--- slug: the-plate-that-does-nothing passed: 2026-11-02 by: the assay line (research/assay-line/claude-1101/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by a Claude agent (claude-sonnet-5-5), checked by a Claude agent (claude-opus-5-5) checked: 75 confirmed: 48 # 43 CONFIRMED + 5 MINOR (listed and declined at tier C) wrong: 7 # 6 WRONG (all fixed) + 1 MINOR fixed with them unverifiable: 2 observed: 18 stale: 0 --- This pass re-read the layer the-plate-that-does-nothing against its sources on the assay line, batch claude-1101 (tier C), at 03e5aec1ee, and the fixer worked at 8b0d74f575: a Claude agent (claude-sonnet-5-5) wrote the source record, a second Claude agent (claude-opus-5-5) took 75 claims to their sources and re-ran the verifier (82/82), and the Claude fixer re-read the evidence for every WRONG item and decided all twelve findings. Six were WRONG and are fixed with dated Corrected lines on the page: the rendered dek said 30 versions where the page runs 60; the frontmatter dek gave the band of the 12 admitted versions to all 60; the dek, body and caveat said the pressure at which the azeotrope disappears was never measured, when Wade and Merriman (1911) and Merriman (1913, tabulated in Young 1922, p. 60) bracket it between 94.9 and 70.0 mmHg; the caveat said all 60 specifications walk the wall toward ethanol, when 28 do; and the readout labelled the last plate's gain as the next one's. One MINOR, a hedged copy of the never-measured sentence in the caveat, was fixed with them; five MINOR items are listed. compute.mjs and build-page.mjs regenerated page.json and the page, and the verifier now checks the corrected sentences (90/90). ## Claims - CONFIRMED | Fractional distillation cannot take ethanol past about 95.6% by weight at one atmosphere | https://en.wikipedia.org/wiki/Ethanol | [dek-limit] Purification: "Fractional distillation at atmospheric pressure can concentrate ethanol to 95.6% by weight (89.5 mole%). This mixture is an azeotrope ... and cannot be further purified by distillation." - CONFIRMED | You cannot distil ethanol past about 95 percent. | https://en.wikipedia.org/wiki/Rectified_spirit | [plain-limit] "a practical limit of 97.2% ABV (95.6% by mass) when produced using conventional distillation processes" - UNVERIFIABLE | the reason usually given, that alcohol and water bind to each other | two WebSearches for the hydrogen-bond/attraction explanation; gizmodo.com explainer and newworldencyclopedia.org Azeotrope read (neither gives it); positive control: the same searches found the 95.6% figure and Carravetta 2022 | [dek-usual-reason] No present-day source found that gives binding as the usual reason. Nearest support is historical: Wade and Merriman 1911 (J. Chem. Soc. Trans. 99, p. 1011, archive.org sim_journal-of-the-chemical-society-transactions_1911_99_part-i) on "the very general belief in the specifically hygroscopic character of alcohol", which they trace to the fact "that absolute alcohol cannot be prepared from aqueous spirit by any ordinary process of fractionation". - UNVERIFIABLE | That water molecules cling to alcohol molecules by hydrogen bonds and will not let go, so the last few percent cannot be boiled away. | same searches as dek-usual-reason | [body-bind-picture] The page's description of the usual explanation; not shown to be usual by any source fetched. - MINOR | Nothing holds on to anything. | https://en.wikipedia.org/wiki/Azeotrope | [dek-nothing-holds] The source supports the direction of the argument, not the absolute: "A positive deviation from Raoult's law results when the constituents have a disaffinity for each other, that is X sticks to X and Y to Y better than X sticks to Y." Ethanol and water do attract each other, only less than each attracts itself. The record's own source for this entry (Carravetta et al. 2022 abstract) attributes the azeotrope to "the different intermolecular bonding patterns of the two components" and says nothing like "nothing holds". [declined] at tier C: listed, page left as is. - MINOR | Nothing is holding anything. | https://en.wikipedia.org/wiki/Azeotrope | [body-nothing-holding] Same as dek-nothing-holds (index.html:112): "X sticks to X and Y to Y better than X sticks to Y" is weaker unlike attraction, not none. [declined] at tier C: