{"id":"a63fa2df-bcd0-4f46-9963-2dafab093b19","arxiv_id":"2506.12121","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The observed spins of black holes in X-ray binaries are almost all high and can be described by a beta distribution with alpha=5.66 and beta=1.09, with spin precision improving for more massive, closer, and more inclined systems.","lead":"This paper reanalyzes 245 NuSTAR spectra of 36 black hole X-ray binaries to map how measured spins relate to binary properties and to expose correlations between spectral model parameters. It reports that almost all of these black holes spin rapidly, fits their spin distribution with a beta function, and releases the full fitting tables for community use.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported beta distribution for XB spins is not yet demonstrated: Section 3.1 does not specify whether the fit uses 36 independent source posteriors or 245 non-independent spectra, nor how measurement uncertainties are propagated, so the claimed incompatibility with GWTC-3 may be a statistical…","rationale":"The paper deserves credit for the uniform reanalysis pipeline, the public data release, and the explicit robustness tests in Sections 4.2 and 4.6. The central quantitative claim, however, is the beta-distribution description of XB spins and its incompatibility with GW spins. That claim rests entirely on the population inference in Section 3.1, which is described in one paragraph with no likelihood, prior, or statement of whether the 36 source-level measurements or the 245 individual spectra enter the fit. Because the 245 spectra are highly non-independent, and because the Table 1 spin measurements have asymmetric credible intervals, the reported alpha=5.66, beta=1.09 and the apparent separation from GWTC-3 cannot be evaluated from the text alone. The reader's identified weakest assumption, disk truncation, would bias measured spins low if incorrect, so it does not threaten the high-spin conclusion in the same way; an unsound population fit could invalidate the headline result even if every individual spin is correct. The proposed hierarchical reanalysis using per-source posterior samples and a one-observation-per-source cross-check would settle the question without new observations.","tokens_in":38389,"tokens_out":9426,"duration_ms":123544,"concrete_test":"Download the Zenodo table (doi:10.5281/zenodo.15801174) and rerun the Section 3.1 population inference as a hierarchical model: for each of the 36 sources, use the full posterior sample (or at minimum the reported asymmetric credible interval) as the likelihood for that source, with a beta-distributed population prior on P(a); as a robustness check, repeat using only one randomly selected NuSTAR observation per source. Report the posterior on (alpha,beta), a posterior-predictive p-value for the beta form, and the overlap or significance of the XB-vs-GWTC-3 comparison. If the alpha-beta credible interval is broad, the beta form is rejected, or the one-observation-per-source fit shifts the mode by more than 0.1, then Eq. (1) and the incompatibility claim require revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.1 states that a Bayesian inference was performed on 'the measurements in this work' and reports Eq. (1) with alpha=5.66, beta=1.09, but the text does not describe the likelihood, the prior, the data units, or whether the input is the 36 source-level measurements from Table 1 or the 245 individual spectral fits. This distinction is load-bearing because the 245 spectra contain many repeated observations of the same 36 systems, so treating them as independent would inflate the effective sample size and make the XB distribution appear artificially narrow and the separation from GWTC-3 artificially significant. The source-level values in Table 1 themselves carry asymmetric credible intervals (e.g., AT 2019wey a=0.91+0.08-0.20; MAXI J1848-015 a=0.8+0.2-0.7); if the population fit uses only point estimates, the posterior on alpha and beta is overconfident and the mode near 0.98 is not justified. The central claim that the XB spin distribution is beta(5.66,1.09) and incompatible with GW spin distributions fails if the population inference is not specified and reproduced with proper uncertainties and independence.