{"id":"8038b21c-9e7c-491d-8903-8c1ad8aa174c","arxiv_id":"2607.20606","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Excess Na I D absorption, interpreted as neutral-gas outflows, is detected in 20 of 811 galaxies at z>3 and rises steeply with stellar mass and quiescence.","lead":"By mining 811 low-resolution JWST spectra of distant galaxies, this study finds excess sodium absorption — a marker of cold, neutral gas — in 20 massive galaxies in the universe's first two billion years. It offers the first statistical census of early neutral-gas outflows, with detection rates rising toward the most massive and star-formation-shut-down galaxies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Residual Na I D excess may be a stellar-template artifact: the paper asserts, but does not test, that non-solar [Na/Fe] cannot produce the 4–16 Å EWs, so the gas/outflow census is not yet secure.","rationale":"The reader's weakest assumption already lists template mismatch among possible contaminants, but alongside He I and polynomial effects. I separate those: He I emission should appear in emission in the residual and lower the measured EW, and the polynomial is not localized enough to create a 4–16 Å feature, so the decisive branch is whether the stellar continuum model itself absorbs the signal. The paper's central quantitative evidence is the 0–2 Å stellar EW vs 4–16 Å gas EW; this is only as strong as the templates. Since the paper concedes it masks Na I D before fitting and uses fixed solar abundance ratios, the residual is not an independent measurement of gas. A [Na/Fe]-variable refit is the natural, low-cost check. I do not see a reason to reject: the stack of non-golden massive dusty galaxies shows a residual ~1.7 Å, consistent with a real weaker population, and the highest EW detections are unlikely to be purely stellar. The concern therefore keeps the paper at CONDITIONAL, not ACCEPT/REJECT.","tokens_in":20487,"tokens_out":9882,"duration_ms":91916,"concrete_test":"Re-run the full Prospector analysis for the 811 galaxies with a stellar library that includes non-solar [Na/Fe] (e.g., C3K/ATLAS9 or E-MILES with enhanced sodium), allowing [Na/Fe] as a free parameter, and re-measure residual Na I D EWs and S/N with the same mask and jitter procedure. A decisive check: if the number of golden-sample galaxies with S/N>3 drops by more than ~5, or the median residual EW falls below 3 Å, the gas/outflow interpretation is not robust; if the 20 detections survive with similar EWs, the template-mismatch concern is settled. For additional ground truth, obtain medium-resolution (R~1000) NIRSpec grating spectra for 5–10 of the lowest-EW golden galaxies and check whether resolved Na I D absorption is present after a full spectral fit.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Every derived quantity (detection fraction, quiescence trend, outflow rate) depends on the residual flux deficit near 5893 Å after subtracting the Prospector/FSPS stellar model. At R~100, a 4 Å rest-frame EW is only a few-percent deficit over the ~59 Å resolution element, and the quoted stellar EW of 0–2 Å comes from the same solar-abundance MILES/FSPS models used to define the continuum. MILES does not vary [Na/Fe], and Na D is sensitive to surface gravity, metallicity, and abundance ratios; old, metal-rich stellar populations can have substantially stronger Na D than the fitted templates. The paper's §4.2 statement that the observed EW 'cannot be explained' by stellar absorption even under unknown abundance patterns is not supported by any calculation or alternate-template fit. Because quiescent galaxies are the most likely to contain old, metal-rich stellar populations with enhanced Na D, a template underestimate of even 2–4 Å would selectively move quenched systems into the golden sample, inflating the 43% incidence. He I 5877 contamination (§3.3) would add flux and bias EWs low, and the 10th-order polynomial is smooth over the masked line region; neither explains a false 4–16 Å absorption, so the unresolved premise is the stellar template itself.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a systematic search for Na I D absorption in 811 galaxies at 3<z<7 using JWST/NIRSpec prism spectroscopy drawn from the public DAWN JWST Archive and the Mirage or Miracle survey. The authors fit each spectrum together with multi-band photometry using Prospector/FSPS, mask and then subtract the best-fit stellar continuum, and measure residual rest-frame equivalent widths. After jitter-rescaling the errors and visual inspection, 20 galaxies form a golden sample with S/N>3 and EW 4–16 Å. The paper derives detection fractions (~2.5% overall, rising with stellar mass and quiescence, reaching ~11% among the most massive galaxies and ~43% among the most quenched massive bin), performs a stacking analysis, and interprets the detections as neutral outflows, estimating mass outflow rates of 4–12 Msun/yr. The central conclusion is that neutral outflows are likely already present and important in massive galaxies during the first 2 Gyr of cosmic history.","tokens_in":1628,"tokens_out":1853,"duration_ms":123840,"significance":"The