{"id":"261451a0-f73b-44b2-9dc1-c992dd4e3adc","arxiv_id":"2501.07569","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"A model-based calibration gives effective temperatures, ionizing fluxes, and intrinsic colors for 0.1 solar-metallicity O-type and early-B supergiant stars, and warns that HeII-ionizing fluxes of late-O dwarfs are highly uncertain.","lead":"Astronomers produced the first calibration chart for very metal-poor massive stars, converting spectral types into temperatures, ionizing light output, and colors at one-tenth of the Sun's metal content. The chart matters because such stars powered the early universe and are now being observed in nearby dwarf galaxies and with the James Webb Space Telescope.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The un-published adaptation of spectral classification criteria to Z=0.10 Zsun (Sect. 2.2) is the load-bearing gap: a one-subtype shift in model SpT propagates through all Teff, ionizing-flux, colour, and extinction results.","rationale":"The paper's central claim is to provide a first reference framework for XMP OB stars, so the assignment of spectral types to synthetic spectra is the pivot on which all tables and derived claims turn. The reader correctly identified that this pivot is deferred to an unpublished companion paper. I agree with that identification: the internal consistency of the FASTWIND/CMFGEN SED comparisons and the external agreement with Ramachandran et al. (2021) and Telford et al. (2023) are real supporting evidence, but they test only a few ionizing-flux points, not the whole SpT-dependent calibration. A one-subtype misclassification would shift Teff by several kK and propagate through ionizing fluxes, colours, and the Sextans A extinction map; the paper even acknowledges that B-type classification is fragile at this metallicity. The qHeII bimodality is a secondary but related vulnerability, since the high-qHeII branch depends on which wind-strength models survive the Sect. 2.3 selection and are binned into the same subtype. However, the classification gap is the more fundamental single concern, and fixing it would also allow the qHeII claim to be checked model-by-model. The requested concrete test is feasible: it requires the authors to publish the classification rules and grid assignments, which they have already promised in the data-availability statement, plus an independent reclassification of a sample. Since the reader's CONDITIONAL verdict already reflects this concern, my independent stress test does not change the verdict; it reinforces it. I therefore mark verdict_should_be as UNCHANGED.","tokens_in":46763,"tokens_out":6974,"duration_ms":75950,"concrete_test":"Ask the authors to publicly release, or provide for review, the exact classification rules used for O and B subtypes at 0.10 Zsun, including all line-intensity and line-ratio thresholds, the He I 4471 / He II 4541 grid, the adopted detectability cuts, and the full SpT/LC assignment for all 13,700 grid models. Then have two independent classifiers re-assign spectral types to a blinded random sample of the model spectra (R=2500, vsini=70 km/s, with and without the adopted noise) using those published rules. If the re-assignment differs by one subtype or more for a non-negligible fraction (e.g. >5%) of models in any (SpT, LC) bin, recompute Tables 1-2 and Eqs. 3-6 with the corrected subtypes and quantify the resulting shifts in Teff, ionizing fluxes, and colours, and in the Sect. 5 extinction map.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The calibration is built by assigning each FASTWIND synthetic spectrum a spectral subtype using classification criteria 'adapted to 0.10 Zsun metallicity following Lennon (1997)'s strategy (see Lorenzo et al. in prep)' (Sect. 2.2). That adaptation is not specified in this paper. The paper itself states that B-type classification is severely hampered by weak Si and Mg lines at this metallicity and low S/N, forcing reliance on He lines and an assumed S/N=100 detectability limit; O-star classification likewise depends on the assumed microturbulence (Appendix A) and on the detectability threshold I_lambda = 1/(S/N) calibrated from injected noise (Figs. 3-4). Because every downstream result (Teff scale Eq. 3, qHI/qHeI fits Eqs. 5-6, the qHeII bimodality, photometric Tables 2 and B1-B3, and the Sextans A E(B-V) map in Sect. 5) is averaged over models binned by the assigned subtype, an error of even one subtype shifts the central reference values by