{"id":"7c02c9de-739e-4b11-a7a7-e276b1f88f5b","arxiv_id":"2502.03684","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Single-epoch quasar black hole masses are about 1.5 times too high on average when iron emission is ignored, and the paper builds a half-million quasar catalog with iron-corrected masses.","lead":"Using more than 55,000 DESI quasars, this paper applies an iron-emission correction to single-epoch black hole mass estimates and reports that standard masses are about 1.5 times too high on average, and up to 5 times too high for super-Eddington sources. If the correction holds, published quasar black hole masses, including some at z≥6, would shrink, which would ease the puzzle of how early black holes grew so massive.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim inherits the Du & Wang (2019) iron coefficient without independent RM verification; the MgII term is calibrated circularly against that same correction, so the factor-of-1.5 and factor-of-2.3 overestimation claims stand or fall with the -0.35 coefficient.","rationale":"The reader identified the same load-bearing assumption: the paper adopts Du & Wang (2019) without new RM data, and all conclusions inherit that coefficient. My stress-test adds specificity: the MgII calibration is not merely 'weakened by circularity' but is, to first order, a direct transfer of the adopted Hbeta iron coefficient to UV wavelengths, because RFe,Hbeta and RFe,MgII are correlated and the reference masses are already iron-corrected. The paper deserves credit for candidly disclosing the lack of new RM data and for providing internal systematics checks, but those checks cannot establish external validity. The catalog and calibration values are still useful as a consistent self-calibrated system, which supports keeping the conditional verdict rather than rejecting. The proposed test, refitting the same sample with a range of iron coefficients and with alternative published R-L relations, would quantitatively settle how load-bearing the -0.35 assumption is. No ad hominem is intended: the critique is about the logical structure of the calibration and the sensitivity of the conclusions to the adopted external relation.","tokens_in":27901,"tokens_out":1796,"duration_ms":15747,"concrete_test":"Re-derive the headline overestimation factors and the super-Eddington fraction on the same DESI Hbeta sample using two alternative Hbeta R-L relations: (i) the canonical Bentz et al. (2013) relation with no iron term, and (ii) the Wang & Woo (2024) relation, with the iron coefficient varied from 0 to -0.50 in steps of 0.05 while holding the other coefficients fixed. If the mean overestimation factor changes from 1.5 to below 1.2, or if the super-Eddington fraction drops below 2%, when the coefficient is moved within the plausible range of published RM-based values, then the central quantitative claims are not robust to the assumed iron correction.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The entire quantitative edifice, including the 1.5x mean overestimation, the 5% super-Eddington fraction, the -0.34RFe MgII term, and the z>=6 factor-of-2.3 implication, is built on Equation 5, log(RHbeta/ltd) = 1.65 + 0.45 log l44 - 0.35 RFe,Hbeta. No new reverberation-mapping measurement is presented (as the paper candidly states). Worse, the Hbeta-MgII calibration in Equation 9 is circular: the reference masses MHbeta,corr already contain the -0.35RFe suppression, and RFe,Hbeta correlates with RFe,MgII (r=0.52, slope 1.16, Fig. 11). Thus the fitted d=-0.34+/-0.02 for MgII is largely a propagated imprint of the adopted Hbeta iron correction rather than an independent empirical determination. The reported 'confirmation' of the iron-Eddington correlation in Figure 6 is also affected, since corrected mass enters the Eddington ratio by construction and mass depends on RFe; the paper notes this, but the claimed super-Eddington fraction remains model-dependent. If the true coefficient of RFe in the Hbeta R-L relation were, say, -0.20 instead of -0.35 (a change within plausible systematic disagreements among RM samples), the mean mass shift would shrink from 0.16 dex to roughly 0.09 dex, the super-Eddington fraction would drop, and the z>=6 overestimation factor would fall from 2.3 to about 1.5. The external checks offered in Section 4.1, including the d=-0.29 result after systematics, are internal consistency checks and cannot validate the external calibration.