{"id":"78659c76-e6ac-49b4-8930-a5ba4380f48a","arxiv_id":"2607.15595","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"A blind FAST survey of 41 HI absorbers re-detected the known OH absorber PKS 1413+135, found no new OH, and set [OH]/[HI] upper limits near 1e-8.","lead":"Astronomers used China's FAST telescope to search for hydroxyl (OH) gas in 19 systems known to absorb neutral hydrogen, and found no new OH absorbers. They set the tightest upper limits yet on how much molecular gas such systems can hold, a benchmark for future radio surveys.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Combined-sample [OH]/[HI] limit is a ratio of separately weighted stacks, not a mean of per-source ratios; this may bias the headline 0.90e-8 limit low and overstate the 'strongest constraints' claim.","rationale":"The reader's weakest assumption concerns the physical conversion parameters (Tex, Ts, cf, Δv), which are standard, explicitly stated, and affect all limits proportionally. A more load-bearing issue is the statistical construction of the combined [OH]/[HI] limit. The paper derives it as a ratio of two separately weighted stacks, a method that can produce a value lower than any meaningful population average when the noise and column-density distributions differ across subsets. The combined limit is the most striking number in the abstract, and it is not internally consistent with the associated and intervening limits. This is not a question of external assumptions but a potential methodological artifact in the headline result. The associated and intervening limits are individually valuable and probably still the strongest in their categories, so the paper can be conditionally accepted if the combined limit is recomputed or properly caveated. The concrete test using survival analysis on per-source limits would settle whether the combined value is biased.","tokens_in":27474,"tokens_out":21326,"duration_ms":230810,"concrete_test":"Recompute the combined-sample [OH]/[HI] 3σ upper limit by first deriving a per-source upper limit for each of the 19 systems from its own OH optical-depth limit and HI column density, then combine these individual limits using the same survival-analysis / Bayesian censored-regression approach used in Section 6.1 (e.g., linmix on the log ratios). Compare the resulting upper limit to the stacked value of 0.90e-8. If the survival-based combined limit exceeds 0.90e-8 by a significant amount (e.g., >1.3e-8), the stacked combined limit is biased and the abstract's headline constraint should be revised or re-presented with this caveat.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim of the strongest constraints on [OH]/[HI] leans on the combined-sample limit of <0.90e-8, which is lower than both the associated (<1.66e-8) and intervening (<1.42e-8) limits. This unusual hierarchy arises because the combined value is computed in Section 6.2 as the ratio of two separately stacked weighted means: the stacked OH column density (from ⟨τ_OH⟩ with weights 1/[C · rms_i^2]) and the stacked HI column density (from ⟨τ_HI⟩ with analogous weights). The OH stack is dominated by low-noise, mostly intervening systems, while the HI stack includes high-N_HI associated systems, inflating the denominator without proportionally affecting the numerator. The ratio of weighted means is not the mean of the per-source [OH]/[HI] ratios and can be systematically lower than the typical value. The paper does not acknowledge this bias, and it is not covered by the stated assumptions on Tex, Ts, cf, or Δv. If the combined limit is biased downward, the 'strongest constraints to date' claim—which in the abstract highlights this lowest value—would be correspondingly overstated.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the third installment of the FAST HI 21-cm absorption blind survey. It adds 394.4 hr and 1622.1 deg^2 of 2024 CRAFTS/FASHI data plus FATHOMER, detecting three known and four new HI absorbers, bringing the catalog to 41 systems. For 18 of these systems (19 sightlines including the HVC toward NVSS J090150+030422) with OH transitions in the FAST band, it conducts an OH 18-cm search, re-detecting OH 1612 MHz absorption and 1720 MHz emission toward PKS 1413+135 and finding no new OH absorbers. Survival analysis with Bayesian censored regression finds no statistically significant N_OH-N_HI correlation or redshift evolution. Stacking yields 3-sigma upper limits N_OH < 4.93, 1.64, 1.72 x 10^12 T_ex/c_f cm^-2 K^-1 and [OH]/[HI] < 1.66e-8, 1.42e-8, 0.90e-8 for