{"id":"88201d2a-91f8-4591-a102-223dfe82b2b4","arxiv_id":"2508.20484","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Potassium-to-iron and potassium-to-calcium ratios in seven extremely metal-poor stars show ~0.1 dex scatter, implying K production in massive stars or supernovae is decoupled from the processes driving Na/Mg variation.","lead":"Astronomers measured potassium in seven ancient, extremely metal-poor stars and found its abundance relative to iron and calcium is remarkably uniform, with almost no scatter. This suggests potassium in the early universe came from a process that barely varies between stars, unlike sodium-to-magnesium ratios which change a lot.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central K-vs-Na/Mg contrast is not reproducible from the paper's own Table 5: quoted 1.45/0.74 dex [Na/Mg] scatter is inconsistent with the tabulated LTE values; the load-bearing comparison needs verification.","rationale":"The K-abundance measurements and homogeneous analysis are a valuable observational contribution, and the paper is appropriately cautious in discussing the K-Mn anticorrelation. However, the main inference is a differential claim: small [K/Ca] scatter versus large [Na/Mg] scatter. Such differential claims are sensitive to how both scatters are measured. The [K/Ca] side is clear from Table 4, but the [Na/Mg] side is not: combining Table 5 values gives a much smaller LTE scatter than the 1.45 dex stated in Section 5.1, and the NLTE-corrected values are not shown. This is a reproducibility gap internal to the paper, more immediate than the external K NLTE-grid extrapolation the reader identified. The fix could be as simple as reporting the corrected ratios, but until then the central contrast is not quantitatively established. Since the issue is resolvable by clarification/recomputation and the underlying K data remain useful, the reader's CONDITIONAL verdict remains appropriate; I would not move it to REJECT or ACCEPT without the missing information.","tokens_in":27278,"tokens_out":15399,"duration_ms":151412,"concrete_test":"Recompute [Na/Mg] for the seven K-detected stars from Table 5 and require the authors to provide, in machine-readable form, the NLTE-corrected Na and Mg abundances (or the individual corrections applied via NiLITE/Lind et al. 2022). Then recompute the [Na/Mg] scatter with and without SMSS J085924.06-120104.9. If the NLTE-corrected scatter among these seven stars is ≲0.4 dex, or if it collapses below ~0.3 dex when the outlier is removed, the claimed contrast with the ~0.1 dex [K/Ca] scatter is not established and the central inference would need to be weakened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on a differential statement: ~0.1 dex scatter in [K/Fe] and [K/Ca] versus a much larger ~0.7 dex (NLTE-corrected) scatter in [Na/Mg]. The K/Ca side is internally consistent: Table 4 gives [K/Ca]NLTE mean 0.11 and σ=0.12. But the Na/Mg side is not reproducible from the data presented. Using the LTE [Na/Fe] and [Mg/Fe] values in Table 5 for the same seven K-detected stars, [Na/Mg] = (−0.45, −0.42, −0.62, +0.01, −0.19, −0.57, −0.06), giving a sample standard deviation of only ~0.25 dex and a full range of ~0.6 dex—not the quoted 1.45 dex. The paper does not tabulate the NLTE-corrected Na and Mg abundances/corrections, nor state whether the 0.74 dex value refers to these seven stars or to a different sample. Without that, the contrast could be inflated by a single outlier (e.g., SMSS J085924, which has the highest [Na/Fe] and whose removal drops the LTE σ to ~0.22 dex) or by an unsupported application of external NLTE grids. This is more directly load-bearing than the K NLTE-grid extrapolation: it is an internal, checkable inconsistency in the quantitative basis of the central argument.