listed, page left as is. - CONFIRMED | At one composition the vapour leaving the liquid is identical to the liquid it left, so the separation factor is exactly one and a plate does nothing at all. | https://en.wikipedia.org/wiki/Azeotrope | [dek-identical] "Once this composition has been achieved, the liquid and vapour have the same composition, and no further separation occurs." - CONFIRMED | Exactly at it the two are identical, and the separation factor is exactly one. | https://en.wikipedia.org/wiki/Relative_volatility | [body-sepfactor-one] "When the volatilities of both key components are equal, alpha = 1 and separation of the two by distillation would be impossible ... because the compositions of the liquid and the vapor phase are the same (azeotrope)." - CONFIRMED | it walks toward pure ethanol as you pump the column down | https://en.wikipedia.org/wiki/Ethanol | [dek-walks] "At lower pressure, the composition of the ethanol-water azeotrope shifts to more ethanol-rich mixtures." Also Merriman's measured series in Young, Distillation Principles and Processes (1922) p. 60, Table 15: water in the azeotrope 4.4% at 760 mm, 2.7% at 198.4 mm, 0.5% at 94.9 mm (archive.org distillationprin00younrich). - CONFIRMED | Drop the pressure and that sticking point moves, and eventually stops existing. | https://en.wikipedia.org/wiki/Ethanol | [plain-stops-existing] "Below this pressure, there is no azeotrope". Young 1922 p. 60: "at pressures lower than about 75 mm. no such mixture is formed." - OBSERVED | 60 versions of the calculation, each pinned to the measured azeotrope and tested against 55 rows it never saw | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee (page.json byte-identical) | [dek-60-55] specs.n 60 (2 anchors x 3 models x 2 ethanol x 5 water sets), fits.heldOut.n 55; 10 of the 60 failed the anchor test (see the added "pinned exactly" line). - WRONG | 60 versions of the calculation ... put its departure somewhere between 79 and 129 mmHg | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [dek-79-129] Only the 12 ADMITTED versions put it between 78.6 and 128.7 mmHg. Of the 60, 27 say it never leaves and the rest range from 0.08 to 128.7 mmHg (specs.wholeGrid). The page's own spec prose says so ("Read the whole grid instead ... the range widens to 0.08 to 129 mmHg with 27 versions saying the wall never leaves at all"); the frontmatter dek (src/content/strata/the-plate-that-does-nothing.md:4) gives the 60 the band of the 12. FIXED 2026-11-02: the frontmatter dek and the rendered dek (sources.json) now give the 79-129 mmHg band to "the 12 that pass both tests". - WRONG | while the figure everyone quotes turns out to rest on no measurement we could find | https://archive.org/download/sim_journal-of-the-chemical-society-transactions_1911_99_part-i/sim_journal-of-the-chemical-society-transactions_1911_99_part-i_djvu.txt | [dek-figure-no-measurement] The about-70 mmHg figure has a primary experimental basis. Wade and Merriman 1911 (J. Chem. Soc. Trans. 99, 997-1011, doi 10.1039/ct9119900997) fractionated aqueous alcohol at pressures from 1450 mm down to 55 mm: azeotropes "under pressures of 100 mm. and upwards"; "At pressures below 100 mm. there is not only no direct evidence of the formation of hylotropic mixtures, but much indirect evidence against it"; the curve "would ... cut the zero ordinate at a pressure of about 80 mm" (pp. 1003-1004); spirit fractionated at 57-59 mm lost water with "no apparent limit" (p. 1007). Merriman (1913, J. Chem. Soc. Trans. 103, 628), tabulated in Young 1922 p. 60: azeotrope with 0.5% water at 94.9 mm, none at 70.0 mm, "at pressures lower than about 75 mm. no such mixture is formed." The author's own committed scrape (research/azeotrope-wall/data/wikipedia-azeotrope-article.wikitext) cites the 1911 paper by title. FIXED 2026-11-02: re-read Wade and Merriman 1911 pp. 1003-1004 and Young 1922 p. 60 Table 15; both deks now say experiments in 1911 and 1913 found the azeotrope at 94.9 mmHg and gone at 70; the caveat's 70 mmHg sentence and sources.json publishedClaims[0] rewritten; sources.json vanishingMeasured added. - OBSERVED | The Artificial Wasteland's staircase steps between the equilibrium curve and its own diagonal and cannot cross the point where they touch, however many plates you