---
slug: the-accountant-we-cant-read
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: 149
confirmed: 83
wrong: 35 # 14 WRONG plus 21 MINOR
unverifiable: 6
observed: 25
stale: 0
---

This was the directed correction pass for the-accountant-we-cant-read at f5e5f84f34, using the 37-item queue-1014c triage, the Codex source record, and the prior Codex and Claude checker rereads. The fixer re-read the evidence for each edited item, corrected the authored sources and regenerated only this layer's embedded data and published data copy. The 35 accepted items retain their triage levels; K10 and K14 are rejected with evidence. The remaining checker-confirmed claims and recorded observations are retained below with their source limits, alongside six untriaged claims the Codex checker could not verify. Their presence is not a fresh confirmation or a claim that unresolved evidence has been settled. MINOR repairs are recorded here without page correction lines.

## Claims

- MINOR | [X1] its language is unknown and probably has no living relative. | https://en.wikipedia.org/wiki/Minoan_language | Classification identifies no demonstrated close relative; it gives no probability that no living relative exists. Changed the dek to no known relative. The dek-derived PNG and WebP are rebuilt at landing; the categorical no-living-relative copy in the-signal-never-sent remains open below.
- MINOR | [X2] A Minoan palace scribe, thirty-six centuries dead, kept his accounts in a script that has never been deciphered. | https://lineara.xyz/items/HT13.html | HT 13 Context dates the inscription to 1500-1450 BCE, approximately 3,500 years before 2026; this does not date the scribe's death. Dated the account to about 3,500 years ago and removed the supposed death date.
- MINOR | [X3] A real human mistake, 3,600 years old, still legible across a language barrier no one has crossed. | https://lineara.xyz/items/HT13.html | The same HT 13 date is approximately 3,500 years ago. Its commentary reports a mismatch without assigning a cause. Corrected the age in prose and data, and described a discrepancy rather than an established human mistake.
- WRONG | [X4] Their exact values were a puzzle of their own until 2021, when a Bologna team (Corazza, Ferrara, Montecchi, Tamburini & Valério) recovered them with a computational, statistical and typological model | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Corazza et al., pp. 3 and 8, narrow possible assignments using premises and constraints; H and A remain tentative. Replaced recovery with proposed assignments in HTML, dek, markdown and data comments; added the fraction Corrected line.
- WRONG | [X5] Most Linear A tablets are accounts: a list of entries, each with a number, and at the bottom one recurring word, KU-RO, followed by a final figure. | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content ; https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Steele 2024, p. 97, distinguishes totalling records, processed copies and preliminary notes, with the last category the majority. Restricted KU-RO endings to some totalling accounts in prose, descriptions and data; added the frequency Corrected line.
- MINOR | [X6] HT 88 Six people, each owing one. Total: six. KI-RO [...] owed (deficit) KU-PA3-PA3 1 KA-JU 1 KU-PA3-NU 1 PA-JA-RE 1 SA-MA-RO 1 DA-TA-RE 1 KU-RO [...] total 6 | https://lineara.xyz/items/HT88.html | Younger, HT 88 .4-6, gives six unit entries totaling 6 and calls them probably people by name. Qualified the introduction and data headline; the generated ledger and source-list captions share that headline.
- MINOR | [X7] HT 88 [...] Six people, each owing one. Total: six. [...] transcription · SigLA cross-check | https://lineara.xyz/items/HT88.html | HT 88 commentary qualifies the people interpretation; six entries and total 6 are unchanged. Qualified the shared caption used by the ledger and source list.
- WRONG | [X8] HT 122 A grand total of two totals [...] off by one. KU-RO (a.8) 31 KU-RO (b.5) 65 PO-TO-KU-RO 97 | https://lineara.xyz/items/HT122a.html ; https://sigla.phis.me/document/HT%20122a/ | HT 122 a.8 records KU-DA 1 separately after KU-RO 31; b.6 accounts for 97 as 65 + 31 + 1. Added KU-DA 1 as a separate component and fixed this layer's ledger, data, descriptions, prose, markdown and verifier; added the HT 122 Corrected line. Copies in the-machine-made-of-months and the There, Here corpus remain open below.
- WRONG | [X9] HT 122 [...] A grand total of two totals [...] off by one. [...] transcription · SigLA cross-check | https://lineara.xyz/items/HT122a.html | The linked HT 122 commentary includes the extra unit and balances at 97. Replaced the off-by-one caption through the shared data headline; shares the HT 122 Corrected line.
- WRONG | [X10] should read 31 + 65 = 96, and instead reads 97: | https://lineara.xyz/items/HT122a.html | 96 is only the sum of the two subtotals. The recorded KU-DA unit brings the grand total to 97. Replaced the expected-96 prose with 31 + 65 + 1 = 97; shares the HT 122 Corrected line.
- WRONG | [X11] the “grand total” summing two sub-totals | https://lineara.xyz/items/HT122a.html ; https://www.worldhistory.org/Linear_A_Script/ | The grand-total gloss remains supported; the displayed HT 122 computation must also include KU-DA 1. Specified two subtotals plus the separate unit; shares the HT 122 Corrected line.
- MINOR | [X12] summed across a real landholding account, fractions and all: | https://lineara.xyz/items/HT123+124a.html | HT 123+124 is classified as specialized commodities and explicitly labels the displayed column OLIV; this source supplies no landholding identification. Used olive account and four entries in the prose and repeated data note.
- MINOR | [X13] they are names of people or places in a lost tongue. | https://lineara.xyz/items/HT88.html | HT 88 says probably people by name; HT 122 says probably places by name. Changed the general identification to possible personal or place names.
- MINOR | [X14] The language is unknown and unclassified, probably a language isolate, and certainly not Greek. | https://en.wikipedia.org/wiki/Minoan_language | The classification source identifies an unknown, unclassified language with no demonstrated relative; it does not give an isolate probability. Replaced probably an isolate with no demonstrated relative in prose and the data summary.
- MINOR | [X15] In 1974 David Packard pointed one of the first computers ever turned on an undeciphered script at Linear A | https://sites.google.com/view/minoan-script/volume-ii/vol-ii-part-i | The cited history distinguishes computer work in the 1967 thesis from the 1974 monograph. Dated publication of the computer study to 1974, not first use of a computer.
- MINOR | [X16] The modern neural decipherers that cracked Linear B and Ugaritic in 2019 simply don't apply here | https://aclanthology.org/P19-1303.pdf | Luo et al. 2019, p. 3146, report 67.3% of Linear B cognates translated; the model needs a known related-language corpus. Replaced cracked with cognate recovery in already deciphered scripts and the paper's 67.3% result.
- WRONG | [X17] 33 checks, all green | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | The June 19 birth log records 36/36. The pre-edit verifier reproduced 36 ok, 0 fail; the revised verifier produces 39 ok, 0 fail. Corrected the displayed count to 39 and dated the original 33-versus-36 error in a Corrected line.
- MINOR | [X18] that the decimal orders reconstruct each recorded value | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | The verifier decomposes only [56, 130, 31, 93, 20], all integers. Narrowed the markdown coverage statement and verifier heading to five integer examples.
- WRONG | [X19] the corpus editor reads them as the scribe's | https://lineara.xyz/items/HT13.html ; https://lineara.xyz/items/HT122a.html | HT 13 commentary reports 130.5 versus 131 without assigning its cause; HT 122 commentary explains a balanced grand total. Removed the claimed editor attribution and causal language throughout this layer; added the shared attribution Corrected line. Copies assigning HT 13's discrepancy to the scribe in is-it-writing, the-signal-never-sent and the There, Here corpus remain open below.