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a homogeneous high-resolution abundance analysis of K I resonance lines at 766.49 and 769.90 nm in 18 extremely metal-poor stars, using Subaru/HDS spectra and MARCS/Turbospectrum spectral synthesis with LTE and externally computed NLTE corrections. K is detected in seven stars with [Fe/H] < -3.0; for the remaining stars only upper limits are obtained. For the seven detections, the authors report a small scatter in both [K/Fe] and [K/Ca] (σ ≈ 0.13 and 0.12 dex, respectively, in LTE and NLTE), while claiming a much larger 0.74 dex scatter in [Na/Mg] after NLTE correction. They interpret this contrast as evidence that K production in massive stars or supernovae is decoupled from the processes causing Na/Mg variations. The paper then compares the observed K, Sc, V, Mn ratios with rotating and non-rotating massive-star and CCSN yield models and finds that no single model reproduces all odd-Z abundances simultaneously.","tokens_in":27640,"tokens_out":5534,"duration_ms":65505,"significance":"If substantiated, the small intrinsic scatter in [K/Fe] and [K/Ca] at [Fe/H] < -3 would provide a genuinely new observational constraint on K nucleosynthesis, especially when contrasted with abundance ratios such as [Na/Mg] that do vary. The merit of the paper lies in its homogeneous treatment of a very difficult measurement: weak K I lines in extremely metal-poor stars, careful telluric subtraction, consistent stellar parameters, and use of modern published NLTE grids. However, the paper's central differential claim is currently not reproducible from the tabulated data. The [Na/Mg] scatter quoted in the abstract and Section 4.3 cannot be recovered from Table 5 for the seven K-detected stars, and the NLTE-corrected Na and Mg values on which the 0.74 dex figure rests are not provided. The K-side result, by contrast, is well documented and robust in its internal statistics. Because the scientific significance depends on the contrast between the K and Na/Mg scatters, this is a load-bearing gap that must be repaired before the claim can be accepted.","major_comments":[{"comment":"The central differential claim is not reproducible. For the seven stars with detected K lines, the LTE [Na/Mg] ratios computed from Table 5 are -0.45, -0.42, -0.62, +0.01, -0.19, -0.57, -0.06 (for CS 22189-0009, CS 22942-0002, CS 22172-0002, SMSS J085924.06-120104.9, CS 30339-0073, CS 22950-0046, CS 22949-0048). Their sample standard deviation is ~0.25 dex and the full range is ~0.63 dex, not the 1.45 dex quoted in the text. The paper neither tabulates the NLTE corrections for Na and Mg nor states whether the post-NLTE 0.74 dex figure refers to these same seven stars or to a different sample. Since the paper's main conclusion is a differential statement about small K scatter versus large Na/Mg scatter, the authors must either provide the NLTE-corrected values for the same seven stars, or explicitly restrict the claim to what the tabulated data support and discuss the influence of the out","section":"§4.3, Table 5, Fig. 7"},{"comment":"The absolute values of [K/Fe] and [K/Ca] used in the model comparisons in Section 5 carry no estimate of systematic error from the adopted Reggiani et al. (2019) NLTE grid. The grid is applied at metallicities as low as [Fe/H] ≈ -3.8, where the published grid was not designed to be validated, and the corrections depend on interpolated stellar parameters and [K/Fe]. The small scatter in the K ratios is not strongly affected by the choice of LTE versus NLTE (σ = 0.13 in both), but the absolute [K/Ca] values compared with yield models in Figure 9 are directly affected by any systematic offset in the grid. Please quantify the sensitivity of the adopted NLTE corrections to the grid's parameter range and include a conservative systematic uncertainty, or state explicitly why such an uncertainty does not affect the conclusions.","section":"§3.4.2, §4.1"},{"comment":"Even after the reproducibility issue in the first comment is fixed, the claim that the [Na/Mg] scatter is significantly larger than the K scatter needs a statistical test that accounts for measurement uncertainties. With n = 7 and per-star errors of order 0.1-0.2 dex, a raw standard deviation of 0.74 dex versus 0.13 dex is suggestive but not formally established unless intrinsic scatter is estimated (e.g., by comparing χ² or using an F-test or a likelihood-based intrinsic-scatter estimator). The paper currently reports only raw standard deviations and does not separate intrinsic scatter from measurement noise. Please provide such an estimate for both [K/Ca] and [Na/Mg] for the same sample.","section":"§5.1"}],"minor_comments":[{"comment":"The column headings 'e[K/Fe]' and 'mNLT E' are garbled; they should read 'ε[K/Fe]' and '[K/Fe]NLTE'. Also, the table reports individual 766 nm and 769 nm measurements but not how the final [K/Fe] was combined when both lines were detected; please