{"id":"e1b74a82-41a2-4cfb-9666-594886cf50b1","arxiv_id":"2506.04262","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An English translation of Boltzmann's 1898 H-curve paper with typo corrections and an appendix formula for normalized hump frequency.","lead":"This preprint provides an English translation of Boltzmann's 1898 paper about the H-curve, a plot that illustrates how a statistical quantity fluctuates over time. It corrects typographical errors in the original equations and adds a short appendix with a new formula for the frequency of large humps when the draw probability is biased.","discovery_kind":"incremental","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Rendered 'N = 1000·2n' is inconsistent with the claimed ~4000 humps; it must be N = 1000·2^n, and the translation or appendix needs correction or verification.","rationale":"The paper's central value is a faithful, corrected translation, and the reader correctly identified translation fidelity as the main uncertainty. My stress-test found a concrete place where the rendered mathematics fails internal consistency. Boltzmann's own calculation—about 2000 black runs and 2000 white runs when N is chosen suitably—imposes N=1000·2^n, but the preprint's full text renders 'N = 1000 · 2n' in both the body and the appendix. Since the expected number of runs is ~4N/2^n, the linear reading gives a vanishing count for large n, while the exponential reading gives 4000. I do not find the appendix formula otherwise objectionable: for N large, the expected number of maximal-hump starts is (2N+1)[p^n+(1-p)^n], so normalizing by its value at p=1/2 gives exactly 2^{n-1}[p^n+(1-p)^n]; that part is a standard calculation and is not a problem. The concrete expression is corrigible, so I would not reject the paper; the conditional verdict stands, sharpened to require verifying this one expression against the original before publication. If the original indeed has 2^n, the author should fix the rendered expression and the appendix's repetition; if not, a correction footnote is needed. In either case the inconsistency in the preprint is real and must be resolved. This is not an attack on the author, and the rest of the translation appears careful—for example, the point-count and summation-index footnotes are sensible and internally consistent.","tokens_in":7697,"tokens_out":8779,"duration_ms":86525,"concrete_test":"Consult the original Boltzmann 1898 paper (Mathematische Annalen 50, pp. 325–332, available via GDZ/Springer) and check the displayed expression at the hump-count passage. If it reads N=1000·2^n, correct '2n' to '2^n' in both Section 2.2 and the appendix; if it reads N=1000·2n, add a correction footnote explaining that the expected number of length-n monochromatic runs is about (2N+1)·2^{1-n}, which equals ~4000 only when N scales as 2^n. As an independent numerical check, evaluate the expectation for n=20 with N=1000·2^20 versus N=1000·2·20; only the former gives a value near 4000.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the translation's key illustration (Section 2.2, after the definition of humps), the text reads 'N = 1000 · 2n' and then claims that over 2N+1 fair draws there will be about 2000 runs of n black and about 2000 runs of n white. With N=1000·2n and n large, the expected number of monochromatic runs of length n is (2N-n+2)·2·(1/2)^n = O(n/2^n), which is not ~4000 for any 'very large' n. The only way to get ~4000 occurrences is N=1000·2^n, because then the expectation is about (2000·2^n)·2·2^{-n} = 4000. The same apparent typo is repeated in the appendix, where N(1/2)≈4000 is cited 'for N=1000·2n'. This is load-bearing because the illustrative hump-count and the normalization in the appendix formula both depend on the exponential form. It is also exactly the kind of mathematical expression whose fidelity the paper promises to have corrected. If the original Boltzmann article contains N=1000·2^n, then this preprint has introduced or reproduced a typographical error; if the original contains 2n, then Boltzmann's numerical claim is wrong and the translation should flag it. Either way, the manuscript as rendered is internally inconsistent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript provides an English translation of Boltzmann's 1898 paper \"Ueber die sogenannte H-Curve\", together with translator's and author's footnotes that correct several apparent typographical errors (e.g., the summation limit in the definition of y and the number of points B). The translation is followed by concluding remarks and an appendix in which the author claims that, for an urn with bias p=1/2+epsilon, the normalized number of occurrences of the largest possible hump satisfies N~(p) approximately equal to 2^(n-1)[p^n+(1-p)^n], with two figures for n=900. The core of the paper is thus a historical translation plus a small probabilistic addendum.","tokens_in":7948,"tokens_out":10185,"duration_ms":94668,"significance":"If the translation is accurate, the paper fills a genuine gap by making this late Boltzmann paper accessible in English; the footnotes are useful and the translation reads coherently. The