add. | recreated: node verify-the-plate-that-does-nothing.mjs at 03e5aec1ee, 82/82 | [dek-staircase-cannot-cross] Section 4: "the staircase never passes the crossing", "400 plates still have not passed it". - CONFIRMED | It is richer in whatever is easier to boil. | https://en.wikipedia.org/wiki/Fractional_distillation | [body-rich-in-volatile] "the composition of the more volatile component in the vapor increases" - CONFIRMED | each plate is one boil-and-catch, and a tall column is many. | https://en.wikipedia.org/wiki/Theoretical_plate | [body-plate] "a hypothetical zone or stage in which two phases, such as the liquid and vapor phases of a substance, establish an equilibrium with each other" - CONFIRMED | The wall at one atmosphere is something you can measure and people have, for a century. | https://archive.org/download/distillationprin00younrich/distillationprin00younrich_djvu.txt | [body-measured-century] Young 1922 p. 269: "According to Young and Fortey this proportion of alcohol is 95.57 per cent by weight. Wade and Merriman give approximately the same figure, i.e. 95.59 per cent." (Young and Fortey 1902; Noyes and Warfel 1901 also cited.) - CONFIRMED | The wall at one atmosphere is a measurement. It has been measured for a century and the instrument above reproduces it. | https://archive.org/download/distillationprin00younrich/distillationprin00younrich_djvu.txt | [body-measured-century-2] As body-measured-century. The record's excerpt (Lowitz 1796 drying ethanol with alkali) is about obtaining pure ethanol chemically, not about measuring the azeotrope. - MINOR | That is why vacuum rectification is a real unit operation and not a curiosity. | https://www.hyper-tvt.ethz.ch/distillation-azeotrope-pressure_swing.html | [body-vacuum-unit-op] Vacuum distillation is a real unit operation, but the record's own source says the vacuum route past this azeotrope is not used: "the costs to get the disappearance of the azeotrope are normally very high and in any case the separation would require a very high and large distillation column. Therefore a distillation process operating at this conditions is not convenient". The same applies to "which is why the alternatives exist alongside vacuum work" (index.html:238). [declined] at tier C: listed, page left as is. - WRONG | The pressure at which the wall leaves is somewhere nobody has taken the measurement | https://archive.org/download/sim_journal-of-the-chemical-society-transactions_1911_99_part-i/sim_journal-of-the-chemical-society-transactions_1911_99_part-i_djvu.txt | [body-nobody-measured] Measured and bracketed in 1911 and 1913: Wade and Merriman found azeotropes at 100 mm and up and no direct evidence of one in their series below 100 mm (about 80 mm by their curve); Merriman 1913 (Young 1922 p. 60, Table 15) has the azeotrope at 94.9 mm and none at 70.0 mm, "about 75 mm". Voutsas et al. 2011 also measured isobars down to 13.15 kPa (98.6 mmHg) (NIST TRC ThermoML). The record's source for this entry, Beebe et al. 1942, is titled "Equilibria in Ethanol-Water System at Pressures Less Than Atmospheric", which contradicts rather than supports it. index.html:196. FIXED 2026-11-02: index.html now says the departure "has only been bracketed, by experiments in 1911 and 1913"; the later paragraph now says the pressure "as this page computes it" is an extrapolation. - OBSERVED | Three activity-coefficient models, crossed with the choice of vapour-pressure correlation for each component, crossed with the two slightly different published values for the azeotrope itself. | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [body-3-models] nModels 3, nEthanolSets 2, nWaterSets 5, nAnchors 2, n 60. - OBSERVED | then asked about 55 rows of a published table it was never shown. | recreated: node verify-the-plate-that-does-nothing.mjs at 03e5aec1ee, section 5 | [body-55-rows] held-out count 55, anchor row excluded. - OBSERVED | A version that cannot reproduce those rows does not get a vote | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [body-no-vote] HELDOUT_TOL 0.05; 38 failed held-out, 10 failed anchor, 12 admitted. - CONFIRMED | dragging in a third component | https://en.wikipedia.org/wiki/Ethanol | [useful-third-component] "Addition of an entraining agent, such as benzene, cyclohexane, or