paper addresses an important open question—whether cold neutral outflows exist and are dynamically important at z>3—using a much larger sample than previous medium-resolution studies. If the residual Na I D is genuinely non-stellar gas absorption, the result would be a valuable statistical census, and the mass/SFR/sSFR trends plus the stacking analysis provide a foundation for follow-up. Strengths include the use of public archival data, a transparent continuum-subtraction and error-rescaling procedure, visual inspection of candidates, explicit caveats about He I contamination, and an attempt to connect the prism measurements to medium-resolution outflow samples. The main limitation is that the reality of the residual absorption as a gas-phase signature is not independently established, and the paper's own caveats in Sec. 3.3 do not cover all contamination paths.","major_comments":[{"comment":"The load-bearing claim that the 4–16 Å residual near 5893 Å is gas rather than stellar absorption is not backed by a test. The Prospector+FSPS/MILES setup fixes relative abundances to the solar pattern and MILES does not sample varying [Na/Fe]; stellar Na I D is very sensitive to [Na/Fe], metallicity, age, and surface gravity. The reported stellar EW of 0–2 Å is measured from the same template family used to remove the continuum, so it is a model assertion, not a bound on template error. Because the quiescent, massive systems that dominate the 43% incidence are the ones most likely to have enhanced [Na/Fe], even a 2–4 Å residual stellar contribution would move them into the golden sample. I ask for a concrete test: fit the residuals with variable-[Na/Fe] stellar libraries or leave [Na/Fe] free in Prospector; or calibrate the maximum stellar Na I D equivalent width from local quiescent ga","section":"§3.1, §4.2, Fig. 6"},{"comment":"The paper quotes two different quenched detection fractions without reconciling them. The abstract and conclusions give ~43% for massive quenched systems; Section 5.2 gives f_QU=13.2% (7 out of 53 massive galaxies below the SFMS). If these use different definitions (e.g., SFMS-offset vs. absolute sSFR<1e-11), this needs to be stated explicitly with the parent-sample sizes in each bin, and the stronger number should not appear as a headline without that definition. As written, the key quantitative result is internally inconsistent and cannot be evaluated by the reader.","section":"Abstract; §5.2; §6.1–6.2"},{"comment":"The EW>5 Å outflow criterion is based on a small medium-resolution compilation: 11 of 14 galaxies above the threshold are blueshifted. The transfer of this threshold to prism data is not demonstrated: the prism EWs are measured on heavily blended, lower-limit residuals, while the literature EWs come from kinematically decomposed medium-resolution profiles. At minimum, propagate the Fig. 10 distribution through the prism resolution and He I contamination model to estimate P(outflow | EW_prism>5 Å); otherwise the conclusion that the golden-sample detections are mostly tracing neutral outflows goes beyond what the evidence supports. The authors' caveat about the lack of kinematics is appropriate, and the conclusion could be softened consistently.","section":"§5.1, Fig. 10"}],"minor_comments":[{"comment":"The He I line wavelength is given as 5877 Å in Sec. 3.3 and 5876 Å elsewhere; please use a single value consistently.","section":"§3.3/§4.3"},{"comment":"The column header 'S/N50A V' is a formatting artifact; it should be two separate columns. Also, state explicitly that the Na I D EWs are rest-frame values measured after stellar subtraction.","section":"Table 1"},{"comment":"Specify the bin boundaries and the number of parent and golden-sample galaxies in each sSFR bin, especially for the sSFR<1e-11 bin, so the 43% fraction can be evaluated.","section":"Fig. 5"},{"comment":"The golden sample includes one broad-line AGN (rubies-egs52-v4_prism-clear_4233_9809) at z=5.70. Show the key statistics and stacks with and without this source to demonstrate that the conclusions are not driven by it.","section":"Table 1, Fig. 2"},{"comment":"Eq. (8) is a useful summary, but its normalization should be derived explicitly from Eqs. (4)–(7) so that the reader can check the conversion from EW to Mdot.","section":"§5.3, Eq. (8)"},{"comment":"The mean jitter factor of 1.54 is reported; please also give the range or distribution, since the quoted EW uncertainties in Table 1 may be underestimated if the jitter varies across the sample.","section":"§3.4"}],"recommendation":"major_revision","confidential_remarks":"The archival sample and the overall approach are in scope and potentially important. The main risk is the stellar-template issue: if the authors can supply a non-solar-abundance test showing that stellar Na I D cannot reproduce the golden-sample EWs, I would be willing to reconsider. The 43% vs 13.2% inconsistency in the quenched fraction must be fixed, and the EW-threshold argument should be made more quantitative. No concerns about citation or novelty disclosure beyond what is in the report."