appendix's formula is a simple, parameter-free expectation that follows from elementary run-counting, and the quantitative claim is falsifiable. However, because the original German text is not included and the mathematical notation in the arXiv text is ambiguous, the paper's stated aim of correcting typos cannot be independently checked. More seriously, the appendix's figures are inconsistent with the stated formula, so the technical addendum needs verification before the paper can be accepted as a reliable scholarly contribution.","major_comments":[{"comment":"If the expressions are read literally, they are mathematically inconsistent with the surrounding claims. With N=1000·2n, the expected number of monochromatic runs of length n among 2N+1 fair draws is approximately 4N/2^n = 8000n/2^n, which tends to zero for large n, not the asserted 4000. The claimed count requires N=1000·2^n. Since the manuscript's own purpose is to fix mathematical typos, the author should verify these expressions against Boltzmann's original and correct the notation in both the translation and the appendix; a short derivation of the run-count estimate would also remove ambiguity.","section":"§2.2 (passage 'N = 1000·2n') and Appendix"},{"comment":"With n=900, the stated formula gives N~(0.502) approximately equal to 2^899 times (0.502^900 + 0.498^900), which is on the order of 10^271, while Fig. 1 plots values around 5×10^6. The formula and the figure are incompatible by an enormous factor, so the graph cannot have been generated from the displayed expression. The author should state precisely how the figures were computed, supply a derivation or source for the formula, and correct either the text or the figures.","section":"Appendix, Figs. 1 and 2"},{"comment":"The central claim of the manuscript is that the translation is faithful and that the footnotes correct actual errors in Boltzmann's original (e.g., footnote 6). Without a reproduction of the original German text, or at least a detailed list of the original expressions and the corresponding corrections, the reader cannot distinguish a genuine correction from an introduced error. This is especially important because the ambiguous notation in the present text (major comment 1) directly affects a passage whose fidelity is promised.","section":"Introduction and §2.1"}],"minor_comments":[{"comment":"There are several grammatical slips, such as 'there are a n equal number' and other small wording issues; these should be cleaned up before publication.","section":"Abstract and main text"},{"comment":"The quantity N(p) is not defined precisely: it should be identified as an expected number of runs (or a limiting frequency), and the dependence on the finite values of N and n should be stated explicitly.","section":"Appendix"},{"comment":"The figure captions do not describe what is plotted, whether it is the formula, a simulation, or an asymptotic curve, nor do they state the values of N and n used; these details are needed to make the figures reproducible.","section":"Figs. 1 and 2 captions"},{"comment":"Footnote 5 correctly corrects the number of points to 2N-n+2, but the main text in the same passage still reads '2N-n+1 points'; this discrepancy should be resolved in the text itself.","section":"Footnote 5"},{"comment":"Reference [18] lacks a title, and a few other references would benefit from complete titles or page ranges; the author should check the reference list against the journal's style.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The historical translation may well be sound, but the appendix appears not to have been carefully checked: the figures are inconsistent with the stated formula by orders of magnitude, and the notation around 2^(n-1) and 2^n is ambiguous in the rendered text. I would ask the editor to require the author to supply the derivation and, ideally, the code or data behind the figures before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is an English translation of Boltzmann's 1898 'Ueber die sogenannte H-Curve', with footnotes correcting apparent typos. That is genuinely useful: historians and teachers get a readable version of a paper that is often cited but less often read, and the corrections (e.g., the summation index in footnote 6) look right. The accompanying appendix adds a formula for the normalized frequency of maximum humps when the white-ball probability p differs from 1/2, N~(p) ≈ 2^(n-1)[p^n + (1-p)^n]. That is a straightforward consequence of the urn model—each length-n window is monochromatic with probability p^n+(1-p)^n—so it is a minor extension, but it is new and correctly normalized.\n\nThe main problem is a load-bearing typo in the translation itself. The text sets N = 1000·2n, then claims about 2000 runs of n black and 2000 of n white, hence ~4000 humps. For large n, the expected number of monochromatic runs of length n in 2N+1 draws is (2N-n+2)[p^n+(1-p)^n], which with N=1000·2n is O(n/2^n)—nothing like 4000. The number only works if N=1000·2^n. The appendix repeats the same 'N=1000·2n' when citing N(1/2)≈4000. Since the paper's stated purpose is to correct typographical errors, this is exactly the kind of error it should have caught, and it undermines the illustrative calculation. Either Boltzmann's original had 2^n and this rendering introduced the typo, or the original had 2n and the numerical claim is wrong and should be flagged. The manuscript as it stands is internally inconsistent.