approximately the bin spacing: up to 2-3 kK in Teff, up to about 0.02-0.04 mag in colours, and potentially much more for qHeII. The headline claim that 0.10 Zsun OB stars are 1-6 kK hotter than Galactic analogues is exactly the kind of statement that depends on the adapted classification. Since the adaptation and the full classified grid are not public, no external check is currently possible. This is a correctness and reproducibility risk, not merely a presentation issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs a model-based reference framework for extremely metal-poor (0.10 Zsun) OB-type stars from a grid of 13,700 FASTWIND synthetic spectra. The models are assigned spectral types and luminosity classes using optical classification criteria adapted to low metallicity (the adaptation itself is deferred to a companion paper), then filtered for plausibility using the wind-momentum-luminosity relation and the Eddington factor, leaving 5,150 models. Averaging over models with compatible spectral morphology yields calibrations of Teff (Eq. 3), log qHI and log qHeI (Eqs. 5 and 6), upper limits on log qHeII, and photometric colours and bolometric corrections in several filter systems (Tables 2, B1-B3), supplemented by CMFGEN-based corrections for hydrogen level dissolution (Sect. 4.1). The headline results are that 0.10 Zsun OB stars are 1-6 kK hotter than Galactic analogues, produce higher H I and He I ionizing fluxes, that log qHeII is bimodal in mid- and late-O types with the high-flux branch not captured by some population-synthesis calibrations, and that the Sextans A extinction is non-negligible and uneven (Sect. 5).","tokens_in":47158,"tokens_out":6463,"duration_ms":69094,"significance":"If the calibration holds up, it fills a genuine gap: there is no existing SpT-based scale for stellar parameters and intrinsic colours of XMP massive stars, and the paper's openly stated methodology (classify, filter, average over degenerate models) is well suited to quantifying the intrinsic scatter of each subtype. The authors deserve credit for several concrete strengths: the pipeline is described in detail with a flowchart (Fig. 1); the averaging over the model grid explicitly addresses the degeneracy between Teff, log g and wind strength; the ionizing-flux scale agrees with independent observational estimates from Ramachandran et al. (2021) and Telford et al. (2023) for the H I and He I fluxes (Fig. 12); cross-code checks against CMFGEN and TLUSTY bound the model-dependence of the ionizing fluxes (Sect. 3.2); the photometric corrections for hydrogen level dissolution are a careful, non-obvious refinement; and the paper makes falsifiable predictions (hotter Teff scale, bluer (U-B), bimodal qHeII, non-uniform reddening in Sextans A) that can be tested with future observations.","major_comments":[{"comment":"The adapted spectral classification criteria are the load-bearing step of the entire calibration, but they are deferred to 'Lorenzo et al. in prep' rather than specified. The paper states that the classification follows Sota et al. (2011) and Negueruela et al. (2024) 'adapted to 0.10 Zsun metallicity following Lennon (1997)'s strategy', with only the detectability threshold (I_lambda = 1/(S/N)) and the adopted microturbulence documented. A one-subtype error in the assigned SpT propagates into every downstream result: the slopes in Eq. (3) are ~2.6-3.9 kK per subtype, the qHI and qHeI fits in Eqs. (5)-(6) shift by amounts comparable to the quoted uncertainties, colors change by up to ~0.02-0.04 mag, and the qHeII upper limits in Table 1 can move by several dex because of the steep bimodal branches (Sect. 3.2.1, Fig. 9). The paper itself acknowledges that B-type classification is 'severely hampered' by weak Si and Mg lines even at S/N = 100, forcing reliance on He lines. Since none of the adapted criteria are given and the classified grid is not released, no external check of the classification is currently possible; this is a correctness and reproducibility risk, not a presentation issue. I request that the adapted criteria be specified in an appendix or that the companion paper and the full classified grid be made available before acceptance.","section":"Section 2.2"},{"comment":"The physical filtering that selects which models enter the calibration depends on radii that are themselves adopted from other calibrations. For O stars, radii are taken from Martins et al. (2005a) and scaled to 0.10 Zsun assuming constant luminosity at fixed spectral