- WRONG | [X20] 31 + 65 = 96, but the grand total reads 97 [...] off by 1. | https://lineara.xyz/items/HT122a.html ; git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | The original runtime used only the two subtotals; the source also records KU-DA 1. Runtime now renders that separate unit, sums all components and reports 97 balances; shares the HT 122 Corrected line.
- WRONG | [X21] (added) the same discrepancies the corpus editor flags. | https://lineara.xyz/items/HT13.html ; https://lineara.xyz/items/HT122a.html | Only HT 13 is reported as a discrepancy in these two commentaries; HT 122 includes the extra unit. Replaced the plural attribution with the distinct HT 13 and HT 122 results; shares the attribution Corrected line.
- MINOR | [K1] A Minoan palace scribe, thirty-six centuries dead, kept his accounts in a script that has never been deciphered. | https://lineara.xyz/items/HT13.html | HT 13 Context dates the inscription to 1500-1450 BCE, approximately 3,500 years before 2026; this does not date the scribe's death. Dated the account to about 3,500 years ago and removed the supposed death date.
- MINOR | [K2] A real human mistake, 3,600 years old, still legible across a language barrier no one has crossed. | https://lineara.xyz/items/HT13.html | The same HT 13 date is approximately 3,500 years ago. Its commentary reports a mismatch without assigning a cause. Corrected the age in prose and data, and described a discrepancy rather than an established human mistake.
- WRONG | [K3] essentially conventional… it is impossible to say whether the homomorphic syllabograms were also homophones. | https://mnamon.sns.it/index.php?page=Scrittura&id=19&lang=en | Mnamon, Possibilities of reading, limits the statement to most cases and immediately describes exceptions. Restored the exact qualification and named and linked Mnamon; added the quotation Corrected line.
- MINOR | [K4] the most-repeated element (~15 times). | https://lineara.xyz/LinearAInscriptions.js | A fresh parse of the cited corpus finds SA-SA-RA forms on 14 inscriptions and A-TA-I-*301 forms on 14; exact spellings have different frequencies. Removed the undefined approximate count and superlative from the formula gloss, closing prose and data; used recurring element.
- WRONG | [K5] Their exact values were a puzzle of their own until 2021, when a Bologna team (Corazza, Ferrara, Montecchi, Tamburini & Valério) recovered them with a computational, statistical and typological model | http://web.archive.org/web/20241219084450id_/https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Corazza et al., pp. 3 and 8, narrow possible assignments using premises and constraints; H and A remain tentative. Replaced recovery with proposed assignments in HTML, dek, markdown and data comments; added the fraction Corrected line.
- WRONG | [K6] Most Linear A tablets are accounts: a list of entries, each with a number, and at the bottom one recurring word, KU-RO, followed by a final figure. | https://lineara.xyz/LinearAInscriptions.js | Steele 2024, p. 97, distinguishes totalling records, processed copies and preliminary notes, with the last category the majority. Restricted KU-RO endings to some totalling accounts in prose, descriptions and data; added the frequency Corrected line.
- MINOR | [K7] the meaning of one Linear A word is certain: ku-ro… must mean something like ‘total.’ | https://www.ics21.org/abstract-of-2nd-round/12-611/ | The cited secondary abstract continues the Chadwick quotation through the alternative so much. Restored the quotation's ending and identified and linked the secondary quotation source.
- MINOR | [K8] the numeral [after it equals] the sum of the tallies in the list. | https://en.wikipedia.org/wiki/Linear_A | Wikipedia's source phrase begins with a numeral representing and continues with the sum of the tallies. Restored the source wording and specifically attributed it to Wikipedia.
- MINOR | [K9] summed across a real landholding account, fractions and all: | https://revistas.usal.es/index.php/0544-3733/article/download/2888/2923 | HT 123+124 is classified as specialized commodities and explicitly labels the displayed column OLIV; this source supplies no landholding identification. Used olive account and four entries in the prose and repeated data note.
- CONFIRMED | [K10] Ester Salgarella, Cambridge (creator of the SigLA database) | https://www.archaeology.wiki/blog/2021/02/18/linear-a-still-keeps-its-secrets/ ; https://sigla.phis.me/about.html | The triage reread reports that the interview calls Salgarella the creator in collaboration with Castellan. The fixer could not reopen that interview (403), but SigLA's own About page independently identifies both creators. A co-creator is a creator; the page does not claim sole authorship. REJECTED: retained creator. The collaboration cited in the triage does not contradict it.
- MINOR | [K11] A century of confident “decipherments” has read Linear A as half the languages of the ancient world, no two agreeing: | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | The Aeon essay describes Best following the Semitic line, Finkelberg and Brown following the Anatolian line, and renewed Etruscan proposals. Replaced no two agreeing and the corresponding markdown descriptions with competing proposals without an accepted solution.
- MINOR | [K12] The modern neural decipherers that cracked Linear B and Ugaritic in 2019 simply don't apply here | https://aclanthology.org/P19-1303.pdf | Luo et al. 2019, p. 3146, report 67.3% of Linear B cognates translated; the model needs a known related-language corpus. Replaced cracked with cognate recovery in already deciphered scripts and the paper's 67.3% result.
- WRONG | [K13] 33 checks, all green | research/linear-a-arithmetic/verify.mjs | The June 19 birth log records 36/36. The pre-edit verifier reproduced 36 ok, 0 fail; the revised verifier produces 39 ok, 0 fail. Corrected the displayed count to 39 and dated the original 33-versus-36 error in a Corrected line.
- CONFIRMED | [K14] The fraction values beyond J = ½ are proposed, not settled | http://web.archive.org/web/20241219084450id_/https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Corazza et al., p. 3, distinguish deduced J from proposed E/F; p. 8 describes relative consensus, not settled decipherment. REJECTED: retained proposed, not settled. The triage correctly distinguishes consensus and model premises from certainty.
- MINOR | [K15] (added) Embedded tablet data the page serves (index.html:384) and data.mjs:73,152: HT 88 "museum":"HM 11"; HT 122 "museum":"HM 1110" | https://lineara.xyz/items/HT88.html ; https://lineara.xyz/items/HT122a.html | The retrieved HT 88 and HT 122 headings give HM 1312 and HM 1366 respectively. Corrected both museum identifiers in canonical data and regenerated the embedded and published source copies.
- MINOR | [K16] (added) HT 123+124 entry "DA-TU", with "cross-checked, sign for sign, against the peer-published SigLA palaeographic database. Every tablet below links both." (index.html; data.mjs:113) | https://sigla.phis.me/document/HT%20123a/ | Younger reads DA-TU in HT 123+124 a.6-7; SigLA HT 123a sign 31 is da and sign 32 is ki. Kept the chosen Younger reading and disclosed DA-TU versus DA-KI in HTML, markdown and data comments; removed sign-for-sign agreement wording.
- CONFIRMED | [N001 undeciphered] Linear A, the script of Minoan Crete, has never been deciphered | https://sigla.phis.me/paper.pdf | Checker reread: PDF p. 1: "to date still undeciphered". Read the PDF directly after the tool reported UNREADABLE.
- CONFIRMED | [N002 unknown-language] The language behind Linear A is unknown. | https://en.wikipedia.org/wiki/Minoan_language | Checker reread: The article says "unknown and unclassified"; the first paragraph also leaves open whether both early Cretan scripts encoded the same language.
- CONFIRMED | [N003 no-relative] its language has no known relative | https://en.wikipedia.org/wiki/Minoan_language | Checker reread: "does not appear to be a close relative of any known language" supports the absence of a demonstrated relative.
- CONFIRMED | [N005 dates] Linear A is the writing of Minoan Crete, in use from about 1800 to 1450 BC. | https://sigla.phis.me/paper.pdf | Checker reread: SigLA paper p. 1 dates the script "ca. 1800-1450 BCE". The page retains the approximation.