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"Using the 245 NuSTAR spectra of 36 BH X-ray binaries previously analyzed in Draghis et al. (2024), this paper presents a population-level study of measured spins and spectral-fit parameters. It reports Spearman correlations between spin uncertainty and system properties (mass, distance, inclination), identifies parameter degeneracies in relativistic reflection fits (e.g., q1--a, R--Gamma, log xi--Gamma), and fits the observed XB spin distribution with a beta distribution alpha=5.66, beta=1.09, concluding that the XB distribution is incompatible with the low-spin distribution inferred from GWTC-3 and that most XB BHs must have formed with high natal spins. The full fitting table is released on Zenodo.","tokens_in":38648,"tokens_out":7767,"duration_ms":205614,"significance":"The paper's main value is as a resource: a uniformly reduced, publicly released table of 245 NuSTAR spectral fits and spin measurements, plus a set of cautionary examples about low-SNR degeneracies (low-q1/low-a vs high-q1/high-a) and the influence of absorption lines and disk density. The comparison of measured spins with the Fragos & McClintock (2015) accretion-spinup ceiling is an interesting and falsifiable argument for high natal spins. However, the two headline population claims--the beta-distribution characterization and the XB/GW incompatibility--are not yet established because the statistical inference is under-specified and the input sample is not independent. The paper is honest about many limitations and does not overstate the exploratory correlation analysis, but the central 'incompatible with GWTC-3' statement needs a properly specified and reproducible hierarchical fit before it can carry the weight placed on it.","major_comments":[{"comment":"The Bayesian inference that produces alpha=5.66 and beta=1.09 is not described. The text does not state the likelihood, the prior on (alpha, beta), the input data (36 source-level values from Table 1, 36 full posteriors, or the 245 per-spectrum fits), how asymmetric 1-sigma intervals are propagated, or how the posterior draws shown as thin blue lines in Figure 4 are generated. The beta distribution is only defined on [0,1], whereas individual fits in Figure 6 extend to negative a, so the treatment of negative/retrograde values must be stated. Without these details, and without credible intervals or a goodness-of-fit statistic for alpha and beta, the claim that the XB distribution 'is well approximated' by beta(5.66,1.09) cannot be evaluated or reproduced; this is the load-bearing step for the XB/GW comparison.","section":"Section 3.1, Eq. (1)"},{"comment":"The correlation analysis treats the 245 spectra as independent samples even though they are repeated observations of 36 sources. The quoted Spearman coefficients and their +/- uncertainties therefore ignore clustering; for example, Figure 8 panels (e) and (f) report rho=0.32+/-0.01 and 0.57+/-0.01 on 245 points, while Figure 9 shows clear source-level structure in the same parameter combinations. A source-resampling or mixed-effects analysis is needed to establish whether the trends are within-source or between-source and whether the reported significance survives. The same issue applies to the spin-uncertainty correlations in Figure 3, although those use one value per source.","section":"Section 4, Figures 6 and 8"},{"comment":"The paper states that the two distributions are 'clearly distinct', but the XB distribution is explicitly the observed distribution with no selection-function correction, while the GWTC-3 distribution is selection-corrected. This asymmetry is acknowledged in the text but not accounted for in the conclusion. Because the XB sample is selected by outburst activity, Eddington fraction, detection of reflection, and successful spin constraint, the observed high-spin excess could be partly a selection artifact. The authors should either model the XB selection function or restrict the claim to the observed sample with the selection caveat carried through the abstract and conclusions.","section":"Section 3.1, Figure 4"},{"comment":"The assumption that the inner disk radius equals the ISCO in every fitted observation is the physical link between the reflection fits and the spin a. The defense in Section 4.1, based on hardness-intensity diagrams for nine sources (Figure 10), is indirect and relies on the absence of an obvious hardness trend in the spin constraints rather than a direct test of Rin/ISCO. If some hard-state disks are truncated, the inferred spins are biased low, and the beta distribution in Eq. (1) describes biased measurements rather than true spins. A direct test, such as letting Rin vary in a subset of spectra and quoting the change in fit statistic or the posterior on Rin/ISCO, would make the population claim much stronger.","section":"Sections 