associated, intervening, and combined samples, claimed as the strongest constraints to date.","tokens_in":1221,"tokens_out":1265,"duration_ms":171808,"significance":"The paper's strengths are the first OH search in a blind HI-selected sample, the re-detection of PKS 1413+135, the use of survival statistics appropriate for ~85% censoring, and stacking that improves on prior limits. If the estimator issues are resolved, the sample and upper limits provide a valuable benchmark for future OH surveys. The paper is honest about the assumptions (T_ex = 10 K, T_s = 100 K, c_f = 1, Delta-v = 30 km/s) and about the large uncertainties in the correlation analysis.","major_comments":[{"comment":"The combined [OH]/[H i] limit is calculated as the ratio of two separately stacked weighted means, with OH weights 1/(C rms_OH^2) and H i weights 1/(C rms_HI^2). Because the weighting is not identical, this quantity is not the mean of per-source [OH]/[H i] ratios; the resulting combined limit (0.90e-8) is lower than both the associated (1.66e-8) and intervening (1.42e-8) limits, which the text does not explain. This suggests that the OH stack is dominated by low-noise intervening sightlines while the H i stack includes high-N_HI associated systems. Please define the estimator explicitly and either justify it as an upper limit on a physically meaningful average abundance, or stack per-source ratio upper limits and report those; if the combined number is not a valid mean ratio, the abstract's headline constraint should be qualified.","section":"Section 6.2, 'OH abundance relative to H i'"},{"comment":"The censored-regression fits are performed only for associated and intervening absorbers; the 11 systems with undetermined classifications are plotted but omitted from the fits. This exclusion is not stated in the text, and no robustness test is given. Because these systems could differ systematically (e.g., in redshift or host properties), the conclusion of no statistically significant correlation/evolution formally applies only to the classified subset. Please state the exclusion explicitly and assess its impact on the conclusions.","section":"Section 6.1 and Figure 13"}],"minor_comments":[{"comment":"The abstract says 19 HI absorption systems were searched for OH, but Section 5 says 'the remaining 18 absorbers' and Table 5 lists 19 rows including the HVC toward NVSS J090150+030422. Please clarify systems vs. sightlines and use consistent counts.","section":"Abstract and Sections 1, 5"},{"comment":"In the PKS 1413+135 paragraph, 'N_OH ~ 0.15 T_s 10^15 cm^-2 K^-1' should use T_ex rather than T_s, since this is an OH column density; this appears to be a typo.","section":"Section 5.1"},{"comment":"The term 'blind' for the OH search should be qualified. The OH search is not a blind sky survey but a targeted search within a blind HI-selected sample; as written, 'first blind survey to detect OH absorption' could be overstated.","section":"Abstract and Section 1"},{"comment":"Please state explicitly whether the 'combined' sample includes the 11 systems with undetermined classifications and how they are weighted; currently 'combined' is ambiguous.","section":"Section 6.2"},{"comment":"Equation (1) assumes Gaussian uncorrelated noise. Please specify whether the 1 km/s rebinned channels used for the stacked RMS remain uncorrelated and whether baseline systematics are included in the jackknife error estimates.","section":"Section 3.2, Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"This is a useful observational paper. The main load-bearing issue is the combined stacking estimator in Section 6.2; it is fixable by revising the estimator or adding a clear interpretational caveat. The sample-count wording and the 'blind' terminology should also be corrected. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a genuinely useful survey paper — four new HI absorbers, the first FAST OH search, and stacking limits that improve on earlier work by an order of magnitude — but the headline \"strongest constraints\" is a bit stronger than the statistics actually support. The combined-sample [OH]/[HI] limit of <0.90e-8 is a ratio of two separately stacked weighted means, not a typical per-source ratio, and it comes out lower than either subsample limit. That is a red flag, and the paper doesn't acknowledge it. If the OH stack is dominated by quiet intervening sightlines while the HI stack includes high-column associated systems, the ratio is biased low. The combined number should be presented with that caveat, or replaced by the subsample limits as the headline.