- UNVERIFIABLE | [N008 adaptation-date] Around 1450 BC the Mycenaean Greeks borrowed its signs to write their language; that adaptation is Linear B | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Steele 2024 chronology, p. xvi: "LM II (1450–1400), LM IIIA 1 (1400–1375)" and "Linear B emerging in later phase." These are attestation horizons, not a dated borrowing event. Also read Mnamon and the Britannica republication, which gives a different conjectural adaptation date. Derivation from Linear A is supported; the particular event date remains uncertain.
- CONFIRMED | [N009 ventris-1952] in 1952 Michael Ventris cracked it | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Checker reread: Salgarella: "successfully deciphered in 1952" and "Michael Ventris". The surrounding sentence identifies Mycenaean Greek.
- CONFIRMED | [N010 known-greek-anchor] because a known language, Greek, sat behind it | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Checker reread: Salgarella says Ventris demonstrated that Linear B "encoded a very early form of Greek". The page is a fair explanation of the known-language anchor, not a claim that language knowledge was the only technique used.
- CONFIRMED | [N011 borrowed-sounds] Because Linear B grew out of Linear A, scholars lend each Linear A sign the sound of its Linear B look-alike. | https://mnamon.sns.it/index.php?page=Scrittura&id=19&lang=en | Checker reread: Mnamon: "more than 60 homomorphic syllabograms" are "often transliterated with the phonetic values of Linear B". This paragraph immediately warns that these are conventional readings.
- CONFIRMED | [N012 pronounce] So we can pronounce Linear A. | https://www.repository.cam.ac.uk/bitstreams/635fb0e4-fed6-4dd0-a369-16067170b87f/download | Checker reread: Steele and Meissner say it is legitimate "in principle and as an approximation" to use the Linear B sound values. Read with the page's immediately following conventional-reading and sound-drift caveats, this is supported, not a claim of certain ancient pronunciation.
- CONFIRMED | [N014 sounds-drift] the sounds may have drifted when the script was borrowed. | https://www.repository.cam.ac.uk/bitstreams/635fb0e4-fed6-4dd0-a369-16067170b87f/download | Checker reread: Steele and Meissner: the Minoan language "did not have the same phonological inventory as the Mycenaean Greek encoded in Linear B". Approximate transfer and possible adaptation are explicit.
- UNVERIFIABLE | [N015 libation-repeated] Here is its most-repeated text, a formula carved on stone offering tables across the island: | https://mnamon.sns.it/index.php?page=Scrittura&id=19&lang=en | Checker reread: Mnamon says offering-table texts "are repetitively formulaic". That does not rank this text against every other text in the corpus. Positive control: searching the retrieved Mnamon article finds "offering tables" and "repetitively formulaic". Read the formula discussion in Steele 2024 too; a comparative frequency count was not supplied.
- CONFIRMED | [N016 libation-dozens] the libation formula at the top recurs on dozens of altars | https://en.wikipedia.org/wiki/Linear_A | Checker reread: The retrieved Linear A article reports "41 have Linear A inscriptions" among the libation tables, and the following sentence identifies their standardized libation formula. "Dozens" is reasonable; this does not validate the separate ~15 word count.
- CONFIRMED | [N017 opening-formula] the opening word of the formula. | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Steele 2024 gives "a-ta-i-*301-wa-ja" first in the Kato Symi bowl transcription and discusses its variants among the formulae. It supports a conventional opening, not a unique spelling on every offering.
- CONFIRMED | [N018 unknown-301] *301 is a sign with no agreed sound at all | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Steele labels "A 301 (not attested in Linear B)" and leaves *301 numerical in the formula. Mnamon explains the conventional borrowing of sounds. This sign has no agreed phonetic value from a Linear B counterpart.
- CONFIRMED | [N020 goddess-guess] A-SA-SA-RA-ME, a recurring element, has been guessed to be a goddess's name. | https://en.everybodywiki.com/Libation_Formula | Fixer update after the checker pass: The prior checker confirmed the tentative goddess interpretation only; the frequency superlative is corrected under K4.
- CONFIRMED | [N021 asherah-guess] Guessed by some to be a goddess-name, “Asasara,” even linked to Semitic Asherah | https://en.everybodywiki.com/Libation_Formula | Checker reread: The source explicitly attributes the Asherah proposal to Jan Best: "the ancient Semitic goddess Asherah". The page labels the connection speculative.
- CONFIRMED | [N022 deity-disputed] whether it names a god at all is disputed. | https://en.everybodywiki.com/Libation_Formula | Checker reread: The same source says the Asasara interpretation "was dismissed by M. Pope". This supports disputed deity identification; it does not endorse either proposed meaning.
- CONFIRMED | [N023 decimal] Linear A counts in a plain decimal system | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Corazza et al., p. 1: "Linear A employs a decimal system". The four orders are then specified.
- CONFIRMED | [N024 additive] one mark per power of ten, repeated, and the orders added together. | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Corazza et al., p. 1: "The system is cumulative and additive"; its example 6352 uses repeated signs for each power of ten.
- CONFIRMED | [N025 fraction-signs] Halves and smaller portions get their own signs. | https://mnamon.sns.it/index.php?page=Scrittura&id=19&lang=en | Checker reread: Mnamon lists "17 for fractions (of uncertain value)" alongside four numeral signs; Corazza et al. independently identifies J as one-half.
- CONFIRMED | [N027 secure-half] The securest of all, and the only fraction we'll lean on below, is J = ½. | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Corazza et al., pp. 2–3 derive J = 1/2 from HT104 and PE1 and fix it among the model premises: "The values of J, E and F are 1/2, 1/4 and 1/8, respectively." Every demonstrated non-integer quantity is a half.
- CONFIRMED | [N028 fraction-optimality] letting the corpus's own constraints pick the assignment that makes the most distinct fractions expressible. | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Read the optimality methods and simulations in Corazza et al. Counting unrepresentable values is an explicit criterion. This is a fair short description of one selection criterion, although it is not the whole model and does not make its results certain.
- CONFIRMED | [N030 sum-general] Add the entries. The sum equals that final figure. | https://en.wikipedia.org/wiki/Linear_A | Checker reread: The Linear A article says KU-RO regularly precedes "a numeral representing the sum of the tallies in the list." HT88 and the olive column of HT123 independently instantiate it. The next movement explicitly allows discrepancies, so this is not read as an exceptionless law.
- CONFIRMED | [N031 kuro-meaning] KU-RO means “total.” | https://en.wikipedia.org/wiki/Linear_A | Checker reread: The article calls KU-RO the most secure word, "meaning "total""; Younger's two clean examples and Salgarella's HT118 caption independently support it.
- CONFIRMED | [N032 kuro-first-read] You just read a word of a dead language off pure addition [...] the same way the word was first read. | https://en.wikipedia.org/wiki/Linear_A | Checker reread: The article explains the inference from the numerical sum following KU-RO. Recreated 1+1+1+1+1+1 = 6. The sentence describes that arithmetic inference, not a newly identified first discoverer.
- CONFIRMED | [N034 chadwick-role] the Linear B co-decipherer John Chadwick | https://kiwix.tristanhavelick.com/content/wikipedia_en_all/A/Linear_A | Checker reread: The older encyclopaedia mirror identifies Ventris "(with John Chadwick)" as deciphering Linear B. The attribution is conventional and supported.
- CONFIRMED | [N035 most-secure] Wikipedia calls KU-RO the most secure word in the script. | https://en.wikipedia.org/wiki/Linear_A | Fixer update after the checker pass: The earlier generic attribution is now specific; its confidence wording is unchanged.