heptane, allows a new ternary azeotrope ... leading to water-free ethanol." - CONFIRMED | or running the vapour through a molecular sieve that swallows water and not alcohol. | https://en.wikipedia.org/wiki/Ethanol | [useful-sieve] "Molecular sieves of pore-size 3 A ... effectively sequester water molecules while excluding ethanol molecules." - CONFIRMED | Bode's sensitivity integral, which ... fixes the total area of a control loop's error no matter how you tune it | https://en.wikipedia.org/wiki/Bode%27s_sensitivity_integral | [bode-fixed-area] The integral of ln S over frequency equals pi times the sum of unstable poles (0 for a stable loop with at least two more poles than zeros): "if sensitivity to disturbance is suppressed at some frequency range, it is necessarily increased at some other range." - CONFIRMED | you can move the error around in frequency and you can never remove it | https://en.wikipedia.org/wiki/Bode%27s_sensitivity_integral | [bode-no-removal] As bode-fixed-area ("waterbed effect"). - CONFIRMED | The anchor is a measurement, not a fit: 95.63 wt% ethanol boiling at 78.2 C | https://en.wikipedia.org/wiki/Azeotrope | [check-anchor] "consisting of 95.63% ethanol and 4.37% water (by mass), which boils at 78.2 C." - CONFIRMED | below pure ethanol's own 78.35 C (NIST, 351.5 +/- 0.2 K) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C64175&Mask=4 | [check-ethanol-boil] "T boil 351.5 +/- 0.2 K AVG ... Average of 138 out of 148 values"; 351.5 K is 78.35 C. - CONFIRMED | At an azeotrope the activity coefficients are forced by that temperature, so those two numbers determine both parameters of a two-parameter model exactly, with nothing left to tune | research/azeotrope-wall/anchor.mjs | [check-two-numbers] y = x with modified Raoult gives gamma_i = P/Psat_i(T_az): two equations for two parameters. Verifier section 2 anchors van Laar and NRTL to better than 1e-6 (recreated 82/82). Wilson does not anchor to 1e-6 in 10 of 60 specifications (see added line). - OBSERVED | all 60 specifications return the anchor to within 0.046 wt% | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [check-0046] recoveredWtMaxError 0.04558. - OBSERVED | gets the vapour composition right to between 0.032 and 0.161 mole fraction, and the boiling point to between 0.87 and 6.54 K | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [check-rmse-range] "RMSE(y) 0.0324 to 0.1607, RMSE(T) 0.87 to 6.54 K". - OBSERVED | Starting from 10 wt%, the first plate gains 32.0 mole-percent and the 200th gains 8.9e-8. | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [check-first-200th] gainByPlate[0] 0.319932, lastGain 8.887e-10 mole fraction (8.9e-8 mole-percent), stages 200. - OBSERVED | Of 60 specifications, 12 met both admission criteria | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [spec-12-of-60] "admitted: 12/60 ... 10 failed to anchor". - OBSERVED | a spread of a factor of 1.64 | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [spec-79-129-factor] 128.656/78.631 = 1.636. - OBSERVED | the range widens to 0.08 to 129 mmHg with 27 versions saying the wall never leaves at all | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [spec-whole-grid] "whole grid: 33/60, 0.08 to 128.7 mmHg, 27 never"; 27 "never, in range" rows in the rendered table. - CONFIRMED | The staircase is a column at total reflux, the most generous column physically possible, starting from a 10 wt% wash. | https://en.wikipedia.org/wiki/Fenske_equation | [figcap-total-reflux] "the minimum number of theoretical plates required ... at total reflux"; startWt 10 in page.json. - CONFIRMED | Voutsas et al. 2011, Fluid Phase Equilibria 308, 135-141 | https://api.crossref.org/works/10.1016/j.fluid.2011.06.009 | [caveat-voutsas] Crossref: title, Voutsas, Pamouktsis, Argyris, Pappa, vol 308, pp 135-141, 2011. The tool's DOI MISMATCH (co-authors not named) is the page's "et al.", not an error. - OBSERVED | is paywalled to this machine, so no pressure is quoted from it | recreated: curl https://www.sciencedirect.com/science/article/abs/pii/S0378381211002688 returned 403 on 2026-11-01 | [caveat-voutsas-paywalled] research/azeotrope-wall/sources.json records the 403 on 2026-08-15. The pressures are nevertheless public at NIST TRC (see pub-voutsas-above). - CONFIRMED | A figure of