{"id":"a9b238f3-23b8-4a35-a236-ea49088fa5ce","arxiv_id":"2501.16190","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The first large sample of 2D stellar velocity maps of ultra-diffuse galaxies shows a mix of rotating and non-rotating systems, suggesting two kinematic classes.","lead":"This paper uses MUSE integral-field spectroscopy to measure how stars move in 18 ultra-diffuse and low-surface-brightness galaxies in the Hydra I cluster, and finds that a minority of them rotate. It is the first large spectroscopic census of such faint galaxies, and the rotation pattern may help tell two proposed formation histories apart.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-class conclusion rests on a qualitative R/IR/NR classification that is not reproducible from the quoted ΔV values, and the 'non-rotating' class is an upper limit rather than a positive detection.","rationale":"The reader's weakest assumption correctly focuses on the rotation classification. I agree partially: the deeper issue is not only that some ΔV values are ~2σ, but that the classification is not reproducible from the published numbers and that the 'non-rotating' class is defined as an absence of evidence. The paper deserves credit for a homogeneous MUSE sample, transparent caveats, and Monte Carlo validation of σeff (Paper I), and the Appendix B maps allow independent checks. However, Section 6's two-class conclusion goes beyond what the classification demonstrates: the ΔV values show a roughly continuous distribution with no stated bimodality test, and the NR galaxies' detection limits are not quantified. My proposed re-analysis with a fixed gradient-significance threshold would settle whether the visual classification (and hence the two-class claim) is reproducible. I therefore keep the reader's CONDITIONAL verdict rather than upgrading or rejecting: the central idea is plausible and worth testing, but the classification must be made objective and the non-rotating class must be shown to be a genuine non-detection, not a sensitivity limit.","tokens_in":30060,"tokens_out":9630,"duration_ms":84779,"concrete_test":"Re-analyze the published Voronoi-binned VLOS maps with an objective criterion: fit a linear velocity gradient to each map (excluding bins with S/N<5), compute the gradient amplitude ΔV_fit and its uncertainty from the fit covariance, and classify a galaxy as rotating only if ΔV_fit/σ > 3 and the gradient direction is within 30° of the photometric major or minor axis. Compare the resulting assignments with Table 2 and test whether the R/IR and NR groups remain distinct in ΔV_fit. If UDG4 or UDG10 are reclassified as rotating, or if the 3σ upper limits of NR galaxies overlap the R/IR range, the two-class claim is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim in Section 6 ('two classes ... galaxies showing mild rotation and those not showing evidence of rotation along any axis') depends entirely on the R/IR/NR classification in Section 5.1 and Table 2. That classification is qualitative: no threshold is defined for a 'significant velocity difference,' and the reported ΔV values do not separate the classes. UDG4 is NR with ΔV=18±6 km/s (3σ), while UDG8 (IR) and LSB8 (IR) have ΔV=20±2 and 20±5 km/s; UDG10 is NR with ΔV=18±6 km/s. Conversely, UDG7 and UDG12 are classified as rotating with ΔV=41±21 km/s, only ~2σ. So the labels are not a monotonic function of ΔV significance. The NR class is defined as 'no evidence of rotation along any axis,' i.e. an upper limit, not a positive detection. With per-bin S/N=6-11 (Section 4.2) and velocity profiles that 'do not extend beyond Reff' (Section 5.1), these galaxies could host undetected rotation. The paper itself acknowledges that a measured velocity difference could be induced by tidal distortion rather than internal kinematics. Consequently, the observed ΔV distribution (≈5-45 km/s) shows no demonstrated bimodality, and the two-class conclusion is not established by the presented classification.