{"id":"04da5b84-681a-45b2-b7b0-1146bc80da18","arxiv_id":"2412.07023","paper_version":3,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":6,"one_line_summary":"A proposed 229/48 x 63/57 x 56/22 gear train would make the Draconic pointer turn 242.00029 times per Saros, and revised ecliptic limits are then used to classify the preserved eclipse events.","lead":"This paper proposes a revised set of gears for the Antikythera Mechanism that would have tracked the Moon's draconic cycle, and uses that scheme to classify the eclipse events engraved on the Saros spiral. The scheme is a reconstruction: none of the new gears survive, and the authors adjust eclipse limits to match the preserved inscriptions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central ratio depends on three unpreserved gears and a shrinkage-corrected b1 count; with a second equally valid gear set and limits fitted to the inscriptions, the 'mechanical generation' claim is underdetermined.","rationale":"The reader's weakest-assumption analysis correctly identifies the load-bearing point: the revised Draconic gearing depends on unpreserved gears and a shrinkage-corrected b1 count selected to make the Saros ratio work. My independent reading confirms this is the most serious weakness. The paper's own text repeatedly calls s1, s2, and t1 'hypothetical,' and Equation (2) shows that a different gear set can produce the same numerical result, which undermines the claim that the specific proposed train was actually present. The calibration of the ecliptic limits to the preserved events adds a second, compounding issue: the match between the model and the inscriptions is not an independent test of the model because the limits were adjusted to fit those same inscriptions. The paper does contain real evidence — the preserved gear r1 on Fragment D, the correct arithmetic of the gear ratios, and the explicit acknowledgement that the scheme is a hypothesis — but the central claim that the eclipse events were 'calculated by pure mechanical processing and not documented observed events' is an overstatement given the current evidence. A focused metrological test on the X-ray data could settle whether the proposed train is physically possible; until then, the reader's REJECT verdict stands.","tokens_in":30581,"tokens_out":3923,"duration_ms":43114,"concrete_test":"Perform a metrological analysis of the X-ray tomography volumes of Fragments A and D: (i) measure gear-b1's tooth count from the preserved tooth spacing on its pitch circle without applying a shrinkage correction; (ii) test whether a shaft carrying s1 (57) and s2 (56) can be physically located between r1 (63) on shaft-a and t1 (22) on the right side of the casing without intersecting preserved components. If b1's independently measured count is not 229/228, or if the 57/56/22 train cannot be placed in the available space, Equations (1) and (2) are numerical coincidences rather than a reconstruction of the mechanism.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim — that the Antikythera Mechanism mechanically pre-calculated the engraved eclipse events rather than recording observed ones — rests on Equation (1), which requires unpreserved gears s1 (57 teeth), s2 (56 teeth), t1 (22 teeth), and an assumed b1 tooth count of 229 (or 228 in option 2). None of these is attested: s1, s2, and t1 are entirely hypothetical, and b1 = 229 is obtained by applying a uniform 3% shrinkage correction to published measurements ranging from 216 to 231, with the value chosen to make the Saros ratio 242.000. Section 3.2 then offers a second, equally admissible gear set (Equation 2: 228/48, 63/34, 61/40) that reaches nearly the same ratio with different unpreserved gears; this multiplicity shows the tooth counts are fitted, not derived from preserved evidence. Furthermore, the ecliptic limits in Section 5 are explicitly 'revised' to match the preserved eclipse events, so the reported agreement between DracoNod-V2 and the inscriptions is partly circular. The four residual mismatches (Table 3) are then absorbed by invoking 'gearing errors,' making the scheme difficult to falsify. Even granting the arithmetic of Equations (1)-(2), the strong conclusion that the events were generated by pure mechanical processing is not supported without physical evidence for the hypothetical gears or an independent determination of b1's tooth count.