Liber Deuteronomii

Merged cruces for the whole work. Translated by five parallel Opus stints 2026-08-31 (chunks 0000–0004, 0005–0009, 0010–0014, 0015–0019, 0020–0023); merged and plate-gated the same day. Per-stint detail: cruces-0000.md, cruces-0005.md, cruces-0010.md, cruces-0015.md, cruces-0020.md — this file is the reconciled top level, not a replacement for them.

Attributed by Migne to Walafrid Strabo; in fact Anselmus Laudunensis et schola (per cruces-0000.md §1, following the corpus's standing correction).

24 chunks, 23,141 Latin words → English at 1.44–1.60×. Apparatus across the whole work: 3 [sic:] · 3 [var:] · 1 [cj:] · 1 [d:] · 2 [nt:] · 2 [ed:], against 346 [n:] notes and 240 column anchors, every count reconciled 1:1 against the Latin twins by verify-english.mjs.

§1 · Conventions held across the whole work

Following src/english/8968/cruces.md §1 (the Glossa baseline), unbroken by any stint:

• VERS. n.-- passes through verbatim; a fragmentary lemma is rendered as the phrase it stands for; Migne's attribution sigla ((RAB.), (ISID.), (AUG., quaest. 6.), (GREG., lib. VII Moral., cap. 10.)) ride as ordinary [n: …], unanglicized.
• CAPUT N. → CHAPTER N. (spelled-out numerals normalized, per the corpus's standing rule — mirror the plate's own form, not a fixed style).
• Greek rides verbatim, untranslated, with its Latin gloss translated as ordinary prose: χιλιάρχους, ἐκατοντάρχους, πεντεκοσιάρχους, δεκάρχους, δευτερονόμιον, Μετονυμικῶς.
• doctor → teacher, never "doctors" (8976 §V via 8947). Checked at merge across all five stints: the whole work is uniform — zero occurrences of untranslated "doctor" survive in any English chunk body (one frontmatter incipit: field quotes the Latin verbatim, correctly untouched).
• usque ad (the abridgment formula) → as far as, italic, matching 8947/8976/8978. Censused in 0000–0004: 27 occurrences, 20 the formula and 7 ordinary prepositions — including one inside an italic scripture run wearing the formula's own typography (0452B), which is the shape the census exists to catch.
• Second person singular in the Decalogue exposition and scripture lemmata renders thou; plural renders you. No addressee's number was altered anywhere in the work (checked stint by stint).
• Etymological glosses keep their Latin lemma beside the English, because the gloss is the etymology: avigeria/avigaria, cartalo (Pattern 9, an attested spelling of cartallum, not a defect), the Cassian vice-names (gastrimargia, philargyria, acedia, cenodoxia kept in Latin; ira/tristitia/superbia/fornicatio translated).
• Hebrew place- and tribe-names stand as Migne spells them, not normalized to familiar English shapes.
• (Josue XX) at 0472A is the work's only inline italic locator outside the [n:] system; left untagged (no [f:]) per Pattern 4's single-occurrence exception, checked at merge — no other stint introduced a second one that would force the question.

§2 · ⭐⭐⭐ THE PLATE GATE — closed this session, 9 columns read, 6 markers overturned

Every [sic:]/[var:] the five stints fired without a plate read (nine candidates, on cols 0445D, 0452A, 0454C, 0471C, 0476C, 0481A, 0481C ×2, 0496C) was read against Gallica ark:/12148/bpt6k5505319w (PL 113 second witness), IIIF native-resolution crops, page map f = ceil(column/2), calibrated against the printed corner numbers on f242 (reads 483/484) and reconfirmed at f223 (445/446), f226 (451/452), f227 (453/454). Full record: data/plate-reads.json → works["8945"]. Result:

3 CONFIRMED as Migne's own printing (all still stand, still carry a marker):

6 REFUTED — Corpus Corporum's own corruption, not Migne's — all withdrawn and patched (data/tei-patches/8945.json), five of them landing in one tight span (0471C–0481C):

⭐ Combined with stint 4's own find at 0492B (pre→per, also Corpus Corporum's, caught and correctly left unmarked before this session), that makes 7 corruptions total closed in this one work — a rate that keeps confirming 9003's finding (CLAUDE.md rule 8a): a plate read overturns roughly as often as it confirms, and corpus frequency or context licenses neither outcome.

Every withdrawn marker's English was corrected to render the word plainly, no marker. node scripts/plate-gate.mjs 8945 now passes clean: 8 markers stand, all 16 recorded reads (9 this session + 7 from stint 4's own reading) license them.