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuinely useful first census of Na I D at z>3 from prism spectra, and the authors are honest about their limits. But the central interpretation—that these are outflows—rests on an untested assumption about stellar templates, and the paper's claim that stellar absorption can be ruled out is asserted, not demonstrated. That is the load-bearing point, and it needs to be fixed before the headline numbers are trusted.\n\nWhat is new: 811 galaxies at 3<z<7, 20 robust detections, detection fraction rising with mass and quiescence (2.5%, ~11%, ~43%). The stacking analysis is the strongest part: selecting purely on mass + dust + S/N, they recover absorption even when excluding the golden sample. Error treatment is careful (jitter rescaling, visual inspection, AGN diagnostics). The EW measurements are direct, and the paper is forthright that He I contamination makes them lower limits.\n\nWhere it gets soft: Sec 4.2 concludes the observed EWs cannot be explained by stellar models \"even when allowing for systematic uncertainties in the stellar models due to, e.g., unknown abundance patterns\"—but no such test is presented. The MILES/FSPS setup fixes solar abundance ratios; Na D is sensitive to [Na/Fe] and surface gravity. Old, metal-rich stellar populations can have substantially stronger Na D than the 0–2 Å the templates predict. If the template is off by a few Å, the low-EW detections (4–6 Å) become suspect, and because quenched galaxies are the most likely to host old, Na-enhanced populations, the striking 43% fraction is exactly where a selection bias would bite. This is not a nitpick; it is the difference between a gas census and a template artifact.\n\nOther soft spots are smaller. The 5 Å outflow threshold is transferred from 14 medium-resolution galaxies with no kinematic check at R~100; the paper concedes this. The outflow rates are linear rescalings with hand-adopted constants (R_out=1 kpc, v_out=200 km/s) and factors of 3–10 uncertainty. Detection fractions lack confidence intervals given a heterogeneous archival parent sample.\n\nBottom line: the data and pipeline are a real contribution, and the stacking makes a plausible case that some excess Na I D is there. But the outflow interpretation, and especially the quiescent fraction, need a quantitative test of the stellar template. I'd send this to a serious referee, but require that test before acceptance.","headline":"Useful census, but the outflow claim rests on an untested stellar-template assumption; a quantitative test of non-solar [Na/Fe] is needed before the headline fractions are trusted.","tokens_in":21452,"tokens_out":3476,"would_cite":true,"duration_ms":53179,"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":"The paper establishes that neutral-gas outflows, traced by Na I D absorption, were already present and dynamically important in massive galaxies during the first two billion years of cosmic history, based on a survey of 811 galaxies at z>3","keywords":["galaxy outflows","neutral gas absorption","Na I D doublet","high-redshift galaxies","JWST NIRSpec prism","galaxy quenching","dusty star-forming galaxies","early universe"],"falsifier":"Take the 20 detected galaxies into medium-resolution NIRSpec spectroscopy and resolve the Na I D doublet from the nearby He I line. If the feature splits into He I emission or shows a non-blueshifted, narrow absorption component, or if it disappears when stellar templates with non-solar abundance ratios are used, the outflow census as interpreted here would not stand; conversely, confirming broad blueshifted Na I D absorption in most of the 20 systems would settle the interpretation.","tokens_in":20173,"feed_emoji":"💨","tokens_out":4302,"duration_ms":38089,"temperature":0.7,"pith_summary":"The paper tries to establish that neutral-gas outflows, traced by sodium absorption, were already present and dynamically important in massive galaxies during the first two billion years of cosmic history. Working from 811 galaxies at 3<z<7 with JWST prism spectra, the authors detect strong Na I D absorption in 20 systems, almost all massive and at z<5. The detection fraction climbs steeply with stellar mass and quiescence, reaching about 43% among massive quenched galaxies, and the measured equivalent widths (4–16 Å) exceed anything stellar atmospheres alone can produce. Interpreting these absorptions as neutral outflows yields mass outflow rates of about 4–12 solar masses per year, exceeding the star formation rate for roughly one third of the detections. A sympathetic reader would take this as the first statistical census of cold neutral outflows at z>3.","feed_headline":"Neutral outflows were already common in massive early galaxies","feed_subtitle":"A survey of 811 early galaxies finds Na I D absorbers in 20, with rates up to 43% among massive quenched systems","key_machinery":"The central object is the Na I D doublet at 5890 and 5896 Å, a resonance transition that forms only in cold, shielded, neutral gas. Because the doublet is unresolvable at R~100 prism resolution, the paper measures the equivalent width of the blended feature in continuum-subtracted