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses more than 55,000 DESI quasars at 0.25<z<0.8 with spectral fits to Hβ and MgII. Adopting the Du & Wang (2019) iron-corrected Hβ R–L relation (Eq. 5), it computes iron-corrected Hβ masses for 10,202 objects and finds that canonical single-epoch masses are overestimated by about 0.16 dex on average and by up to about 0.7 dex for the most iron-strong objects, with the super-Eddington fraction rising from 0.44% to 5.3%. It then calibrates an MgII-based mass formula containing an RFe,MgII term (Eq. 9, d=-0.34) against the corrected Hβ masses using a 954-object subsample, and applies this formula to roughly 490,000 DESI quasars at 0.6<z<1.6. The paper argues that if the same iron correction applies at z≥6, earlier high-redshift mass measurements would be overestimated by a factor of about 2.3, easing the tension in early supermassive black hole growth.","tokens_in":28327,"tokens_out":8669,"duration_ms":82887,"significance":"If the adopted iron-corrected R–L relation is correct, the paper would establish a systematic downward correction to single-epoch quasar black hole masses, change inferred Eddington-ratio distributions, and soften the high-redshift seed-growth tension. The paper's strengths are its large and homogeneous DESI sample, careful spectral fitting with explicit uncertainty treatments, systematics tests on fitting windows and templates, and a public catalog with machine-readable mass estimates. These resource aspects are valuable. The main caveat is that no new reverberation-mapping data are presented, so the external validity of the adopted iron coefficient is inherited from Du & Wang (2019), and the MgII calibration and the claimed iron–Eddington correlation are both entangled with the same adopted iron term.","major_comments":[{"comment":"The regression of log MMgII,corr on RFe,MgII uses log MHβ,corr as the target, and log MHβ,corr contains the term -0.35 RFe,Hβ by construction through Eq. (5). Since RFe,Hβ and RFe,MgII are correlated (Fig. 11 lower-left, r=0.52, slope=1.16), the fitted d=-0.34±0.02 is largely the Hβ iron coefficient propagated onto the MgII iron scale, rather than an independent empirical determination that the MgII R–L relation requires an iron term. The robustness check in §4.1 yielding d=-0.29±0.02 is an internal consistency check of the same propagation and cannot validate the MgII iron correction externally. The paper should frame Eq. (9) as a transfer calibration that assumes Eq. (5), and the Abstract's statement that 'The new relation adds an extra term of -0.34RFe' should be moderated accordingly.","section":"§3.2, Eq. (9)"},{"comment":"The claim that iron strength is 'confirmed' as a good tracer of the Eddington ratio is evaluated using log λEdd,corr, whose denominator is log MHβ,corr and therefore contains the -0.35RFe,Hβ term from Eq. (5). This mechanically injects a positive correlation between log λEdd,corr and RFe,Hβ; the parenthetical remark that the correlation is 'enhanced' by the formula understates the effect. The paper should provide a quantitative decomposition or repeat the correlation using uncorrected Eddington ratios, and the abstract/conclusion language should be softened unless the correlation survives that test.","section":"§3.1, Fig. 6"},{"comment":"The central quantitative claims—the mean overestimation of 0.16 dex, the super-Eddington fraction of about 5%, and the high-redshift overestimation factor of 2.3—are algebraically inherited from the adopted Du & Wang (2019) iron-corrected R–L relation and from the zero-point/slope differences between Eq. (5) and the canonical relation. The paper explicitly states that no new RM measurements are added, so the external validity of the -0.35 iron coefficient is entirely assumed. The §4.1 tests probe fitting, templates, and host-galaxy corrections, but not the external validity of Eq. (5). I recommend adding a sensitivity analysis that varies the iron coefficient and luminosity slope over the range spanned by published RM samples, and stating in the Abstract that the overestimation factors are conditional on Eq. (5).","section":"§3.1, Eq. (5); §4.3"}],"minor_comments":[{"comment":"The caption lists 'comparison between the uncorrected MMgII and