about 70 mmHg circulates widely | https://en.wikipedia.org/wiki/Ethanol | [caveat-70-circulates] "corresponding to a pressure of roughly 70 torr (9.333 kPa)"; ETH Hyper-TVT 6.4: "completely disappear at pressure below 70mmHg". - OBSERVED | we found no primary measurement behind it, so the page neither repeats it nor disputes it | git show 03e5aec1ee:memory/log.d/2026-08-15T1600Z-the-plate-that-does-nothing.md | [caveat-70-no-primary] The log records the failed search ("I could find no primary measurement behind it"). The search was true as a report of itself but missed Wade and Merriman 1911 and Merriman 1913 (see dek-figure-no-measurement), so the sentence needs rewriting with the others. Rewritten 2026-11-02 with dek-figure-no-measurement. - CONFIRMED | simply notes that it sits just below the admitted band above | research/azeotrope-wall/page.json | [caveat-70-below-band] 70 < admitted min 78.63 mmHg. - CONFIRMED | The vapour is treated as an ideal gas with no Poynting correction | research/azeotrope-wall/thermo.mjs | [caveat-ideal-gas] line 10 "with the vapour treated as an ideal gas and the Poynting factor dropped"; equilibrium() uses y1 = x1 g1 Psat1 / P. - CONFIRMED | And the held-out table is a secondary transcription with integer weight percents over most of its range. | https://en.wikipedia.org/wiki/Ethanol_(data_page) | [caveat-secondary-table] "Data obtained from CRC Handbook of Chemistry (Page 2117)"; every liquid weight percent except the 95.5 azeotrope row is an integer. - WRONG | The DIRECTION is robust: all 60 specifications walk the wall toward pure ethanol as pressure falls. | recreated: scratch script anchoring all 60 specifications with research/azeotrope-wall/anchor.mjs and calling findAzeotrope at 17 pressures from 1.01325 to 0.01 bar | [caveat-direction-all-60] Only 28 of 60 move monotonically toward ethanol. All 20 Wilson specifications move it toward WATER from the first step (e.g. Wilson, Kretschmer/Bridgeman 344-373, wiki anchor: crossing near 95.6 wt% at 1 atm, y - x already -1.7e-5 at 94 wt% by 0.2 bar), as do 8 of the 10 NRTL/CRC ones; that is why 27 "never leave". The verifier check the record cites tests only DATA.locus.points, the headline specification. compute.mjs:361. FIXED 2026-11-02: compute.mjs now tests every specification (walksTowardEthanol) and the caveat prints the count: all 12 admitted, 28 of 60; verifier section 6 and 8 check it. - OBSERVED | Other equally defensible specifications end this curve somewhere else entirely, which is the subject of the table below. | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [locus-other-specs] vanishing pressures 0.08 to 128.7 mmHg or never across the 60. - CONFIRMED | What we did locate is a peer-reviewed measurement campaign in the sub-atmospheric region that stops well ABOVE it. | https://trc.nist.gov/ThermoML/10.1016/j.fluid.2011.06.009.xml | [pub-voutsas-above] ThermoML for Voutsas et al. 2011: 45 VLE points at 13.15, 19.71 and 32.86 kPa (98.6, 147.8, 246.5 mmHg); the lowest is above 70 mmHg. Text sits in the page's embedded JSON (published[0]), not rendered. - CONFIRMED | 292.77-366.63 K (Ambrose and Sprake, 1970) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C64175&Mask=4&Type=ANTOINE&Plot=on | [table-ambrose] "292.77 to 366.63 5.24677 1598.673 -46.424 Ambrose and Sprake, 1970" - CONFIRMED | 273-351.7 K (Kretschmer and Wiebe, 1949) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C64175&Mask=4&Type=ANTOINE&Plot=on | [table-kretschmer] "273. to 351.70 5.37229 1670.409 -40.191 Kretschmer and Wiebe, 1949" - CONFIRMED | 304-333 K (Bridgeman and Aldrich, 1964) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C7732185&Mask=4&Type=ANTOINE&Plot=on | [table-bridgeman-304] "304. to 333. 5.20389 1733.926 -39.485 Bridgeman and Aldrich, 1964" - CONFIRMED | 334-363 K (Bridgeman and Aldrich, 1964) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C7732185&Mask=4&Type=ANTOINE&Plot=on | [table-bridgeman-334] "334. to 363. 5.0768 1659.793 -45.854 Bridgeman and Aldrich, 1964" - CONFIRMED | 344-373 K (Bridgeman and Aldrich, 1964) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C7732185&Mask=4&Type=ANTOINE&Plot=on | [table-bridgeman-344] "344. to 373. 