--- slug: the-invariant-of-relabeling passed: 2026-10-14 by: the assay line (research/assay-line/queue-1014c/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex and a Claude agent checked: 138 confirmed: 98 wrong: 37 unverifiable: 0 observed: 0 stale: 3 --- This was the directed repair pass for the-invariant-of-relabeling at f5e5f84f34, using the queue-1014c triage after the Codex and Claude source checks. The fixer reread the evidence before editing, reran the page calculations, and carried forward the checkers' distinct CONFIRMED claims, excluding verdicts superseded by triage. There is one line for each of the 42 triage items, including repeated findings; all 17 MINOR lines count in wrong, while the 20 WRONG lines retain their level. The seven dated corrections cover WRONG findings only. The quotation remains the cipher input, with its closing factual overclaim explicitly corrected beside it. The two Shannon findings were rejected on the printed formula. The JSON-LD and verification dates use the triage's exceptions. No other layer or shared file was edited. ## Claims - MINOR | X1: the index of coincidence that broke the Vigenère cipher | https://pages.mtu.edu/~shene/NSF-4/Vigenere.pdf | [ic-broke-vigenere] MTU, pp. 1-2, places the published Kasiski attack in 1863 and Friedman's IC method in 1922. The later attack is valid; the plain description now says IC can break Vigenère. - WRONG | X2: Linear B cracked from pure structure | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [linear-b-pure-structure] Cambridge process.pdf, pp. 3-4, combines the structural grid with guessed a, Cypriot ni and Cretan names. The June 19 layer log separately supports the claimed cross-seam tie. - MINOR | X3: Michael Ventris guessed that four often-repeated sign-groups were Cretan place-names | https://ignca.gov.in/Asi_data/17617.pdf | [historical-guessed-towns] Read Documents in Mycenaean Greek, printed pp. 21-22 (PDF pp. 58-59). Its table has five towns, including Lyktos and all four named here. Those four remain valid examples. - WRONG | X4: Wrong L: still a mixture, still flat. | git show 60ea23c371:research/invariant-of-relabeling/artifact.json | [wrong-period-flat] Executed the page script: mean column IC is 0.07254864311315923 at 7 and 0.07134216589861751 at 14. Multiples also keep each column on a single shift. - MINOR | X5: In a cipher the language is known | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [cipher-known-language] Ca' Foscari's key-recovery procedure compares with the plaintext language's characteristic IC. English is supplied in this example, not required to be known by the definition of a cipher. - WRONG | X6: the cipher's residual ambiguity is astronomical | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [cipher-ambiguity-column] Ca' Foscari's recovery loop tries 0 through 25. The local vigenere function adds one key-letter shift modulo 26. Each known-period column has 26 keys; 26! remains valid for unrestricted alphabet substitutions. - WRONG | X7: 26! per column | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [cipher-26factorial-column] Ca' Foscari's recovery loop tries 0 through 25. The local vigenere function adds one key-letter shift modulo 26. Each known-period column has 26 keys; 26! remains valid for unrestricted alphabet substitutions. - WRONG | X8: so you need a whole frequency table to break it. | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [whole-frequency-table] Ca' Foscari recovers relative shifts and then tries 26 possible first-key letters, with a reference text as an alternative. Algebraically c = p + k mod 26 gives k from one known plaintext/ciphertext letter pair. - CONFIRMED | X9: Shannon counts the redundancy of English in bits | https://jontalle.web.engr.illinois.edu/uploads/537.F18/Papers/Shannon50b.pdf | [shannon-redundancy] REJECTED with evidence: read and visually inspected Shannon50b.pdf, PDF p. 5, printed p. 252 [130]. Its bound R >= log2(27) + sum(q_i log2(q_i)) is an unnormalized binary-log entropy deficit, in bits per character. Percentage redundancy elsewhere does not exclude this absolute quantity. The directing triage's rejection holds. - WRONG | X10: the difference is only the size of the symmetry group. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [difference-only-group] Cambridge process.pdf, pp. 3-4, uses phonetic and geographical clues. The models compare unrestricted substitution with a complete grid; actual Vigenère columns use 26 shifts. Group size alone does not describe the difference between the historical decipherments. - MINOR | X11: plaintext (L=1)0.0714 | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [plaintext-value] Executed the exact page code: the unchanged passage has 444 letters and IC 14054/196692 = 0.0714518129868017, giving 0.0715 at four decimals. Both existing live readouts were already correct. - WRONG | X12: At L=1 the text is in the clear and the shape is fully English. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [clear-at-one] The page's seed key begins T. At L=1 it applies shift 19 and begins T VBIAXK VTG VATGZX; its IC equals the plaintext IC but its letters are encrypted. - MINOR | X13: Raise L and the whole-text IC slides down toward the random floor | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [slider-monotonic] Executed rotatingCipher: L=8/9 gives IC 0.0409/0.0413, and L=11/12/13 gives 0.0394/0.0406/0.0423. The trend is not monotone. - MINOR | X14: Every cipher number is recomputed live. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [every-cipher-live] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - MINOR | X15: Every figure live-recomputed and checked offline: research/invariant-of-relabeling/verify.mjs · 17/17. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [every-figure-offline] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - MINOR | X16: Every number recomputed live and checked offline (verify.mjs, 17/17). | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [dek-every-number] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - MINOR | X17: Run both; every figure recomputed live. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [og-every-figure] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - MINOR | X18: Run both, every number checked live. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [meta-every-number] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - WRONG | X19: That stratum cracks Linear B from pure structure and named this very tie as its cleanest cross-seam edge. