catalogue measurements, then checked the selection against spectra from several facilities, including Hubble, Keck, the MMT, JWST, and follow-up observations with the Gran Telescopio Canarias. Spectroscopy separates light by wavelength, allowing astronomers to check whether the expected emission lines are present.

Based on photometric redshift probabilities, the authors estimate that about 84% of the hydrogen-beta-plus-oxygen candidates and 77% of the hydrogen-alpha candidates fall in the intended redshift ranges. Spectra were available for only 27 candidates: 19 in the first group and eight in the second. The main text reports confirmation rates of 94.1% and 62.5%, respectively. These checks cover only a small portion of the samples, particularly the hydrogen-alpha group, so they cannot establish the status of every candidate.

Compact and active—but not unusually small compared with peers

The galaxies have low stellar masses, with typical values around one hundred million times the Sun’s mass, and typical sizes of about 1.6 kiloparsecs. Their specific star-formation rates—the rate of forming stars relative to the stellar mass already present—are around 10⁻⁸ per year. Compared with galaxies matched in redshift, brightness, and mass, the EELGs are generally fainter, less massive, and more strongly star-forming.

A qualification matters: although these systems are compact in absolute terms, the authors do not find robust evidence that they are significantly more compact than the comparison galaxies. Nor do their tests find evidence of significant active-galactic-nucleus contamination. However, the available diagnostics cannot rule out every heavily obscured active nucleus, so the result is best described as no evidence for a significant contribution—not proof that none are present.

A statistical fit, adjusted for the survey’s mass-completeness limit, indicates that the characteristic stellar mass of the selected population changes with redshift. This suggests an evolutionary trend within the sample and the period studied; it is not a direct record of how individual galaxies grew. Comparisons across studies also depend on selection criteria and analysis methods. The authors note that some literature samples do not align closely with their results, while a comparison using a similar selection method agrees better.

Analogues, not time machines

The researchers argue that their galaxies share important measured properties with extreme emitters found at redshifts above 5, including low stellar masses and strong emission. That makes them potential intermediate-redshift analogues: accessible examples for investigating processes that may also occur in early galaxies. But resemblance does not show that these particular systems emitted enough escaping ionizing radiation to reionize the universe. The study selects galaxies through line-related colors; it does not directly measure their contribution to reionization.

There are additional uncertainties. Broadband color excess can be confused by other spectral features or galaxies at different redshifts. The authors caution that their line estimates may be high if continuum features or neighboring lines contribute to the measured excess. Most candidates lack spectroscopic confirmation, and star-formation estimates from different methods can disagree because they trace activity over different timescales. The paper calls for deeper spectroscopy to test the full sample and better characterize its members.

The manuscript says it was accepted for publication in Astronomy & Astrophysics in November 2026. The source available for this article is the manuscript and abstract, not an independent assessment of the findings. There is also an inconsistency in the supplied text: the abstract gives a hydrogen-beta-plus-oxygen spectroscopic confirmation rate of 85%, while the main text reports 94.1%—18 confirmations among 19 candidates with spectra. The figures should not be silently combined or treated as equivalent. Overall, the study presents cosmic-noon EELGs as promising systems for further study, while leaving open how directly they represent the galaxies responsible for reionization.