- CONFIRMED | [N037 one-certain-word] The one Minoan word we are sure of, arithmetic handed us for free: KU-RO, “total.” | https://en.wikipedia.org/wiki/Linear_A | Checker reread: Read as the pull quote about the especially secure KU-RO inference, supported by the article and the Chadwick quotation. The page itself later identifies deficit as a second accounting word, so I do not treat this as a literal complete dictionary-size claim.
- CONFIRMED | [N040 ledger-ht123] HT 123+124 Four olive consignments. Total: ninety-three and a half. OLIV KI-TA-I 31 OLIV PU-VINa 31½ OLIV SA-RU 16 OLIV DA-TU 15 KU-RO OLIV [...] total 93½ | https://lineara.xyz/items/HT123+124a.html | Checker reread: Compared every row: KI-TA-I 31, PU-VINa 31 J, SA-RU 16, DA-TU 15, and KU-RO OLIV 93 J. Younger says "the total (KU-RO) of OLIV, 93 J is correct". Recalculation gives 93.5. This concerns the olive column, not the omitted *308 column.
- CONFIRMED | [N041 citation-ht123] HT 123+124 [...] Four olive consignments. Total: ninety-three and a half. [...] transcription · SigLA cross-check | https://lineara.xyz/items/HT123+124a.html | Checker reread: The Younger page and SigLA document resolve. Younger identifies the tablet and a four-entry OLIV column totaling 93 J. The caption is supported; landholding is audited separately.
- CONFIRMED | [N042 ledger-ht13] HT 13 A wine account that misses its own total by a half. RE-ZA 5½ TE-TU 56 TE-KI 27½ KU-ZU-NI 18 DA-SI-*118 19 I-DU-NE-SI 5 KU-RO VIN [...] total 130½ | https://lineara.xyz/items/HT13.html | Checker reread: Compared every displayed word and number with Younger .2–7, including the conventional damaged reading at RE-ZA. 5.5+56+27.5+18+19+5 = 131; recorded KU-RO is 130 J. The wine logogram and discrepancy are supported.
- CONFIRMED | [N043 citation-ht13] HT 13 [...] A wine account that misses its own total by a half. [...] transcription · SigLA cross-check | https://lineara.xyz/items/HT13.html | Checker reread: The link resolves to HT13 with VINa and the explicit note "KU-RO here records 130.5, but the numbers total 131." The SigLA cross-check link also resolves.
- CONFIRMED | [N046 younger-half] KU-RO here records 130.5, but the numbers total 131. | https://lineara.xyz/items/HT13.html | Checker reread: Exact sentence appears in Younger's HT13 commentary. Confirmed independently of the page and source-record copy.
- CONFIRMED | [N047 ht13-131] The entries come to 131. | https://lineara.xyz/items/HT13.html | Checker reread: Younger says "the numbers total 131"; independently summed 5.5, 56, 27.5, 18, 19 and 5.
- CONFIRMED | [N048 ht13-scribe] The scribe wrote KU-RO 130½. | https://lineara.xyz/items/HT13.html | Checker reread: Younger reads ".7 KU-RO 130 J" and glosses this as 130.5. The page's 130½ matches.
- CONFIRMED | [N051 kiro] Two words, total and deficit, both deciphered by counting. | https://www.cambridge.org/core/blog/2021/10/20/writing-images-reading-syllables-script-creation-in-bronze-age-crete/ | Checker reread: Salgarella's Cambridge blog caption gives "ku-ro = total" and "KI[-ro] = deficit" in a numerical account. Opened it directly when curl was blocked. The contextual inference is supported.
- CONFIRMED | [N052 kiro-owed] KI-RO is what is still owed | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Steele 2024, p. 65: "ki-ro, which we know marks a deficit". This supports the page's owed/deficit explanation.
- CONFIRMED | [N053 olive-logogram] olives, a logogram we can read because the same little leafy-twig sign survived into Linear B | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Steele 2024, p. 62 identifies "olives" as AB122 and says these usages "carry on this usage into Linear B". The olive logogram in the cited tablet is explicit.
- CONFIRMED | [N055 commodity-pictures] the numbers and the commodity pictures (olives, wine, figs, grain, a man) are secure | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content ; https://crewsproject.wordpress.com/2018/04/20/crews-display-replica-linear-a-tablet/ | Checker reread: Steele lists "VIR ‘man’, GRA ‘grain’", "OLIV ‘olives’", and "VIN ‘wine’" in the shared logograms. The CREWS replica-tablet article supplies the figs/NI evidence. These are conventional commodity identifications, not translations of the accompanying names.
- CONFIRMED | [N056 commodity-continuity] the pictures because they passed into the deciphered Linear B. | https://api.repository.cam.ac.uk/server/api/core/bitstreams/14838f02-195b-442c-a988-d8db87773c45/content | Checker reread: Read Steele's discussion around the shared AB logograms and the explicit statement that olive/olive-oil usages continue into Linear B. The page gives the standard basis for these glosses.
- CONFIRMED | [N058 arithmetic-proof] On HT 13 we can compare an intended total with its entries. | https://lineara.xyz/items/HT13.html | Fixer update after the checker pass: The original checker treated this as inference from the near-match, not a quotation or a determination of who caused the discrepancy. The revised page states that limitation.
- CONFIRMED | [N059 rest-dark] are very nearly the whole of what we securely hold. | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Checker reread: The recorded excerpt, "The second obstacle is the quality of the Linear A evidence", does not support the claim by itself. Read the complete Aeon discussion of the understood numerical/accounting material and the unread language, corroborated by Steele. The broad qualified summary is supported; replace the excerpt.
- CONFIRMED | [N061 britannica-unknown] the language written in Linear A remains unknown. | https://www.britannica.com/topic/Linear-A ; mirror: https://www.ceramicstudies.me.uk/linear_a_b_ebtxt.html | Checker reread: Britannica blocks direct retrieval. Its publicly republished text at ceramicstudies.me.uk contains the exact sentence ending "the language written in Linear A remains unknown." This is a transparent mirror verification, not a claim to have read the current blocked origin.
- CONFIRMED | [N062 salgarella-read] at the moment we can ‘read’ the Linear A script but we still cannot understand it. | https://www.archaeology.wiki/blog/2021/02/18/linear-a-still-keeps-its-secrets/ | Checker reread: The Archaeology.wiki interview contains the exact words; the page closes the quotation with a full stop where the interview continues into attribution. The tool's punctuation mismatch is not a substantive misquotation.
- CONFIRMED | [N064 attempt-gordon] Cyrus H. Gordon West Semitic not accepted | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script ; https://gredos.usal.es/bitstream/handle/10366/133930/On_some_recent_attempts_to_identify_line.pdf?sequence=1 | Checker reread: Aeon identifies Gordon's Semitic hypothesis. Facchetti's primary critique calls the older approaches, including Gordon, "unsuccessfully arisen". Together they support the proposal and lack of acceptance, beyond the narrower N1 excerpt.
- UNVERIFIABLE | [N065 attempt-palmer] Leonard Palmer & al. Anatolian / Luwian lost support | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script ; https://gredos.usal.es/bitstream/handle/10366/133930/On_some_recent_attempts_to_identify_line.pdf?sequence=1 | Checker reread: Aeon identifies "the Luwian hypothesis" and "Leonard Palmer (1958)". Facchetti describes the old approaches as unsuccessful, but neither establishes a history of support being gained and then lost. Same-text positive control: Palmer found twice in Aeon, "lost support" zero times. Change to "unproven" or supply reception history.