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a revised Draconic gearing for the Antikythera Mechanism, correlating Fragment D with Fragment A and introducing three unpreserved gears (s1, s2, t1). Equation (1) yields 242.0002901 turns of the Draconic pointer per 223 synodic months, matching the Saros relation. The authors use this gearing with a recalibrated DracoNod-V2 program to regenerate the eclipse events of the Saros spiral, claiming that the events were mechanically pre-calculated rather than recorded observations. They also propose an eclipse classification based on Eudoxus papyrus and the Back Plate Inscription, and report a revised set of ecliptic limits and sector boundaries for the Draconic scale.","tokens_in":30921,"tokens_out":4466,"duration_ms":45148,"significance":"If the central claim were established, it would settle a long-standing question in Antikythera research: whether the engraved eclipse sequence was an observational record or a mechanically generated prediction. The paper has the merit of making an explicit, quantitative gear-ratio proposal (Eqs. 1–3), engaging with the physical deformation and shrinkage of the fragments, and providing a detailed, letter-by-letter comparison of the Back Plate Inscription with a running simulation. However, the strength of these virtues is limited by the fact that the proposed gear train contains three entirely hypothetical wheels, the key gear-b1 tooth count is assumed after a uniform shrinkage correction, and the simulation's ecliptic limits are fitted to the very inscriptions the paper claims to predict.","major_comments":[{"comment":"The 242-turn ratio is achieved by setting b1 = 229 (or 228 in Eq. 2) and by postulating gears s1 = 57, s2 = 56, t1 = 22. Table 1 shows measured b1 estimates between 216 and 231 teeth; the value 229 is not independently derived but is chosen 'to improve its precision' (Section 3.1). Since Eq. (2) reaches nearly the same ratio with a different set of unpreserved gears (63/34, 61/40), the tooth counts are evidently fitted rather than fixed by evidence. The central claim of the paper therefore rests on an underdetermined parameter set.","section":"§3.1, Eq. (1)"},{"comment":"The ecliptic limits in Section 5 are explicitly revised 'in order to improve the best match to the preserved eclipse events.' Consequently, the agreement between DracoNod-V2 and the inscriptions in Table 3 is to a significant degree a calibration artifact. The four residual mismatches (Cells 130, 166, 177, 189) are attributed to 'gearing errors' without a quantitative error budget, which prevents the reader from testing whether the proposed gearing plus errors is actually consistent with the data. As it stands, the argument is circular.","section":"§5, Table 3"},{"comment":"The physical feasibility of the proposed gear train is asserted but not demonstrated. The paper does not provide dimensional data from Fragment A or Fragment D showing that gears with 57, 56, and 22 teeth fit in the space between the casement and mesh with gear-r1 as drawn. Since these gears are entirely hypothetical, the claim that this is the Mechanism's original Draconic gearing requires evidence that the parts existed and could mesh, not merely that the ratio works.","section":"§3.1, Fig. 2"},{"comment":"The paper's conclusion that 'the eclipse events are calculated by pure mechanical processing and that they are not documented observed events' (Abstract) is not supported by the evidence presented. The preserved events are used to calibrate both the gearing (b1 tooth count) and the ecliptic limits; showing that a fitted model reproduces the calibration data does not demonstrate that the ancient instrument generated the events mechanically.","section":"Abstract, §8 (Epilogue)"}],"minor_comments":[{"comment":"The phrase 'the Mechanism s parts' contains a typo; it should read 'the Mechanism's parts.'","section":"Abstract"},{"comment":"The transition from Eq. (2) to Eq. (3) is confusing: Eq. (2) uses (61/40), but Eq. (3) replaces t1 = 40 with t1 = 20 to double the turns; this should be stated more explicitly for the reader.","section":"§3.2, Eqs. (2)–(3)"},{"comment":"The definitions of 'Ecliptic Zone A' and 'Ecliptic Zone B' are given in a paragraph below Fig. 7; consider presenting them in a table or as a bulleted list to improve clarity.","section":"§5, Fig. 7"},{"comment":"The sector labels 'NL, NF, NCA, SCA, SF, SL' are used without an explicit glossary; define