2 and 4.1"}],"minor_comments":[{"comment":"The definition of the gamma function contains a typo: 'r^{-t}' should be 'e^{-t}'.","section":"Eq. (1)"},{"comment":"Section 2 states that 36 of the 245 spectra required the zxipcf component for complex obscuration, while Section 4.6 states that 96 spectra required an absorption Gaussian line and 149 did not; these numbers need to be reconciled or explicitly described as different diagnostics.","section":"Sections 2 and 4.6"},{"comment":"The phrase 'modes and +/-1 sigma of the mean distributions' is confusing; please specify whether the vertical lines are the mode and central credible interval of the population distribution itself.","section":"Section 3.1, Figure 4"},{"comment":"Several panels use the placeholder symbol 'square' as the y-axis label (e.g., Figure 8 panels e, f, i, o); these should be replaced with the actual parameter name (Gamma or another label).","section":"Figures 6 and 8"},{"comment":"The footnote marker '3 represents an inclination estimate based on dips' should be typeset as a superscript to match the table entries and to avoid confusion with the numeric value 3.","section":"Table 1 footnote"},{"comment":"The caption for the theoretical curves from Fragos & McClintock (2015) does not state the assumed accretion efficiency or spin-up prescription; adding one sentence would help readers interpret the comparison.","section":"Figure 2, panel (g)"}],"recommendation":"major_revision","confidential_remarks":"The paper reuses the authors' own published dataset (Draghis et al. 2024), so the novelty is primarily in the population-level analysis and the release of the fit table. The beta-distribution claim is the main new scientific result and is currently under-specified; I recommend requiring the full inference specification and a source-resampling treatment before acceptance. The paper's self-citations are appropriate in context, and the Zenodo data release is a genuine asset to the community."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this before citing it as evidence for high natal BH spins: the beta distribution that anchors the population claim is fit in a way that is not fully specified, and the comparison with GW spins is observed-vs-intrinsic, not apples-to-apples. That said, the paper ships a genuinely useful uniform dataset and is unusually honest about its limitations.\n\nWhat is actually new: the beta parametrization of the XB spin distribution, the Spearman correlations between spin uncertainty and mass/distance/inclination, and the source-independent correlation matrix of spectral-fit parameters. The Zenodo release of all 245 fits is a real contribution. The density experiments with reflionx HD are careful and suggest spin is robust to disk-density assumptions.\n\nThe main soft spot is Section 3.1. The Bayesian inference for P(a) is described in one sentence. It does not state whether the inputs are the 36 source-level measurements from Table 1 or the 245 individual spectra, does not give the likelihood or prior, and does not say how the asymmetric credible intervals are propagated. If it uses point estimates, the beta parameters are overconfident; if it pools the 245 spectra, repeated observations of the same 36 systems inflate the effective sample size. Either way, alpha=5.66, beta=1.09 and the mode near 0.98 are not reproducible as written. The paper does note that the blue curve is the observed distribution and the GW curve corrects for selection effects, so the headline incompatibility is not a clean comparison. The correlations in Figure 3 are computed on at most 36 sources, with literature values of mixed quality; they are labeled moderate/mild, but the rho uncertainties are not derived clearly.\n\nNone of this is fatal. The paper is careful where it knows to be careful, and it explicitly acknowledges that the sample may be biased and that low spins come preferentially from low-SNR spectra. The q1-a degeneracy discussion and the absorption-line experiment are informative. The stress-test concern lands: the population fit needs to be specified properly, ideally with a hierarchical model that uses the full posteriors.