\n\nA second, lesser issue: the appendix formula is stated without derivation. It is simple enough that an expert can fill it in, but for a pedagogical paper a three-line derivation would help. The plain-text rendering of 2^(n-1) as '2(n−1)' is also confusing, though that may be an artifact of the arXiv format.\n\nI cannot check translation fidelity without the German original, which the preprint does not include. That is an inherent limitation of any translation, not a fatal flaw.\n\nWho is this for? Historians of statistical mechanics, and teachers who want a primary-source supplement. It does not change our scientific understanding. The paper deserves peer review—a referee with access to the original can verify the translation and sort out the 2n/2^n question—but it needs a correction before publication.","headline":"A useful translation of Boltzmann's H-curve paper, but a likely 2n/2^n typo in the key illustration needs fixing before it can be trusted.","tokens_in":8437,"tokens_out":2864,"would_cite":false,"duration_ms":26048,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["01.65.+g","05.20.-y"],"model":"deepseek-v4-flash","headline":"This paper makes Boltzmann's 1898 essay on the H-curve available in English, corrects typographical errors in its equations, and adds a formula for the frequency of its largest humps under biased draws.","keywords":["H-Curve","H-theorem","Boltzmann statistics","entropy","urn model","quasitangent","scientific translation"],"falsifier":"Compare the translation line by line with the original in Mathematische Annalen 50 (1898) 325–332, especially the corrected summation index in the definition of $y$, and run a Monte Carlo urn simulation with $n=900$ and $p=0.51$, counting maximal humps and checking the count against $2^{899}(0.51^{900}+0.49^{900})$; either mismatch would show the corrections or the formula are wrong.","tokens_in":7479,"feed_emoji":"📜","tokens_out":10034,"duration_ms":83668,"temperature":0.7,"pith_summary":"The paper provides an English translation of Boltzmann's 1898 article \"Ueber die sogenannte H-Curve,\" with footnotes correcting typographical errors in the original equations. The H-curve is a plot of a quantity related to entropy, built from a simple urn experiment: draw black and white balls with replacement and record, in each sliding window of $n$ draws, the deviation of the white-ball count from $n/2$. Boltzmann uses this curve to argue that an entropy-like quantity can fluctuate, lack an ordinary tangent, and nevertheless have quasitangents, so ordered states overwhelmingly evolve toward disorder while reversibility and recurrence remain possible. The translator adds an appendix claiming that when the probability of drawing a white ball is $p=1/2+\\varepsilon$, the normalized number of maximal humps obeys $\\tilde{N}(p) \\approx 2^{n-1}[p^n + (1-p)^n]$, so a small bias sharply increases the number of large humps.","feed_headline":"Boltzmann's H-curve paper translated, corrected, gets bias formula","feed_subtitle":"Read Boltzmann's own 1898 argument that entropy curves fluctuate, and a new formula for how bias multiplies humps.","key_machinery":"The central object is the H-curve of the lottery, defined by plotting points with abscissa $x=k/n$ and ordinate $y=\\overline{1-2a_k/n}$, where $a_k$ counts white balls in the $n$ draws $Z_k,\\dots,Z_{k+n-1}$. The argument runs on the contrast between single-step chords, whose slope is always $\\pm 2$, and longer chords, which approach a quasitangent; for a point leaving a maximal hump the quasitangent has slope $-1$. A continuous version is built from smoothed functions $f_k(t)$, with the translator correcting the summation range to $k=-N+n$ through $k=N$. The appendix's formula for the number of maximal humps under biased draws supplies the paper's quantitative extension.","core_discovery":"The central claim is that Boltzmann's 1898 demonstration of the H-curve's properties deserves a modern, corrected English text. In the translated essay, Boltzmann constructs the H-curve from a fair urn, shows that adjacent points have chord slopes $\\pm 2$ while chords over intermediate scales approach quasitangents (slope $-1$ leaving a maximal hump), and concludes that humps of finite height become vanishingly rare as their height grows. The translator's corrections fix typos in the equations, and the appended remark claims the normalized count of maximal humps is approximately $2^{n-1}[p^n + (1-p)^n]$ when the draw probability is $p$.","pith_inferences":["The appendix formula could be tested by direct simulation of the urn experiment; if it holds, it provides an explicit large-deviation rate for the tallest humps in a sliding-window Bernoulli sequence.","Because the predicted count grows so steeply with $p-1/2$, numerical H-curve experiments would be highly sensitive