- CONFIRMED | [N066 attempt-georgiev] Vladimir Georgiev | https://grbs.library.duke.edu/index.php/grbs/article/download/11991/4031 | Checker reread: Nagy 1963 p. 210 n. 24: "A proponent of Greek in A has been V. Georgiev". His 1963 proposal differentiated HT Greek from other material. The rendered row's separate "met silence" quotation is audited under entry 71.
- CONFIRMED | [N067 attempt-nagy] Gregory Nagy “Greek-like elements” not a decipherment | https://grbs.library.duke.edu/index.php/grbs/article/download/11991/4031 | Checker reread: Nagy's conclusion says "attempted to show evidence for some Greek-like elements in A". The original paper is explicitly limited; it is not a successful general decipherment.
- CONFIRMED | [N068 attempt-best] archaic Phoenician | https://kiwix.tristanhavelick.com/content/wikipedia_en_all/A/Linear_A | Checker reread: The older encyclopaedia mirror identifies Jan Best's 2001 proposal that Linear A encoded "an archaic form of Phoenician". Facchetti independently critiques Best's proposals. This confirms attribution, not truth of the proposal.
- CONFIRMED | [N069 attempt-facchetti] Facchetti & al. Tyrrhenian / Etruscan-kin unproven | https://gredos.usal.es/bitstream/handle/10366/133930/On_some_recent_attempts_to_identify_line.pdf?sequence=1 | Checker reread: Facchetti, section 4: "a simple suggestion" that cannot "be, at the moment, adequately verified or refused". Read the primary PDF through its working handle URL; the N1 URL served an anti-bot page.
- CONFIRMED | [N070 nagy-title] Greek-like elements | https://grbs.library.duke.edu/index.php/grbs/article/download/11991/4031 | Checker reread: Nagy's original article is headed "GREEK-LIKE ELEMENTS IN LINEAR A" and repeats the phrase in its modest conclusion. The shortened quoted title is accurate.
- UNVERIFIABLE | [N071 georgiev-silence] Vladimir Georgiev Greek “met silence” | https://grbs.library.duke.edu/index.php/grbs/article/download/11991/4031 ; exact-phrase web search; https://www.cambridge.org/core/journals/language/article/abs/les-deux-langues-des-inscriptions-cretoises-en-lineaire-a-par-vladimir-georgiev-linguistique-balkanique-71-pp-104-sofia-academie-bulgare-des-sciences-1963/069E41A1435753F408541ECD7400B983 | Checker reread: The N1 claim field only names Georgiev, while its quote field and rendered row contain "met silence". Checked that actual assertion. Search of Nagy finds Georgiev and his publications but not that phrase; exact-phrase web searches also failed. A 1964 scholarly review of his 1963 work exists, so silence is not self-evident. No attributable quotation was established.
- CONFIRMED | [N072 best-criticism] Jan Best archaic Phoenician widely criticised | https://gredos.usal.es/bitstream/handle/10366/133930/On_some_recent_attempts_to_identify_line.pdf?sequence=1 | Checker reread: N1's claim field repeats the language name rather than the rendered judgement "widely criticised". Checked that judgement too. Facchetti calls Best-related ideas "absolutely groundless" and supplies critical references. The proposal and criticism are supported, though the claim field should be repaired.
- CONFIRMED | [N074 no-bilingual] no bilingual (no Rosetta Stone) | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Checker reread: Salgarella: "we do not (yet) have a bilingual inscription like the Rosetta Stone". No conflicting identified bilingual appeared in the authoritative sources read.
- CONFIRMED | [N075 tiny-corpus] a tiny corpus of short documents (about 1,400, most only a few signs) | https://aeon.co/essays/without-a-rosetta-stone-can-linguists-decipher-minoan-script | Checker reread: Aeon: "does not exceed 1,400 inscriptions" and extensive fragmentary preservation. Corazza gives 1,527 under another enumeration; "about 1,400" is a rounded corpus estimate, not an exact updated census. The source discussions support the small, mostly short corpus.
- CONFIRMED | [N076 no-proven-relative] no proven relative to lean a guess against. | https://en.wikipedia.org/wiki/Minoan_language | Checker reread: The classification source reports no apparent close relative among known languages. This careful "proven relative" wording is supported and preferable to the probability claims in entries 4 and 60.
- CONFIRMED | [N078 packard-nine] nine fictitious decipherments, the Linear B sound-values shuffled at random onto the signs, to test whether the “real” reading was statistically any less meaningless. | https://sites.google.com/view/minoan-script/volume-ii/vol-ii-part-i | Checker reread: The source explicitly says "nine fictitious decipherments" and explains restricted reassignment within frequency bands, followed by a tenth run using the conventional values. The page's brief random-control description is supported; the source actually reports an advantage for the conventional reading.
- CONFIRMED | [N080 neural-relative] they work by leaning on a known relative, and Linear A has none. | https://aclanthology.org/P19-1303.pdf | Checker reread: Luo et al. explicitly require "a non-parallel corpus in a known related language". Read "has none" with its antecedent "known relative". This supports the limitation of this model, not a universal claim that no future computational method could apply.
- CONFIRMED | [N081 canonical-data] The canonical numbers live in research/linear-a-arithmetic/data.mjs | git show 37137afa5c:research/linear-a-arithmetic/data.mjs | Checker reread: Pinned data.mjs exports TABLETS, NUMERALS and FRACTIONS. Its tablet quantities exactly match the JSON consumed by the page; this confirms the repository architecture, not the historical correctness of every choice.
- OBSERVED | [N083 md-check-count] 39 internal checks | research/linear-a-arithmetic/verify.mjs | Fixer update after the checker pass: Recreated after repairs: node research/linear-a-arithmetic/verify.mjs gives 39 ok, 0 fail.
- OBSERVED | [N084 all-sums-checked] every sum re-checked in code | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: The source and historical log record the additions; reproduced them from the pinned data and with the downloaded page's inline script. This confirms the author's computational event, not omitted historical inputs such as KU-DA 1.
- OBSERVED | [N085 verifier-additions] does the additions itself and confirms each one | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: Verifier reduces the entry lists and compares sums with its expectations. Recreated all 36 checks successfully. It independently adds the supplied data, but uses the same historically incomplete HT122 input.
- OBSERVED | [N086 verifier-clean-sums] HT 88 sums to KU-RO 6 and HT 123+124 to KU-RO 93½ exactly | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: Recreated 1+1+1+1+1+1 = 6 and 31+31.5+16+15 = 93.5. Both results are also explicitly supported by Younger.
- OBSERVED | [N087 verifier-ht13] HT 13 overshoots its KU-RO by exactly ½ | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: Recreated 5.5+56+27.5+18+19+5 = 131; 131-130.5 = 0.5. The historical reading independently agrees.
- OBSERVED | [N088 verifier-ht122] HT 122 balances at 97 with KU-DA 1 included. | research/linear-a-arithmetic/verify.mjs | Fixer update after the checker pass: The original observed subtraction omitted a source component. The corrected verifier includes KU-DA 1 and confirms 31 + 65 + 1 = 97; see X8.
- CONFIRMED | [N089 only-half] Only the securest fraction (J = ½) appears in any sum. | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: Every non-integer entry in TABLETS has frac J, and the only fractional remainders are 0.5. The other proposed fraction values are not used in the ledgers.
- CONFIRMED | [N090 drift-guard] A drift guard confirms the page embeds numbers byte-identical to the dataset. | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: The drift guard compares JSON.stringify on embedded TABLETS, FRACTIONS and NUMERALS against data.mjs. Mutating the embedded 130.5 to 132.5 produces exit 1, "35 ok, 1 fail". This guard does not validate narrative text.
