{"id":"bddd3017-1b20-4897-9429-5a17622198e5","arxiv_id":"2608.05288","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The composition of small S-complex near-Earth asteroids shifts from LL-chondrite-like to L-chondrite-like below about 40 meters, with H and LL parent bodies constrained to be smaller than about 18 meters.","lead":"Astronomers measured near-infrared spectra of 80 near-Earth asteroids spanning 4 to 343 meters and found that composition shifts with size: L-chondrite-like objects dominate below roughly 40 meters, matching meteorite fall statistics. This is the first observational evidence that the well-known mismatch between asteroid spectra and meteorite falls is a size effect.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sample representativeness is the load-bearing risk: the 15-object 4-31 m bin is a target-of-opportunity sample with no completeness function, and the paper's own NEOMOD source-region analysis is not used to correct the reported H/L/LL fractions.","rationale":"The paper presents a genuinely interesting dataset and the qualitative trend is consistent with an independent dynamical prediction, so I do not see an internal inconsistency or a reason to reject. The most load-bearing vulnerability is not the multinomial classifier or the assumed albedos; it is that the target list is a convenience sample of close-approaching NEOs. The paper's own Section 6 shows the sub-100 m sample is dominated by ν6 orbits, and ν6 is the resonance predicted to deliver Massalia-family L chondrites. If the discovery and targeting process overweights such orbits, the observed L excess could be produced without any true size-composition trend. The source-region probabilities are computed but never used to correct the reported fractions, so the quantitative claims (31-49 m crossover, <18 m H/LL limits) rest on an implicit assumption of representativeness. A forward-model test with NEOMOD and an explicit selection filter would settle this directly. Since the reader already identified this weakness and issued CONDITIONAL, my independent read does not change the verdict; it sharpens the condition under which the claim would be accepted: demonstrate that the observable target list is representative, or correct the fractions using the selection function.","tokens_in":15946,"tokens_out":7560,"duration_ms":634338,"concrete_test":"Use the NEOMOD simulator to generate the full synthetic NEO population with H≥24.7, assign each object an ordinary-chondrite subtype using the Marsset et al. (2024) source-region mapping (ν6/Massalia→L, Flora/Nysa→LL, Koronis/Phocaea→H), apply the IRTF target-selection filter used in this program (apparent V≤18.5, favorable geocentric distance and solar elongation), and compare the predicted observable H/L/LL fractions with Table 2's 12.8/62.1/25.1. If the selection filter alone moves a constant-composition null (43/47/10) toward the observed fractions, the size-dependence claim is not robust; if the observed fractions are reproduced only when an intrinsic size-composition trend is included, the representativeness concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central size-dependence claim rests on the smallest bin in Table 2 (24.7≤H≤29.2, 15 objects, diameters ~4-31 m). These objects were not drawn from a survey with known completeness; they were targeted because they made close, bright approaches. If close-approach accessibility preferentially samples ν6-resonance orbits, and the paper's own Section 6 finds ν6 contributes ~64% of its sub-100 m sample, then the apparent jump in L-chondrite-like objects (62.1±12.4%) and the deficits of H (12.8±8.8%) and LL (25.1±11.3%) relative to meteorite falls could be a selection artifact rather than a physical size-composition trend. The NEOMOD source-region probabilities are computed but never used to re-weight the compositional fractions, so the reported numbers are raw sample proportions. The crossover size (31-49 m) and the ~18 m upper limits are derived from these raw fractions, so the quantitative conclusions are contingent on the unquantified representativeness of the target list. The two-subgroup comparison (26.6±6.6% vs 48.4±8.6% L) reduces but does not remove this concern, because both subgroups come from the same target-of-opportunity pool and the same selection mechanism could vary with size.