to random-number-generator bias, so the formula could serve as a calibration test for simulations.","The same sliding-window construction may transfer to general fluctuation questions, such as the distribution of local maxima of moving averages in stochastic processes, connecting Boltzmann's lottery to modern extreme-value statistics.","Extending the formula to continuously drawn H-curves or to time-dependent probabilities would give predictions about entropy-curve humps in non-equilibrium settings, though the paper does not pursue that extension."],"forward_implications":["Readers can now follow Boltzmann's own argument in English, including the corrected definitions, without needing the 1898 German original.","The appendix formula implies that at $p=0.51$ and $n=900$ the normalized count of maximal humps reaches tens of millions, so a tiny bias changes the H-curve's appearance dramatically.","Boltzmann's construction shows that a curve can have ordinary tangents on infinitesimal scales and quasitangents on intermediate scales, so the gas H-curve's lack of an ordinary tangent does not block analysis of its fluctuations.","The H-curve's symmetry under reversal of the draw order means any property proved for increasing abscissae holds equally for decreasing abscissae, matching the time-reversal symmetry of molecular motion.","The Ehrenfest urn recurrence estimate of $2^N$ steps gives a concrete timescale for return to the initial state, quantifying Zermelo's recurrence objection."],"supporting_citations":[{"why":"the original 1898 article whose text and equations are translated and corrected","marker":"[1]"},{"why":"Loschmidt's reversibility objection that motivates Boltzmann's defense of entropy increase","marker":"[2]"},{"why":"Zermelo's recurrence objection that the H-curve's fluctuations address","marker":"[5]"},{"why":"Boltzmann's earlier reply to Zermelo, which the translated paper complements","marker":"[8]"},{"why":"the Ehrenfest urn model used in the concluding remarks to estimate recurrence time as 2^N","marker":"[18]"},{"why":"background source for the Ehrenfest model and Boltzmann's statistical ideas","marker":"[10]"}],"fun_headline_variants":["Boltzmann's H-curve paper: English translation plus typo fixes","H-curve gets corrected translation and bias formula","Boltzmann's 1898 H-curve essay, now in clear English","New formula for hump bias from Boltzmann's H-curve paper","Translated H-curve work yields hump count formula"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The translation must faithfully represent Boltzmann's original German text and equations; because the original is not reproduced alongside, a mistranslation or a missed typo would undermine the value of the corrections and the appendix formula.","fun_headline_variants_meta":{"raw":{"variants":["Boltzmann's H-curve paper: English translation plus typo fixes","H-curve gets corrected translation and bias formula","Boltzmann's 1898 H-curve essay, now in clear English","New formula for hump bias from Boltzmann's H-curve paper","Translated H-curve work yields hump count formula"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000254,"raw_usage":{"total_tokens":1476,"prompt_tokens":763,"completion_tokens":713,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":379,"completion_tokens_details":{"reasoning_tokens":625}},"tokens_in":379,"tokens_out":713,"duration_ms":6505,"temperature":1.0,"reasoning_tokens":625,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:24:11.387339+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the translation line by line with the original in Mathematische Annalen 50 (1898) 325–332, especially the corrected summation index in the definition of $y$, and run a Monte Carlo urn simulation with $n=900$ and $p=0.51$, counting maximal humps and checking the count against $2^{899}(0.51^{900}+0.49^{900})$; either mismatch would show the corrections or the formula are wrong.","supporting_citations":[{"cited_title":"Ueber die sogenannte H -Curve","cited_arxiv_id":null,"evidence_quote":"the original 1898 article whose text and equations are translated and corrected"},{"cited_title":"¨Uber die Beziehung eines allgemeinen mechanischen Satzes zum zweite n Hauptsatze der W¨ armetheorie.Sitzungsberichte der Kaiserlichen Akademie der Wissensch aften","cited_arxiv_id":null,"evidence_quote":"Loschmidt's reversibility objection that motivates Boltzmann's defense of entropy increase"},{"cited_title":"Ueber einen Satz der Dynamik und die mechanische W¨ armetheorie.Annalen der Physik , 57:485–494, 1896","cited_arxiv_id":null,"evidence_quote":"Zermelo's recurrence objection that the H-curve's fluctuations address"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Boltzmann's earlier reply to Zermelo, which the translated paper complements"},{"cited_title":"Physikalische Zeitschrift , 8:311–314, 1907","cited_arxiv_id":null,"evidence_quote":"the Ehrenfest urn model used in the concluding remarks to estimate recurrence time as 2^N"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"background source for the Ehrenfest model and Boltzmann's statistical ideas"}],"review_version":1}