- CONFIRMED | [N092 tablet-crosscheck-links] that every tablet cites a transcription **and** a SigLA cross-check | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: The verifier asserts "every tablet cites a source and a cross-check". All four records have both, and all eight target documents were retrieved successfully. This establishes citation presence and reachability, not complete agreement between editorial readings.
- OBSERVED | [N093 six-agents] Six adversarial research sub-agents | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The June 19 log explicitly records "Six adversarial research sub-agents". Step 7 accepts this committed account of the event; no second investigation of the old session is required.
- OBSERVED | [N094 research-fanout] plus their own fan-out | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The same log adds "(plus their own fan-out)". Supported as an author observation, without inferring an exact additional agent count.
- OBSERVED | [N095 ht13-changed-work] the single most important correction was that **HT 13 does *not* add up** | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The log records "The load-bearing correction: HT 13 does NOT add up." Its conventional reading independently sums to 131. The historical change in the author's process is supported.
- OBSERVED | [N096 early-snippet] an early search snippet had it balancing to 130½ via a **fringe re-segmentation** (Perdih) | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The log says an early snippet balanced the tablet only using a "fringe re-segmentation". The Serafimov/Perdih paper independently prints altered quantities 28.5 and 16.5 and a 130.5 sum. This supports the event and explains the change.
- OBSERVED | [N097 moved-example] so the tablet moved from "clean example" to the page's honesty centerpiece. | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The log explicitly says the tablet moved from "clean example" to "the honesty centerpiece". This is an authorial-workflow observation, not an external archaeological claim.
- CONFIRMED | [N098 perdih-reading] which a fringe “translation” actually re-segments the tablet to erase | https://korenine.si/zborniki/zbornik08/seraf_ht13.pdf | Checker reread: Read Serafimov and Perdih 2008: their transcription gives teki 28.5 and kudoni 16.5, unlike Younger's 27.5 and 18, and says "The sum of all numbers is 130.5". The page correctly distinguishes the alternative reading.
- CONFIRMED | [N099 gorila-editors] The transcriptions are the standard GORILA (Godart & Olivier) readings as edited by John G. Younger | https://lineara.xyz/items/HT13.html | Checker reread: Younger's commentary names GORILA and its volume/page references and carries "Commentary © John Younger". Facchetti's bibliography independently credits GORILA to Godart and Olivier. These establish edition and attribution, not all interpretations added by the page.
- UNVERIFIABLE | [N100 younger-retired] Younger's own site was retired by the University of Kansas in 2024 and is offline | https://www.academia.edu/117949722/Younger_JG_Linear_A_folder_introduction ; https://linguistics.stackexchange.com/questions/51236/whatever-happened-to-this-linear-a-website | Checker reread: Tried the Academia URL and alternate page forms with curl and web retrieval. Curl returned 403; web retrieval did not expose the introduction. N1 supplies a plausible first-person 2024 excerpt but I could not independently read it. A Stack Exchange discussion corroborates that the old site disappeared, and the old host failed DNS here; neither verifies the precise institutional decision/date. Marked unverified, not false.
- OBSERVED | [N101 source-procedure] Transcriptions and notes came from lineara.xyz and were compared with SigLA. | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Fixer update after the checker pass: The historical log records that comparison. K16 now discloses a disagreement; a recorded comparison is not proof of complete agreement.
- CONFIRMED | [N102 sigla-description] peer-published SigLA palaeographic database | https://sigla.phis.me/paper.pdf ; https://www.fluxus-editions.fr/gla5-salg.php | Checker reread: Read the SigLA paper and the publisher's chapter page https://www.fluxus-editions.fr/gla5-salg.php, DOI 10.36824/2020-graf-salg. The abstract supports the database purpose, and the proceedings publication supports "peer-published", which the excerpt alone cannot establish.
- CONFIRMED | [N103 every-tablet-links] Every tablet below links both. | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: The source-list builder creates one transcription and one SigLA link for each of the four tablet records. The reader harness produces twelve source entries, including the four paired tablet entries; every tablet link was retrieved.
- CONFIRMED | [N104 schematic-glyphs] The numeral glyphs in Movement II are schematic of the decimal system (vertical = unit, horizontal = ten, circle = hundred), not tracings of particular sign-forms. | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: Read numeralSVG: it constructs vertical lines for units, horizontal lines for tens and circles for hundreds. No tablet-image tracings are used. The public statement of a schematic representation matches the implementation.
- CONFIRMED | [N106 fraction-exclusion] and are kept out of the sums. | git show 37137afa5c:research/linear-a-arithmetic/data.mjs | Checker reread: TABLETS contains only integers and J halves; values for E, F, B, D and K in FRACTIONS never enter the demonstrated sums. Recomputed the ledgers.
- CONFIRMED | [N108 checks-page-bytes] This layer's check reads the bytes this page serves and asserts against them. | git show 37137afa5c:research/linear-a-arithmetic/verify.mjs | Checker reread: Verifier reads index.html and extracts its script#data block before asserting equality of three objects. The claim of reading page bytes is literally true. It must not be mistaken for a check of prose; the step 6 mutation demonstrates that limitation.
- OBSERVED | [N109 empty-files-test] with every file this layer ships emptied, it goes red. | git show 37137afa5c:research/verifier-independence/depends-1.json | Checker reread: The pinned depends-1.json record reports exitCode 1, depends true, timedOut false for this verifier with shipped files emptied. Recreated an empty page in scratch: exit 1, "33 ok, 1 fail". Supported as a recorded and reproducible observation.
- OBSERVED | [N110 last-verified] The check placard dates the original verification to 19 June 2026. | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Fixer update after the checker pass: The historical creation log supports that date. The present October checks are reported separately in this record and its repair report.
- OBSERVED | [N111 date-published] "datePublished":"2026-06-19" | git show 37137afa5c:src/content/strata/the-accountant-we-cant-read.md | Checker reread: Frontmatter date is 2026-06-19T20:30:00Z and birth commit abe1a262132 plus the committed log support publication on June 19. Confirmed as a publication event.
- OBSERVED | [N112 date-modified] JSON-LD dateModified is 2026-10-14. | public/strata/the-accountant-we-cant-read/index.html | Fixer update after the checker pass: Set to the date of these substantive repairs; the Corrected lines date the revision.
- OBSERVED | [N113 reader-ledger] Sum the columns yourself and watch them hit the total | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: Downloaded the page into an initially empty reader directory, then ran its inline script in a local minimal DOM harness. Its sum-all callbacks produce 6, 93.5 and 131; HT122 displays 96 vs 97 from the chosen subtotals. The functionality works. The HT122 historical interpretation is separately wrong.
- OBSERVED | [N114 numeral-input] type a number 1–9999: | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: The input is wired to the renderer and clamps integer values to 0–9999. Harness checks 1, 56, 130 and 9999 all pass. Allowing an additional zero does not break the advertised 1–9999 range.
- UNVERIFIABLE | [N115 reference-1] Younger, Linear A Texts in phonetic transcription (GORILA; site retired 2024) | https://www.academia.edu/117949722/Younger_JG_Linear_A_folder_introduction ; https://lineara.xyz/items/HT13.html | Checker reread: The source title and mirror transcription attribution are corroborated, but I could not reopen Younger's introduction to independently confirm "site retired 2024". Same retrieval limitation and corroboration as entry 100, not a claim that N1 fabricated the excerpt.
- CONFIRMED | [N116 reference-2] Corazza, Ferrara, Montecchi, Tamburini & Valério 2021 [...] fraction values (JAS 125:105214) | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Read the paper front page: all five authors, Journal of Archaeological Science 125 (2021), article 105214 and DOI 10.1016/j.jas.2020.105214 match. Crossref tool result: DOI MATCH. Online availability in September 2020 does not make the 2021 journal citation wrong.