type, iterated to convergence; for B supergiants, radii come from a linear regression over literature analyses. The WLR and Gamma_e filters then determine which 5,150 models survive, and all averages in Table 1, Eqs. (3), (5)-(6), and the qHeII maxima are computed over the surviving set. A systematic error in the assumed radii (for example, if the constant-luminosity scaling is not accurate at fixed spectral type) would change which models pass the filters and therefore shift the calibrated means and the stated min-max ranges, even though the per-model radii are varied by 20%. The 20% scatter probes random uncertainty but not the systematic offset. I ask for a sensitivity test in which the adopted radii are globally varied by plus and minus 10% and the calibration outputs are recomputed, and for a test of the chosen weak-wind luminosity threshold (log L/Lsun <= 5.6).","section":"Section 2.3"},{"comment":"The empirical support for the headline '1-6 kK hotter' claim is weaker for O dwarfs than the text suggests. In Fig. 8, the two Leo A dwarfs (K1, O9.5 V and K7, O9.7 V from Gull et al. 2022) lie about 4 kK below the new XMP scale and are compatible with the Galactic scale for class V, and the IC 1613 O8.5 I star (Garcia & Herrero 2013) also deviates substantially. The paper explains these discrepancies as possible misclassification of the observed stars, but the same classification logic is being applied to the synthetic spectra without external calibration, so the attribution of the discrepancy to the observations is not fully convincing. The agreement with observation is strong for the ionizing fluxes (Fig. 12) and for early-B supergiants, but for O dwarfs the XMP-vs-Galactic offset is currently a model-prediction resting on the unvalidated classification. Please either add a quantitative test (e.g., spectral fitting of one benchmark star with the new grid) or discuss explicitly what remains of the dwarf temperature offset if the Leo A dwarfs are taken at face value.","section":"Figure 8; Section 3.1"}],"minor_comments":[{"comment":"The summary uses the abbreviation 'S11' for Sota et al. (2011) while the main text cites the full name; please make the citation style consistent.","section":"Section 6"},{"comment":"The table and appendix heading refer to 'WFPC3' filters; the instrument is WFC3 (as in Sect. 4). Please correct this typo.","section":"Appendix B, Table B3"},{"comment":"The E(B-V) annotations (0.0, 0.5, 1.0, etc.) overlap heavily with the HI contours and the galaxy image, making the map difficult to read. Consider providing the per-star E(B-V) values in a machine-readable table and enlarging the color scale.","section":"Figure 23"},{"comment":"Please state explicitly the units of the filter zero points ZP_T used in Eq. (8) and specify the adopted value of the solar bolometric magnitude in Eq. (9), since the BC_V values in Tables 2 and B1 are quoted to 0.01 mag and are used with M_V.","section":"Equations (8) and (9)"},{"comment":"The CMFGEN grid used for the level-dissolution corrections (Fig. 18) is described as having one mass-loss value per Teff-logg pair; stating the adopted values of log Qwind in that grid would make the quoted 0.001 mag insensitivity to this parameter more transparent.","section":"Section 4.1"},{"comment":"The data availability statement says 'All data are incorporated into the article and its online supplementary material', but the full classified grid (SpT, LC, Teff, log g, log Qwind per model) is not part of the article and the classification criteria are in a companion paper. Please clarify or amend this statement, and consider releasing the classified grid to enable reproduction of Eqs. (3), (5), and (6).","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The central issue for the editor: the paper's most important methodological step, the metallicity-adapted spectral classification of the synthetic grid, is contained in an unpublished companion paper by the same group. The calibration is therefore not independently checkable in its current form. I would consider conditioning acceptance on either (a) the companion paper being posted and made available to the referees before final acceptance, or (b) moving the essential classification criteria into an appendix, even in condensed form. The science is well within the scope of MNRAS and the paper is likely to be highly cited as a reference work, so this is a fixable major-revision situation rather than a rejection. There is also a minor mismatch between the data-availability claim and the actual availability of the classified grid that should be reconciled."