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper combines 25 new near-infrared spectra of S-complex NEOs with 55 previously published spectra to build an 80-object sample spanning absolute magnitudes 20.0 to 29.2 (roughly 4-343 m). The sample is subdivided by H magnitude into two and three subgroups, each object is assigned an H, L, or LL chondrite-like composition using band parameters and a multinomial logistic regression classifier, and the observed subtype fractions are corrected for classifier confusion. The authors report that LL-chondrite-like objects dominate the largest subgroups, L-chondrite-like objects become dominant in the smallest subgroup (24.7 ≤ H ≤ 29.2, 62.1 ± 12.4%), and the H-chondrite fraction remains far below the meteorite-fall proportion, leading to an upper size limit of ~18 m for H and LL parent bodies and a crossover to L dominance at ~31-49 m. Section 6 further analyzes the source regions of the sub-100 m sample with NEOMOD.","tokens_in":16191,"tokens_out":6158,"duration_ms":52422,"significance":"The paper addresses a long-standing discrepancy between the LL-dominated compositions of larger S-complex NEOs and the H/L-dominated ordinary chondrite fall statistics. Its strengths include new spectroscopic observations of 25 small NEOs, an explicit confusion-matrix correction for classification bias, and a NEOMOD-based source-region analysis. If the size-composition trend is real, it constitutes the first observational confirmation of the predicted increase in L-chondrite-like objects at small sizes and provides useful size constraints on meteorite parent bodies. However, the quantitative claims currently rest on a target-of-opportunity sample without a demonstrated completeness function, and some of the headline limits are not statistically robust.","major_comments":[{"comment":"The central size-composition trend is computed from raw proportions in target-of-opportunity subgroups, and no correction or bounding argument is given for selection effects. The smallest bin (24.7 ≤ H ≤ 29.2, 15 objects) is not drawn from a survey with known completeness; these objects were observed because they made close, observable approaches. Section 6 shows that the sub-100 m sample is dominated by ν6-resonance delivery (~64%), and the paper notes an apparent concentration of small L-chondrite-like NEOs below the ν6 resonance. If orbital accessibility preferentially selects ν6-delivered material, the observed rise in L fraction and the derived crossover at H ≈ 24.5 could be in part a selection artifact. The NEOMOD probabilities are used to characterize source regions but are not used to re-weight the H/L/LL fractions. The authors should either demonstrate that the selection function is composition-independent or provide re-weighted estimates and show how the crossover and upper limits change.","section":"Section 3, Table 2; Section 6"},{"comment":"The claimed upper size limit of ~18 m for LL-chondrite parent bodies is not statistically supported by the reported numbers. The smallest-bin LL fraction is 25.1 ± 11.3%, which is only about 1.3σ above the 10% meteorite-fall proportion. The statement that 'the size threshold ... has still not been reached' and the resulting ~18 m upper limit therefore rest on a non-significant excess. The H-chondrite deficit (12.8 ± 8.8% vs 43%) is significant, but the LL limit should be removed or replaced with a statement that the LL fraction is statistically consistent with the fall proportion.","section":"Section 5, Table 2; Section 7"},{"comment":"The quoted crossover diameter of 40 m (1σ range 31-49 m) is obtained by linear interpolation between the median diameters of the two adjacent bins, but the propagation described in the text appears to include only the uncertainty in the L-chondrite fractions, not the systematic uncertainty in converting H to diameter from assumed geometric albedos (0.15-0.25, Section 3). Since diameter scales as albedo^{-1/2}, the assumed albedo range alone changes median sizes by roughly 10-30%, comparable to the quoted 31-49 m range. The authors should state whether albedo systematics are included in the quoted range and, if not, re-derive the crossover with them.","section":"Section 5, Fig. 9"},{"comment":"The multinomial logistic regression classifier uses three input features (ol/(ol+px), Fa, Fs), but several objects in Table 4 have no measured BAR and therefore no ol/(ol+px) (e.g., 2023 PM, 2023 VR4, 2024 BH, 2024 OL1, 2025 OL1). The paper does not state how missing features are handled in the classifier or whether these objects were excluded. Because these objects are still assigned H/L/LL types in Table 3, the classification pipeline for missing data must be described; otherwise the reported fractions, including those in the smallest bin, are not reproducible.","section":"Section 5, Table 4; Section 5, ML classifier description"}],"minor_comments":[{"comment":"The phrase 'first observational evidence' is strong given the selection