{"id":"c9c48600-09b0-4b88-9132-12ffab2a3ce4","arxiv_id":"1908.01508","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"From 1974-2014 Yakutsk data, the rate of change of air shower maximum with energy is nonlinear, indicating a changing cosmic ray mass composition over 10^16 to 5.7x10^19 eV.","lead":"The Yakutsk cosmic ray observatory reports how deep air showers reach into the atmosphere for cosmic rays between 10^16 and 5.7x10^19 eV, using 40 years of Cherenkov light data. The depth changes at different rates at different energies, which the authors interpret as a change in the average mass of cosmic ray nuclei.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fluctuation confirmation fails a self-consistency check: Eq. (6)-(7) applied to Tables II-III gives 43-49 g/cm^2, not the quoted 57-63 g/cm^2 peak.","rationale":"The central claim is that the nonlinear elongation rate, together with a maximum in Xmax fluctuations near 10^17-10^18 eV and a decrease above 5x10^18 eV, indicates a changing cosmic-ray composition. The elongation rate and mean Xmax values are direct measurements and are broadly consistent with other experiments, so I do not see a fundamental argument-level flaw there. The most vulnerable point is the fluctuation confirmation, because it depends on subtracting an instrumental resolution estimated from a linear fit to Monte Carlo simulations. The reader identified this same correction as the weakest assumption. My stress test goes further: applying the paper's own Eq. (6)-(7) to the tabulated widths does not reproduce the quoted 57-63 g/cm^2 peak; it yields corrected values around 43-49 g/cm^2. That is an internal numerical inconsistency rather than merely a concern about external model bias. If the plotted fluctuation values are in fact raw, the paper's quoted fluctuation range is misleading and the composition interpretation tied to that range would need to be redone. If the plotted values are corrected, the table definitions or the subtraction formula in the manuscript need revision. This does not overturn the qualitative composition-change conclusion, because the mean Xmax and elongation rate still show the same trend, but it does remove the fluctuation analysis as a clean independent confirmation. The appropriate response is to require the authors to provide the corrected table/figure and to re-state the fluctuation-based composition inference, which is exactly the conditionality that the reader already imposed. Hence I do not change the reader's CONDITIONAL verdict; I strengthen the technical basis for that conditionality.","tokens_in":9273,"tokens_out":9670,"duration_ms":106999,"concrete_test":"Reproduce the corrected sigma(Xmax) values for every bin in Tables II and III using Eq. (6) and Eq. (7), and compare them point-by-point with the values plotted in Fig. 7(b). Specifically, at 2.68x10^17 eV, determine whether the plotted point is 62.3 g/cm^2 (raw) or ~44 g/cm^2 (corrected); at 1.09x10^18 eV, check whether it is 60.9 g/cm^2 or ~47 g/cm^2. If the plotted points are the raw widths, the claimed 57-63 g/cm^2 mixed-composition interpretation must be re-derived from the corrected 43-49 g/cm^2 values and the text revised; if the plotted points are corrected, the table column definitions and the application of Eq. (7) need to be reconciled. This single check settles whether the fluctuation-based confirmation is internally consistent as reported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's independent confirmation of a composition change rests on the detector-width subtraction in Eq. (6)-(7). The text and Fig. 7(b) quote a maximum of sigma(Xmax) = 57-63 g/cm^2 in the range 2x10^17-2x10^18 eV, interpreting it as mixed composition with high p+He content. But applying the authors' own correction to the measured widths in Tables II and III gives a different result. For example, at 2.68x10^17 eV, Eq. (6) gives <sigma_app> = 38.5 - 10*log10(0.268) ~ 44.2 g/cm^2; with the tabulated sigma_meas = 62.3 