\n\nWhat is actually new: the four new HI absorbers are new measurements, the OH search over 19 systems is the first blind search with FAST, and the survival analysis with ~85% censoring is a step up from Zheng et al. The upper limits on [OH]/[HI] from the associated and intervening stacks (1.66e-8 and 1.42e-8) do look like the tightest yet in this line of work. The paper is careful with the statistical treatment. The Gaussian fitting with BIC, the jackknife errors on the stacks, and the explicit statement of assumptions on Tex, Ts, cf, and line width are all to the good. The re-detection of PKS 1413+135 is a sanity check, not a discovery, so calling the survey \"the first blind survey to detect OH absorption\" is technically true but overstates what happened.\n\nThe main soft spots: first, the combined-stack ratio issue above; it needs a sentence in the methods or a sensitivity test. Second, the quoted limits scale linearly with the assumed Tex, Ts, and covering factor, so \"strongest constraints\" is conditional on those choices; the authors do say this, but the abstract reads as if it's absolute. Third, the sample-count wording (19 systems vs 18 sources) is slightly confusing, and the HVC of NVSS J090150+030422 complicates the counting. These are minor.\n\nWho this is for: people working on cold gas absorption, HI surveys, and molecular gas at high redshift. It's a solid catalog paper and a useful benchmark for SKA/FAST follow-up. It deserves peer review. I'd send it out, but I would ask the authors to address the combined-stack interpretation before it goes to press.","headline":"Solid survey paper with genuinely new HI absorbers and the tightest [OH]/[HI] limits to date, but the combined-sample limit is a ratio of weighted stacks that may be biased low, so the headline claim is slightly oversold.","tokens_in":28310,"tokens_out":5263,"would_cite":true,"duration_ms":62270,"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":"A blind FAST search for OH 18-cm absorption in 19 HI-selected systems re-detects the known absorber toward PKS 1413+135, finds no new ones, and places the tightest limits to date on [OH]/[HI].","keywords":["OH 18-cm absorption","HI 21-cm absorption","molecular gas in galaxies","spectral stacking","survival analysis","radio absorption lines","hydroxyl radical","blind survey"],"falsifier":"Take any one of the 19 HI absorbers, integrate with FAST in ON-OFF mode until the 3-sigma sensitivity on the OH 1667 MHz integrated optical depth is three times deeper than the current stacked limit. A detection at the expected redshifted frequency would falsify the claim that these systems contain little OH. Alternatively, a direct measurement of OH excitation temperature (e.g., from the 1612/1667 ratio in PKS 1413+135) below 10 K would rescale the quoted column-density limits upward.","tokens_in":27390,"feed_emoji":"📡","tokens_out":6004,"duration_ms":64044,"temperature":0.7,"pith_summary":"The paper extends a blind HI 21-cm absorption survey on FAST to also search for OH 18-cm absorption in the 19 systems whose redshifted OH lines fall in the telescope band. It re-detects the known OH absorption toward PKS 1413+135 — the first blind survey to catch OH absorption — but finds no new OH absorbers. Stacking the non-detections yields 3-sigma upper limits on the OH-to-HI abundance ratio, [OH]/[HI] < 1.66e-8, < 1.42e-8, and < 0.90e-8 for associated, intervening, and combined samples, under stated assumptions. A survival analysis of column densities finds no statistically significant correlation between N_OH and N_HI, nor redshift evolution of their ratio. The paper argues this establishes an unbiased benchmark for molecular-gas studies in HI-selected systems.","feed_headline":"No new OH absorbers in blind survey; limits drop to 1e-8","feed_subtitle":"Stacking 19 FAST HI absorbers caps hydroxyl abundance near Milky Way diffuse-gas levels — a blind-survey benchmark for molecular gas.","key_machinery":"The central machinery is the OH 18-cm ground-state transitions (1612, 1665, 1667, 1720 MHz) observed toward a parent sample of HI 21-cm absorbers found in a blind drift-scan survey. Because OH absorption is typically optically thin, non-detections are converted into 3-sigma upper limits on integrated optical depth; these are combined across sources by spectral stacking, which weights each OH spectrum by completeness and noise, and by Bayesian censored regression that treats detections and upper limits together. The load-bearing comparison is the