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents MUSE integral-field spectroscopy for a sample of ultra-diffuse galaxies (UDGs) and low-surface-brightness (LSB) galaxies in the Hydra I cluster, as part of the LEWIS Large Programme. The authors measure stacked 1D velocity dispersions σ_eff, construct 2D stellar velocity fields via Voronoi binning, extract velocity profiles along photometric axes, and use the semi-amplitude ΔV to classify galaxies as rotating (R), rotating along an intermediate axis (IR), non-rotating (NR), or unconstrained (U). From these measurements they derive dynamical masses and dark-matter content, compare the sample with the Faber-Jackson relation, compute λ_R for five galaxies, and conclude that UDGs in Hydra I split into rotating and non-rotating classes, implying different formation channels.","tokens_in":30332,"tokens_out":3928,"duration_ms":34090,"significance":"If the central result holds, this is the first homogeneous, large census of spatially resolved stellar kinematics for UDGs in a cluster environment, providing a unique observational benchmark for formation scenarios (e.g., born versus tidal UDGs). The paper also delivers the first 2D velocity maps for a sample of UDGs, carefully characterizes the MUSE LSF, and uses Monte Carlo perturbations to estimate parameter uncertainties. These strengths make the paper potentially valuable for the UDG community, provided that the kinematic classification and the headline numbers are made robust and internally consistent.","major_comments":[{"comment":"The abstract claims '7 out of 18 UDGs in LEWIS show a mild rotation', but Table 2 lists only four UDGs (UDG1, UDG7, UDG8, UDG12) with R or IR classifications; the remaining three rotators (LSB6, LSB7, LSB8) are LSB galaxies, not UDGs. The counts of 7 rotating, 5 non-rotating, and 6 unconstrained are therefore inconsistent with the table for the UDG subsample. This needs to be corrected in the abstract and Section 7, and the implications for the 'two classes of UDGs' conclusion must be re-evaluated.","section":"Abstract and Table 2"},{"comment":"The R/IR/NR classification is not reproducible from the quoted ΔV values because no quantitative significance threshold is defined and the assignments are not a monotonic function of ΔV/σ_ΔV. For example, UDG4 (NR) has ΔV = 18 ± 6 km/s (3σ), while UDG8 (IR) has ΔV = 20 ± 2 km/s (10σ), but UDG7 and UDG12 (R and IR) have ΔV = 41 ± 21 km/s (~2σ). The paper should adopt an objective criterion (e.g., a minimum ΔV/σ_ΔV, a comparison with noise from randomized bin shuffling, or a kinematic position-angle fit) and apply it uniformly; otherwise the two-class conclusion in Section 6 rests on a fragile qualitative judgment.","section":"Section 5.1, Table 2"},{"comment":"The 'NR' class is defined as 'no evidence of rotation along any axis', which is an upper limit rather than a positive detection, given per-bin S/N = 6–11 (Section 4.2) and velocity profiles that do not extend beyond Reff (Section 5.1). The paper itself acknowledges that a measured velocity difference 'could be induced by an external tidal distortion rather than the internal stellar kinematics'. The conclusion that LEWIS UDGs 'can be grouped, on average, into two classes' therefore overstates the evidence: the observed ΔV distribution shows no demonstrated bimodality, and the non-rotating sample could contain undetected rotation. A quantitative treatment of the detection limits and a noise-injection test are needed to support the two-class claim.","section":"Section 5.1 and Section 6"},{"comment":"UDG10 is listed as NR in Table 2 with ΔV = 18 ± 6 km/s, but it is omitted from the NR list in Section 5.1 (which names UDG4, UDG9, UDG11, and LSB5). This discrepancy affects the census and the abstract count of '5 do not have evidence of any rotation'. Please clarify whether UDG10 is classified as NR or unconstrained and update the text and figure accordingly.","section":"Table 2, Section 5.1"}],"minor_comments":[{"comment":"UDG7 and UDG12 both have ΔV = 41 ± 21 km/s and the identical uncertainty; please verify that this is not a typographical copy error, and report the actual measured values and errors.","section":"Table 2"},{"comment":"The caption of Figure 2 states that red and blue lines mark the photometric major and minor axes, but the text in Section 4.2 refers to 'apertures parallel to the photometric major and minor axes'; please clarify whether the velocity profiles are extracted in apertures parallel to the axes or along the axes themselves.","section":"Section 4.2, Figure 2"},{"comment":"The Wolf et al. (2010) formula is quoted in text but not numbered; considering that the paper uses both σ_eff and V_rms for different galaxies, please number the equation and explicitly state the replacement of σ with V_rms in the rotating cases, as well as the caveat that the formula is strictly derived for dispersion-supported systems.","section":"Section 5.3, Equation (5.3)"},{"comment":"The λ_R values are derived for only five galaxies, all LSBs and UDGs combined, and the figure shows no distinction between UDGs and LSBs; the text should avoid overinterpreting the 'rotation-supported' label for the sample as a whole, given the small number and the