g/cm^2, Eq. (7) gives sigma_phys ~ sqrt(62.3^2 - 44.2^2) ~ 44 g/cm^2. At 1.09x10^18 eV, sigma_phys ~ sqrt(60.9^2 - 38.1^2) ~ 47 g/cm^2. These corrected values are 43-49 g/cm^2 across the claimed peak region, not 57-63 g/cm^2. The quoted 57-63 values match the raw measured column, not the detector-corrected fluctuations. Thus either Fig. 7(b) plots raw widths (making the peak after subtraction substantially lower than stated) or the table definitions and Eq. (6)-(7) application are inconsistent. Additionally, below ~5x10^16 eV the tabulated widths (53-57 g/cm^2) are within about one standard deviation of the estimated instrumental resolution, so the quadrature-subtracted physical fluctuation is essentially unconstrained and cannot support the low-energy heavy-composition interpretation. This is load-bearing because the fluctuation analysis is presented as the confirmation of the elongation-rate composition change; if its quantitative basis is wrong, that independent support is not yet established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports Xmax measurements from 1974-2014 Yakutsk Cherenkov array data over 10^16 to 5.7×10^19 eV. It derives elongation rates of 48±6, 78±5, 63±6, and 50±7 g/cm^2 per decade in four energy intervals and interprets the non-monotonic elongation rate as evidence for a composition change from heavier nuclei at low energy to proton-rich at ~10^18 eV and back to heavier at the highest energies. The paper also computes <lnA> by interpolating measured <Xmax> between proton and iron endpoints from QGSJetII-04 simulations, and it claims that a maximum in Xmax fluctuations around 2×10^17-2×10^18 eV provides independent confirmation of the composition change. The central problem is that the fluctuation confirmation appears quantitatively inconsistent with the paper's own correction formulas and tabulated values.","tokens_in":9704,"tokens_out":5959,"duration_ms":54521,"significance":"The dataset is valuable: it covers four decades in energy with a single technique and is compared with Auger, TA, HiRes, Tunka, and LOFAR results. The elongation-rate measurement is an important anchor for the transition region, and the composition inference via QGSJetII-04 interpolation is a legitimate, model-dependent procedure rather than a circular fit. However, the claimed independent confirmation from fluctuations is not supported by the paper's own equations and tables, and the reported <lnA> values carry only statistical errors. If the fluctuation correction issue is resolved and systematic uncertainties are propagated, this could become a useful contribution; in its current form the central confirmation claim is not quantitatively reliable.","major_comments":[{"comment":"The claimed maximum of physical fluctuations sigma(Xmax)=57-63 g/cm^2 in the range 2×10^17-2×10^18 eV is not what the paper's own correction produces. Applying Eq. (6) at E0=2.68×10^17 eV gives <sigma_app>=38.5-10*lg(0.268)=44.2 g/cm^2; with sigma_meas=62.3 g/cm^2 from Table II, Eq. (7) gives sigma_phys≈44 g/cm^2. At E0=1.09×10^18 eV, the corrected value is ≈47 g/cm^2. These corrected values are 43-49 g/cm^2 across the claimed peak, not 57-63 g/cm^2. The quoted 57-63 range matches the uncorrected sigma_meas column in Tables II-III. Thus either Fig. 7(b) plots raw measured fluctuations rather than physical ones, or the table definitions and Eq. (7) are inconsistent. Since the fluctuation maximum is presented as the independent confirmation of the composition change, this inconsistency is load-bearing and must be resolved.","section":"II.C, Eq. (6)-(7), Tables II-III, Fig. 7(b)"},{"comment":"The tabulated sigma_meas values at low energy (e.g., 53.6 g/cm^2 at 3.5×10^16 eV) are nearly equal to the instrumental width from Eq. (6) (≈53 g/cm^2 at that energy), so the quadrature-subtracted sigma_phys is ≈7 g/cm^2 with an uncertainty comparable to or larger than the value. The low-energy fluctuation data therefore cannot meaningfully constrain the physical