caveats; consider softening to 'observational evidence' unless the sample representativeness is established or explicitly corrected.","section":"Abstract"},{"comment":"The caption should state explicitly that the reported percentages are the bias-corrected Ptrue values and should define the error bars as the quadrature combination of Wilson-score sampling uncertainties and confusion-matrix systematic uncertainties.","section":"Table 2 caption"},{"comment":"For the source-region ratios in Figures 11-13, please state the number of objects per subtype used in each ratio and whether the subtype assignments used in these figures are the same hard classifications as in Table 3 or probability-weighted assignments.","section":"Section 6"},{"comment":"The slit width appears as '0.8”' with a right quotation mark; this should be typeset as 0.8 arcsec.","section":"Section 2"},{"comment":"Several in-press references (Marsset et al. 2026; Vokrouhlický et al. 2026) would benefit from arXiv identifiers or DOI links to help readers assess the cited results.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The manuscript contains valuable new data and a plausible physical trend, but its quantitative conclusions depend on an unmodeled selection function and on at least one statistically weak inference (the LL upper limit). These issues are fixable within the manuscript's scope by re-weighting or reframing the claims, so I recommend major revision rather than rejection. If the authors cannot access a complete survey sample or estimate selection biases, the headline size limits should be presented as tentative rather than as established constraints."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline is that this paper delivers the first observational evidence that S-complex NEO composition depends on size: LL-dominated at a few hundred meters, L-dominated below ~40 m, with the L fraction crossing the meteorite fall fraction around 31-49 m. If that holds, it resolves a long-standing tension between NEO spectroscopy and fall statistics. The analysis is mostly careful: the spectral reduction follows established procedures, the multinomial logistic regression is trained on meteorite standards, and they correctly invert the confusion matrix to correct observed fractions. The uncertainties are propagated from both sampling and misclassification, and they are explicit that the 18 m limits for H and LL are upper bounds, not detections of the parent-body population itself. The source-region analysis with NEOMOD is a nice addition and matches the dynamical prediction that small L-like NEOs come preferentially from the nu6 resonance.\n\nThe main soft spot is exactly the one flagged in the stress-test: the smallest bin (15 objects, ~4-31 m) is a target-of-opportunity sample with no completeness function. Objects that make close, bright approaches are not a random draw from the NEO population, and orbital accessibility correlates with source region. The paper computes NEOMOD source-region probabilities but does not use them to re-weight the reported H/L/LL fractions, so the quoted numbers are raw sample proportions. The crossover size and the 18 m limits are derived from those raw proportions. This does not kill the central trend - the two-subgroup comparison (26.6% vs 48.4% L) also moves in the right direction, and the error bars on the largest and smallest groups barely overlap - but it does mean the quantitative size boundaries are less secure than the qualitative conclusion.\n\nA second, smaller issue: the linear interpolation between two bin medians to get the crossover relies on a monotonicity assumption that is reasonable but unproven. They acknowledge this.\n\nOverall, this is a solid, honest paper. It deserves peer review. The referee should push for a sensitivity analysis that re-weights the fractions using the NEOMOD probabilities, or at least a discussion of how selection could bias the trend. I would not desk-reject it.\n\nBest,\n[Your name]","headline":"First credible observational evidence for a size-composition trend in S-complex NEOs, but the quantitative size limits rest on a small target-of-opportunity sample with an unquantified selection function.","tokens_in":16816,"tokens_out":1885,"would_cite":true,"duration_ms":17457,"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 presents the first observational evidence that the composition of S-complex near-Earth objects depends on size, with L chondrite-like bodies replacing LL-dominated compositions at about 31-49 meters.","keywords":["near-Earth objects","S-complex asteroids","ordinary chondrites","H chondrites","L chondrites","LL chondrites","near-infrared