\n\nWho is this for? Anyone working on BH spin measurements or population synthesis will want the dataset and the parameter-correlation results. The population claim should not be cited as established until Section 3.1 is pinned down. I would send it to a referee: the data release alone justifies that, and the population claim could matter if properly demonstrated.","headline":"A valuable data release and transparent exploratory analysis, but the headline beta-distribution claim is under-specified and the XB/GW incompatibility is not yet demonstrated.","tokens_in":39204,"tokens_out":2095,"would_cite":true,"duration_ms":24895,"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":"X-ray binary black holes spin near maximum, a uniform NuSTAR sample shows.","keywords":["black hole spin","X-ray binaries","relativistic reflection","NuSTAR","beta distribution","gravitational waves","natal spin","accretion disk"],"falsifier":"Take one high-quality NuSTAR spectrum from the selected Eddington range, leave the inner disk radius free instead of fixing it to the ISCO, and check whether the best fit prefers a radius significantly outside the ISCO; if it does, that spin is biased low and the beta(5.66, 1.09) distribution is not a clean natal-spin measurement. A second check would be to re-fit the faint spectra that currently yield low spin with absorption-line components and see whether the low-spin solutions disappear.","tokens_in":38193,"feed_emoji":"🕳️","tokens_out":5402,"duration_ms":64767,"temperature":0.7,"pith_summary":"This paper argues that the black holes in X-ray binaries almost all spin very fast, and that their measured spins follow a beta distribution with shape parameters alpha = 5.66 and beta = 1.09. The distribution is built from 245 uniform relativistic-reflection fits to NuSTAR spectra of 36 systems, so it is meant to be a systematic, source-independent characterization rather than a collection of heterogeneous measurements. The central stakes are formation physics: at the short orbital periods many of these systems have, accretion alone cannot have spun the holes up to the measured values, so the rotation must have been largely present at birth. That makes the observed spin distribution a direct constraint on stellar collapse and binary evolution, and it stands in sharp contrast to the low-spin distribution inferred from gravitational-wave mergers.","feed_headline":"Most black holes in X-ray binaries spin near maximum","feed_subtitle":"A uniform analysis of 245 NuSTAR spectra finds a spin distribution incompatible with gravitational-wave mergers.","key_machinery":"The load-bearing object is the uniform sample itself: 245 NuSTAR spectra of 36 accreting black hole X-ray binaries, each fit with six flavors of the relxill relativistic-reflection model, selected by deviance information criterion and processed through MCMC posteriors. The named identity carrying the population claim is the $\\beta$ distribution P(a) = Gamma($\\alpha$+$\\beta$)/(Gamma($\\alpha$)Gamma($\\beta$)) $a^{{alpha-1}}$(1-a)^{$\\beta$-1} with $\\alpha$ = 5.66 and $\\beta$ = 1.09, which summarizes the observed spin sample. The comparison that makes the claim consequential is the same Bayesian-inference procedure applied to the GWTC-3 black-hole merger spins, yielding two distributions that barely overlap.","core_discovery":"The authors claim that the observed spin distribution of the 36 X-ray binary black holes is a $\\beta$ distribution, P(a) proportional to $a^{{alpha-1}}$(1-a)^{$\\beta$-1} with $\\alpha$ = 5.66 and $\\beta$ = 1.09, peaking near a approximately 0.97 and incompatible with the spin distribution inferred from GWTC-3, whose mode is around 0.18. The incompatibility is presented as the first such comparison built from X-ray measurements made with one uniform pipeline, and therefore not an artifact of mixing different model assumptions. The paper further claims that high spins in systems with short orbital periods exceed the maximum spin that accretion can deliver, so these black holes must have formed rotating near their maximum rate; that spin-measurement precision increases with black hole mass and decreases with distance and with lower reflection counts; and that low or negative individual spin fits are confined to faint spectra with low reflection strength, where the model cannot distinguish high-spin/high-emissivity from low-spin/low-emissivity solutions.","pith_inferences":["If the same pipeline were applied to a sample selected without the Eddington-fraction cut, or to AGN spins measured by reflection, the inferred beta parameters might shift, which would test whether the high-spin peak is a property of X-ray binaries or of the reflection method itself.","The tentative spin-emissivity degeneracy suggests that some low-spin measurements are algorithmic artifacts of faint spectra; high-resolution microcalorimeter spectra could distinguish real low spins from unmodeled absorption features.","A physical consequence