clarify.","section":"Table 4"},{"comment":"HE 0130-1749 and HE 0132-2439 are said to be excluded from the analysis, yet HE 0132-2439 appears in Tables 2, 4, and 5. Please indicate clearly which stars are in the final sample and whether the excluded stars are shown only for completeness.","section":"§2.1, Tables 1, 2, 5"},{"comment":"The statement that a weighted correlation coefficient of -0.99 has a bootstrap p-value of 0.59 is surprising and likely needs a detailed explanation of the weighting and resampling procedure. With seven points, a correlation of this magnitude would normally be significant. Since the anti-correlation is described as tentative, please provide enough information for the reader to assess the p-value.","section":"§4.4, Fig. 8"},{"comment":"The paper would benefit from a machine-readable table of the NLTE-corrected abundances for all elements (or at least the Na and Mg corrections used in Section 4.3), so that the central differential claim can be independently recomputed by readers.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript reports a valuable homogeneous K abundance sample, but the main claim is currently undercut by an internal inconsistency in the quoted [Na/Mg] scatter, which does not match the paper's own Table 5 for the seven K-detected stars. This is fixable with a supplementary table and a re-analysis of the same seven stars, but without it the central conclusion is unsupported. The K abundances themselves appear to be carefully derived, and the small scatter in [K/Fe] and [K/Ca] may survive further scrutiny even if the Na/Mg contrast weakens."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this paper gives you something real—the first homogeneous K abundance sample in EMP stars below [Fe/H]=-3, with seven detections and a scatter in [K/Fe] and [K/Ca] that is genuinely small (~0.12–0.13 dex). That part is worth having. But the headline contrast with [Na/Mg] scatter does not survive a check against the paper's own Table 5, and that is the crux of the paper.\n\nWhat is new: they did the telluric and NLTE corrections carefully, report upper limits sensibly, and don't overstate the K–Mn anti-correlation (p=0.59). The measured K values are consistent with earlier work and extend it to a regime where K detections were sparse.\n\nThe problem: the abstract and Section 4.3 claim ~0.7 dex scatter in [Na/Mg] after NLTE correction, with LTE scatter \"as large as 1.45 dex\". For the seven K-detected stars, Table 5 gives [Na/Mg] values of -0.45, -0.42, -0.62, +0.01, -0.19, -0.57, -0.06. The sample standard deviation is ~0.25 dex, not 1.45. The paper never tabulates the NLTE-corrected Na and Mg abundances or corrections, so the 0.74 dex figure cannot be reproduced. If the true Na/Mg scatter is 0.2–0.3 dex, the contrast with the K scatter shrinks to a factor of two or three, not the order-of-magnitude difference the text claims. That weakens the argument that K production is decoupled from the processes driving Na/Mg variation. It may still be true, but this paper hasn't shown it.\n\nThere are minor issues too: seven stars with statistical errors of 0.1–0.2 dex means the small-scatter claim rests on a small sample, and the K NLTE corrections come from an external grid with systematic uncertainties not included. Those are acceptable caveats, not fatal.\n\nWho should read it: anyone working on EMP star abundances or massive star nucleosynthesis. The K measurements are a useful addition and I'd cite them. But I'd want the authors to fix or qualify the Na/Mg claim before building on it.\n\nRecommendation: send it to peer review, but ask for the NLTE-corrected Na and Mg values, a proper calculation of the [Na/Mg] scatter from the actual sample, and a revised interpretation if the scatter is not 0.7 dex. This is an honest, careful observational paper with one number in it that does not add up.","headline":"New K abundances in EMP stars are a real contribution, but the paper's key Na/Mg scatter claim doesn't match its own Table 5 and needs verification before the astrophysical conclusion can be trusted.","tokens_in":28219,"tokens_out":4351,"would_cite":true,"duration_ms":43100,"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":"Potassium tracks calcium tightly in seven extremely metal-poor stars, with scatter at the measurement-error level, while sodium varies.","keywords":["potassium