⬜ Still open, not blocking ship (no marker was fired on these, so rule 8a's gate does not reach them): three [var:] candidates logged and withheld pending a plate read — 0468C explorationis for expletionis (Lev. 23:16), 0468D habebit/(I Cor. V) for habemus/2 Cor. 5:1, 0470A Supra for Super cathedram (Matt. 23:2), all in cruces-0005.md §9; and 0503D's prae fulgorem case in cruces-0020.md §7, which would take a marker under Pattern 12's second test but was declined because a case ending is exactly what the transcription is least trustworthy on. A future session may read cols 0468 and 0503 and settle them; nothing about shipping this work depends on it.

§3 · The one [d:], the one [cj:], and the two [nt:]

• [d:] at 0484D–0485A — a doubled clause plate-verified as Migne's own printing (Carnalis vita polluta ex carnali immunditia set twice); the intrusive copy identified by argument, not position (Pattern 11a). Detail: cruces-0015.md.
• [cj:] at 0461A — damnant for damna (nom. pl.), Migne's word kept, the reading offered beside it (Pattern 18, additive). Detail: cruces-0005.md §0005. A second candidate cj, terna for aeterna at 0490D–0491A, was plate-verified and licensed the same way; detail in cruces-0015.md.
• [nt:] at 0475A — Migne's note on Adam and the tree is editorial prose, not a locator (Pattern 15), translated in place. Detail: cruces-0010.md §3.
• [nt:]-adjacent: two [ed:]s for the SAME digitization hole, at 0449B (Choreb ( ), chunk 0001) and 0493B (Sinai/Seir/Pharan ( ), chunk 0020) — Corpus Corporum dropped the Hebrew type inside empty parentheses in both places, the identical class already recovered once for the sibling Glossa 8944 (data/hebrew-patches/8944.json). Both are marked, neither is silently rendered, and both stints independently reached for the same marker ([ed:]) without coordinating — checked at merge and found consistent, no reconciliation needed. ⬜ Open, cheap: one Gallica page render each (f225 for 0449, f248 area for 0493) would recover the Hebrew and retire both [ed:]s into data/hebrew-patches/8945.json; not done this session because it is new work, not a gate the ship test requires closed.

§4 · Negation audit — full census, both directions, whole work

Every non · nec · neque · nisi · ne · nemo-family · nullus-family · nihil · nunquam · nusquam in all 24 chunks was located and checked against its English exponent, chunk by chunk, by the stint that translated it (0000–0004: 125 tokens; 0020–0023: full census by type). None was dropped, none was added, and no word was rendered as its opposite to rescue a sentence that would not otherwise construe — including the sites where carrying the negation makes the English harsher or odder than a smoothed reading would (catalogued in cruces-0000.md §5, cruces-0020.md §10). Where a printed sentence would not construe even with its negation carried faithfully, the defect is logged as a crux and left unrepaired, never patched invisibly into the translation.

§5 · The polarity read — three blind Opus readers, all 24 chunks, no overlap gaps

Read after the plate gate closed (§2), each reader assigned a disjoint range (0000–0007, 0008–0015, 0016–0023), each explicitly barred from reading this file or any cruces-NNNN.md before forming findings. 19 of 24 chunks came back clean. Total contradiction-class sites across the three reads: 13 (some newly found, most reconfirming sites already logged above or in the per-stint files). One required an English fix; two candidate "our defect" flags were reviewed and declined on closer parse.

Fixed — the one real translator error found:

• @0471C (0010.md) — post perfectionem bonorum operum requiem aeternam exspectat in sabbatismo had been rendered "he looks after the perfection of good works for eternal rest," which folds post into a phrasal verb ("looks after") and turns requiem aeternam from the object of exspectat ("awaits") into a purpose clause. Corrected to "after the perfection of good works he awaits eternal rest in the sabbath-rest." This sentence had already been touched once this session (the withdrawn [sic: coversans], §2) — the tail was left garbled when that marker came out; this closes it.

Reviewed and declined — proposed as "our defect," found to be a faithful literal parse on re-reading:

• @0464B (0006.md), twice, same gloss — one reader flagged nec praeter potestatem suam ostendit esse, quae ita contingunt (rendered "nor does he show that the things which so happen are beyond his own power") as mis-scoped negation, proposing "he shows... are not beyond his power." Declined on re-parse: nec negates the finite verb ostendit, not the prepositional phrase — "he does not show [them] to be beyond his power" is the grammatically indicated reading and the current English matches it exactly. The apparent tension with the following "But he subjoins why he permits it" is not a contradiction: declining to assert these events lie outside God's power is compatible with going on to explain why he permits them. Same gloss, same column, a second reader flagged ad cognoscendum ... non illi qui omnia novit ("for the knowing... not for him who knows all things") as a dangling contrast. Declined: cognoscendum is impersonal ("for it to be known," not "so that he may know"), and the English's "for the knowing... not for him who [already] knows" completes the contrast as printed. Both left unchanged; no crux needed, since nothing here misrepresents Migne's text.