spectroscopy","pre-atmospheric parent bodies"],"falsifier":"Run a completeness-corrected near-infrared spectroscopic survey of every S-complex NEO discovered in a defined magnitude-limited volume between roughly 4 and 70 m, with no preselection by approach distance, and compare the H, L, and LL fractions: the claimed sub-18 m upper limits require H and LL fractions to rise toward 43% and 10% at smaller sizes, and the claimed 31-49 m crossover requires L to sit near 47% in that unbiased sample.","tokens_in":15707,"feed_emoji":"☄️","tokens_out":10735,"duration_ms":90609,"temperature":0.7,"pith_summary":"Ordinary chondrites, the most common meteorite type, come in three subtypes — H, L, and LL — but near-Earth asteroids that resemble them have looked mostly LL-like, even though LL chondrites are only 10% of meteorite falls. The paper argues that the mismatch is a size effect: previous studies looked at asteroids too large to be the immediate parents of meteorites. Using near-infrared spectra of 80 S-complex near-Earth objects spanning roughly 4 to 343 meters, it shows that LL dominance disappears at tens of meters, L-like objects take over around 31-49 meters at the same 47% share seen among meteorite falls, and H and LL parent bodies must be smaller than about 18 meters. This is the first reported observational evidence that the composition of S-complex NEOs depends on size, and it locates the meteorite-producing population at sizes that current spectroscopic surveys have barely sampled.","feed_headline":"LL dominance vanishes near 40 meters, L chondrites take over","feed_subtitle":"A spectral survey of sub-100 m near-Earth objects matches asteroid compositions to meteorite falls for the first time.","key_machinery":"The argument is carried by a size-binned compositional census. Eighty S-complex NEOs are split by absolute magnitude, converted to diameter using an assumed albedo, and each spectrum is reduced to band centers and a band-area ratio, then to fayalite ($\\mathrm{Fa}$), ferrosilite ($\\mathrm{Fs}$), and olivine/(olivine+pyroxene) values. A multinomial logistic regression trained on those compositional variables assigns each object a probability of being H, L, or LL, and the classifier's confusion matrix is inverted so the reported subtype fractions are corrected for misclassification. The crossover diameter is obtained by linear interpolation between the median diameters of the two adjacent magnitude bins that bracket the 47% L-chondrite fall fraction.","core_discovery":"The paper establishes a size-composition trend in S-complex near-Earth objects, the silicate-rich asteroids thought to be the parent bodies of ordinary chondrites, and reads it as a record of where ordinary chondrite meteorites come from. In the $20.0\\le H\\le 22.1$ subgroup (roughly 91-343 m), LL chondrite-like objects make up $65.4\\pm7.5\\%$ of the sample; in the $24.7\\le H\\le 29.2$ subgroup (roughly 4-31 m), the L chondrite-like fraction rises to $62.1\\pm12.4\\%$, matching the 47% L-chondrite share of meteorite falls, while the H and LL fractions ($12.8\\pm8.8\\%$ and $25.1\\pm11.3\\%$) still exceed the fall proportions of 43% and 10%. A linear interpolation between adjacent bins places the L crossover at $H=24.5$, corresponding to about 31-49 m. Because the H and LL fractions have not yet approached their meteoritic proportions even at 4-31 m, the paper sets an upper size limit of about 18 m for the pre-atmospheric parent bodies of H and LL chondrites, and concludes that the long-standing overabundance of LL-like NEOs is a size-dependent phenomenon rather than a contradiction of meteorite statistics.","pith_inferences":["Because the sample's smallest objects were selected by close-approach opportunity rather than by a completeness-corrected survey, the size trend could be sharpened or weakened once detection biases are modeled; the paper reports source-region statistics but does not use them to re-weight the compositional fractions.","The crossover size is probably not a universal constant: if young-family transport sets it, the threshold should depend on family age and resonance geometry, so older families may produce a different transition diameter.","If H and LL parent bodies really are mostly below about 18 m, the H and LL meteorite flux should be tied to the smallest end of the NEO size distribution, and fireball networks may be a better probe of their abundances than NEO spectroscopy."],"forward_implications":["The apparent overabundance of LL-like S-complex NEOs among objects larger than about 100 m no longer needs a special explanation; it is the large-size end of a trend that reaches meteorite-fall proportions