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper provides the first systematic calibration of Teff, log g, ionizing fluxes, colors, and bolometric corrections for O and early-B stars at 0.1 Zsun. That alone makes it useful. The methodology is mostly transparent: they parse a large FASTWIND grid, filter out unrealistic models with WLR and Eddington constraints, apply CMFGEN corrections for hydrogen level dissolution, and validate their ionizing fluxes against independent analyses (Ramachandran 2021, Telford 2023). The agreement there is a real point in its favor. I'd also credit the honest reporting of ranges per subtype\\u2014they don't pretend one parameter set fits each spectral type, which is a genuine improvement over earlier calibrations.\n\nThe main soft spot is exactly what the stress-test flags: the spectral classification adaptation is not in this paper. Section 2.2 says the criteria follow Lennon (1997)'s strategy \"see Lorenzo et al. in prep\" without specifying the actual line-ratio thresholds. They give detectability limits, S/N choices, and argue microturbulence changes classifications by less than 0.5 subtypes, but the core mapping from line ratios to subtypes is load-bearing and unpublished. A one-subtype shift propagates through the Teff scale, the ionizing flux fits, the colors, and the Sextans A extinction map. The model grid is also not public, so no external check is currently possible. I wouldn't call this fatal\\u2014the paper is self-consistent and the physical arguments are sound\\u2014but it is a real reproducibility risk, and I'd want it fixed or at least acknowledged more prominently.\n\nThe qHeII bimodality claim is the most attention-grabbing part. \"Up to 4 orders of magnitude\" sounds dramatic, but the mechanism is well-established (wind-dependent ground-state population of HeII), and they demonstrate it cleanly in Figures 14\\u201315. My take: it's a plausible warning for population synthesis codes, not a measured result. The absolute numbers should be treated as model predictions until more observational constraints on Mdot in XMP O stars exist.\n\nThe Sextans A extinction map is a nice demonstration but not the main contribution; it inherits the classification uncertainty and probably also blends in binarity, disks, and unresolved clusters, which the authors themselves acknowledge.\n\nBottom line: this deserves a serious referee. The gap is not in the physics; it's in the missing classification details and unavailable grid. I'd ask for the companion paper or at least a release of the classified grid before publication, but the work itself is honest and internally consistent.\n\nFor a reading group, I'd say maybe\\u2014it's a good case study in how model-based calibrations can be rigorous yet difficult to reproduce. I'd cite it if I worked on XMP massive stars, but I'd want the classification details first.","headline":"A genuinely useful first calibration for 0.1 Zsun OB stars, with a real reproducibility gap: the classification criteria are deferred to an unpublished companion paper, and every number in the tables inherits that uncertainty.","tokens_in":47711,"tokens_out":2050,"would_cite":true,"duration_ms":24787,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Extremely metal-poor O and B stars are hotter and more ionizing than their Milky Way counterparts, with a HeII-ionizing output that depends on wind strength as much as on spectral type.","keywords":["extremely metal-poor OB stars","FASTWIND models","spectral classification","effective temperature scale","ionizing photon fluxes","HeII-ionizing flux","intrinsic colours","Sextans A extinction"],"falsifier":"Take a sample of ~20 XMP O dwarfs with high-S/N spectra, classify them with the paper's criteria, and measure $T_{\\mathrm{eff}}$ independently from Balmer line profiles and the SED; if O9 V stars consistently come out near 30 kK instead of the calibrated ~34.5 kK, the $T_{\\mathrm{eff}}$ scale is wrong. Alternatively, resolve an XMP HII region whose ionizing late-O dwarf has a wind measured from UV P Cygni profiles: the model predicts $\\log q_{\\mathrm{HeII}}$ jumps by up to 4 dex across similar subtypes as wind strength changes, so a single-valued high $q_{\\mathrm{HeII}}$ for all O8-9 V stars would refute the bimodality claim.","tokens_in":46555,"feed_emoji":"🔭","tokens_out":8558,"duration_ms":76911,"temperature":0.7,"pith_summary":"Stars