- CONFIRMED | [N117 reference-3] Steele & Meißner 2017 [...] backward projection of sound values | https://www.repository.cam.ac.uk/handle/1810/265228 | Checker reread: Cambridge repository metadata identifies Steele and Meissner and the title "From Linear B to Linear A: The problem of the backward projection of sound values". Read the deposited text and confirmed the 2017 bibliography.
- CONFIRMED | [N118 reference-4] Wikipedia [...] Linear A | https://en.wikipedia.org/wiki/Linear_A | Checker reread: The linked article is titled Linear A and concerns the Minoan script described. The target is reachable and the basic date range is present.
- CONFIRMED | [N119 reference-5] World History Encyclopedia [...] Linear A Script | https://www.worldhistory.org/Linear_A_Script/ | Checker reread: The linked World History Encyclopedia article is titled "Linear A Script: The Undeciphered Script of the Minoans" and supplies the cited administrative glosses.
- CONFIRMED | [N120 reference-6] Encyclopædia Britannica [...] Linear A | https://www.britannica.com/topic/Linear-A ; mirror: https://www.ceramicstudies.me.uk/linear_a_b_ebtxt.html | Checker reread: Direct Britannica retrieval is blocked, but the named article and relevant language-status text are preserved in the Britannica republication. This verifies the cited text through a mirror, not origin availability.
- CONFIRMED | [N121 reference-7] CREWS Project (Cambridge) [...] figs / NI / nikuleon | https://crewsproject.wordpress.com/2018/04/20/crews-display-replica-linear-a-tablet/ | Checker reread: The Cambridge CREWS post says Athenaeus, quoting Hermonax, records "nikuleon" as "a special Cretan word for the fig" and discusses NI. Read the full post; the reference labels its actual subject.
- CONFIRMED | [N122 reference-8] Packard, Minoan Linear A (1974) | https://books.google.com/books/about/Minoan_Linear_A.html?id=vax3kwoscWQC | Checker reread: Google Books metadata: "Minoan Linear A", David W. Packard, University of California Press, 1974, 272 pages. The book exists with the named authorship and date.
- CONFIRMED | [N123 fraction-full-citation] Corazza, Ferrara, Montecchi, Tamburini & Valério (2021), "The mathematical values of fraction signs in the Linear A script: a computational, statistical and typological approach," Journal of Archaeological Science 125:105214 | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: The paper title, five authors, 2021 volume 125 and article number 105214 match the front page and DOI metadata. The title's capitalization difference does not alter the citation.
- OBSERVED | [N124 standalone-route] An immersive layer; it renders as its own standalone page at the `href` above | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: Recreated with curl: the href returns HTTP 200 and the standalone document, including the expected h1 and inline ledger program. It is not the Markdown body.
- OBSERVED | [N125 index-not-body] its body is not rendered. | git show 37137afa5c:src/content/strata/the-accountant-we-cant-read.md | Checker reread: The downloaded route contains the page heading but not the Markdown-body phrase "An immersive layer; it renders as its own standalone page". That heading is the same-text positive control. Read both pinned source files and the live response.
- OBSERVED | [N126 related-grid] That page's freshest open edge was 'Linear A [...] the silence, as a negative result' | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The N1 sibling-file pointer does not itself supply the quoted follow-on. The author's committed June 19 log does record "Linear A" and "the silence, as a negative result" as the prior page's open edge. Accepted as the recorded author observation under step 7, with the source pointer corrected.
- CONFIRMED | [N127 related-cipher] There it is the index of coincidence, a statistic that breaks a cipher before the key is known. | git show 37137afa5c:src/content/strata/the-cipher-couldnt-hide.md | Checker reread: Read the sibling Markdown at the pinned commit: it identifies the index of coincidence and breaking a cipher without knowing its key. The related-page description matches its stated subject.
- CONFIRMED | [N128 related-invariant] That page's thesis was that every decipherment exploits an invariant of relabeling. | git show 37137afa5c:src/content/strata/the-invariant-of-relabeling.md | Checker reread: Read the sibling Markdown at the pinned commit: its stated thesis is decipherment via an invariant of relabeling. This line reports that thesis; it is not a new universal theorem established by this audit.
- CONFIRMED | [N129 related-number] There, grouped data deceives because the grouping discards information | git show 37137afa5c:src/content/strata/every-number-honest.md | Checker reread: Read the sibling Markdown at the pinned commit: its grouping examples expose information lost by aggregation. This is a fair description of that page's content.
- OBSERVED | [N130 runtime-ht88] ✓ the entries sum to 6 = KU-RO 6. It balances. | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: Recreated the actual sum-all callback from downloaded HTML: "the entries sum to 6 = KU-RO 6. It balances." Independently matches Younger's six one-unit entries.
- OBSERVED | [N131 runtime-ht123] ✓ the entries sum to 93½ = KU-RO 93½. It balances. | git show 37137afa5c:public/strata/the-accountant-we-cant-read/index.html | Checker reread: Recreated the actual sum-all callback: "the entries sum to 93½ = KU-RO 93½. It balances." Independently matches Younger's olive-column figures.
- OBSERVED | [N132 runtime-ht13] Entries sum to 131, but KU-RO reads 130½; a discrepancy of ½. | public/strata/the-accountant-we-cant-read/index.html | Fixer update after the checker pass: The revised runtime reports the difference without assigning its cause; browser verification passes on desktop and mobile.
- CONFIRMED | [N134 numeral-legend] 1 unit · vertical stroke 10 ten · horizontal stroke / dot 100 hundred · circle 1000 thousand · circle with rays | https://cris.unibo.it/retrieve/e1dcb336-9b0e-7715-e053-1705fe0a6cc9/Montecchi_JAS_125_2021.pdf | Checker reread: Corazza et al., p. 1 describes the four magnitudes and the four matching sign forms. Read the source and the generated legend. All four pairs match.
- CONFIRMED | [N135 gorila-rejects-perdih] that the GORILA/Younger reading rejects | https://lineara.xyz/items/HT13.html | Checker reread: Younger reads TE-KI 27 J and KU-ZU-NI 18, while Serafimov/Perdih prints 28.5 and 16.5. "Rejects" here describes the incompatible reading adopted by the page, not an alleged published Younger rebuttal of that paper.
- CONFIRMED | [N136 packard-concordance] Packard's 1974 computer concordance | https://books.google.com/books/about/Minoan_Linear_A.html?id=vax3kwoscWQC | Checker reread: Google Books contents include numerical analyses, sign frequencies, transliteration material and "Random Decipherments of Linear A". Together with the computer-method history, this supports the concordance description and 1974 publication.
- OBSERVED | [N137 sibling-quotation] That page's freshest open edge was 'Linear A [...] the silence, as a negative result' | git show 37137afa5c:memory/log.d/2026-06-19T2030Z-the-accountant-we-cant-read.md | Checker reread: The committed creation log records the quoted follow-on from the sibling page. Same historical observation as entry 126, with the appropriate log source already supplied by N1.
- CONFIRMED | [N138 added-findspot] (added) Real tablets from the Haghia Triada archive | https://lineara.xyz/items/HT13.html ; https://lineara.xyz/items/HT88.html ; https://lineara.xyz/items/HT123+124a.html ; https://lineara.xyz/items/HT122a.html | Checker reread: Each tablet has a Haghia Triada findspot in the retrieved metadata; SigLA also identifies that findspot. N1 records the tables but does not give this explicit geographical claim its own entry.
- CONFIRMED | [N139 added-six-entries] (added) Add its six entries: | https://lineara.xyz/items/HT13.html | Checker reread: Counted RE-ZA, TE-TU, TE-KI, KU-ZU-NI, DA-SI-*118 and I-DU-NE-SI in the cited transcription. Six quantity-bearing entries are displayed and sum to 131. This reader-visible count was not an explicit claim field in N1.