them at first use.","section":"§7, Table 5"},{"comment":"The reference 'Voulgaris et al., 203b' appears to be a typo for 'Voulgaris et al., 2023b'; please correct this and check for similar citation errors throughout.","section":"§4"}],"recommendation":"reject","confidential_remarks":"The manuscript is a speculative reconstruction paper whose central claim is not supported by the evidence as presented. The strongest result, the arithmetic of Eq. (1), is correct but does not identify the actual gearing. The paper might be publishable as a hypothesis if reframed with explicit caveats and a falsifiable error model, but in its current form the conclusions outrun the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a detailed, internally coherent proposal for the Draconic gearing of the Antikythera Mechanism. The two gear-train equations and the six-sector eclipse classification tied to the Eudoxus papyrus are genuinely new, and the authors are transparent that some gears (s1, s2, t1) are hypothetical. They also build physical bronze reconstructions and engage carefully with the X-ray data and the inscriptions. The treatment of the cursive omega is a concrete, checkable contribution.\n\nThat said, the central claim does not hold up as stated. Equation (1) depends on b1 having 229 teeth and on three unattested gears with tooth counts 57, 56, and 22. The b1 value is a shrinkage-corrected estimate chosen from a range of measured values, and the hypothetical gears are introduced specifically so the train yields 242.000 draconic turns per Saros. In other words, the known Saros relation is built into the gear counts, so the 'prediction' that the pointer returns after 242 turns is not an independent result. Equation (2) offers an equally valid alternative configuration, which makes the specificity of the reconstruction weaker.\n\nSection 5 is a bigger problem. The ecliptic limits are explicitly recalibrated to improve the match with the preserved eclipse events. Using the same data to set the boundaries and then reporting agreement with those events is circular. The four residual mismatches are then attributed to gearing errors, which is plausible only if the error budget is independently constrained. The paper gives typical error magnitudes but does not show they specifically cover these cases.\n\nSo the strong conclusion that the eclipse events were generated by pure mechanical processing rather than recorded observations is not supported by the current evidence. A more careful version would separate what is attested from what is assumed, treat the gearing as a hypothesis, and either find independent support for the hypothetical gears or soften the claim.\n\nWho benefits from this paper? Someone working on the Antikythera Mechanism or on ancient astronomical instruments will find the proposed gear trains and the classification scheme a useful prompt for further testing. It deserves a serious referee, because the topic is important and the paper is substantive, even if the referee is likely to demand major revisions or a narrower conclusion.","headline":"A serious but over-fitted reconstruction: the new gear ratios and six-sector eclipse classification are worth engaging with, but the central claim of pure mechanical generation rests on unpreserved gears and on limits fitted to the same inscriptions they are said to predict.","tokens_in":647,"tokens_out":929,"would_cite":false,"duration_ms":41198,"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 revised 229-tooth gear train reproduces the Antikythera Mechanism's Saros eclipse events from gear ratios alone, the paper argues.","keywords":["Antikythera Mechanism","Draconic gearing","eclipse prediction","Saros spiral","ecliptic limits","gear errors","Fragment D","ancient Greek astronomy"],"falsifier":"A decisive check would be to measure every surviving gear-tooth arc on Fragment A and reconstruct the original $b1$ tooth count from the measured radii; if that count comes out below 228 (several current measurements give 216–226), or if a physical bronze model of the proposed $s1$–$s2$–$t1$ gear cluster cannot fit between the internal and external casements on the right side of Fragment A, the proposed gearing relation is falsified. In addition, an independent recalculation using the exact ancient value 5458 synodic = 5923 Draconic, which gives 241.998 rather than 242.000 turns per Saros, would show whether the paper's