S/N limits.","section":"Section 5.4, Figure 5"},{"comment":"There are several typographical and formatting issues, including 'V oronoi' with a space, 'di fferent' and 'a ffected' with extra spaces, and inconsistent notation for S/N (e.g., 'S/N' vs 'S /N'). A careful proofreading pass is recommended.","section":"Throughout"},{"comment":"The text says that for S/N < 10 the estimate of σ_eff is 'heavily overestimated', but Figure 3 (left panel) includes only constrained-fit galaxies for the LEWIS sample; please state explicitly how the unconstrained and intermediate-fit galaxies are treated in the physical interpretation of σ_eff, beyond their exclusion from the FJ and Mdyn/L analyses.","section":"Section 4.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is based on a valuable and unique dataset, and the methodological care (LSF measurement, Monte Carlo perturbations) is commendable. However, the headline claim of '7 out of 18 UDGs' is numerically inconsistent with Table 2, and the R/IR/NR classification is not reproducible without a quantitative criterion. These issues are fixable, but they are central to the paper's main conclusion, so I recommend major revision rather than rejection. I would also suggest that the authors provide a table with the individual ΔV, its uncertainty, the S/N per bin for the velocity profile, and the maximum radius of the profile, so that readers can independently assess the classification. The paper would also benefit from an explicit statement about the availability of the derived data products (velocity maps and profiles) in a public repository."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the dataset: before this, only NGC 1052-DF2 and DF44 had resolved stellar kinematics for UDGs. LEWIS II gives velocity fields and profiles for a dozen-plus UDGs and LSBs in one cluster, plus σeff, dynamical masses, and λR for five galaxies. That is a real advance, and the data handling looks careful: custom masks, LSF measured from the data, pPXF with Monte Carlo perturbations, and the paper is transparent about many of its own limitations. Credit where due.\n\nThe soft spots are real and mostly where the reader put them. First, the abstract says '7 out of 18 UDGs' show mild rotation, but Table 2 lists only four UDGs with R or IR flags (UDG1, UDG7, UDG8, UDG12); the other three rotators are LSB6, LSB7, LSB8. And UDG8 itself has Reff = 1.40 kpc and μ0 = 23.2, so it is not a strict UDG by the van Dokkum criteria — the UDG-specific count is closer to three. The abstract overstates the headline.\n\nSecond, the R/IR/NR classification is qualitative, and I agree with the stress-test that the quoted ΔV values do not separate the classes. UDG4 is NR with ΔV = 18±6 km/s (3σ), while UDG8 (IR) has ΔV = 20±2 and LSB8 (IR) has 20±5. UDG7 is R with 41±21, barely 2σ. The NR class is defined as no evidence along any axis — an upper limit, not a detection. With per-bin S/N of 6–11 and profiles that mostly stop at Reff, the non-rotating bucket is consistent with undetected rotation. So the 'two classes' conclusion in Section 6 is not actually supported by the presented classification. The paper acknowledges tidal distortion and limited radial coverage, then draws the strong conclusion anyway.\n\nThe sub-resolution σeff issue is minor: relying on Paper I's validation from the same team is normal in a series, though not independently reproduced. More annoying is that no velocity maps or derived products are offered for release; for a sample this faint, independent checking would be valuable.\n\nBottom line: this is a solid dataset paper with an overreaching interpretation. The existence of coherent velocity gradients in some cluster UDGs is plausible and worth stating, but the bimodal rotating/non-rotating split is not demonstrated by the quoted numbers. I would send it to review with a request for a rewritten abstract, a quantitative threshold for rotation significance, and ideally public velocity products. The data deserve to be in the literature.