abundances","extremely metal-poor stars","NLTE corrections","K I resonance lines","spectral synthesis","massive star nucleosynthesis","core-collapse supernovae","odd-Z elements"],"falsifier":"Observe the K I 766/769 nm lines in 20 to 30 additional extremely metal-poor stars with [Fe/H] < -3 at similar or higher signal-to-noise and measure [K/Ca]; if the scatter exceeds about 0.2-0.3 dex, the claim of uniform potassium production fails. Independently, recomputing NLTE corrections for K I with updated collision rates at low metallicity would settle whether a shift greater than 0.3 dex in [K/Fe], varying with Teff and log g, resolves the apparent uniformity into an artifact.","tokens_in":27158,"feed_emoji":"🔭","tokens_out":7785,"duration_ms":81751,"temperature":0.7,"pith_summary":"This paper measures potassium abundances in seven of the most iron-poor stars known, stars with less than a thousandth of the Sun's iron. It finds that the ratios [K/Fe] and [K/Ca] are enhanced over solar and, more importantly, nearly identical from star to star: the scatter is about 0.1 dex, the same size as the measurement uncertainty. In the same stars, [Na/Mg] varies by about 0.7 dex. If each star's composition samples the ejecta of only one or a few early massive stars, this contrast implies that potassium production in massive stars or their supernovae is independent of the varying stellar properties that drive sodium-magnesium differences.","feed_headline":"Potassium tracks calcium tightly in 7 extremely metal-poor stars","feed_subtitle":"Tight K/Ca in ancient, iron-poor stars points to a potassium source untouched by stellar mass and rotation.","key_machinery":"The central object is the pair of K I resonance lines at 766.49 and 769.90 nm in high-resolution spectra. The analysis pipeline is spectral synthesis under LTE with MARCS model atmospheres, followed by interpolation of the Reggiani et al. (2019) NLTE correction grid in Teff, log g, [Fe/H], vmic, and [K/Fe]. The load-bearing comparison is the contrast between the small scatter in [K/Fe] and [K/Ca] (~0.1 dex) and the large scatter in [Na/Mg] (~0.7 dex after NLTE correction): two element ratios from the same stellar environment, one nearly constant and one highly variable. This contrast is what separates a potassium production channel that is robust to progenitor properties from the processes s","core_discovery":"The paper reports a homogeneous potassium abundance analysis of seven extremely metal-poor stars ([Fe/H] < -3.0) using high-resolution spectra of the K I resonance lines at 766.49 and 769.90 nm, with LTE spectral synthesis followed by NLTE corrections from a published grid. After correction, [K/Fe] has a mean of 0.35 dex with a standard deviation of 0.13 dex, and [K/Ca] has a mean of 0.11 dex with a scatter of 0.12 dex — scatter as small as the typical measurement uncertainty. In contrast, [Na/Mg] after NLTE correction shows a scatter of about 0.74 dex. The paper interprets the narrow K/Ca distribution as evidence that K and Ca are co-produced in a process insensitive to progenitor mass, rot","pith_inferences":["By extension, if a larger EMP sample reproduces the ~0.1 dex scatter in [K/Ca], the cleanest interpretation is that K/Ca is set by the neutrino-driven electron fraction in the innermost supernova ejecta, where progenitor mass and rotation play only a secondary role — a testable prediction for multidimensional core-collapse simulations.","By extension, the same K I line pair could be measured in globular-cluster giants where a K-Mg anti-correlation has been reported; a uniform K/Ca there would bridge the EMP-star result and the cluster phenomenon.","By extension, a targeted experiment would be to extend the analysis across a wider range of Teff and log g at fixed [Fe/H]; if the [K/Ca] scatter grows with stellar parameters, part of the observed uniformity is a consequence of the homogeneous stellar-parameter window rather than nucleosynthesis.","By extension, the K-Mn anti-correlation hint suggests a falsifiable discriminator: if it survives in larger samples, K and Mn must be produced in complementary regimes (for example, proton-rich versus neutron-rich ejecta), which existing one-dimensional yield models do not capture."],"forward_implications":["If the small scatter in [K/Fe] and [K/Ca] is correct, potassium becomes a tracer of the final evolutionary stages of