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [parent-named-tie] Cambridge process.pdf, pp. 3-4, combines the structural grid with guessed a, Cypriot ni and Cretan names. The June 19 layer log separately supports the claimed cross-seam tie. - STALE | X20: "dateModified":"2026-06-19" | git show 4eef9d2d149 -- public/strata/the-invariant-of-relabeling/index.html | [date-modified] The layer was revised by 4eef9d2d149 on August 1 and 08d0604448d on September 28. June 19 was no longer the latest modification date. - STALE | X21: Last verified 19 June 2026, when it was written. | git show a2b8a9e4259:research/verifier-independence/depends-2.json | [last-verified] Read a2b8a9e4259:research/verifier-independence/depends-2.json, which records this verifier exiting 0 after the June 19 run. Ran the revised layer verifier on October 14: 17/17, exit 0. - WRONG | X22: at a single name is enough to make the whole arrangement speak aloud at last. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [passage-single-name] Executed the grid and both buttons: zero, one and two distinct-row/column pins leave 36, 4 and 1 assignments. Cambridge describes multiple guesses and deductions. The composed passage exactly matches the verifier input, but its closing sufficiency claim is false. - MINOR | X23: which is exactly why a single lucky name could do so much. | https://ignca.gov.in/Asi_data/17617.pdf | [historical-one-name-cause] Documents, p. 22, supports substantial propagation from guessed values to 31 signs. It does not make an erroneous Vigenère key-space comparison the exact historical cause. - WRONG | X24: the difference is exactly why the harder-looking problem needed only a name or two. | https://ignca.gov.in/Asi_data/17617.pdf | [one-or-two-historical] Read Documents, pp. 21-22: five-town table, propagation to 31 signs, and further vocabulary and inflection checks. The source does not prove either that two names sufficed or that exactly five were necessary. - WRONG | X25: then let one anchor break the rest | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | [one-anchor-rest] Executed the grid and both buttons: zero, one and two distinct-row/column pins leave 36, 4 and 1 assignments. Cambridge describes multiple guesses and deductions. The composed passage exactly matches the verifier input, but its closing sufficiency claim is false. - MINOR | X26: index of coincidence 0.0714 | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [generated-ic] Executed the exact page code: the unchanged passage has 444 letters and IC 14054/196692 = 0.0714518129868017, giving 0.0715 at four decimals. Both existing live readouts were already correct. - MINOR | X27: key length L = 1 whole-text IC 0.0714 | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | [generated-whole-initial] Executed the exact page code: the unchanged passage has 444 letters and IC 14054/196692 = 0.0714518129868017, giving 0.0715 at four decimals. Both existing live readouts were already correct. - MINOR | K1: the index of coincidence that broke the Vigenère cipher (aw-plain placard, as one of "Two famous cracks as one move") | https://pages.mtu.edu/~shene/NSF-4/Vigenere.pdf | [ic-broke-vigenere] MTU, pp. 1-2, places the published Kasiski attack in 1863 and Friedman's IC method in 1922. The later attack is valid; the plain description now says IC can break Vigenère. - WRONG | K2: Linear B cracked from pure structure (joins list; relates note) | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [linear-b-pure-structure] Cambridge process.pdf, pp. 3-4, combines the structural grid with guessed a, Cypriot ni and Cretan names. The June 19 layer log separately supports the claimed cross-seam tie. - MINOR | K3: Michael Ventris guessed that four often-repeated sign-groups were Cretan place-names | https://ignca.gov.in/Asi_data/17617.pdf | [historical-guessed-towns] Read Documents in Mycenaean Greek, printed pp. 21-22 (PDF pp. 58-59). Its table has five towns, including Lyktos and all four named here. Those four remain valid examples. - WRONG | K4: Wrong L: still a mixture, still flat. | public/strata/the-invariant-of-relabeling/index.html; research/invariant-of-relabeling/artifact.json | [wrong-period-flat] Executed the page script: mean column IC is 0.07254864311315923 at 7 and 0.07134216589861751 at 14. Multiples also keep each column on a single shift. - WRONG | K5: the cipher's residual ambiguity is astronomical - 26! per column - so you need a whole frequency table to break it. | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [cipher-26factorial-column] Ca' Foscari's recovery loop tries 0 through 25. The local vigenere function adds one key-letter shift modulo 26. Each known-period column has 26 keys; 26! remains valid for unrestricted alphabet substitutions. - WRONG | K6: so you need a whole frequency table to break it. | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | [whole-frequency-table] Ca' Foscari recovers relative shifts and then tries 26 possible first-key letters, with a reference text as an alternative. Algebraically c = p + k mod 26 gives k from one known plaintext/ciphertext letter pair. - CONFIRMED | K7: Shannon counts the redundancy of English in bits | https://jontalle.web.engr.illinois.edu/uploads/537.F18/Papers/Shannon50b.pdf | [shannon-redundancy] REJECTED with evidence: read and visually inspected Shannon50b.pdf, PDF p. 5, printed p. 252 [130]. Its bound R >= log2(27) + sum(q_i log2(q_i)) is an unnormalized binary-log entropy deficit, in bits per character. Percentage redundancy elsewhere does not exclude this absolute quantity. The directing triage's rejection holds. - MINOR | K8: plaintext (L=1) 0.0714 | public/strata/the-invariant-of-relabeling/index.html; recreated: node on the page's ic() and PLAINTEXT | [plaintext-value] Executed the exact page code: the unchanged passage has 444 letters and IC 14054/196692 = 0.0714518129868017, giving 0.0715 at four decimals. Both existing live readouts were already correct. - WRONG | K9: At L=1 the text is in the clear and the shape is fully English. | public/strata/the-invariant-of-relabeling/index.html | [clear-at-one] The page's seed key begins T. At L=1 it applies shift 19 and begins T VBIAXK VTG VATGZX; its IC equals the plaintext IC but its letters are encrypted. - MINOR | K10: Every cipher number is recomputed live. | public/strata/the-invariant-of-relabeling/index.html | [every-cipher-live] Read the HTML and rendering code: dynamic histograms, IC and grid counts coexist with static labels. The standalone verifier reads no HTML and does not certify every prose figure. - STALE | K11: "dateModified":"2026-06-19" | public/strata/the-invariant-of-relabeling/index.html; git show 4eef9d2d149; git show 08d0604448d | [date-modified] The layer was revised by 4eef9d2d149 on August 1 and 08d0604448d on September 28. June 19 was no longer the latest modification