\n\nFor peer review: yes — a serious referee can fix the framing without discarding the measurements.","headline":"The first large 2D stellar kinematic census for UDGs is genuinely new and worth publishing, but the rotating/non-rotating two-class claim is not established by the data as presented.","tokens_in":31040,"tokens_out":3076,"would_cite":true,"duration_ms":27775,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"MUSE integral-field spectroscopy reveals that the ultra-diffuse galaxies of the Hydra I cluster split into mildly rotating and non-rotating classes.","keywords":["ultra-diffuse galaxies","low-surface-brightness galaxies","Hydra I cluster","integral-field spectroscopy","stellar kinematics","galaxy rotation","dark matter content","Faber-Jackson relation"],"falsifier":"Point deeper or higher spectral-resolution observations at the six unconstrained galaxies and the weakest rotators, reaching per-bin S/N above 15 and velocity fields beyond one effective radius. If the roughly 40 km s$^{-1}$ gradients disappear, wander off the photometric axes, or align with tidal features and isophotal twists, the two-class claim would lose its observational support.","tokens_in":29816,"feed_emoji":"🌌","tokens_out":14701,"duration_ms":120172,"temperature":0.7,"pith_summary":"The paper uses MUSE integral-field spectroscopy to measure, homogeneously, the 2D stellar kinematics of 18 ultra-diffuse and low-surface-brightness galaxies in the Hydra I cluster. Its central goal is to decide whether UDGs form one population or several by asking whether they rotate. The authors find two kinematic classes: seven galaxies show mild rotation with semi-amplitude $\\Delta V\\sim25$–$40$ km s$^{-1}$, five show no rotation along any axis, and six are too faint to constrain. If true, this means a single cluster environment has hosted at least two different formation histories, and rotation must be included when estimating UDG masses, since rotation can inflate velocity-dispersion estimates by up to an order of magnitude.","feed_headline":"Ultra-diffuse galaxies come in rotating and non-rotating types","feed_subtitle":"First homogeneous 2D kinematic census of Hydra I UDGs finds mild rotation in seven of eighteen.","key_machinery":"The central machinery is the spatially resolved stellar velocity field, obtained by Voronoi-tessellating MUSE spaxels (average S/N 6–11 per bin) and fitting every binned spectrum with pPXF, then extracting velocity profiles along the photometric major and minor axes. The semi-amplitude $\\Delta V$ of these profiles is the quantity that sorts galaxies into rotation along the major axis (R), rotation along an intermediate axis (IR), no rotation (NR), or unconstrained (U). A 1D stacked spectrum inside the effective radius yields $\\sigma_{\\rm eff}$, and dynamical masses are computed from $\\sigma_{\\rm eff}$ or, for the five brightest rotating galaxies, from the luminosity-weighted second velocity moment $V_{\\rm rms}$. The value of $\\lambda_R$, the projected specific angular momentum, then gives the rotation-versus-dispersion support for those five galaxies.","core_discovery":"On the paper's own terms, the discovery is that ultra-diffuse galaxies in a cluster are not one kinematically uniform population. From the 2D stellar velocity maps, 7 of the 18 LEWIS galaxies show mild rotation with $\\Delta V\\sim25$–$40$ km s$^{-1}$ (three along the photometric major axis, four along an intermediate axis), 5 show no evidence of rotation, and 6 are unconstrained because their spectra are too faint. The paper presents this as the first large census of velocity profiles for UDGs and argues that the existence of rotating and non-rotating classes implies different formation pathways. It also finds $\\sigma_{\\rm eff}$ peaking near 20–30 km s$^{-1}$, dynamical mass-to-light ratios $M_{\\rm dyn}/L_{V,\\rm eff}\\sim10$–$100$ $M_\\odot/L_\\odot$, and no clear correlation between kinematics and cluster-centric distance or infall stage.","pith_inferences":["If the rotation is intrinsic, the cleanest testable extension is that the rotating and non-rotating classes should differ in globular-cluster system rotation and star formation histories; the planned LEWIS stellar-population and globular-cluster analyses can check whether the two classes have distinct ages or globular-cluster content.","Because six of eighteen galaxies were unconstrained and several others yielded only two spatial bins, seven of eighteen is likely a lower limit on the rotating fraction; deeper data could make the split sharper or reveal a continuum.","Rotation along an intermediate axis in roundish galaxies may be a signature of triaxiality or tidal perturbation rather than a disk; comparing the kinematic axis with isophotal twists and tidal tails would test whether the IR class is a true kinematic class or an artifact.","The two-class claim predicts that independent measurements of dark matter content should separate the classes if they formed through different channels; the paper does not test that prediction, but GC dynamics from the same MUSE data could."],"forward_implications":["Two kinematic classes in the same cluster mean formation models must produce both rotation-supported and