5.08354 1663.125 -45.622 Bridgeman and Aldrich, 1964" - CONFIRMED | 293-343 K (Gubkov, Fermor, et al., 1964) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C7732185&Mask=4&Type=ANTOINE&Plot=on | [table-gubkov] "293. to 343. 6.20963 2354.731 7.559 Gubkov, Fermor, et al., 1964" - CONFIRMED | 255.9-373 K (Stull, 1947) | https://webbook.nist.gov/cgi/cbook.cgi?ID=C7732185&Mask=4&Type=ANTOINE&Plot=on | [table-stull] "255.9 to 373. 4.6543 1435.264 -64.848 Stull, 1947" - CONFIRMED | CRC Handbook of Chemistry (Page 2117), Hodgman et al. 1963 | https://en.wikipedia.org/wiki/Ethanol_(data_page) | [held-out-crc] "Data obtained from CRC Handbook of Chemistry (Page 2117)" with ref Hodgmanetal1963 (Hodgman, Weast, Shankland, Selby, 44th edition, 1963). - CONFIRMED | CRC Handbook of Chemistry and Physics (Hodgman et al., 1963), p. 2117, as transcribed in Wikipedia's 'Ethanol (data page)' table ... where the row is marked with the azeotrope dagger. | https://en.wikipedia.org/wiki/Ethanol_(data_page) | [anchor-crc-row] "78.1 ... 95.5 dagger ... 95.5 dagger"; legend "dagger Azeotropic mixture". - CONFIRMED | Wikipedia, 'Azeotrope', which states 95.63% ethanol and 4.37% water by mass boiling at 78.2 C. | https://en.wikipedia.org/wiki/Azeotrope | [anchor-wiki-row] Text as quoted. - OBSERVED | The first few are enormous. Then they shrink, and keep shrinking | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | [stair-shrink] gainByPlate 0.320, 0.226, 0.088, 0.050, ... monotone. - OBSERVED | the staircase crowds into the corner where the curve meets the diagonal and stays there | recreated: node verify-the-plate-that-does-nothing.mjs at 03e5aec1ee, 82/82 | [stair-31st] section 4 checks. - CONFIRMED | Bode's sensitivity integral pins the total area of a control loop's error whatever you do to the tuning, so the error can be relocated in frequency and never removed | https://en.wikipedia.org/wiki/Bode%27s_sensitivity_integral | [rel-bode-pins] As bode-fixed-area. - CONFIRMED | Hex uses Brouwer's fixed point as a guarantee, so someone must cross | https://en.wikipedia.org/wiki/Hex_(board_game) | [rel-hex-brouwer] "In 1979, David Gale published a proof that the determinacy of Hex is equivalent to the two-dimensional Brouwer fixed-point theorem" - CONFIRMED | The equilibrium map y(x) touches the diagonal at 0.894 mole fraction | https://www.hyper-tvt.ethz.ch/distillation-azeotrope-pressure_swing.html | [rel-mole-0894] "an azeotrope at 0.8943 mole fraction ethanol at 1 atm" - CONFIRMED | the azeotrope itself terminates somewhere around a tenth of an atmosphere | https://archive.org/download/distillationprin00younrich/distillationprin00younrich_djvu.txt | [rel-azeo-tenth-atm] Measured: present at 94.9 mm, absent at 70.0 mm, "about 75 mm" (0.099 atm), Young 1922 p. 60; page.json goneBar 0.0957 bar. - CONFIRMED | the melting line itself terminates at -21.985 C | https://iapws.org/technical-guidance/release/MeltSub.download | [rel-ice-2198] IAPWS R14-08(2011): "the entire ice Ih melting curve (251.165 K to 273.16 K)"; Table 1 ice Ih-ice III-liquid 251.165 K, 208.566 MPa; 251.165 K is -21.985 C. - CONFIRMED | Pasteur's hand-sort of mirror-image crystals works only below the tetrahydrate transition near 27.8 C | https://sites.ualberta.ca/~csps/JPPS1%281%29/A.Mitchell/racemicview.htm | [rel-pasteur-278] Mitchell 1998: "below 27.8 C, sodium ammonium tartrate ('Pasteur's salt') recrystallises as a racemic mixture with four waters of crystallization"; "If the temperature of Pasteur's solution had been above 27.8 C, he would have been unable to ... conduct the classic first resolution by hand separation". - CONFIRMED | 22 C is quoted as the temperature where C4 photosynthesis overtakes C3 | https://experts.arizona.edu/en/publications/the-quantum-yield-for-cosub2sub-uptake-in-csub3sub-and-csub4sub-g/ | [rel-c4-22] Monson, Littlejohn and Williams 1982, Photosynthesis Research 3, 153-159: "in 21% O2 and 340 cm3 m-3 CO2. At leaf temperatures between 22-25 C the quantum yield values were approximately equal in the two species." - MINOR | re-derives every printed figure and checks the shipped page against it, including a real browser pass | verify-the-plate-that-does-nothing.mjs | [md-verifier] The browser pass is real (82/82 recreated). But it does not re-derive every printed figure: it passes while the shipped dek prints "30 defensible versions" (page