the authors leave implicit is that if most X-ray binary black holes are born spinning fast while most merging black holes appear to spin slowly, the two populations may trace different mass or metallicity channels, not just different measurement techniques.","The density experiments imply that fitting with densities above log n = 20 lowers Fe abundance and inclination but not spin; extending this to a full sample could turn the bimodal Fe abundance into a diagnostic of unmodeled disk density."],"forward_implications":["If the beta distribution is correct, most stellar-mass black holes in X-ray binaries are born with near-maximal rotation, and supernova and binary-evolution models must reproduce that angular momentum before any accretion occurs.","Accretion spin-up cannot explain the fastest rotators at short orbital periods, so mass transfer is not the origin of the high spins; the observed values are close to natal values.","The X-ray and gravitational-wave spin distributions are genuinely different, or at least are not reconciled by uniform X-ray systematics, motivating searches for selection effects and formation-channel differences.","Spin measurements will be most precise for massive, nearby, bright systems with high reflection counts, while the low-spin tail of the distribution should be treated cautiously because it comes from faint spectra.","The published 245-fit dataset becomes a community resource for testing model degeneracies and future parameter correlations."],"supporting_citations":[{"why":"Supplies the 245 uniform NuSTAR fits of 36 systems that this paper reanalyzes and builds its population statements on.","marker":"Draghis et al. (2024)"},{"why":"Provides the GWTC-3 merger spin distribution and the Bayesian inference approach used for the comparison in Section 3.1.","marker":"Abbott et al. (2023)"},{"why":"Gives the theoretical maximum spin from accretion as a function of orbital period, used to argue that high spins at short periods require natal rotation.","marker":"Fragos & McClintock (2015)"},{"why":"Earlier comparison of X-ray and gravitational-wave spins that this work expands with a uniform X-ray sample.","marker":"Fishbach & Kalogera (2022)"},{"why":"Establishes the fitting pipeline and reflection-strength definition underlying the 36 spin measurements.","marker":"Draghis et al. (2023b)"},{"why":"Reports a high measured spin for a newborn black hole, cited as evidence that high spin is natal.","marker":"Draghis et al. (2023c)"},{"why":"Introduces the density-enhanced reflection model used in the density experiments of Section 4.2.","marker":"Tomsick et al. (2018)"}],"fun_headline_variants":["X-ray binaries host black holes spinning near maximum","Uniform NuSTAR survey finds X-ray binary black holes spin fast","Black hole spin distribution in X-ray binaries peaks near maximum","NuSTAR: X-ray binary black holes spin near maximum","Most black holes in X-ray binaries spin near maximum"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that in every selected observation the accretion disk reaches the innermost stable circular orbit, so the fitted inner radius can be read directly as spin; if some disks are truncated, those spins are biased low and the beta distribution would not describe true natal spins.","fun_headline_variants_meta":{"raw":{"variants":["X-ray binaries host black holes spinning near maximum","Uniform NuSTAR survey finds X-ray binary black holes spin fast","Black hole spin distribution in X-ray binaries peaks near maximum","NuSTAR: X-ray binary black holes spin near maximum","Most black holes in X-ray binaries spin near maximum"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000576,"raw_usage":{"total_tokens":2752,"prompt_tokens":1014,"completion_tokens":1738,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":1657}},"tokens_in":630,"tokens_out":1738,"duration_ms":14033,"temperature":1.0,"reasoning_tokens":1657,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:57:52.298498+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take one high-quality NuSTAR spectrum from the selected Eddington range, leave the inner disk radius free instead of fixing it to the ISCO, and check whether the best fit prefers a radius significantly outside the ISCO; if it does, that spin is biased low and the beta(5.66, 1.09) distribution is not a clean natal-spin measurement. A second check would be to re-fit the faint spectra that currently yield low spin with absorption-line components and see whether the low-spin solutions disappear.","supporting_citations":[],"review_version":1}