OH-to-HI column-density ratio [OH]/[HI] derived by converting optical-depth limits into column densities under fixed excitation and spin temperature","core_discovery":"In the survey's own terms, the result is that OH 18-cm absorption is rare in unbiased, HI-selected systems. The only detection is the previously known absorber toward PKS 1413+135, whose 1612 MHz absorption and 1720 MHz emission are re-detected. All other searched spectra are non-detections; spectral stacking turns these into 3-sigma upper limits of [OH]/[HI] < 1.66e-8, < 1.42e-8, and < 0.90e-8 (associated, intervening, combined) under adopted Tex=10 K, Ts=100 K, covering-factor-unity, 30 km/s assumptions. The survival analysis finds no statistically significant N_OH-N_HI correlation or redshift evolution, suggesting earlier reported trends came from targeted, molecule-biased samples.","pith_inferences":["Editorial inference: if deeper integrations still show no OH in these systems, the simplest reading is that the bulk of HI-selected absorbers are not molecule-rich; this would lower estimates of the 'dark molecular gas' fraction in such sightlines.","Editorial inference: the PKS 1413+135 re-detection, now reclassified as intervening, suggests that red, dusty sightlines (V-K > 6) are the high-yield subset of HI absorbers; a testable extension is to compare stacked [OH]/[HI] limits for red versus blue HI absorbers within the same survey.","Editorial inference: the quoted limits scale linearly with the assumed OH excitation temperature and inversely with covering factor. If OH excitation in these diffuse gas clouds is actually colder than 10 K, the true OH content could be higher; measuring Tex or cf along one absorber would calibrate the whole stacking result."],"forward_implications":["Every known extragalactic OH absorber also shows HI 21-cm absorption, so HI-selected samples provide an efficient route to finding OH absorbers; this survey demonstrates that route blind.","Because the stacked limits sit at or below the diffuse-Galactic value of [OH]/[HI] ~ few x 1e-8, the typical HI absorber in this sample contains very little OH, meaning molecular gas is either confined to small clumps or absent.","The absence of a significant N_OH-N_HI correlation or redshift evolution in the unbiased sample indicates previously reported correlations were likely driven by targeted, dust- and molecule-biased samples.","Wide instantaneous bandwidth instruments can search HI and OH absorption simultaneously, so future large drift-scan surveys can expand the census of molecular absorbers across redshift."],"fun_headline_variants":["Blind OH survey: no new absorbers, tightest limits yet","FAST blind search: OH absorption rare in HI systems","Stacking 19 HI absorbers caps hydroxyl abundance","No new OH detections; limits reach 1e-8 abundance","First blind OH survey sets benchmark for molecular gas"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The headline upper limits assume optically thin OH lines, an OH excitation temperature of 10 K, an HI spin temperature of 100 K, unit covering factors, and a 30 km/s line width; if the real excitation temperature is lower or the covering factor is below unity, the limits weaken proportionally.","fun_headline_variants_meta":{"raw":{"variants":["Blind OH survey: no new absorbers, tightest limits yet","FAST blind search: OH absorption rare in HI systems","Stacking 19 HI absorbers caps hydroxyl abundance","No new OH detections; limits reach 1e-8 abundance","First blind OH survey sets benchmark for molecular gas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000215,"raw_usage":{"total_tokens":1415,"prompt_tokens":1042,"completion_tokens":373,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":786,"completion_tokens_details":{"reasoning_tokens":289}},"tokens_in":786,"tokens_out":373,"duration_ms":3825,"temperature":1.0,"reasoning_tokens":289,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T22:49:42.583492+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take any one of the 19 HI absorbers, integrate with FAST in ON-OFF mode until the 3-sigma sensitivity on the OH 1667 MHz integrated optical depth is three times deeper than the current stacked limit. A detection at the expected redshifted frequency would falsify the claim that these systems contain little OH. Alternatively, a direct measurement of OH excitation temperature (e.g., from the 1612/1667 ratio in PKS 1413+135) below 10 K would rescale the quoted column-density limits upward.","supporting_citations":[],"review_version":1}