date. - WRONG | K12: at a single name is enough to make the whole arrangement speak aloud at last. (the passage shown in Movement I) | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | [passage-single-name] Executed the grid and both buttons: zero, one and two distinct-row/column pins leave 36, 4 and 1 assignments. Cambridge describes multiple guesses and deductions. The composed passage exactly matches the verifier input, but its closing sufficiency claim is false. - WRONG | K13: Linear B's grid ... so one guessed place-name cuts 36 readings to 4 and a second forces the whole arrangement to speak. (dek; meta and JSON-LD: "cuts 36 readings of the grid to 4 and a second forces the rest") | src/content/strata/the-invariant-of-relabeling.md; research/invariant-of-relabeling/verify.mjs | [metadata-collapse] The page sets R=C=3 and enumerates nine-sign permutations, yielding 36, 4 and 1. Cambridge process.pdf, p. 1, describes 87 phonetic signs in the actual script. The counts are toy results. - WRONG | K14: the difference is exactly why the harder-looking problem needed only a name or two. | https://ignca.gov.in/Asi_data/17617.pdf | [one-or-two-historical] Read Documents, pp. 21-22: five-town table, propagation to 31 signs, and further vocabulary and inflection checks. The source does not prove either that two names sufficed or that exactly five were necessary. - WRONG | K15: then let one anchor break the rest (aw-plain placard: "Both seize a structure that relabeling the symbols cannot move, then let one anchor break the rest") | public/strata/the-invariant-of-relabeling/index.html; git show 881d6d962f4 | [one-anchor-rest] Executed the grid and both buttons: zero, one and two distinct-row/column pins leave 36, 4 and 1 assignments. Cambridge describes multiple guesses and deductions. The composed passage exactly matches the verifier input, but its closing sufficiency claim is false. - CONFIRMED | A cipher called unbreakable for three centuries | https://pages.mtu.edu/~shene/NSF-4/Vigenere.pdf | Carried from N2-codex [cipher-three-centuries]: The PDF says "for nearly three centuries" and names Kasiski's 1863 publication. Approximately three centuries describes the historical interval, not the cipher's present age. - CONFIRMED | a 300-year-old cipher | https://pages.mtu.edu/~shene/NSF-4/Vigenere.pdf | Carried from N2-codex [cipher-three-hundred-years]: Same historical interval as entry 1; "a 300-year-old cipher" describes its age at the famous break. The source discusses sixteenth-century origins and the 1863 publication. - CONFIRMED | a script unread for three thousand years | https://www.classics.cam.ac.uk/research/projects/mycep/decipherment | Carried from N2-codex [script-three-millennia]: The supplied excerpt dates the writing, rather than explicitly dating its unread interval. Supplemented by Cambridge's project page: "c.1400-1200 B.C.E." and "The decipherment was achieved in 1952". Approximately 3,150 years from the end of use supports the rounded claim. - CONFIRMED | a Bronze Age script | https://www.classics.cam.ac.uk/research/projects/mycep | Carried from N2-codex [script-bronze-age]: The source says "a script used in Bronze Age Crete and Greece (c.1400-1200 B.C.E.)". - CONFIRMED | Alice Kober showed that Linear B's signs fall on a grid | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [kober-grid]: The source says Kober was "able to begin constructing a ‘grid’ showing the relationships between signs." This establishes the method, not a completed grid of every sign. - CONFIRMED | you can tell which signs share a consonant and which share a vowel from how words inflect, with no idea what any sign says. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [kober-inflection]: The discussion of inflection and the bu-o-no / bu-o-na example explains a shared consonant without knowing its value; the subsequent grid explains the vowel relation. The excerpt and surrounding pages support the principle. - CONFIRMED | the man who broke Linear B | https://www.classics.cam.ac.uk/research/projects/mycep/decipherment | Carried from N2-codex [ventris-broke-linear-b]: Cambridge says "The decipherment was achieved in 1952 by a young British architect, Michael Ventris." - CONFIRMED | That is the whole shape of the 1952 decipherment, in miniature. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [1952-shape]: Read with HTML:330 and :407, which explicitly call this a toy. Cambridge describes structural relationships, guessed values, then propagation. The miniature represents that sequence; it is not a literal reconstruction of every historical step. - CONFIRMED | Knossos, Amnisos, Tylissos, Phaistos. | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [historical-towns-list]: All four places occur in the p.22 table. Source-record transcription issue: the scan prints "Τυλισός Κνωσός Ἀμνισός", while N1 gives "Τυλισσός Κνωσσός Ἀμνισσός". The same row's "Λύκτος Φαιστός" matches, a visual positive control. The English place-names on the page are sound. - CONFIRMED | Each pinned a few rows and columns | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [historical-propagation]: Printed p.22 says "By the inevitable ‘chain-reaction’ thirty-one of the signs on the grid would thereby receive fixed phonetic values." Cambridge p.4 explains the row and column propagation. - CONFIRMED | signs that appeared in no place-name then came out forced, spelling real Greek | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [historical-tripod-forced]: Printed p.22 describes "thirty-one" propagated values and further Greek vocabulary; Cambridge p.4 says "others could immediately be deduced." The record's p.5 tripod confirmation alone does not establish this derivation, but the earlier discussion does. - CONFIRMED | ti-ri-po, "tripod." | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [tripod-gloss]: Printed p.43 gives "τρίπους (Hom. τρίπος) appears as ti-ri-po". Independently, https://lrc.la.utexas.edu/lex/master/2032 lists "ti-ri-po" and "tripod". N1's Cambridge excerpt is about "ti-ri-po-de = tripode", the dual, so it is not the right excerpt for the singular gloss. The page's gloss is correct. - CONFIRMED | Real Linear B has ~90 signs | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [linear-b-sign-count]: Cambridge says "Linear B has 87 different phonetic signs". Approximately 90 is justified for the phonetic inventory under discussion; the same source separately describes ideograms. - CONFIRMED | A simple substitution cipher swaps every letter for another symbol | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [substitution-definition]: The source says the substitution "maps a letter always to the very same letter" and represents