intrinsic width, and the conclusion of a heavier composition at low energy cannot be supported by the fluctuation analysis there.","section":"II.C, below ~5×10^16 eV"},{"comment":"The reported <lnA> values include only the 'stat. error' column. The systematic uncertainties described in Section II.B — the 21% absolute calibration error, the 23-26% energy scale uncertainty, and the per-shower Xmax reconstruction uncertainty of 15-55 g/cm^2 — are not propagated into <lnA> or into the quoted composition fractions. Because the interpolation in Eq. (8) is nearly linear in Xmax, an unpropagated systematic of ~20 g/cm^2 in Xmax translates into a shift of order 0.5 in <lnA>, which is comparable to the bin-to-bin changes interpreted as composition changes. The authors should quote a systematic uncertainty on <lnA> or justify that the relative trend is unaffected.","section":"II.D, Eq. (8), Tables II-III"},{"comment":"The highest-energy elongation-rate value and the claim of a heavier composition above ~5×10^18 eV rest on very few events: the last two bins contain 7 and 3 events, with Xmax uncertainties of 15 and 23 g/cm^2, respectively. The derived <lnA> values at 3.74×10^19 and 5.69×10^19 eV (1.57±0.64 and 1.69±0.97) are consistent with a constant composition within about 1.5 sigma. The text should temper the high-energy conclusion or provide a sensitivity check with different binning.","section":"III, Table III"}],"minor_comments":[{"comment":"Reference [53] has a stray double bracket in the text: '1994-2010 [53]]' should be corrected.","section":"References"},{"comment":"The table header '10 18·5.7·1019 eV' is unformatted and should read '10^18 - 5.7×10^19 eV'.","section":"Table III header"},{"comment":"The entry '0,64' in the last row uses a comma as the decimal separator; it should be '0.64' for consistency with the rest of the paper.","section":"Table III"},{"comment":"The spelling of the hadronic model is inconsistent ('QGSJetII-04', 'QGSjetII-04', 'QGSjetII-03'); one standard spelling should be used throughout.","section":"Throughout"},{"comment":"Figure 3 is not explicitly cited in the running text; a citation should be added where atmospheric transparency is discussed.","section":"Section II.A"},{"comment":"The uncertainties on the four elongation-rate values are given without stating how they were derived (fit method, correlated versus uncorrelated errors); a brief description would improve reproducibility.","section":"Section II.C"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle is the inconsistency between the fluctuation correction formulas and the plotted/claimed sigma(Xmax) peak. If the authors can clarify exactly which quantity is shown in Fig. 7(b) and show corrected values, the confirmation claim may be recoverable. The absence of systematic errors on <lnA> is also a significant gap that should be addressed before publication. The high-energy claims rest on very sparse statistics and should be softened accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick take: this is a useful dataset paper with a plausible composition trend, but the fluctuation analysis has an internal inconsistency that should prevent acceptance in its current form.\n\nWhat's actually new: the 1974-2014 Yakutsk Cherenkov sample, with binned Xmax and sigma(Xmax) from 1e16 to 5.7e19 eV. That is roughly four decades of energy and forty years of data. The mean Xmax values are consistent with Auger, TA, HiRes, Tunka, and LOFAR within errors, which is a meaningful independent check. The elongation-rate measurement (48±6, 78±5, 63±6, 50±7 g/cm^2 per decade) shows a clear nonlinearity, and the inference about composition changing from heavy at low energy to proton-rich near 1e18 and heavier again at the highest energies is in line with the field.