claimed error of +0.00029 turns per Saros is genuinely negligible across the spiral's 50-event span.","tokens_in":30330,"feed_emoji":"⚙️","tokens_out":6195,"duration_ms":60062,"temperature":0.7,"pith_summary":"The paper argues that a missing train of four gears—three of them entirely hypothetical—once sat on the right side of the Antikythera Mechanism and drove a Draconic pointer giving the Moon's latitude relative to the ecliptic. With gear counts chosen so that one Saros equals about 242.000 Draconic turns, the eclipse events engraved on the Saros spiral, including events in cells whose inscriptions are lost, can be generated by turning the mechanism itself rather than by recording observed eclipses. If this is right, the Mechanism was a self-contained eclipse computer: preserved and missing eclipse events, and the classification words on the Back Plate, follow from gearing alone. The paper also uses a 3% shrinkage correction for the largest gear to explain why two engraved eclipse pairs are absent from the mechanical prediction and two extra events appear: marginal ecliptic-limit positions plus gear errors.","feed_headline":"Lost Antikythera gears make eclipse events purely mechanical","feed_subtitle":"A revised 229-tooth gear train reproduces preserved and missing Saros eclipse events from mechanism motion alone.","key_machinery":"The load-bearing object is the proposed Draconic gear train $b1 \\to a1 \\to r1 \\to s1 \\to s2 \\to t1$, with tooth counts 229, 48, 63, 57, 56, and 22. This train locks the Draconic cycle to the Saros so that 223 synodic months produce almost exactly 242 turns of the Draconic pointer. A second object is the revised ecliptic-limit scale, divided into a Node area and six unequal sectors labeled with the Greek magnitude words Minor, Medium, Large, Greatest, and Major, which converts the pointer's position into an eclipse classification. A third object is the 3% linear shrinkage estimate derived from the 354 versus 365 calendar-ring hole count, which is what justifies raising the annual gear's tooth count to 229.","core_discovery":"The central claim is that the Antikythera Mechanism encoded the Draconic lunar cycle through a revised gear train expressed by Equation (1): $18.029787234 \\times \\{(229/48) \\times (63/57) \\times (56/22)\\} = 242.0002901$ Draconic-pointer turns per Saros, with a second option using a 228-tooth $b1$ gear. The train starts from the annual gear $b1$, runs through the preserved crown gear $a1$ and the Fragment D gear $r1$, then through three unpreserved gears $s1$, $s2$, and $t1$, ending at a Draconic pointer mounted on the right side of the casing. With revised ecliptic limits—6.5° South and 22.5° North around Node-A and 5.7° North and 15.3° South around Node-B—a recalculation reproduces the preserved eclipse events and supplies events from lost cells. The paper concludes that the eclipse events are therefore calculated by pure mechanical processing and are not documented observed events, and that the four discrepancies (two missing and two additional events) are explained by marginal pointer positions and endogenous gear errors.","pith_inferences":["Beyond the paper: the same shrinkage argument used to raise the $b1$ tooth count to 229 could be tested against other gear diameters in Fragment A; if shrinkage is not uniform, the chosen tooth counts are an average rather than a unique reconstruction.","Beyond the paper: the claim that the events are 'not documented observed events' is stronger than the evidence requires; the mechanism could still have been calibrated or checked against earlier Babylonian eclipse records even if its daily operation was purely mechanical.","Beyond the paper: the six-sector classification scheme is a testable model—applying it to another Saros series and comparing the predicted sector order with an independent ancient eclipse catalog would show whether the sector widths in Table 5 are stable or merely fitted to this one spiral.","Beyond the paper: a bronze reconstruction with the proposed gear counts would directly reveal whether the 22-tooth end gear turns smoothly at the claimed 242-turn ratio or whether accumulated gear errors exceed the paper's assumed ±4° pointer uncertainty."],"forward_implications":["If Equation (1) is correct, the Saros spiral's 64 eclipse events can be regenerated purely by rotating the mechanism's input, with no need for an observational record of