its key as an alphabet permutation. - CONFIRMED | It cannot change how often the masks recur. | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [frequency-preservation]: The source says "This preserves the statistics of the plaintext". A bijection permutes the counts and cannot alter recurrence multiplicities. - CONFIRMED | The index of coincidence is the chance that two letters picked from the text are the same | https://pages.mtu.edu/~shene/NSF-4/Tutorial/VIG/Vig-IOC.html | Carried from N2-codex [ic-probability]: MTU defines IC as "the probability of two randomly selected letters being equal." The displayed implementation samples distinct positions, using N(N-1). - CONFIRMED | The index of coincidence is invariant over all 26! ways to swap the letters of the alphabet | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [ic-all-permutations]: The source's statistical preservation statement and IC's sum of n_i(n_i-1) establish invariance under any permutation of the 26 counts. This is a mathematical deduction, not a claim to have enumerated 26! trials. - CONFIRMED | 26! ≈ 4×10²⁶ ways to relabel the alphabet. | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [factorial-number]: The source explicitly gives 26! and approximately 4 x 10^26. Recomputed exactly: 403291461126605635584000000. - CONFIRMED | the cipher hides the letters; it cannot hide the language. | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [cannot-hide-language]: In the surrounding discussion this means preserving a language's frequency shape. The source says "This preserves the statistics of the plaintext". It does not prove that a single IC value uniquely identifies a language, and this pass does not read it as that stronger claim. - CONFIRMED | The Vigenère cipher defeats the attack above by rotating L alphabets in turn | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | Carried from N2-codex [poly-rotation]: The example repeats "FLUTE" and uses its five shifts on successive letters. This is the described rotation of L alphabets. - CONFIRMED | the same letter becomes different symbols at different positions. | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | Carried from N2-codex [poly-position]: The source says "one letter is not always mapped to the same one" and shows I encrypting to C and M in different positions. Repeated key letters may still reuse an alphabet. - CONFIRMED | That genuinely flattens the lumpiness toward random. | https://pages.mtu.edu/~shene/NSF-4/Tutorial/VIG/Vig-IOC.html | Carried from N2-codex [poly-flattening]: The cited example reports "The IC reduces to 0.041989". The page's own VENTRIS example also falls from 0.0714518129868017 to 0.04261484961259228. This supports the qualitative effect, not monotonicity for every possible key. - CONFIRMED | it repeats every L letters | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | Carried from N2-codex [poly-repeats]: The source's repeated key establishes period L, possibly with a smaller fundamental period if the key itself repeats. - CONFIRMED | If L is the true period, each column was enciphered with a single alphabet, so its own IC springs back to English. | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | Carried from N2-codex [true-period-columns]: The source requires the subtexts' IC to be close to the language's characteristic IC. At the true period each column has one Caesar shift; the page's L=7 mean is 0.07254864311315923. - CONFIRMED | The first column-set to read English is the period. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [first-english-period]: For this supplied ciphertext, the smallest tested L above the displayed threshold 0.055 is seven. The external teaching source describes this heuristic. It is not an unconditional theorem about arbitrary short ciphertexts. - CONFIRMED | Once the period is known, each column is a plain Caesar shift and a frequency table reads off its letter | https://secgroup.dais.unive.it/teaching/cryptography/polyalphabetic-ciphers/ | Carried from N2-codex [caesar-columns]: A Vigenere column is one shift. The source's recovery routine searches shifts 0 through 25 and compares mutual frequencies. The page's description is a valid frequency-analysis route, with the usual sampling limitation. - CONFIRMED | E is commonest | https://pages.mtu.edu/~shene/NSF-4/Tutorial/VIG/Vig-IOC.html | Carried from N2-codex [e-commonest]: The source's table explicitly puts "E (13.11%)" first among its most frequent letters. This is a general English frequency reference, not a promise about every short passage. - CONFIRMED | In a lost script, the labels and the language are unknown. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [lost-script-unknown-language]: For the Linear B case being contrasted, Cambridge says "no-one could read the script, and no-one knew what the language of the inscriptions was." - CONFIRMED | The index of coincidence is William Friedman's, 1922 | https://www.nsa.gov/portals/75/documents/news-features/declassified-documents/friedman-documents/publications/folder_233/41761039080018.pdf | Carried from N2-codex [friedman-date]: Opened the NSA PDF through the web tool after curl returned 403. Printed p.1, footnote 1 states the earlier paper was "published in 1922 by the Riverbank Laboratories, Geneva, Ill." Authorship and publication year check out. - CONFIRMED | the grid is Kober's and Ventris's, 1945–52. | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [kober-ventris-date]: The supplied p.15 excerpt says "between 1943 and her premature death in 1950", which alone does not establish 1945. Supplement: https://www.journals.uchicago.edu/doi/10.2307/499698 dates Kober's inflection paper "1 April 1945". Cambridge shows Ventris's "Work Note 17" of February 1952. The broad 1945-52 development interval is supported, not a claim that the finished grid appeared in 1945. - CONFIRMED | Three runnable movements: | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [three-movements]: Read and executed the pinned inline program with a minimal local DOM. Its three movement sections operate: scrambling, a key-length slider and period scan, and a grid with two anchor controls. - CONFIRMED | It is identical, to twelve decimal places, under every one of the 26! ≈ 4×10²⁶ ways to relabel the alphabet. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [ic-twelve-decimals]: IC depends only on the integer count multiset, so any bijection preserves it exactly. The verifier checks 500 distinct substitutions at tolerance 1e-12 and all pass. The universal mathematical assertion follows from the sum, not from that finite sample alone. - CONFIRMED | random floor (1/26) | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [random-floor]: For independent uniform letters, equality probability is 26(1/26)^2 = 1/26. "Floor" is the population benchmark; finite sample estimates may fall below it. - CONFIRMED | random floor (1/26)0.0385 | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [floor-value]: 1/26 = 0.038461538461538464, which rounds to the printed 0.0385 at four decimals. - CONFIRMED | key length L = | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [period-slider]: The slider ranges from 1 to 14 and THEWORDFORBOTH supplies exactly 14 key letters; slice(0,L) supplies the chosen prefix. - CONFIRMED | Cut the ciphertext (enciphered with a real 7-letter key) into L columns | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [real-seven-key]: The period scan uses vigenere(PLAINTEXT, KEY), where KEY is "VENTRIS" and its seven letters do not form a repeated shorter key. - CONFIRMED | The key is VENTRIS | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [key-ventris]: The pinned script assigns var KEY = "VENTRIS"; the verifier and artifact use the same key. - CONFIRMED | Here is the principle as a toy you can run: a 3×3 grid of signs, the rows consonants, the columns vowels. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [toy-grid]: The code uses R=3, C=3 and nine abstract marks; sameRow and sameCol encode the two stated relations. No claim of reconstructing the actual tablets is made in this sentence. - CONFIRMED | It leaves the grid fixed up to relabeling the rows among themselves and the columns among themselves. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-row-column]: For a full grid with the row and column relations distinguished, every automorphism is an independent row permutation and column permutation. The pinned enumerator and an independent 9! enumeration both give 36. - CONFIRMED | the grid of shared rows and columns, fixed over all of them. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-invariant]: The two equivalence relations are preserved by all independent row/column relabelings. These are exactly the relations tested in countSymmetries. - CONFIRMED | Counted by brute force, every relation-preserving permutation enumerated | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [grid-count-enumerated]: countSymmetries recursively tries unused destination cells and rejects a branch only when it violates an already assigned pair's row or column relation. Thus every admissible full permutation is counted. - CONFIRMED | the same routine the verifier runs. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [same-grid-routine]: Compared the browser implementation with verifier section 4: both recursively assign unused destinations, preserve both pair relations, and constrain anchors to fixed points. The browser specializes the verifier's r,c parameters to 3,3. - CONFIRMED | one guessed place-name pins specific sounds and cuts 36 readings to 4; a second forces the grid. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-one-anchor]: For this explicit toy, fixing cell 0 leaves 2! times 2! = 4 possibilities; also fixing cell 4 leaves 1! times 1! = 1. Calling each pin a place-name is the toy's stated interpretation. - CONFIRMED | The grid here is a toy, not the tablets. | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [grid-not-tablets]: The instrument has nine abstract signs, R=C=3, and fixed-point pins. HTML:407 explicitly states its illustrative scope. It is not a historical 87-sign implementation. - CONFIRMED | the same routine in research/invariant-of-relabeling/verify.mjs (§4). | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-verifier-section]: The cited file exists at the pin; section 4 contains countSymmetries and the three grid checks. Compared it with the browser routine. - CONFIRMED | It agrees with r!·c! because the math says it must, but the page shows the count, not the claim. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-formula-match]: The algorithm counts admissible permutations instead of returning factorial(r)*factorial(c). The factorial formula is used separately for comparison; 3!3!=36, 2!2!=4, 1!1!=1. - CONFIRMED | Two names in fresh rows and columns force every sound. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-fresh-rows]: Within the stated 3x3 toy, pins 0 and 4 occupy distinct rows and columns and leave exactly one assignment. This is conditional on the chosen grid, not a rule for arbitrary larger grids. - CONFIRMED | The grid is most of the way home on a single guess. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-home-message]: A single pin removes 32 of 36 possible assignments, or 88.9 percent. "Most of the way" is qualitative and consistent with that reduction; four possibilities still remain. - CONFIRMED | One place-name pins a sign to a sound | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [grid-one-pin-message]: The first button sets anchors=[0], fixing that sign's row and column. In this explicit toy one place-name represents one fixed-point constraint. - CONFIRMED | the signs simply don't say which sound they are. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-no-sounds]: With no assigned sound labels, the full 3x3 relation structure permits 36 relabelings. Thus the relations do not choose a particular labeling. - CONFIRMED | = (3-1)!·(3-1)! = 2!·2! | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-factorial-one]: (3-1)! times (3-1)! = 2! times 2! = 4. Verified by enumeration and printed in the one-pin state. - CONFIRMED | = (3-2)!·(3-2)! = 1 | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-factorial-two]: (3-2)! times (3-2)! = 1! times 1! = 1. Verified by enumeration and printed in the two-pin state. - CONFIRMED | Plug in the four place-names and watch the forced signs spell Greek, step by step. | git show 60ea23c371:public/strata/the-grid-that-spoke-greek/index.html | Carried from N2-codex [parent-grid-runnable]: The parent page contains four town selectors, a row/column propagation routine, a grid and Greek word outputs. This accurately describes the linked instrument. It is not independent historical corroboration of its simplified reconstruction. - CONFIRMED | The index of coincidence, worked end to end on a real message, with a companion film. | git show 60ea23c371:public/strata/the-cipher-couldnt-hide/index.html | Carried from N2-codex [parent-cipher-film]: The parent contains an actual composed plaintext, encrypted message and recovery engine, plus a video source film.mp4. Live HEAD for that film returns 200, video/mp4, content-length 8023729. "Real message" here means the instantiated demonstration, not an archival intercepted message. - CONFIRMED | The same weapon in a fifth field, cited as kin rather