are classified by the lines in their spectra, but the mapping from spectral type to physical properties has only been calibrated for metal-rich galaxies. This paper builds the first such calibration for extremely metal-poor (XMP) O and early-B stars, at one tenth of the Sun's metallicity, by classifying a grid of 13,700 synthetic stellar atmospheres and averaging the parameters of all models that reproduce each spectral subtype. It concludes that XMP OB stars of a given type are 1-6 kK hotter than their Galactic analogues, emit more hydrogen- and helium-ionizing radiation, and have a HeII-ionizing output that depends on wind strength so strongly that single-value calibrations can miss individual late-O stars by up to four orders of magnitude. A sympathetic reader would care because these stars are the local proxies for the first massive stars in the Universe, and the same numbers feed population synthesis and photoionization codes used for high-redshift galaxies.","feed_headline":"Metal-poor O stars run 6,000 K hotter than Milky Way twins","feed_subtitle":"New 0.1-solar-metallicity OB scale shows HeII ionizing output can swing by a factor of 10,000.","key_machinery":"The central object is the classified model grid: 13,700 FASTWIND synthetic spectra at 0.10 $Z_\\odot$, degraded to $R=2500$ with $v \\sin i = 70$ km/s, assigned spectral subtypes and luminosity classes using criteria adapted from Lennon (1997) and Sota et al. (2011), and filtered to 5,150 realistic models by the wind-momentum-luminosity relation (scaled from the SMC) and an Eddington-factor cut. The load-bearing step is averaging stellar parameters over all models compatible with each (spectral type, luminosity class) pair, which is what turns a grid into a calibration that carries a range of values instead of a single sequence. The colour tables are corrected for hydrogen-series line dissolution using CMFGEN models, and the $Q_{\\mathrm{phot}}$ pseudo-colour relations translate the calibration into reddening-free selection and extinction estimators.","core_discovery":"The central claim is that a grid of 13,700 FASTWIND model atmospheres at $Z = 0.10\\,Z_\\odot$, classified with spectral criteria adapted to that metallicity and filtered to physically realistic winds, yields the first reference calibration of stellar parameters and intrinsic colours for XMP OB stars. On this scale, O dwarfs of a given subtype are up to ~6 kK hotter than the Milky Way scale of Martins et al. (2005a), giants and supergiants are 1-2 kK hotter, and the whole class emits more H i and He i ionizing photons. The He ii-ionizing flux is not a single-valued function of spectral type: within one late-O subtype it splits into a high and a low branch depending on the wind-strength parameter, with differences of up to 4 dex, so the paper reports only upper limits for $\\log q_{\\mathrm{HeII}}$. Applying the calibrated intrinsic colours to the Sextans A OB-star sample yields a patchy extinction map with $E(B-V)$ reaching 0.5-0.6 mag, showing that internal reddening in this 0.10 $Z_\\odot$ dwarf galaxy is non-negligible and uneven.","pith_inferences":["If the bimodal HeII-ionizing behaviour is real, nebular HeII 4686 and HeII 1640 emission in XMP galaxies may serve as a wind diagnostic for late-O dwarfs, complementing UV and near-IR wind tracers.","The same grid-and-classify pipeline could be run at 0.05 and 0.2 $Z_\\odot$ to build a continuous metallicity ladder, which would let population synthesis codes interpolate ionizing fluxes with wind-strength aware libraries rather than scaling the Milky Way scale.","A direct test of the calibration would be to compare the paper's assigned spectral types against an independent XMP spectral atlas: a systematic one-subtype shift would move every $T_{\\mathrm{eff}}$ entry by roughly 3 kK in the opposite direction, because spectral type and temperature are tied through the He line ratios.","The Sextans A extinction map suggests that some apparently blue 'outlier' stars in dwarf-galaxy CMDs may be reddened rather than young, which would matter for star-formation histories inferred from resolved stellar populations."],"forward_implications":["Any use of Milky Way effective-temperature scales on 0.10 $Z_\\odot$ O and B stars will under-estimate their temperatures, and the under-estimate grows toward dwarfs, reaching ~6 kK.","Population synthesis codes that adopt a single wind-strength per spectral type will mis-estimate the HeII-ionizing photon output of