individual eclipses.","The four mismatches between engraved and predicted events become a diagnostic: they bound the endogenous gear errors, roughly ±1° to ±4° in pointer position, and give a quantitative quality criterion for functional reconstructions.","The revised Draconic scale maps the Back Plate's magnitude words onto sectors of the ecliptic zone, so the classification letter attached to an event follows from where the Draconic pointer sits at that event.","The same gearing supplies the Moon's second coordinate: the Draconic pointer, combined with the Lunar Disc pointer, gives a 3D lunar position and also makes star occultations by the Moon predictable from the mechanism alone."],"supporting_citations":[{"why":"supplies the preserved gear counts and fragment layout on which the revised gearing is built","marker":"Freeth et al., 2006"},{"why":"gives the Saros spiral cell structure and eclipse index letters that the recalculation must match","marker":"Freeth et al., 2008"},{"why":"proposes the Fragment D correlation and the initial Draconic gearing that this paper revises","marker":"Voulgaris et al., 2022"},{"why":"provides the DracoNod program and gear-error model that the revised ecliptic limits are calibrated against","marker":"Voulgaris et al., 2023b"},{"why":"gives the 354-hole count on the calendar ring used to estimate 3% shrinkage","marker":"Budiselic et al., 2020"},{"why":"reconfirms the improved calendar-ring hole count that supports the same shrinkage estimate","marker":"Woan and Bayley 2024"},{"why":"supplies the torque analysis used to argue the input is the Lunar Disc rather than gear $a1$, motivating the Draconic gearing layout","marker":"Roumeliotis 2018"},{"why":"provides the preserved Back Plate Inscription words that anchor the eclipse classification scheme","marker":"Iversen and Jones 2019"},{"why":"edits the ancient papyrus text used to complete the lost classification words for eclipse magnitude","marker":"Blass 1887"},{"why":"records the ancient ratio 5458 synodic months = 5923 Draconic cycles, the numerical target the gear train approximates","marker":"Toomer 1984"}],"fun_headline_variants":["Antikythera's lost Draconic gears compute eclipses mechanically","Revised gear train makes Antikythera eclipses purely mechanical","New gearing scheme precalculates Antikythera eclipse events","Draconic gear revision turns eclipse prediction into machine math","Antikythera mechanism's Draconic gearing is fully mechanical"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire result rests on the assumption that three completely unpreserved gears with 57, 56, and 22 teeth existed, meshed exactly in the proposed order, and that the original annual gear had 229 teeth after a uniform 3% shrinkage correction; if the real tooth counts or the physical fit on the right side of the casing were different, the 242-turn relation would fail.","fun_headline_variants_meta":{"raw":{"variants":["Antikythera's lost Draconic gears compute eclipses mechanically","Revised gear train makes Antikythera eclipses purely mechanical","New gearing scheme precalculates Antikythera eclipse events","Draconic gear revision turns eclipse prediction into machine math","Antikythera mechanism's Draconic gearing is fully mechanical"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000275,"raw_usage":{"total_tokens":1683,"prompt_tokens":1026,"completion_tokens":657,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":567}},"tokens_in":642,"tokens_out":657,"duration_ms":6018,"temperature":1.0,"reasoning_tokens":567,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:12:12.329762+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check would be to measure every surviving gear-tooth arc on Fragment A and reconstruct the original $b1$ tooth count from the measured radii; if that count comes out below 228 (several current measurements give 216–226), or if a physical bronze model of the proposed $s1$–$s2$–$t1$ gear cluster cannot fit between the internal and external casements on the right side of Fragment A, the proposed gearing relation is falsified. In addition, an independent recalculation using the exact ancient value 5458 synodic = 5923 Draconic, which gives 241.998 rather than 242.000 turns per Saros, would show whether the paper's claimed error of +0.00029 turns per Saros is genuinely negligible across the spiral's 50-event span.","supporting_citations":[],"review_version":1}