dispersion-supported UDGs, so the observed rotation fraction becomes a constraint on the relative rates of those channels.","For rotating UDGs, $\\sigma_{\\rm eff}$ overestimates the true velocity dispersion, so dynamical masses computed from it (as for UDG7 and UDG12) can be wrong by up to an order of magnitude; $V_{\\rm rms}$ is the safer input.","The LEWIS UDGs and LSBs are dark-matter-dominated at levels above dwarf galaxies of the same luminosity, so any successful formation scenario must put these objects in massive halos or explain the high mass-to-light ratios otherwise.","Absence of a clear correlation between rotation and cluster-centric distance or infall epoch argues that internal properties or pre-infall history, rather than current location in Hydra I, set the kinematics.","The $\\lambda_R$ values $0.25$–$0.35$ for five rotating galaxies show they are rotation-supported, consistent with simulated field UDGs, so cluster membership does not automatically erase rotational support."],"supporting_citations":[{"why":"Defines UDGs by central surface brightness fainter than 24 mag arcsec$^{-2}$ and effective radius above 1.5 kpc, setting the sample definition used throughout.","marker":"van Dokkum et al. 2015"},{"why":"Supplies the Voronoi binning algorithm used to co-add MUSE spaxels into the 2D velocity maps.","marker":"Cappellari & Copin 2003"},{"why":"Provides the pPXF spectral fitting code used for all velocity and velocity-dispersion measurements.","marker":"Cappellari 2017"},{"why":"Compiles the existing spectroscopic UDG measurements that the LEWIS results are compared with for $\\sigma_{\\rm eff}$ and dark matter content.","marker":"Gannon et al. 2024"},{"why":"Reports the earlier resolved stellar kinematics of NGC 1052-DF2 with mild rotation, the benchmark this census extends.","marker":"Emsellem et al. 2019"},{"why":"Reports the earlier resolved kinematics of DF44 with no rotation, the other benchmark this census extends.","marker":"van Dokkum et al. 2019a"},{"why":"Derives the mass formula from velocity dispersion and half-light radius used to compute dynamical masses.","marker":"Wolf et al. 2010"},{"why":"Simulation prediction of born versus tidal UDG classes in clusters that the observed two-class kinematics are interpreted against.","marker":"Sales et al. 2020"}],"fun_headline_variants":["Ultra-diffuse galaxies split into rotating and non-rotating types","First big 2D survey finds faint galaxies come in two kinematic types","Hydra I's ultra-diffuse galaxies: some rotate, some don't","Two classes of ultra-diffuse galaxies emerge from cluster census","Faintest galaxies in Hydra I reveal two kinematic natures"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification into rotating and non-rotating systems assumes that small velocity differences in low-signal-to-noise spectra (some only about $2\\sigma$, such as $\\Delta V=41\\pm21$ km s$^{-1}$ for UDG7 and UDG12) trace genuine stellar rotation rather than noise, projection effects, or tidal distortions; the paper itself states that an external tidal distortion could produce the same signal.","fun_headline_variants_meta":{"raw":{"variants":["Ultra-diffuse galaxies split into rotating and non-rotating types","First big 2D survey finds faint galaxies come in two kinematic types","Hydra I's ultra-diffuse galaxies: some rotate, some don't","Two classes of ultra-diffuse galaxies emerge from cluster census","Faintest galaxies in Hydra I reveal two kinematic natures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000868,"raw_usage":{"total_tokens":3860,"prompt_tokens":1143,"completion_tokens":2717,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":759,"completion_tokens_details":{"reasoning_tokens":2621}},"tokens_in":759,"tokens_out":2717,"duration_ms":18911,"temperature":1.0,"reasoning_tokens":2621,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:38:02.522061+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Point deeper or higher spectral-resolution observations at the six unconstrained galaxies and the weakest rotators, reaching per-bin S/N above 15 and velocity fields beyond one effective radius. If the roughly 40 km s$^{-1}$ gradients disappear, wander off the photometric axes, or align with tidal features and isophotal twists, the two-class claim would lose its observational support.","supporting_citations":[{"cited_title":"G., Romanowsky , A","cited_arxiv_id":null,"evidence_quote":"Defines UDGs by central surface brightness fainter than 24 mag arcsec$^{-2}$ and effective radius above 1.5 kpc, setting the sample definition used throughout."}],"review_version":1}