late-O dwarfs by up to four orders of magnitude, which directly affects predictions of HeII emission in low-metallicity galaxies.","At fixed $Q_{\\mathrm{phot}}$, the calibrated $(B-V)_0$ is redder by 0.07 mag than Massey et al. (2000)'s relation, so extinction estimates and candidate cuts shift toward redder intrinsic colours.","Sextans A's internal reddening, with colour excesses up to ~0.6 mag in some sightlines, means that assuming negligible extinction in XMP dwarf galaxies biases derived luminosities, masses, and star-formation rates.","Four orders-of-magnitude in HeII-ionizing output also changes the interpretation of nebular HeII emission: strong HeII lines do not automatically require very hot or evolved stars if a late-O dwarf with a weak wind is present."],"supporting_citations":[{"why":"Supplies the Galactic O-star effective-temperature and radius calibration that is the direct comparison baseline and the source of the radii scaled to 0.10 $Z_\\odot$.","marker":"Martins et al. 2005a"},{"why":"Provides the 0.20 and 0.05 $Z_\\odot$ HeII-ionizing flux scales used by population synthesis codes that the paper claims underestimate late-O XMP dwarfs by up to 4 dex.","marker":"Smith et al. 2002"},{"why":"Empirical SMC wind-momentum-luminosity relation that, scaled to 0.10 $Z_\\odot$, filters unrealistic models from the grid.","marker":"Mokiem et al. 2007b"},{"why":"Defines the spectral standards and observed resolution/rotation properties used to degrade and classify the synthetic spectra.","marker":"Sota et al. 2011"},{"why":"Strategy for adapting spectral classification criteria to low metallicity, on which the XMP classification of the grid is based.","marker":"Lennon 1997"},{"why":"Provides the Sextans A spectroscopic catalogue and photometry that the extinction map is built from, plus the previous $Q_{\\mathrm{phot}}$ selection thresholds.","marker":"Lorenzo et al. 2022"},{"why":"Supplies the KPNO-MOSAIC photometric catalogue and filter system used for observed photometry and synthetic-colour calculations.","marker":"Massey et al. 2007"},{"why":"Describes CMFGEN, the code used to compute corrections for hydrogen line series and level dissolution in the synthetic photometry.","marker":"Hillier & Miller 1998"}],"fun_headline_variants":["HeII flux of metal-poor O stars can swing by 10,000x","New calibration: metal-poor OB stars are up to 6 kK hotter","First reference framework for ultra-metal-poor OB stars","Metal-poor OB colours map uneven dust in Sextans A"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calibration stands on the assignment of spectral types to synthetic spectra using classification criteria adapted to 0.10 $Z_\\odot$, but those adapted criteria are not presented in this paper (deferred to a companion work), and the paper notes that B-type classification is severely hampered by weak Si and Mg lines at this metallicity and low signal-to-noise.","fun_headline_variants_meta":{"raw":{"variants":["HeII flux of metal-poor O stars can swing by 10,000x","New calibration: metal-poor OB stars are up to 6 kK hotter","First reference framework for ultra-metal-poor OB stars","Metal-poor OB colours map uneven dust in Sextans A"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000403,"raw_usage":{"total_tokens":2163,"prompt_tokens":1073,"completion_tokens":1090,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":689,"completion_tokens_details":{"reasoning_tokens":1011}},"tokens_in":689,"tokens_out":1090,"duration_ms":10290,"temperature":1.0,"reasoning_tokens":1011,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:38:48.444698+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a sample of ~20 XMP O dwarfs with high-S/N spectra, classify them with the paper's criteria, and measure $T_{\\mathrm{eff}}$ independently from Balmer line profiles and the SED; if O9 V stars consistently come out near 30 kK instead of the calibrated ~34.5 kK, the $T_{\\mathrm{eff}}$ scale is wrong. Alternatively, resolve an XMP HII region whose ionizing late-O dwarf has a wind measured from UV P Cygni profiles: the model predicts $\\log q_{\\mathrm{HeII}}$ jumps by up to 4 dex across similar subtypes as wind strength changes, so a single-valued high $q_{\\mathrm{HeII}}$ for all O8-9 V stars would refute the bimodality claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Sextans A spectroscopic catalogue and photometry that the extinction map is built from, plus the previous $Q_{\\mathrm{phot}}$ selection thresholds."}],"review_version":1}