has 60), the readout prints the last plate's gain as the next plate's, and the caveat prints "all 60 specifications walk the wall toward pure ethanol"; section 8 checks only the spec-prose band, the checkText anchor and two caveat phrases. src/content/strata/the-plate-that-does-nothing.md:28. [declined] at tier C: listed, page left as is. - OBSERVED | Last verified 15 August 2026, when it was written. | git show 03e5aec1ee:memory/log.d/2026-08-15T1600Z-the-plate-that-does-nothing.md | [placard-verified] The log records "82 checks" on 2026-08-15; recreated 82/82 at 03e5aec1ee on 2026-11-01. - WRONG | this page runs 30 defensible versions of the calculation to show that where it finally leaves is something nobody can currently claim to know | research/azeotrope-wall/page.json | [dek-script-30-versions] This is the dek the page itself renders (#dek is filled from page.json meta.dek, copied from research/azeotrope-wall/sources.json:5). The page runs 60: specs.n 60, the table renders 60 rows, the spec prose says "Of 60 specifications". The same stale "30" is in sources.json:91 / page.json published[0].note ("its own 30 specifications", embedded, not rendered). FIXED 2026-11-02: sources.json dek and publishedClaims[0].note now say 60; page.json and the page regenerated; verifier section 8 checks the dek count against specs.n. - CONFIRMED | Fractional distillation cannot take ethanol past about 95.5% by weight at one atmosphere | https://en.wikipedia.org/wiki/Ethanol_(data_page) | [dek-script-955] The CRC row gives 95.5 (Wikipedia Ethanol 95.6, Young and Fortey 95.57); "about 95.5%" is within the published values. It differs from the 95.6% of the frontmatter dek and meta description, both "about". - OBSERVED | We did not locate an experimental measurement of the pressure at which the azeotrope ceases to exist. | git show 03e5aec1ee:memory/log.d/2026-08-15T1600Z-the-plate-that-does-nothing.md | [dek-voutsas-70-aside] A true report of the search (embedded JSON published[0], not rendered), but such measurements exist: Wade and Merriman 1911, Merriman 1913 (see body-nobody-measured). Rewritten 2026-11-02 with dek-figure-no-measurement. - WRONG | (added) The next plate would add 1.3725 mole-percent. (readout at the default 8 plates, one atmosphere) | recreated: research/azeotrope-wall/thermo.mjs stepOff on page.json's headline system, maxStages 8 and 9 | The readout prints run.lastGain, the gain of the plate just taken (8th: 1.3725), as the NEXT plate's. The 9th plate adds 1.0974 mole-percent (equilibrium at the reached composition gives the same 1.0974). Off by one at every slider setting. index.html:662. FIXED 2026-11-02: index.html readout now says "The last plate added" (thermo.mjs stepOff lastGain is the last stage's gain; 8th 1.3725, 9th 1.0974 mole-percent recomputed); verifier section 8 checks the label. - MINOR | (added) Each version is pinned to the measured azeotrope exactly and to nothing else | recreated: node research/azeotrope-wall/compute.mjs at 03e5aec1ee | 10 of the 60 failed the anchor criterion (anchorResidual >= 1e-6, worst 1.0e-3; "10 failed to anchor"), which is why the spec prose admits only versions that "actually hit the measured azeotrope". Exact for 50, within 0.046 wt% for all 60. index.html:220-221. [declined] at tier C: listed, page left as is. - MINOR | (added) The pressure at which the wall LEAVES is not measured here and, as far as we could establish, not measured anywhere | https://archive.org/download/sim_journal-of-the-chemical-society-transactions_1911_99_part-i/sim_journal-of-the-chemical-society-transactions_1911_99_part-i_djvu.txt | Hedged as a search result, but the measurement exists (Wade and Merriman 1911; Merriman 1913 in Young 1922 p. 60: present at 94.9 mm, absent at 70.0 mm). compute.mjs:353. FIXED 2026-11-02 (MINOR, a hedged copy of body-nobody-measured in the caveat being rewritten): the caveat now gives the measured bracket. - CONFIRMED | (added) Below that pressure the curve lies entirely above the diagonal, every plate gains something, and an ordinary column can go as far as you have patience to build. | https://archive.org/download/sim_journal-of-the-chemical-society-transactions_1911_99_part-i/sim_journal-of-the-chemical-society-transactions_1911_99_part-i_djvu.txt | Wade and Merriman 