massive stars and of supernova explosions, regimes inaccessible to direct observation.","A successful nucleosynthesis model must simultaneously reproduce the mean [K/Fe] ~ 0.35 dex and [K/Ca] ~ 0.11 dex while keeping their scatter at the measurement-error level; models that tie K yield strongly to progenitor mass, rotation, or metallicity are disfavored.","The observed upper limits, mostly [K/Fe] NLTE < 0.8 dex, place ceiling constraints on potassium yields from the earliest supernovae.","The tentative anti-correlation between [K/Fe] and [Mn/Fe], if confirmed with a larger sample, would distinguish neutrino-processed, proton-rich channels from other odd-Z production mechanisms.","The agreement of the mean NLTE [K/Fe] with previous analyses supports the reality of supersolar potassium enhancement at the lowest metallicities."],"supporting_citations":[{"why":"Supplies the precomputed NLTE correction grid for the K I lines; the absolute [K/Fe] values and the size of the scatter depend on this grid.","marker":"H. Reggiani et al. (2019)"},{"why":"Earlier K abundance measurements in EMP stars used as the LTE comparison baseline.","marker":"R. Cayrel et al. (2004)"},{"why":"NLTE K abundances from a re-analysis of the Cayrel sample, providing the comparison baseline for the corrected ratios.","marker":"Y. Takeda et al. (2009)"},{"why":"Establishes hydrostatic oxygen burning as the predicted production site of 39K in massive stars, the hypothesis the observations test.","marker":"S. E. Woosley & T. A. Weaver (1995)"},{"why":"Rotating massive-star yield models whose [K/Ca] predictions are compared with the observed uniform ratios.","marker":"M. Limongi & A. Chieffi (2018)"},{"why":"Non-rotating zero-metallicity supernova yield grid against which the observed odd-Z abundances are checked.","marker":"A. Heger & S. E. Woosley (2010)"},{"why":"Two-dimensional core-collapse supernova model of neutrino-processed ejecta, giving the K/Ca scenario most consistent with the small-scatter interpretation.","marker":"S. Wanajo et al. (2018)"},{"why":"Supplies the NLTE corrections for Na that are used to compute the [Na/Mg] scatter contrasting with K/Ca.","marker":"I. Koutsouridou et al. (2025)"}],"fun_headline_variants":["Tight K/Ca in ancient stars hints at stable potassium source","Extreme stars show potassium and calcium move together","K/Ca scatter minimal in 7 ultra-metal-poor stars","Potassium's tight bond with calcium in early universe","Why potassium stays steady while sodium varies in old stars"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The NLTE correction grid, computed for atmospheric conditions near solar metallicity, is assumed to remain valid when interpolated down to [Fe/H] ~ -3 to -4; if its collision rates or atomic model are wrong at these metallicities, the absolute [K/Fe] and [K/Ca] values and the size of the scatter would shift.","fun_headline_variants_meta":{"raw":{"variants":["Tight K/Ca in ancient stars hints at stable potassium source","Extreme stars show potassium and calcium move together","K/Ca scatter minimal in 7 ultra-metal-poor stars","Potassium's tight bond with calcium in early universe","Why potassium stays steady while sodium varies in old stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000185,"raw_usage":{"total_tokens":1237,"prompt_tokens":903,"completion_tokens":334,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":253}},"tokens_in":647,"tokens_out":334,"duration_ms":4339,"temperature":1.0,"reasoning_tokens":253,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T15:04:36.969992+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Observe the K I 766/769 nm lines in 20 to 30 additional extremely metal-poor stars with [Fe/H] < -3 at similar or higher signal-to-noise and measure [K/Ca]; if the scatter exceeds about 0.2-0.3 dex, the claim of uniform potassium production fails. Independently, recomputing NLTE corrections for K I with updated collision rates at low metallicity would settle whether a shift greater than 0.3 dex in [K/Fe], varying with Teff and log g, resolves the apparent uniformity into an artifact.","supporting_citations":[{"cited_title":"2009, PASJ, 61, 563, doi: 10.1093/pasj/61.3.563","cited_arxiv_id":null,"evidence_quote":"NLTE K abundances from a re-analysis of the Cayrel sample, providing the comparison baseline for the corrected ratios."}],"review_version":1}