than walked as a mask (to keep the portal's five from sprawling). | git show 60ea23c371:src/content/strata/find-what-doesnt-change.md | Carried from N2-codex [fifth-field-kin]: The source layer's relates note contains this exact sentence and explicitly treats cryptanalysis as an extra kinship link rather than a walked example. "Fifth field" is its grouping description, not a measured cipher statistic. - CONFIRMED | An immersive layer; it renders as its own standalone page at the `href` above | git show 60ea23c371:public/strata/the-invariant-of-relabeling/index.html | Carried from N2-codex [standalone-source]: The live href returns the standalone HTML. Its prose and inline JavaScript match the pinned file; only production analytics scripts were added to the head. - CONFIRMED | its body is not rendered. | git show 60ea23c371:src/pages/strata/[...slug].astro | Carried from N2-codex [markdown-unrendered]: getStaticPaths filters with !s.data.href. This layer has an href, so the markdown body does not become the served article. Positive control: its HTML hero is present in the curl capture. - CONFIRMED | This layer's check works the answer out again from scratch, independently of this page. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [check-independent]: The file imports writeFileSync, not readFileSync, and computes its own plaintext, cipher and grid results. The standalone reader run reproduces 17/17 without page files. - CONFIRMED | It can settle the finding; it cannot notice a figure that drifted on this page. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [check-settles-finding]: The check has no page read or page input argument, so it verifies the independent computations while remaining insensitive to prose drift. This accurately states its limit. - CONFIRMED | It is published, and it runs on its own: one file, and Node. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [check-standalone]: Downloaded only the linked verify.mjs into an empty directory. HTTP 200, 15519 bytes. node verify.mjs exits 0 with 17 passed and 0 failed, using only Node built-ins. It writes artifact.json beside itself in scratch. - CONFIRMED | A cipher can change which letter wears which mask, but it cannot change how | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-1]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | often the masks recur, and that stubborn unevenness is a fingerprint no costume | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-2]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | can hide. A script is stranger still, for there both the masks and the language | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-3]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | behind them are unknown, and yet the same handle is there to be seized: the way | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-4]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | the signs lean on one another holds its shape no matter what sounds we later give | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-5]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | them. So we read the structure first and the meaning second, and one lucky guess | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-6]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | at a single name is enough to make the whole arrangement speak aloud at last. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [composed-passage-quote-7]: CONFIRMED only as an exact transcription of the composed PLAINTEXT kept in the page and verifier, compared again by this fixer. The closing single-name factual claim is WRONG and is addressed separately at X22/K12. - CONFIRMED | research/invariant-of-relabeling/verify.mjs | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [verifier-reference]: The pinned file begins with the supplied verifier header and is the exact path linked from the page. The live mirror is byte-identical to this verifier. - CONFIRMED | gaps in the grid, signs that resist it | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [linear-b-grid-gaps]: Read p.23, figure 4. The "Basic values" table has empty cells and a "Homophones" column, plus unassigned numbered signs. The record's two heading words require the figure itself as evidence, which was inspected. - CONFIRMED | a script so lossy one spelling can be two words. | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [linear-b-ambiguous]: Printed p.43 says "More than one equivalent may often be possible within these rules" and describes omitted distinctions. This directly supports possible spelling ambiguity. - CONFIRMED | That is the whole shape of the 1952 crack. | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [historical-whole-crack-message]: Read in its explicitly miniature/toy context, the structure, guesses and propagation sequence matches p.22. It does not assert that nine signs and two pins literally reproduce the historical work; overstatements of sufficiency are separately flagged. - CONFIRMED | signs in no name now read off for free, the way ti-ri-po did. | https://ignca.gov.in/Asi_data/17617.pdf | Carried from N2-codex [historical-for-free-message]: The source describes propagated phonetic values and subsequent Greek readings on p.22; p.43 confirms ti-ri-po. N1's Cambridge p.5 excerpt establishes later confirmation, not this propagation by itself. It should cite the earlier grid explanation as well. - CONFIRMED | A code or a script assigns arbitrary labels to an underlying system | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [codes-arbitrary-labels]: For the substitution cipher and syllabic-grid models on this page, the encipherment or sign labels represent an underlying text or sound inventory. This is the frame's model assumption, not a claim that every writing system is a bijective alphabet. - CONFIRMED | But some structure among the labels is forced by the system, and so it cannot be moved by relabeling. | https://www.classics.cam.ac.uk/system/files/documents/process.pdf | Carried from N2-codex [structure-before-sound]: Cambridge's unknown-sound grid supplies the concrete example: row and column relationships can be retained while the sound names remain unknown. The mathematical invariance is also shown in the toy enumeration. - CONFIRMED | the index of coincidence is the residue of that redundancy a cipher can't erase | https://secgroup.dais.unive.it/teaching/cryptography/monoalphabetic-ciphers/ | Carried from N2-codex [ic-redundancy]: In this monoalphabetic context, statistics are preserved by substitution. IC summarizes single-letter unevenness; it is not the entirety of Shannon's redundancy and the page itself shows that polyalphabetic encryption can change it. - CONFIRMED | Linear B's grid of which signs share a vowel and which share a consonant is invariant over only r!