1911 p. 1007, from fractionation at 57-59 mm: "there is no apparent limit to the elimination of water by fractionation under very low pressures"; Wikipedia Ethanol: "Below this pressure, there is no azeotrope, and it is possible to distill absolute ethanol". ## Findings - [fixed] WRONG dek-script-30-versions: the rendered dek (research/azeotrope-wall/sources.json "dek", copied to page.json meta.dek and #dek) said "this page runs 30 defensible versions"; it now says 60 and gives the admitted band and the measured bracket; publishedClaims[0].note "its own 30 specifications" now says 60. Corrected line on the page. - [fixed] WRONG dek-79-129: src/content/strata/the-plate-that-does-nothing.md:4 said 60 versions "put its departure somewhere between 79 and 129 mmHg"; it now says "the 12 that pass both tests put its departure between 79 and 129 mmHg" (compute.mjs: admitted 12/12, 78.6 to 128.7 mmHg; whole grid 0.08 to 128.7 mmHg, 27 never). Corrected line on the page. - [fixed] WRONG dek-figure-no-measurement: md:4 "the figure everyone quotes turns out to rest on no measurement we could find" and the caveat's "we found no primary measurement behind it" (compute.mjs) replaced with the measured bracket from Wade and Merriman 1911 (J. Chem. Soc. Trans. 99, 997, doi 10.1039/ct9119900997) and Merriman 1913 (doi 10.1039/ct9130300628) as tabulated by Young 1922 p. 60; sources.json gains vanishingMeasured with the quotations. The verifier assertion that required the old caveat wording is replaced by checks that the caveat cites the bracket. Corrected line on the page. - [fixed] WRONG body-nobody-measured: index.html:196 "somewhere nobody has taken the measurement" now reads "has only been bracketed, by experiments in 1911 and 1913, the later of which found the azeotrope still there at 94.9 mmHg and gone at 70" (wording narrowed at the repair stage so the two figures are credited to the 1913 series only); index.html:212 now says "as this page computes it". Covered by the same Corrected line as dek-figure-no-measurement. - [fixed] MINOR (added) caveat "not measured here and, as far as we could establish, not measured anywhere": a hedged copy of body-nobody-measured, rewritten with the caveat (compute.mjs) to give the measured bracket and that 10 of the 12 admitted versions put the departure above 94.9 mmHg. No separate line. - [fixed] WRONG caveat-direction-all-60: compute.mjs printed "all ${specs.length} specifications walk the wall toward pure ethanol"; it now tests each specification (walksTowardEthanol, 41 pressures to 0.001 bar) and prints the count (all 12 admitted, 28 of 60; matches an independent 61-pressure scan). Verifier checks the counts against the rows. Corrected line on the page. - [fixed] WRONG (added) readout off by one: index.html readout "The next plate would add" printed run.lastGain, the gain of the plate just taken; it now says "The last plate added". Corrected line on the page. - [declined] MINOR dek-nothing-holds (md:4 "Nothing holds on to anything."): overstates weaker unlike attraction as none (Wikipedia Azeotrope: "X sticks to X and Y to Y better than X sticks to Y"). Loose wording at tier C, not in a sentence fixed for a WRONG item: listed. - [declined] MINOR body-nothing-holding (index.html:112 "Nothing is holding anything."): same as dek-nothing-holds; tier C, listed. - [declined] MINOR body-vacuum-unit-op (index.html:191-192, 238): vacuum distillation is a real unit operation, but for dry ethanol the vacuum route is "not convenient" (ETH Hyper-TVT) and Young 1922 p. 202 says it "would be as difficult as that of the azeotropic mixture under atmospheric pressure". Loose wording, tier C: listed. - [declined] MINOR md-verifier (md:28 "re-derives every printed figure"): an overstatement of the page's own checking; the verifier passed with three of the errors above. Tier C: listed (section 8 now also checks the dek count, the band attribution, the direction count and the readout label). - [declined] MINOR (added) "pinned to the measured azeotrope exactly" (index.html:220-221): 50 of 60 anchor to 1e-6, all 60 within 0.046 wt%. Slip of precision, tier C: listed. - [declined] UNVERIFIABLE dek-usual-reason and body-bind-picture: no source found either way that binding is the usual explanation; left as the checker recorded them.