REBELS-25, a massive star-forming galaxy at redshift 7.31, appears to contain warm ionised gas and colder neutral gas moving within the same large-scale rotating structure. The study found no separate kinematic component at the locations of clumps seen with JWST, although faint components could still be missed. In the fiducial model, ordered motion outweighed random motion in both tracers: the ordered-to-random ratio, V/σ, was 4.5 for [O iii] 88 μm and 11 for [C ii].
That does not make the galaxy a simple, perfectly circular disc. Residuals left after a rotating model was fitted, warped equal-velocity contours and features in the spectra all indicated non-circular motions. The data could not show whether those motions came from inflows, outflows, interactions or noise.
The result rests on a spatially resolved, multi-wavelength case study of one galaxy. The ALMA data included new Band 8 observations of [O iii] 88 μm and the underlying dust continuum at roughly 0.2 arcsec, or about 1 kpc, alongside higher-resolution [C ii] and 150 μm continuum data at about 0.14 arcsec, or 710 pc. JWST/NIRSpec integral-field observations were also included in the broader comparison.
Many ways to map one galaxy
Across those views, the galaxy did not have one single shape. Ultraviolet and optical emission was highly clumpy and irregular, with its clumps offset from the far-infrared, or FIR, emission by about 2 kpc. The FIR maps instead followed a near-exponential disc, with Sérsic indices, a measure of profile shape, of about 1 to 1.8. [C ii] was the most extended tracer, with an effective radius of about 2.2 kpc; the 90 μm continuum was the most compact, at about 1.1 kpc. Researchers quantified the shapes by fitting two-dimensional Sérsic profiles with models convolved with the telescope's point-spread function.
That difference between the ultraviolet and infrared views also changed the picture of star formation. In four clump apertures, resolved ultraviolet/FIR comparisons estimated that obscured star formation made up roughly 55% to 98% of the total star formation. The obscured share was dominant even in ultraviolet-luminous clumps, so the brightest regions in ultraviolet light were not necessarily the places where the full star-forming activity was easiest to see.
To make that estimate, the team used point-spread-function-matched, JvM-corrected fluxes at 90 and 150 μm. It fitted an optically thin modified-blackbody dust model, corrected for the cosmic microwave background and propagated uncertainty with Markov-chain Monte Carlo. The result depends on assumptions about dust temperature, and the apertures overlap, so the regional luminosities and star-formation rates cannot simply be added to create a global budget.
The chemistry looks broadly even
The gas diagnostics looked more even from place to place. Two emission-line ratios used as indirect interstellar-medium diagnostics, R3 and [O iii]/[C ii], varied by no more than about a factor of two across the resolved regions, and most clumps were consistent with the integrated values. R3-based oxygen abundances ran from about 8.29 to 8.47 on the 12 + log(O/H) scale, corresponding to roughly 0.4 to 0.6 times the Sun's metallicity. Within the stated uncertainties, that gave no strong evidence of a metallicity gradient. The absolute values remain dependent on the calibration used for strong emission lines, dust assumptions and the simplified one-zone model.
Rotation, with important caveats
The rotation estimates came from 3DBAROLO version 1.7, which fits three-dimensional models made of rotating rings while accounting for the telescope beam, then uses Monte Carlo trials to estimate uncertainty. In that fiducial fit, [O iii] 88 μm had a maximum rotation speed of 274 km/s (+89/-87), a velocity dispersion, or spread of random speeds, of 61 km/s (+7/-8), and V/σ of 4.5 (+2.4/-1.8). For [C ii], the corresponding figures were 374 km/s (+86/-91), 33 km/s (+9/-7), and 11 (+6/-5).
Those figures should not be read as a universal warm-versus-cold difference. The comparison changed with the model: DysmalPy gave an ionised-gas dispersion no higher than the cold-gas value, while 3DBAROLO gave an ionised-to-cold dispersion ratio of about two. The comparison is therefore a model-dependent detail, not a settled universal offset between the two phases.
The limits of a one-galaxy view
The study's reach is deliberately narrow. It examines one galaxy, with resolved comparisons drawn from four within-galaxy clump apertures and an integrated spectrum, at roughly kiloparsec resolution. That design can reveal how different tracers behave inside REBELS-25, but it cannot show how common dusty, enriched, rotationally supported discs are across the wider high-redshift population or resolve smaller-scale gradients. Low signal-to-noise also leaves tentative FIR residual clumps, the optical continuum, residuals and possible broad spectral components vulnerable to noise.
For now, the clearest picture is of ordered rotation coexisting with a complicated face: ultraviolet and optical clumps, a compact-to-extended mix of dust and gas emission, hidden star formation and signs of non-circular motion. The authors interpret REBELS-25 as a dusty, chemically enriched, dynamically cold, rotationally supported disc, while leaving open whether its faint clumps are real structures, merging components or noise, and whether its extra motions are inflows, outflows, interactions or noise. Because the analysis covers only one galaxy, it cannot establish how common such systems are.
Paper data and sources
Original title: REBELS-25: multi-phase morphology and kinematics at z = 7.31
Authors: Lucie E. Rowland, Hiddo S. B. Algera, Jacqueline Hodge et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text