\n\nThe soft spot is the fluctuation section. The text says detector smearing was subtracted using Eq. (6)-(7), and Fig. 7(b) reports a peak of 57-63 g/cm^2 in the 2e17-2e18 eV range. But applying Eq. (6) to the energies in Tables II-III and subtracting in quadrature from the tabulated sigma(Xmax) gives corrected values of about 44-48 g/cm^2 in that region, not 57-63. For instance, at 2.68e17 eV, sigma_app is about 44 g/cm^2 and the measured width is 62.3, so the physical width is ~44. The quoted peak matches the raw measured column, not the corrected values. So either the correction is not applied as described or the figure/table definitions are inconsistent. That is load-bearing, because the fluctuation peak is presented as independent confirmation of the composition change. The low-energy bins (below ~5e16 eV) have measured widths close to the estimated resolution, leaving the subtracted physical width essentially unconstrained there; that part of the composition interpretation has no independent fluctuation support.\n\nLesser issues: the per-shower systematic uncertainties (15-55 g/cm^2 in Xmax, 23-26% in energy) are not propagated into the quoted elongation rates or ⟨lnA⟩. The highest-energy bins have 11, 7, and 3 events, so those points carry little weight. The composition inference relies on QGSJetII-04, which is customary, but the resulting model dependence is not discussed in detail.\n\nThe paper deserves a serious referee: the dataset and the mean-Xmax results are worth publishing. But the fluctuation analysis needs a major revision, either to apply the correction properly and report the lower corrected values, or to clearly label the raw widths and avoid the claim of confirmed fluctuations.\n\nRecommendation: send to peer review, but require fixing the fluctuation section before acceptance. The elongation-rate result and the binned data are worth keeping.","headline":"Valuable 40-year Yakutsk dataset with a plausible composition trend, but the fluctuation confirmation in Fig. 7(b) does not match the authors' own correction equations.","tokens_in":10235,"tokens_out":5372,"would_cite":true,"duration_ms":49929,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Yakutsk array data show the cosmic-ray mix swings from heavy to light to heavy across four decades.","keywords":["cosmic rays","mass composition","Xmax","elongation rate","Cherenkov radiation","air shower fluctuations","Yakutsk array","QGSJetII-04"],"falsifier":"Recompute the physical spread of $X_{\\max}$ from the same shower sample using a full Monte Carlo simulation of the Yakutsk array with realistic non-Gaussian measurement errors and event-by-event axis reconstruction, without the linear correction of Eq. (6); if the corrected spread no longer peaks near $2\\times10^{17}$–$2\\times10^{18}$ eV and declines above $5\\times10^{18}$ eV, the fluctuation-based support for a mass-composition change collapses.","tokens_in":9075,"feed_emoji":"🌌","tokens_out":11628,"duration_ms":94976,"temperature":0.7,"pith_summary":"This paper uses 40 years of Cherenkov-light measurements at the Yakutsk array to map the depth of shower maximum $X_{\\max}$ for cosmic rays from $10^{16}$ to $5.7\\times10^{19}$ eV. It finds that the elongation rate—how fast $X_{\\max}$ grows per decade of energy—is not constant: $48\\pm6$, $78\\pm5$, $63\\pm6$, and $50\\pm7$ g/cm$^{2}$ per decade in the successive energy decades. The paper argues this nonlinearity, together with a maximum in $X_{\\max}$ fluctuations near $10^{17}$–$10^{18}$ eV and a decrease above $5\\times10^{18}$ eV, shows the cosmic-ray mass composition changes: heavier nuclei below $10^{17}$ eV, a proton-rich mix around $10^{18}$ eV, and heavier nuclei again above $10^{19}$ eV. The result matters because these are the energies where cosmic rays are thought to shift from Galactic to extragalactic sources, and composition is the main observed handle on that transition.","feed_headline":"Cosmic-ray mix swings heavy-to-light-to-heavy","feed_subtitle":"Forty years of Cherenkov-light showers show the mix turning proton-rich near 10^18 eV, then heavy again.","key_machinery":"The central object is the depth of shower maximum $X_{\\max}$, the