·c! ways to assign sounds | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-rfactorial-cfactorial]: The formal assertion holds for a full grid with known row/column classes and fixed inventories of row and column labels. Read with the explicit toy limitation at HTML:407-411. The verifier does not establish a literal count of all real Linear B decipherment hypotheses. - CONFIRMED | can't tell you which sounds go where. | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [grid-model-relations]: Independent permutations of the three row labels and three column labels preserve both relations and yield 36 possibilities. The structure alone therefore does not select the sounds. - CONFIRMED | a consonant×vowel structure fixes the signs only up to r!·c! sound-assignments | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [full-grid-formula]: For the complete r by c toy with distinguished row and column relations, the count is r!c!. The 3 by 3 instance has been independently enumerated, with the same result. - CONFIRMED | the script's residual ambiguity is far smaller than the cipher's | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [model-smaller-symmetry]: As the explicitly formal comparison of full monoalphabetic S26 and the 3 by 3 toy, 26! is larger than 36. This does not rescue the separate erroneous application to known-period Vigenere columns at entries 34-36. - CONFIRMED | = 3!·3!  (relabel rows, relabel columns) | git show 60ea23c371:research/invariant-of-relabeling/verify.mjs | Carried from N2-codex [initial-grid-factorial]: 3! times 3! = 36. The page's initial formula and independent exhaustive enumeration agree. - CONFIRMED | scramble a real passage's alphabet and watch the by-letter histogram jump | public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [histogram-shape]: renderHist draws the by-letter bars from the substituted text. - CONFIRMED | the one sorted by height never moves. | public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [sorted-fixed]: the sorted bars come from the counts sorted; a relabeling only permutes counts. A scrambled run printed IC 0.0715 again. - CONFIRMED | unique scrambles tried | public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [scramble-counter]: doScramble counts distinct keys. - CONFIRMED | The scan finds 7 with no help. | public/strata/the-invariant-of-relabeling/index.html; research/invariant-of-relabeling/artifact.json | Carried from N2-claude [scan-seven]: first L with coset IC above 0.055 is 7 (0.07255); the DOM-stub run prints period 7. - CONFIRMED | Structure alone: 36 ways to read the grid. | research/invariant-of-relabeling/verify.mjs; public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [grid-initial]: 36 counted live (DOM stub) and in §4. - CONFIRMED | 36 → 4 → 1 is produced by enumerating every permutation of the nine signs that preserves both relations | research/invariant-of-relabeling/verify.mjs | Carried from N2-claude [grid-all-permutations]: §4, rerun. - CONFIRMED | all run in your browser from the same algorithm the verifier checks (17/17) | public/strata/the-invariant-of-relabeling/index.html; research/invariant-of-relabeling/verify.mjs | Carried from N2-claude [every-cipher-algorithm]: same ic, vigenere and coset mean; 17/17 on the pinned copy and on the copy downloaded from https://artwaste.land/checks/research/invariant-of-relabeling/verify.mjs (byte-identical). - CONFIRMED | which itself reuses the definition the parent stratum proved (23/23). | research/index-of-coincidence/verify.mjs | Carried from N2-claude [parent-23]: The checker recorded the parent verifier at 23/23. This fixer read that record; it did not rerun or alter the other layer. - CONFIRMED | The demonstration passage is composed English prose, so no figure here leans on quoting an outside text. | public/strata/the-invariant-of-relabeling/index.html; git show 60ea23c371:memory/log.d/2026-06-19T1100Z-invariant-of-relabeling.md | Carried from N2-claude [composed-passage]: log: "The demo passage is composed prose (no external-quote claim)". - CONFIRMED | The freedom drops from 36 to 4. | research/invariant-of-relabeling/verify.mjs; public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [grid-drop-message]: true (count 4), though this sentence is still typed rather than filled from the count. - CONFIRMED | The structure is solved | public/strata/the-invariant-of-relabeling/index.html | Carried from N2-claude [grid-structure-solved]: gated on n === 1 since 08d0604448d. - CONFIRMED | This page joins two earlier layers | src/content/strata/the-grid-that-spoke-greek.md; src/content/strata/the-cipher-couldnt-hide.md | Carried from N2-claude [joins-two-layers]: dated 2026-06-19T13:30Z and 2026-06-19; this layer 2026-06-19T16:30Z. - CONFIRMED | "datePublished":"2026-06-19" | src/content/strata/the-invariant-of-relabeling.md | Carried from N2-claude [date-published]: frontmatter date 2026-06-19T16:30:00Z. - CONFIRMED | The instrument and the verifier both run 36 → 4 → 1 | research/invariant-of-relabeling/verify.mjs | Carried from N2-claude [august-two-steps]: §4 and the DOM-stub run. - CONFIRMED | smallest English-like period 7 | public/strata/the-invariant-of-relabeling/index.html; research/invariant-of-relabeling/artifact.json | Carried from N2-claude [generated-period]: rendered 7. - CONFIRMED | (added) One Bit of Structure: Two famous cracks that were one move: find the quantity relabelling can't change. | git show 60ea23c371:src/content/strata/liber-primus-measured.md | Carried from N2-claude [added-nearby-kin]: Nearby-layers line on this page, linked to /strata/liber-primus-measured/ (200); verbatim in that layer's relates note to this one. ## Findings - [fixed] X1 (ic-broke-vigenere): The markdown plain description and its HTML placard now say can break; no page correction line for this MINOR precision change. - [fixed] X2 (linear-b-pure-structure): Replaced pure structure in the joins text, relates note, nearby copy and markdown body with structural analysis plus phonetic and place-name guesses. Correction 1 records the false account. - [fixed] X3 (historical-guessed-towns): Changed the historical sentence to including Knossos, Amnisos, Tylissos and Phaistos; no exclusive count and no MINOR correction line. - [fixed] X4 (wrong-period-flat): The period explanation includes the fundamental period and its multiples; nonmultiples generally mix alphabets. Correction 2 gives this scan's values, and the verifier comment matches. - [fixed] X5 (cipher-known-language): Qualified the comparison as In this cipher example; no MINOR correction