corrected MMgII' for both the lower-left and lower-right panels; the lower-right panel as described in the text is a comparison between corrected MMgII and corrected MHβ, so the caption should be corrected.","section":"Fig. 12 caption"},{"comment":"The sentence 'The derived iron-corrected R–L relation for Mg ii is robust' is stronger than the presented evidence: the fitted iron coefficient changes from -0.34 to -0.29 when systematics are added, and the calibration remains tied to the adopted Hβ iron correction. Suggest softening to 'consistent with the adopted Hβ correction'.","section":"§4.1"},{"comment":"When introducing the canonical and iron-corrected R–L relations, the paper should clarify explicitly that the two relations differ in intercept and luminosity slope as well as in the iron term; this matters for interpreting the 0.16-dex mean shift as partly an iron effect and partly a zero-point/slope effect.","section":"§3.1"},{"comment":"The caption states that the figure 'confirms' FWHMFe ~ 0.75 FWHMHβ, but no best-fit line or slope is overlaid or quoted in the caption; adding the fit parameters would make the confirmation quantitative.","section":"Fig. 8 caption"}],"recommendation":"major_revision","confidential_remarks":"The main issue is that the MgII calibration and the iron–Eddington confirmation are not independent of the adopted Hβ iron correction; this is not a fatal flaw for a calibration/catalog paper, but the framing must change and a sensitivity analysis should be added. The paper is within scope, and the catalog and fitting transparency are genuinely useful. I would support publication after major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper is honest about what it is: an application of the Du & Wang (2019) iron-corrected Hβ R-L relation to about 55k DESI quasars, plus a new MgII iron-corrected mass formula fit against those Hβ masses, and roughly a 490k quasar catalog. The Hβ part is algebraically straightforward; the MgII part is new, and the catalog is the real deliverable. The spectral fitting is careful, the uncertainty discussion is thorough, and the authors repeatedly flag that they add no new reverberation mapping and that the high-redshift implication is conditional.\n\nThe soft spot is real and the stress-test note has it right. Equation 5 puts -0.35 RFe,Hβ directly into the Hβ masses, and Equation 9 regresses MgII against those masses. Since RFe,Hβ and RFe,MgII correlate (r=0.52), the fitted d=-0.34 is largely the Hβ iron coefficient transferred to MgII, not an independent empirical determination. The same issue colors the claim of confirming that iron traces Eddington ratio, since corrected mass enters that ratio and depends on RFe by construction. The paper notes this in passing, but it does not rescue the MgII term from being circular. The d=-0.29 robustness check is still internal, so it cannot validate the external calibration.\n\nDoes that sink the paper? Not entirely. The Hβ mass reduction is a legitimate consequence of adopting Du & Wang (2019); the large-sample confirmation of the correlation is useful even if partly built in; and the catalog gives the community a consistent set of iron-corrected masses with enough raw measurements for users to re-derive with their own estimator. What the paper cannot claim is independent establishment of the MgII iron term. That needs RM-based calibration or an external sample not used to construct the Hβ masses.\n\nMy take: this is a catalog and calibration paper that deserves a serious referee, but the authors should be pushed to reframe the MgII result as a consistent extension rather than an independent confirmation, and to state plainly that the -0.35 coefficient is inherited. The high-z implication should stay clearly conditional. I would cite the catalog if I needed homogenized low-z masses; I would not cite the MgII coefficient as an independent measurement.","headline":"A useful large-sample catalog and calibration, but the new MgII iron term is calibrated against masses that already contain the same correction, so the headline factors stand or fall with the adopted Du & Wang -0.35 coefficient.","tokens_in":29069,"tokens_out":2059,"would_cite":true,"duration_ms":19320,"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":"Single-epoch quasar black hole masses are overestimated by about 1.5 times when