New Migne-side candidates, rendered faithfully as printed, no marker fired (rule 8a — no plate read for these columns), logged for a future session:

• @0490B–C (0018.md), VERS. 17 — Recedentes a cultura Dei, invenerunt sibi deos quos patres eorum coluerunt lacks the Vulgate's non before coluerunt (Deut. 32:17, quos non coluerunt patres eorum) and, as printed, asserts the opposite of the verse it glosses. Strong candidate for a dropped non; a minority patristic reading (the "new" gods were really the old Egyptian gods) is possible but less likely given Recedentes a cultura Dei two words earlier. Rendered as printed: "gods whom their fathers had worshipped."
• @0448B (0001.md) — the work's own station count contradicts itself: "they made twenty stations" against three other statements in the same and the next chunk (0451C, 0452D) totalling "one and twenty." 12 + 20 + 9 does not make the 42 the passage insists on either way. Rendered as printed throughout; the arithmetic defect is Migne's.
• @0475C (0012.md) — cujus utilitatem et immunditiam non debemus despicere ("whose usefulness and uncleanness we ought not to despise") contradicts the work's own repeated insistence elsewhere (0474A, 0476A) that Gentile uncleanness is exactly what must be cut away. Likely a dropped propter: cujus utilitatem propter immunditiam non debemus despicere, "whose usefulness we ought not to despise on account of his uncleanness" — which matches the Deut. 22:3 lemma (the stranger's stray goods are still returned). Rendered as printed.
• @0473B (0011.md), low confidence — scitur enim ("for it is known") reads as self-cancelling after ideo ignorari dicitur ("is therefore said to be unknown"); a concessive scitur tamen ("yet it is known") would resolve it, but a loose enim in the glossator's own voice is equally possible. Rendered as printed.

All four are candidates for the same future plate-read pass as §2's open items, not blockers — none was rendered other than what Migne's transcription prints, per rule 8a and Pattern 7a throughout.

§6 · What remains, for a future session

1. Two [ed:] holes (§3) — cheap plate recovery, not urgent.
2. Three withheld [var:] candidates and one withheld case-ending candidate (§2) — cols 0468, 0470, 0503.
3. Four new polarity-read candidates (§5) — cols 0448, 0473, 0475, 0490.
4. Nothing else blocks shipping. verify-english.mjs and plate-gate.mjs both pass clean; the polarity read (§5) is done, its one real fix applied. This work is ready to be built, indexed, and marked ours.

PL 113, cols 0445D–0459C. PRAEFATIO + CAPUT PRIMUM (whole) + CAPUT II, III, IV, V, VI (through VERS. 16). Attributed by Migne to Walafrid Strabo; in fact Anselmus Laudunensis et schola.

⛔ Do NOT merge this into cruces.md by hand — it is one of five range files.

Chunks read and their state. 0000, 0001, 0002, 0003, 0004 were each read line by line against the Latin twin. 0000 and 0004 are clean — nothing in them is a defect, and the few notes below on those two record readings that were checked and let alone, not findings. 0001, 0002 and 0003 carry the substantive entries.

Marker tally for this range: 4 markers — 1 [var:], 2 [sic:], 1 [ed:], against 55 [n:], 53 column anchors, 8 ## heads, 0 guillemets in either language, every one of those four counts at exact parity with the Latin twins (verified mechanically, chunk by chunk, anchors compared in order). 0 [d:], 0 [cj:], 0 [nt:].

§1 · Conventions followed

The Glossa conventions settled in src/english/8968/cruces.md §1 are followed unchanged and were not re-opened: VERS. n.-- passes through verbatim; the lemma is Englished; a fragmentary lemma is rendered as the phrase it stands for; Migne's attribution sigla ((RAB.), (ISID.), (AUG., quaest. 6.), (GREG., lib. VII Moral., cap. 10.)) ride as ordinary [n: …], verbatim and unanglicized. CAPUT PRIMUM. → CHAPTER ONE., the corpus form. (cont.) on a repeated head is our chunking and is carried, untranslated.

Greek rides verbatim, untranslated (8944's rule): χιλιάρχους, ἐκατοντάρχους, πεντεκοσιάρχους, δεκάρχους [0451A], δευτερονόμιον [0447B], Μετονυμικῶς [0457B]. The Latin glosses that stand beside them are translated as ordinary prose.