residuals, and treats the EW magnitude as a proxy for strong outflows based on a comparison with medium-resolution spectra. The supporting machinery is Bayesian stellar-population fitting, which produces a stellar continuum template that is subtracted to isolate the gas contribution, plus a stacking analysis that recovers weaker absorption in physically selected subsamples.","core_discovery":"The core claim is that the residual absorption excess at the Na I D doublet, measured after subtracting the best-fitting stellar continuum, is neutral gas absorption tracing outflows, and that its incidence scales strongly with stellar mass and quiescence. The authors measure equivalent widths of 4–16 Å, show that stellar models contribute less than about 2 Å, and connect large EW values to blueshifted outflowing gas using a calibration from medium-resolution samples. Stacked spectra confirm the trends and yield neutral outflow masses of a few times 10^7 solar masses and outflow rates of 4–12 solar masses per year. They conclude that neutral outflows are likely already present and important","pith_inferences":["Because He I 5877 Å emission is blended with Na I D at prism resolution and fills in absorption in star-forming galaxies, the true incidence of neutral gas absorption is likely higher than the measured 2.5%; the strong quiescent excess may be partly a sensitivity effect, a distinction medium-resolution follow-up can settle.","The quoted mass outflow rates assume a conservative 1 kpc outflow radius; adopting the 3 kpc radii seen in other studies would raise the rates by a factor of three, making the feedback impact on quenching correspondingly stronger.","The same continuum-subtraction and stacking approach could be pushed to wider or deeper prism surveys to map neutral outflows at lower masses or beyond z~7, provided the He I contamination is modeled explicitly rather than only treated as a selection bias.","The measured detection fractions can be read as lower limits on outflow covering fractions; if the true covering fraction is higher, the opening angles of neutral outflows in quenched galaxies at z>3 are likely wider than the ~30° lower limit derived here."],"forward_implications":["Neutral outflows were already common in massive galaxies during the first 2 Gyr, extending the previously established census at z~2–3 to higher redshift.","The detection fraction rises steeply with stellar mass and quiescence: roughly 2.5% overall, about 11% among the most massive galaxies, and about 43% among massive quenched systems.","Na I D equivalent widths of 4–16 Å exceed the stellar-only prediction by a wide margin, so the absorption must come from substantial amounts of neutral gas.","The inferred neutral mass outflow rates of 4–12 solar masses per year exceed the current star formation rate for about 30% of the detected galaxies, implying outflows can regulate the gas content of these systems.","The stacking analysis detects weaker Na I D absorption in massive, dusty galaxies that are not individually detected, suggesting the 20 golden-sample galaxies are the extreme tail of a broader population."],"fun_headline_variants":["Neutral outflows found in massive early galaxies","JWST reveals neutral outflows in early massive galaxies","Early massive galaxies show neutral gas outflows","Neutral outflows common in massive early galaxies","First 2 Gyr: massive galaxies host neutral outflows"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the residual flux deficit near 5893 Å is genuine Na I D absorption by neutral gas, not a blend of He I 5877 Å emission, a stellar-template mismatch from non-solar abundance patterns, or a residual calibration artifact left by the 10th-order polynomial—an ambiguity the authors themselves acknowledge when they caution that their equivalent widths are lower limits.","fun_headline_variants_meta":{"raw":{"variants":["Neutral outflows found in massive early galaxies","JWST reveals neutral outflows in early massive galaxies","Early massive galaxies show neutral gas outflows","Neutral outflows common in massive early galaxies","First 2 Gyr: massive galaxies host neutral outflows"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000502,"raw_usage":{"total_tokens":2381,"prompt_tokens":926,"completion_tokens":1455,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":670,"completion_tokens_details":{"reasoning_tokens":1381}},"tokens_in":670,"tokens_out":1455,"duration_ms":9177,"temperature":1.0,"reasoning_tokens":1381,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T09:50:35.223042+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the 20 detected galaxies into medium-resolution NIRSpec spectroscopy and resolve the Na I D doublet from the nearby He I line. If the feature splits into He I emission or shows a non-blueshifted, narrow absorption component, or if it disappears when stellar templates with non-solar abundance ratios are used, the outflow census as interpreted here would not stand; conversely, confirming broad blueshifted Na I D absorption in most of the 20 systems would settle the interpretation.","supporting_citations":[],"review_version":1}