only below tens of meters.","Most pre-atmospheric parents of H and LL chondrite falls are smaller than about 18 m, so the population that actually feeds these meteorites is nearly invisible to current spectral surveys.","S-complex NEOs smaller than about 40 m should be dominated by L chondrite-like compositions, with L parent bodies concentrated in the roughly 18-60 m range.","Sub-100 m S-complex NEOs are supplied mainly through the $\\nu_6$ resonance (about 64%) and the 3:1 resonance (about 25%), matching the expectation that young-family fragments are delivered through those escape routes.","An object like Chelyabinsk, roughly 20 m across, is probably rare among LL-chondrite precursors if the about-18 m upper limit is correct."],"supporting_citations":[{"why":"Supplies the 55 previously observed S-complex NEO spectra, the Sx subclass definition, and the machine-learning classification pipeline that this paper extends to a larger size range.","marker":"Sanchez et al. 2024"},{"why":"Documents the LL-dominated composition of larger S-complex NEOs and the source-region results that the size-dependence claim must reconcile with meteorite-fall statistics.","marker":"Binzel et al. 2019"},{"why":"Predicted that L chondrite-like NEOs should increase below about 100 m through Massalia-family delivery; this is the specific prediction the new observations test and confirm.","marker":"Marsset et al. 2024"},{"why":"Provides the fayalite and ferrosilite calibrations, the ordinary-chondrite subtype boundaries, and the uncertainty values used to train the classifier.","marker":"Sanchez et al. 2020"},{"why":"Supplies the absolute-magnitude-to-diameter relation used to convert H bins into the reported size ranges.","marker":"Pravec & Harris 2007"},{"why":"Defines the S(IV) subtype region on the Band I center versus Band Area Ratio diagram that anchors ordinary-chondrite identification.","marker":"Gaffey et al. 1993"},{"why":"Provides the NEOMOD source-region probability model used to assign escape regions for the sub-100 m NEOs.","marker":"Nesvorný et al. 2023"},{"why":"Identifies young asteroid families and resonance transport that explain why small objects reach the ν6 resonance efficiently, providing the dynamical rationale for size-dependent composition.","marker":"Brož et al. 2024"}],"fun_headline_variants":["Size limit set for parent bodies of ordinary chondrites","LL chondrite dominance ends at ~40 m, L takes over","Meteorite parent bodies capped at ~18 m","Size-dependent shift in asteroid composition matches meteorites","Where do ordinary chondrites come from? Size answer"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The trend rests on the assumption that the sub-100 m NEOs observed here represent the underlying NEO population, even though they were targeted because they made close approaches; if close-approach accessibility favors certain source-region histories, the size trend could be partly an observing bias.","fun_headline_variants_meta":{"raw":{"variants":["Size limit set for parent bodies of ordinary chondrites","LL chondrite dominance ends at ~40 m, L takes over","Meteorite parent bodies capped at ~18 m","Size-dependent shift in asteroid composition matches meteorites","Where do ordinary chondrites come from? Size answer"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000222,"raw_usage":{"total_tokens":1558,"prompt_tokens":1151,"completion_tokens":407,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":767,"completion_tokens_details":{"reasoning_tokens":328}},"tokens_in":767,"tokens_out":407,"duration_ms":3613,"temperature":1.0,"reasoning_tokens":328,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T16:01:15.806900+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a completeness-corrected near-infrared spectroscopic survey of every S-complex NEO discovered in a defined magnitude-limited volume between roughly 4 and 70 m, with no preselection by approach distance, and compare the H, L, and LL fractions: the claimed sub-18 m upper limits require H and LL fractions to rise toward 43% and 10% at smaller sizes, and the claimed 31-49 m crossover requires L to sit near 47% in that unbiased sample.","supporting_citations":[{"cited_title":"P., DeMeo, F","cited_arxiv_id":null,"evidence_quote":"Documents the LL-dominated composition of larger S-complex NEOs and the source-region results that the size-dependence claim must reconcile with meteorite-fall statistics."},{"cited_title":"J., Burbine, T","cited_arxiv_id":null,"evidence_quote":"Defines the S(IV) subtype region on the Band I center versus Band Area Ratio diagram that anchors ordinary-chondrite identification."}],"review_version":1}