A parenthetical citation Migne prints in the running text, not as an [n:], is kept verbatim Latin like a note: (AUG., quaest. [0457B] 4 in Deut.) at 0004 VERS. 16 — the one such in the range, and it carries a column anchor inside it, kept in place.

Fixed renderings, held constant across the range (the within-work one-voice rule):

usque ad censused per occurrence (addendum §D — decided from what stands on either side, never by rule). 27 occurrences in the range: 20 are the abridgment formula and 7 are not.

The formula (20), rendered as far as: 19 set in Migne's own italics (etc., usque ad) — 0001 ×2 (ID. @0448D, ISID. @0450A) · 0002 ×2 (ISID. @0453A, ISID. @0453C) · 0003 ×2 (GREG. @0456C, RAB. @0456D) · 0004 ×12 (the AUG./ISID./GREG. excerpts, VERS. 16 through VERS. 13) · 0001 @0449C (ISID., Quadraginta anni, etc.) — plus one printed ROMAN, 0002 @0452C etc., usque ad hoc, "Haec est aqua contradictionis", rendered "as far as this".

NOT the formula (7), each an ordinary preposition: 0001 ab exitu … usque quo pervenerunt, A Ramesse usque ad montem Sina, de Cadesbarne usque ad campestria Moab · 0002 ab Aegypto usque ad Jordanem, a squalore vitiorum usque ad damnationem poenarum · 0003 usque ad haereticorum conciliabula · and ⛔ 0002 @0452B de Cadesbarne usque ad torrentem Zareth, which is inside an italic scripture run and is therefore the exact shape of warning D's trap: an ordinary usque ad wearing the formula's italics. Rendered "as far as the torrent Zareth", the sentence's own words.

§2 · ⛔ THE THREE MARKERS IN THIS RANGE STAND ON UNREAD COLUMNS — plate reads owed

node scripts/plate-gate.mjs 8945 names all three. This is the honest state, not an oversight: each is recorded here with its attribution explicitly unsettled, because our Latin is Corpus Corporum's transcription and not Migne's plate, and both marker families make a public claim about the plate. Before the work ships, cols 0445D, 0452A and 0454C must be read (PL 113 patrologiaecurs04migngoog, PDF page = (column + 11) / 2 — pp. 228, 232 and 233; second witness Gallica ark:/12148/bpt6k5505319w, f = (column + 1) / 2), or the markers withdrawn. ⚠ Both [sic:] sites below are in the letter-confusion class that OCR manufactures (g for q, b for d), which is precisely why frequency and context cannot license them.

0452A · Mitte viros gui considerent terram Chanaan — [sic: gui]

Num 13:2, quoted inside an italic scripture run. gui for qui is not a word. Carried under Patterns 7/10 and marked under Pattern 12; no second italic marker, since it sits inside an already-italic lemma, so the marker reads [sic: gui]. ⭐ 7a⁗ applied: the quarantined run is the clause's only relative pronoun, so the English supplies it in plain words beside the marker — Send men [sic: gui] who may view the land. Deleting the marker's contents still leaves a clause with a relative and a verb. Migne prints the same clause correctly eight lines later (Mitte viros qui considerent terram, @0452A, and again qui considerent in the Deuteronomy quotation), which is the internal control.

0454C · Cabesbarne est locus in deserto — [sic: Cabesbarne]

b for d in the place-name, and the lemma it glosses prints Cadesbarne correctly in the same line (De Cadesbarne.), as do eleven other occurrences across 0001–0002. A proper name with a wrong letter cannot be rendered; carried and marked. 7a⁗ check: the run is the sentence's subject, no verb, negation or conjunction rides inside it, and the lemma immediately before it tells the reader what the word is.

0445D · Fasciculus myrrhae delectus mihi — [var: Vulg. dilectus meus mihi, …]

Cant. 1:12. Migne prints no meus, so his lemma reads "a bundle of myrrh, beloved to me" rather than the Vulgate's "my beloved is to me a bundle of myrrh" — and the whole preface is built on the shorter reading: the bundle is the book, and the gloss closes merito est fasciculus dilectus Christo @0447C, "deservedly is it a bundle beloved by Christ." Under the received text the bundle would be Christ himself and the exposition would not stand. The divergence is therefore material and takes a [var:].

⚠ The printed delectus is NOT marked and NOT treated as a divergence. delectus for dilectus is the ordinary medieval e/i alternation (Pattern 9: attested spellings are correct Latin for their period), and the gloss's own dilectus Christo twelve lines later shows the same word is meant. Rendered "beloved", no italics, no marker. Marking it would have buried the real divergence — the missing meus — in noise.

