A new arXiv preprint reports that some high-redshift galaxies may look extremely metal-poor because their emitting gas is much denser than the standard analysis assumes. In a study of four candidates, accounting for that density produced substantially higher oxygen-abundance estimates and led the authors to classify three as possible extremely metal-poor galaxy, or EMPG, impostors.
The difference reached up to about 1.1 dex in the high-density cases. Dex is a logarithmic unit used to report abundance differences. The result does not establish how common such objects are across the wider high-redshift galaxy population; it shows that the density assumption changed the classification of individual candidates in this selected sample.
A narrow test of four candidates
The researchers used ultra-deep, medium-resolution spectroscopy from the James Webb Space Telescope's NIRSpec instrument. The observations used three gratings, with a resolving power of about R=1,000 and exposures of approximately 29 hours, 8 hours and 3 hours for the G140M, G235M and G395M settings.
The screening began with 75 spectra. Of those, 47 were at redshifts of about 4.8 to 9.5. The team then retained 16 objects with R3 below 5, where R3 is the study's oxygen-to-hydrogen emission-line ratio. Finally, it required detections stronger than 3 standard deviations for all three required O iii lines. Four candidates remained.
That sequence makes the scope of the finding important. The final result came from candidates selected for their redshift coverage, low R3 and detections of the three oxygen lines, rather than from an unrestricted sample of high-redshift galaxies.
Measuring density instead of assuming it
The comparison used the traditional direct-temperature method, which treats the gas as low density, and a self-consistent method that measured electron temperature and density together. The newer calculation combined three O iii ratios and selected the best temperature and density from the intersections of their curves. The search covered log(n_e/cm−3) from 0 to 7 in 1,000 steps.
Emission lines were fitted with an emcee Markov-chain Monte Carlo procedure, which samples possible values to estimate uncertainty. The analysis used 32 walkers and 5,000 steps, discarding the first 3,000 as burn-in. The resulting uncertainty ranges were carried through the dust correction, temperature, density and metallicity calculations.
The approach focused on the high-ionization oxygen-emitting zone. That matters because the density measured there need not be the same as the density inferred from lines produced in another part of the gas.
The metallicity estimates moved upward
The self-consistent analysis gave electron temperatures of about 13,000 to 22,000 K. Three candidates had very high [O iii]-zone densities, roughly 10^5 to 10^6 cm−3. Candidate 437 had an upper limit of about 10^5.2 cm−3, while candidate 544 also allowed a low-density solution because its line-ratio uncertainties were large.
Under the traditional low-density treatment, the candidates had inferred metallicities of about 12+log(O/H)=7.0 to 7.2. The self-consistent calculation instead produced values spanning about 12+log(O/H)=7.3 to 8.2. The notation is the oxygen-abundance scale used in the study.
The authors therefore classified three of the four candidates as high-density EMPG impostors. Candidate 437 was the only one with density low enough to be genuine in their interpretation, although candidate 544 remained consistent with that possibility.
The low R3 measurements did not resolve the issue. R3 ranged from 2.8 to 4.3, corresponding to strong-line metallicities of approximately 12+log(O/H)=7.2 to 7.5. The optical-line signal used to flag the candidates as unusually metal-poor was therefore still compatible with the high-density interpretation.
The authors interpret the pattern as a problem with applying the low-density calculation to high-density gas. In the affected cases, the analysis found suppressed [O iii] λ5008 alongside the lower metallicity estimates from the low-density treatment. That explanation is tied to this line-ratio comparison, not presented as a general result for every high-redshift galaxy.
Different parts of the gas may tell different stories
The study also found evidence that the emitting gas may not have one uniform density. In two objects classified as impostors, the density inferred from [O iii] was about 2 dex higher than the density inferred from C iii].
The comparison involved only two impostors, and the analysis did not fully map density across all ionization zones. The authors say the inferred structure of the interstellar medium therefore remains incomplete, even though the contrast raises questions about using one density measurement to represent all of the emitting gas.
A result that needs a larger test
The authors suggest that impostors could represent about 50% to 75% of EMPG candidate samples in this analysis. They describe that as an exploratory, sample-specific estimate and say that a more detailed completeness analysis is needed before the figure can be used to assess prevalence more broadly.
The final sample contained only four candidates from a single lensed field and depended on low R3 plus detections of all three required O iii lines. The temperature-density curves did not converge for candidate 437, and the uncertainties for candidate 544 allowed both high- and low-density interpretations.
The supplied analysis also reports no independent high-ionization density validation; nitrogen lines were not detected. Where a direct [O iii]-zone density was unavailable in comparisons with other work, the study used N IV density as a representative high-ionization tracer.
The researchers tested whether requiring a detection of O iii] λ1666 had itself favored dense objects. In their simplified fixed-flux test, the requirement did not preferentially exclude genuine low-density EMPGs. Even so, that line-detection condition was part of the selection that produced the four-object sample.
The next test will need larger samples with measured completeness and independent high-ionization diagnostics. The authors point to N IV] or ALMA [O iii] measurements as possible checks, along with comparisons across multiple ionization zones to establish how their densities are related.
Those measurements could show whether density corrections alter conclusions drawn from high-redshift mass-metallicity relations, fundamental-metallicity relations and abundance patterns. The four candidates in this preprint cannot answer that population-wide question on their own.
The evidence is still preliminary
The work is identified as arXiv:2608.20339v1 in astro-ph.GA. The acknowledgments report University of Texas at Austin support for Tiger Hsiao and NSF Graduate Research Fellowship support for Ansh R. Gupta under grant DGE-2137420.
For readers interpreting JWST metallicity measurements, the practical warning is straightforward: the assumed electron density can materially affect how emission-line ratios are translated into an EMPG classification. Whether the same effect is widespread remains an open question requiring broader, independently checked samples.
Paper data and sources
Original title: An Optical Illusion: High Electron Densities Create Extremely Metal-Poor Galaxy Impostors
Authors: Tiger Yu-Yang Hsiao, Danielle A. Berg, Steven L. Finkelstein et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text