atmospheric depth at which the number of shower particles is largest, reconstructed for individual showers from the lateral distribution of Cherenkov light. The reconstruction solves a Fredholm integral equation of the first kind with an adaptive method, folding in the electron lateral-angular distribution and measured atmospheric transmission. Shower energy is set by the Cherenkov light flux density at 200 m from the shower axis, and composition is extracted by interpolating the measured average $X_{\\max}$ between the QGSJetII-04 predictions for proton and iron primaries. The fluctuation analysis subtracts a Monte Carlo-based instrumental-smearing correction from the measured spread of $X_{\\max}$ to isolate the physical shower-to-shower fluctuations.","core_discovery":"The central discovery is that the elongation rate $dX_{\\max}/dE$ is nonlinear, with measured values $48\\pm6$, $78\\pm5$, $63\\pm6$, and $50\\pm7$ g/cm$^{2}$ per decade in the ranges $10^{16}$–$10^{17}$, $10^{17}$–$10^{18}$, $10^{18}$–$10^{19}$, and $10^{19}$–$10^{20}$ eV. The rate peaks near $10^{18}$ eV, and there are kinks in the $X_{\\max}$ versus energy relation near the second knee at $\\sim10^{17}$ eV and again near $5\\times10^{18}$ eV. Fluctuations $\\sigma(X_{\\max})$ rise to 57–63 g/cm$^{2}$ between $2\\times10^{17}$ and $2\\times10^{18}$ eV and then fall to about 40 g/cm$^{2}$ above $2\\times10^{19}$ eV. Using the QGSJetII-04 model to interpolate between proton and iron predictions, the paper derives a mean logarithmic mass $\\langle\\ln A\\rangle$ of about 2.0–2.5 below $10^{17}$ eV, falling to near zero around $3.7\\times10^{18}$ eV, then rising again to about 1.6 by $3.7\\times10^{19}$ eV. The paper concludes that the mass composition of cosmic rays changes twice over the observed energy range: heavy at low energies, proton-dominated near $10^{18}$ eV, and heavy again at the highest energies.","pith_inferences":["Editorial inference: the rising-then-falling elongation rate the paper reports is a direct, energy-resolved constraint on hadronic interaction models; any model used to interpret air showers should reproduce a slope that peaks near $10^{18}$ eV without ad hoc energy-dependent tuning.","Editorial inference: the same dataset could be re-analyzed with newer interaction models to test whether the inferred composition shifts are stable or partly a model artifact of QGSJetII-04.","Editorial inference: the fluctuation peak near $2\\times10^{17}$–$2\\times10^{18}$ eV is a clean target for independent verification by other Cherenkov and radio air-shower arrays operating in that energy range; if they do not see the peak after detector corrections, the composition-change claim would need revision."],"forward_implications":["If the nonlinear elongation rate is correct, the transition from Galactic to extragalactic cosmic rays near 0.1–1 EeV is not compositionally neutral: the mix becomes proton-dominated around $10^{17}$–$10^{18}$ eV before turning heavier.","The breaks in the $X_{\\max}$ versus energy relation place the second knee near $10^{17}$ eV and the beginning of the 'dip-bump' region near $5\\times10^{18}$ eV in the composition, not only in the energy spectrum.","Above $10^{19}$ eV the composition is not pure proton; helium, CNO, and iron-group nuclei make up a substantial fraction, which bears on where and how the highest-energy particles are accelerated.","The fluctuation maximum near $10^{17}$–$10^{18}$ eV indicates a mixed composition with a strong proton and helium component, consistent with the mean-depth interpolation."],"supporting_citations":[{"why":"Supplies the reconstruction algorithm that turns measured Cherenkov lateral distributions into individual cascade curves and $X_{\\max}$.","marker":"[7]"},{"why":"Provides the energy-balance method and the Q(200)-to-energy calibration used to assign each shower's energy.","marker":"[10]"},{"why":"Defines the QGSJetII-04 hadronic interaction model whose proton and iron $X_{\\max}$ predictions anchor the composition interpolation.","marker":"[11]"},{"why":"Gives the interpolation formula used to convert measured average $X_{\\max}$ into mean logarithmic mass.","marker":"[12]"},{"why":"Supplies independent $X_{\\max}$ measurements used for cross-checking the average depth values in Fig. 7(a).","marker":"[42]"},{"why":"Supplies a second experiment's $X_{\\max}$ measurements used for the same cross-check in Fig. 7.","marker":"[43]"},{"why":"Underpins the Monte Carlo estimate of instrumental $X_{\\max}$ smearing used in Eq. (6).","marker":"[50]"},{"why":"Supplies the dependence of instrumental smearing on shower geometry and station count used in Eq. (6).","marker":"[51]"}],"fun_headline_variants":["Cosmic-ray composition flips heavy-light-heavy","Yakutsk array reveals proton-rich peak at 1 EeV","Nonlinear elongation points to changing cosmic-ray mass","Heavy to light to heavy: cosmic-ray mix over decades"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that the fluctuation pattern confirms a composition change rests on the detector-smearing correction being right: if the correction subtracted from the measured spread is biased, the peak near $10^{17}$–$10^{18}$ eV and the drop above $5\\times10^{18}$ eV could be artifacts, and the composition argument would lose its independent support.","fun_headline_variants_meta":{"raw":{"variants":["Cosmic-ray composition flips heavy-light-heavy","Yakutsk array reveals proton-rich peak at 1 EeV","Nonlinear elongation points to changing cosmic-ray mass","Heavy to light to heavy: cosmic-ray mix over decades"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00027,"raw_usage":{"total_tokens":1675,"prompt_tokens":1045,"completion_tokens":630,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":661,"completion_tokens_details":{"reasoning_tokens":564}},"tokens_in":661,"tokens_out":630,"duration_ms":7150,"temperature":1.0,"reasoning_tokens":564,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:10:23.934876+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the physical spread of $X_{\\max}$ from the same shower sample using a full Monte Carlo simulation of the Yakutsk array with realistic non-Gaussian measurement errors and event-by-event axis reconstruction, without the linear correction of Eq. (6); if the corrected spread no longer peaks near $2\\times10^{17}$–$2\\times10^{18}$ eV and declines above $5\\times10^{18}$ eV, the fluctuation-based support for a mass-composition change collapses.","supporting_citations":[{"cited_title":"Knurenko, V","cited_arxiv_id":null,"evidence_quote":"Supplies the reconstruction algorithm that turns measured Cherenkov lateral distributions into individual cascade curves and $X_{\\max}$."},{"cited_title":"Knurenko, A","cited_arxiv_id":null,"evidence_quote":"Provides the energy-balance method and the Q(200)-to-energy calibration used to assign each shower's energy."},{"cited_title":"Ostapchenko, Phys","cited_arxiv_id":null,"evidence_quote":"Defines the QGSJetII-04 hadronic interaction model whose proton and iron $X_{\\max}$ predictions anchor the composition interpolation."},{"cited_title":"H¨ orandel, J","cited_arxiv_id":null,"evidence_quote":"Gives the interpolation formula used to convert measured average $X_{\\max}$ into mean logarithmic mass."},{"cited_title":"Bellido, A","cited_arxiv_id":null,"evidence_quote":"Supplies independent $X_{\\max}$ measurements used for cross-checking the average depth values in Fig. 7(a)."},{"cited_title":"Abbasi, M","cited_arxiv_id":null,"evidence_quote":"Supplies a second experiment's $X_{\\max}$ measurements used for the same cross-check in Fig. 7."},{"cited_title":"Dyakonov, A","cited_arxiv_id":null,"evidence_quote":"Underpins the Monte Carlo estimate of instrumental $X_{\\max}$ smearing used in Eq. (6)."},{"cited_title":"Dyakonov, A","cited_arxiv_id":null,"evidence_quote":"Supplies the dependence of instrumental smearing on shower geometry and station count used in Eq. (6)."}],"review_version":1}