## Findings

- [fixed] X1, dek and share images: Changed the dek to no known relative; share card rebuilt at landing from the corrected dek. assay/land.mjs:187-191 rebuilds the PNG and then its WebP thumbnail. Neither image was rendered during this repair.
- [fixed] X1, other-layer copy: On 2026-10-18 public/strata/the-signal-never-sent/index.html:303 was changed from "has no living relative" to "has no known relative", the wording the evidence supports (no dated line for this MINOR); no other copy is in that layer. The same absence stated as fact in this layer's TV narration (public/tv/lib/narration/batch-9.js, "No relatives, no key, no way in.") now says no known relatives, corrected by the directing instance on 2026-10-18. Board note 0606ad is closed.
- [fixed] X2: Dated the account to about 3,500 years ago and removed the supposed death date.
- [fixed] X3: Corrected the age in prose and data, and described a discrepancy rather than an established human mistake.
- [fixed] X4: Replaced recovery with proposed assignments in HTML, dek, markdown and data comments; added the fraction Corrected line.
- [fixed] X5: Restricted KU-RO endings to some totalling accounts in prose, descriptions and data; added the frequency Corrected line.
- [fixed] X6: Qualified the introduction and data headline; the generated ledger and source-list captions share that headline.
- [fixed] X7: Qualified the shared caption used by the ledger and source list.
- [open note:07859a] X8: This layer includes KU-DA 1 and correctly balances HT 122 at 97. On 2026-10-18 the authored copy in the-machine-made-of-months was fixed (src/content/strata/the-machine-made-of-months.md:12 now says a stranger can find the half-unit discrepancy on HT 13), with a dated Corrected 2026-10-18 line on that page and its markdown. The only thing left is the preserved quotation in There, Here (public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1712]), which still says "catch his off-by-one" and needs an annotation through There, Here's annotation mechanism without changing its words.
- [fixed] X9: Replaced the off-by-one caption through the shared data headline; shares the HT 122 Corrected line.
- [fixed] X10: Replaced the expected-96 prose with 31 + 65 + 1 = 97; shares the HT 122 Corrected line.
- [fixed] X11: Specified two subtotals plus the separate unit; shares the HT 122 Corrected line.
- [fixed] X12: Used olive account and four entries in the prose and repeated data note.
- [fixed] X13: Changed the general identification to possible personal or place names.
- [fixed] X14: Replaced probably an isolate with no demonstrated relative in prose and the data summary.
- [fixed] X15: Dated publication of the computer study to 1974, not first use of a computer.
- [fixed] X16: Replaced cracked with cognate recovery in already deciphered scripts and the paper's 67.3% result.
- [fixed] X17: Corrected the displayed count to 39 and dated the original 33-versus-36 error in a Corrected line.
- [fixed] X18: Narrowed the markdown coverage statement and verifier heading to five integer examples.
- [open note:881e37] X19: This layer reports HT 13's half-unit discrepancy without assigning a cause. On 2026-10-18 the is-it-writing relation note (src/content/strata/is-it-writing.md:12) was made cause-neutral, with a dated Corrected 2026-10-18 line on that page and in its markdown. Scribal-slip claims still remain in src/content/strata/the-signal-never-sent.md:13,50; public/strata/the-signal-never-sent/index.html:309,623,665; research/the-signal-never-sent/verify.mjs:148; this layer's TV narration at public/tv/lib/narration/batch-9.js:20 ("So we can catch the scribe's slip."); and the preserved quotation in There, Here (public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1992]). Correct the authored text, narration and verifier labels, and annotate the preserved quotation without changing its words. Round 3 on 2026-10-18 corrected the scribal-slip copies in the-signal-never-sent (dated line there), and the directing instance corrected the TV narration line; what remains is the preserved quotation in There, Here.
- [fixed] X20: Runtime now renders that separate unit, sums all components and reports 97 balances; shares the HT 122 Corrected line.
- [fixed] X21: Replaced the plural attribution with the distinct HT 13 and HT 122 results; shares the attribution Corrected line.
- [fixed] K1: Dated the account to about 3,500 years ago and removed the supposed death date.
- [fixed] K2: Corrected the age in prose and data, and described a discrepancy rather than an established human mistake.
- [fixed] K3: Restored the exact qualification and named and linked Mnamon; added the quotation Corrected line.
- [fixed] K4: Removed the undefined approximate count and superlative from the formula gloss, closing prose and data; used recurring element.
- [fixed] K5: Replaced recovery with proposed assignments in HTML, dek, markdown and data comments; added the fraction Corrected line.
- [fixed] K6: Restricted KU-RO endings to some totalling accounts in prose, descriptions and data; added the frequency Corrected line.
- [fixed] K7: Restored the quotation's ending and identified and linked the secondary quotation source.
- [fixed] K8: Restored the source wording and specifically attributed it to Wikipedia.
- [fixed] K9: Used olive account and four entries in the prose and repeated data note.
- [fixed] K11: Replaced no two agreeing and the corresponding markdown descriptions with competing proposals without an accepted solution.
- [fixed] K12: Replaced cracked with cognate recovery in already deciphered scripts and the paper's 67.3% result.
- [fixed] K13: Corrected the displayed count to 39 and dated the original 33-versus-36 error in a Corrected line.
- [fixed] K15: Corrected both museum identifiers in canonical data and regenerated the embedded and published source copies.
- [fixed] K16: Kept the chosen Younger reading and disclosed DA-TU versus DA-KI in HTML, markdown and data comments; removed sign-for-sign agreement wording.

## Verification and limits

The command results below are the fixer's observations before this repair. The repair's current results and pending board-note handoff are recorded in research/assay-line/queue-1014c/the-accountant-we-cant-read/repair-report.md.

- `node research/linear-a-arithmetic/verify.mjs`: before repairs, 36 ok, 0 fail; after repairs, 39 ok, 0 fail.
- `node research/linear-a-arithmetic/verify-page.mjs`: 32 ok, 0 fail across desktop and mobile, including KU-DA 1 and the balanced HT 122 result.
- `node scripts/check-claims-pass.mjs`: exit 1 because 11 other layers have invalid findings. This layer is absent from the failures; a direct `checkRecord` call returns no errors for its 149 claims and 35 findings.
- Isolated mutation tests: baseline 39/0; displayed 131 changed to 132 fails 38/1; deleting KU-DA 1 or changing it to 2 in both canonical and embedded data fails 37/2; an empty page fails 34/3.
- Six shared WRONG corrections appear in the HTML Corrections section above the check placard and at the end of the markdown body. No MINOR or Updated lines were added.
- The original source notes called the KU-RO corpus count 29/436. The fixer's parse of `transliteratedWords` gives 34/436 tablet-side records, still a minority. These are different counting procedures; no corpus percentage is published as a replacement. Steele p. 97 independently supports the narrowed prose.
- Untriaged, unresolved claims retained without edits: the precise borrowing date for Linear B; the full libation text's frequency rank; Palmer's loss of support; the attributed Georgiev silence; and the institutional/date details of the Younger site retirement in two copies. They are UNVERIFIABLE, not relabelled CONFIRMED to pass the gate.
- The OG HTML source is corrected. Visual inspection shows the existing PNG and WebP use the older, truncated dek ending in probably has no..., not the separate OG HTML design. The markdown source is corrected under X1; share card rebuilt at landing from the corrected dek, with the WebP refreshed by the following landing step. Neither image was rendered or re-encoded during repair. Historical birth-log statements remain historical; unresolved current copies in other layers are the three PENDING findings above.