line. - [fixed] X6 (cipher-ambiguity-column): Separated unrestricted substitution from Vigenère columns in the comparison, verifier and generated artifact. The verifier now counts 26 distinct shifts. Correction 3 states both scopes. - [fixed] X7 (cipher-26factorial-column): Separated unrestricted substitution from Vigenère columns in the comparison, verifier and generated artifact. The verifier now counts 26 distinct shifts. Correction 3 states both scopes. - [fixed] X8 (whole-frequency-table): Frequency scoring is an available method, not a necessary whole table. Correction 4 explains the alternative; the verifier comments match. - [fixed] X10 (difference-only-group): The plain description, metadata and comparison explicitly identify the model analogy and outside evidence. Correction 5 rejects the historical sufficiency inference. - [fixed] X11 (plaintext-value): The static plaintext reference now starts at 0.0715 and is filled from plainIC; the two N1 generated-readout transcriptions now say 0.0715. No MINOR correction line and no plaintext change. - [fixed] X12 (clear-at-one): Replaced the in-the-clear claim with one Caesar shift preserving the frequency shape and IC. Correction 7 states what the key actually does. - [fixed] X13 (slider-monotonic): The note and script comment now state a general trend with fluctuations, with no MINOR correction line. - [fixed] X14 (every-cipher-live): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [fixed] X15 (every-figure-offline): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [fixed] X16 (dek-every-number): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [fixed] X17 (og-every-figure): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [fixed] X18 (meta-every-number): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [fixed] X19 (parent-named-tie): Replaced pure structure in the joins text, relates note, nearby copy and markdown body with structural analysis plus phonetic and place-name guesses. Correction 1 records the false account. - [dated] X20 (date-modified): Set JSON-LD dateModified to 2026-10-14. This field is explicitly exempt from an Updated line when the Corrected lines date the revision. - [dated] X21 (last-verified): Refreshed the placard to Last verified 14 October 2026, by the assay line (17/17). The triage explicitly exempts this verification date from an Updated line. - [fixed] X22 (passage-single-name): Pluralized and qualified the authored prose, renamed the grid heading, and placed an explicit explanation beside the unchanged passage. Correction 6 records the overclaim without changing the cipher input; matching verifier comments and OG source are corrected. - [fixed] X23 (historical-one-name-cause): The comparison now calls the toy an illustration of constraint propagation; no MINOR correction line. - [fixed] X24 (one-or-two-historical): Restricted the two-name result to the nine-sign toy in the dek and comparison. Correction 5 explains that these counts do not establish historical sufficiency. - [fixed] X25 (one-anchor-rest): Pluralized and qualified the authored prose, renamed the grid heading, and placed an explicit explanation beside the unchanged passage. Correction 6 records the overclaim without changing the cipher input; matching verifier comments and OG source are corrected. - [fixed] X26 (generated-ic): The static plaintext reference now starts at 0.0715 and is filled from plainIC; the two N1 generated-readout transcriptions now say 0.0715. No MINOR correction line and no plaintext change. - [fixed] X27 (generated-whole-initial): The static plaintext reference now starts at 0.0715 and is filled from plainIC; the two N1 generated-readout transcriptions now say 0.0715. No MINOR correction line and no plaintext change. - [fixed] K1 (ic-broke-vigenere): The markdown plain description and its HTML placard now say can break; no page correction line for this MINOR precision change. - [fixed] K2 (linear-b-pure-structure): Replaced pure structure in the joins text, relates note, nearby copy and markdown body with structural analysis plus phonetic and place-name guesses. Correction 1 records the false account. - [fixed] K3 (historical-guessed-towns): Changed the historical sentence to including Knossos, Amnisos, Tylissos and Phaistos; no exclusive count and no MINOR correction line. - [fixed] K4 (wrong-period-flat): The period explanation includes the fundamental period and its multiples; nonmultiples generally mix alphabets. Correction 2 gives this scan's values, and the verifier comment matches. - [fixed] K5 (cipher-26factorial-column): Separated unrestricted substitution from Vigenère columns in the comparison, verifier and generated artifact. The verifier now counts 26 distinct shifts. Correction 3 states both scopes. - [fixed] K6 (whole-frequency-table): Frequency scoring is an available method, not a necessary whole table. Correction 4 explains the alternative; the verifier comments match. - [fixed] K8 (plaintext-value): The static plaintext reference now starts at 0.0715 and is filled from plainIC; the two N1 generated-readout transcriptions now say 0.0715. No MINOR correction line and no plaintext change. - [fixed] K9 (clear-at-one): Replaced the in-the-clear claim with one Caesar shift preserving the frequency shape and IC. Correction 7 states what the key actually does. - [fixed] K10 (every-cipher-live): Narrowed the coverage claims in markdown, metadata, the edge note, footer, verifier header and OG source to interactive calculations and independent offline checks. No MINOR correction line. - [dated] K11 (date-modified): Set JSON-LD dateModified to 2026-10-14. This field is explicitly exempt from an Updated line when the Corrected lines date the revision. - [fixed] K12 (passage-single-name): Pluralized and qualified the authored prose, renamed the grid heading, and placed an explicit explanation beside the unchanged passage. Correction 6 records the overclaim without changing the cipher input; matching verifier comments and OG source are corrected. - [fixed] K13 (metadata-collapse): Explicitly qualified the toy in the dek, plain description, meta and JSON-LD descriptions, social descriptions and comparison labels. Correction 5 corrects the attribution. - [fixed] K14 (one-or-two-historical): Restricted the two-name result to the nine-sign toy in the dek and comparison. Correction 5 explains that these counts do not establish historical sufficiency. - [fixed] K15 (one-anchor-rest): Pluralized and qualified the authored prose, renamed the grid heading, and placed an explicit explanation beside the unchanged passage. Correction 6 records the overclaim without changing the cipher input; matching verifier comments and OG source are corrected.