iron emission is ignored, and up to five times for super-Eddington sources.","keywords":["supermassive black holes","quasars","single-epoch mass estimators","iron emission","Eddington ratio","broad-line region","DESI","reverberation mapping"],"falsifier":"Measure reverberation-mapping lags for a sample of quasars covering the same range of $R_{\\mathrm{Fe}}$ and Eddington ratio as the DESI H$\\beta$ sample and check whether, at fixed $L_{5100}$, the broad-line region radius decreases with $R_{\\mathrm{Fe}}$ with a coefficient near $-0.35$; if the lags show little or no iron dependence, the overestimation factors and the $-0.34$ Mg II term would need to be revised.","tokens_in":27683,"feed_emoji":"🕳️","tokens_out":10204,"duration_ms":81540,"temperature":0.7,"pith_summary":"This paper argues that the standard single-epoch recipe for weighing supermassive black holes in quasars carries a systematic offset because it ignores the strength of iron emission, which tracks how fast the black hole is accreting. Using more than 55,000 DESI quasars at $0.25<z<0.8$, the authors apply an iron-corrected relation between broad-line region size and luminosity and find that previous canonical masses are too high by a factor of 1.5 on average, and by up to a factor of 5 for the most super-Eddington objects. The paper also calibrates an iron-corrected Mg II mass estimator and uses it to build a catalog of about 0.5 million quasars. If the same correction applies at high redshift, current mass estimates of $z\\ge6$ quasars would fall by about a factor of 2.3, easing the difficulty of forming billion-solar-mass black holes in the first billion years.","feed_headline":"Iron strength cuts quasar black-hole masses by 1.5x","feed_subtitle":"Adding the iron correction raises the super-Eddington fraction to 5% and would shrink z>6 quasar masses by 2.3x.","key_machinery":"The load-bearing object is the iron-corrected R-L relation: the empirical relation between broad-line region radius and optical/UV luminosity modified by a term proportional to iron strength. For H$\\beta$ the paper adopts $\\log(R_{\\mathrm{H}\\beta}/\\mathrm{ltd}) = 1.65 + 0.45\\log \\ell_{44} - 0.35 R_{\\mathrm{Fe,H}\\beta}$; for Mg II the paper fits $\\log(R_{\\mathrm{MgII}}/\\mathrm{ltd}) \\approx 0.46\\log \\ell_{44} - 0.34 R_{\\mathrm{Fe}}$, and both feed the virial mass formula $M = f R \\Delta V^2 / G$ with $f=1.1$. The iron term does the work of shifting masses down most for the highest-Eddington-ratio quasars, which are also the ones whose canonical masses were most inflated.","core_discovery":"The central claim is that the relative iron strength $R_{\\mathrm{Fe}}$ (the flux ratio of Fe II to the broad emission line) is a usable tracer of the Eddington ratio, and that folding it into single-epoch mass estimates removes a systematic overestimate. Adopting the iron-corrected H$\\beta$ relation $\\log(R_{\\mathrm{H}\\beta}/\\mathrm{ltd}) = 1.65 + 0.45\\log \\ell_{44} - 0.35 R_{\\mathrm{Fe,H}\\beta}$, the authors find uncorrected masses are high by 0.16 dex on average and up to 0.7 dex, raising the super-Eddington fraction from 0.44% to about 5%. For Mg II they calibrate $\\log(M/M_\\odot) = 1.14 + 2\\log(\\mathrm{FWHM}_{\\mathrm{MgII}}/\\mathrm{km\\,s^{-1}}) + 0.46\\log(L_{3000}/10^{44}\\,\\mathrm{erg\\,s^{-1}}) - 0.34 R_{\\mathrm{Fe,MgII}}$ (Eq. 9), producing an iron-corrected Mg II R-L relation with an extra $-0.34 R_{\\mathrm{Fe}}$ term. Applying this to about 490,000 DESI quasars at $0.6<z<1.6$ yields a catalog of corrected masses, and extrapolating to luminous $z\\ge6$ quasars lowers their masses by roughly a factor of 2.3.","pith_inferences":["One could test the Mg II iron term independently with reverberation-mapping lags for Mg II across a wide range of $R_{\\mathrm{Fe}}$; the predicted slope of the lag-luminosity-iron plane is $-0.34$ in $R_{\\mathrm{Fe}}$.","The same logic may extend to C IV-based masses at even higher redshift, where iron-blended UV spectra are common; a UV iron correction could shift the high-redshift quasar mass function beyond what this paper explicitly measures.","Because the correction lowers masses and raises Eddington ratios, it may alter inferences about quasar feedback and the black-hole-galaxy scaling relations; those consequences are not developed in the paper.","The observed FWHM$_{\\mathrm{Fe}} \\sim (3/4)$ FWHM$_{\\mathrm{H}\\beta}$ relation points to an intermediate-width component; fitting that component separately, rather than only adding the iron term, could further improve line-width mass estimators."],"forward_implications":["H$\\beta$-based single-epoch masses in large surveys should be revised downward by about 0.16 dex on average, with larger corrections for iron-strong, narrow-line quasars.","The super-Eddington fraction among low-redshift quasars is about 5%, an order of magnitude above the canonical 0.4%.","Mg II-based masses for $0.6<z<1.6$ quasars need the $-0.34 R_{\\mathrm{Fe}}$ term; without it, masses are overestimated by about a factor of 1.5 on average and 2.3 for super-Eddington objects.","If the correction holds at $z\\ge6$, luminous early-universe quasar masses shrink by about a factor of 2.3, reducing the required seed masses and accretion efficiency.","The DESI catalog of about 0.5 million quasars provides iron-corrected masses together with uncorrected and literature-estimator values for comparison."],"supporting_citations":[{"why":"Supplies the iron-corrected H$\\beta$ R-L relation (Eq. 5) that all corrected H$\\beta$ masses inherit.","marker":"Du & Wang (2019)"},{"why":"Provides the canonical H$\\beta$ R-L relation $\\log(R_{\\mathrm{H}\\beta}/\\mathrm{ltd})=1.53+0.51\\log \\ell_{44}$ used as the uncorrected baseline.","marker":"Du et al. (2018)"},{"why":"The standard reverberation-mapping R-L relation that the canonical relation is nearly identical to, defining the uncorrected scale.","marker":"Bentz et al. (2013)"},{"why":"Supplies the Mg II virial mass formula (Eq. 8) whose zero point is used and against which the corrected estimator is compared.","marker":"Vestergaard & Osmer (2009)"},{"why":"Provides the reverberation-mapping comparison showing super-Eddington AGNs lie below the canonical R-L relation, motivating the iron correction.","marker":"Wang & Woo (2024)"},{"why":"Adds moderate- to high-luminosity AGNs and a shallower R-L slope, used to check the canonical relation and the Eddington-ratio dependence.","marker":"Woo et al. (2024)"},{"why":"Comparison Mg II estimator and the origin of the line-profile offset discussion between FWHM Mg II and H$\\beta$.","marker":"Le et al. (2020)"},{"why":"Recent reverberation-mapping-calibrated Mg II estimator with $c=0.39$, included in the catalog comparison.","marker":"Yu et al. (2023)"}],"fun_headline_variants":["Iron-corrected masses drop 1.5x for DESI quasars","Iron strength fixes quasar mass overestimates","Iron correction cuts quasar black hole masses by 1.5x","New iron-corrected masses shrink z>6 quasar masses by 2.3x","Iron-corrected masses for 500k quasars challenge early universe growth"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the iron-corrected H$\\beta$ R-L relation adopted from earlier reverberation-mapping work, with its $-0.35$ iron coefficient, is the right calibration for the DESI quasars; the paper adds no new reverberation-mapping data, so any error in that coefficient propagates directly into the reported overestimates and the Mg II iron term.","fun_headline_variants_meta":{"raw":{"variants":["Iron-corrected masses drop 1.5x for DESI quasars","Iron strength fixes quasar mass overestimates","Iron correction cuts quasar black hole masses by 1.5x","New iron-corrected masses shrink z>6 quasar masses by 2.3x","Iron-corrected masses for 500k quasars challenge early universe growth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001324,"raw_usage":{"total_tokens":5547,"prompt_tokens":1264,"completion_tokens":4283,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":880,"completion_tokens_details":{"reasoning_tokens":4184}},"tokens_in":880,"tokens_out":4283,"duration_ms":28053,"temperature":1.0,"reasoning_tokens":4184,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T04:06:40.842932+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure reverberation-mapping lags for a sample of quasars covering the same range of $R_{\\mathrm{Fe}}$ and Eddington ratio as the DESI H$\\beta$ sample and check whether, at fixed $L_{5100}$, the broad-line region radius decreases with $R_{\\mathrm{Fe}}$ with a coefficient near $-0.35$; if the lags show little or no iron dependence, the overestimation factors and the $-0.34$ Mg II term would need to be revised.","supporting_citations":[],"review_version":1}