§3 · 0001 — CAPUT PRIMUM

0449B · Qui Hebraice Choreb ( ) — a hole in the digitization. [ed: …]

Our source prints an empty parenthesis in the middle of a Hebrew etymology. This is not a guess about the plate: PL 113's plate read of 2026-08-28 found Corpus Corporum had dropped three Hebrew words Migne sets in the praefatio of the companion Glossa 8944 (recorded in data/plate-reads.json, recovered as data/hebrew-patches/8944.json). The same class, the same volume, the same recension. Marked [ed: …] under Pattern 13a clause 3 — the hole is made visible and nothing is supplied — because unmarked it reads as our build defect, which is exactly the 11055 case.

⬜ Open, and cheap: one page render settles it. Col. 0449 is PDF p. 230 of patrologiaecurs04migngoog (or Gallica f225). If Migne sets a Hebrew word there it belongs in data/hebrew-patches/8945.json and the [ed:] is removed in the same commit. ⚑ A second instance of the same ( ) stands at 0020.md (col. 0496 region, another agent's range) — worth recovering on the same page pass; it is the only other one in the work.

0449C · legis Evangelii impletionem — no et, and none was invented

Quadragenarius … legis Evangelii impletionem significat, with the two genitives simply juxtaposed. The next clause proves both are meant (decem namque ad legem pertinent, quatuor vero ad Evangelium). Rendered asyndetically, "the fulfilment of the law, of the Gospel" — supplying "and" would be exactly the invented conjunction that both blind readers caught in 8968 @0533D. Not marked: nothing is broken type, and the English asserts nothing false.

0450C · Moyses non ex suo cognati sui consilio hoc populo suo persuasit — the missing sed

The sentence prints non ex suo and then cognati sui consilio with no adversative, and the argument plainly wants non ex suo, sed cognati sui consilio: the paragraph's whole point is that the counsel came from Jethro (Exod. XVIII, cited two lines above). Rendered asyndetically and literally — "not from his own, from his kinsman's counsel" — with the negation carried and no conjunction supplied. ⛔ A [cj:] was considered and declined. It is close to the 7561 @0648 shape (a dropped member of a correlative, glossed on the word Migne did print), but the bar is whether the English asserts something false, and an asyndetic rendering asserts nothing; it is merely compressed. Recorded here, where the conjecture belongs.

0449A · [n: (Matth. XXXI; Luc. XIII)] — Migne's citation, printed as he printed it

Matt. 23:38 is meant (there is no Matt. 31). Not corrected in the text: citation repair belongs to data/citation-corrections.json at index time, never to the running prose.

0449C · scripta ad correctionem nostram — 1 Cor 10:11, and NOT marked

Migne prints Omnia enim in figura contingebant illis, scripta ad correctionem nostram for the Vulgate's Haec autem omnia … scripta sunt autem ad correptionem nostram. The quotation is abridged at both ends (an ordinary Glossa habit) and correctio / correptio are near-synonyms whose difference the argument does not touch. Rendered "for our correction". No [var:]: Pattern 14 fires on a divergence that changes the sense, and this one does not. Recorded so a later reader knows it was seen.

0450C · Novit qui sunt ejus qui numerat multitudinem stellarum [n: (II Tim. II)]

A conflation: 2 Tim. 2:19 (Novit Dominus qui sunt eius) welded to Ps. 146:4 (qui numerat multitudinem stellarum), under the 2 Timothy citation alone, and with Dominus dropped. Rendered exactly as printed. No [var:] — a conflation of two verses is not a divergent witness to either, and no reading is lost to a reader who has the citation and the Latin opposite.

0447B/0451A · illis … illis, and πεντεκοσιάρχους for the officers of fifty

Two readings that are odd and are Migne's, rendered as printed with no marker:

• 0000 @0447B — illis enim promissa sunt … illis quadrata, the same pronoun for the two contrasted sets, where sense wants "to the former … to the latter." English "to those … to those" is equally ambiguous and equally faithful, so nothing is concealed.
• 0001 @0451A — the Greek glossing quinquagenarios ("officers of fifty") is πεντεκοσιάρχους, commanders of five hundred. The Greek rides verbatim (§1) and the mismatch stands visible between the two words. It is a fact about this printing, not a translatable one; no [sic:], because the Greek is a well-formed word and the column is unread.

0448C · Undecim vero dies Moyses antequam veniret ad Cadesbarne

Elliptical and hard: the eleven days of Deut. 1:2 are made the span of an oral recapitulation, and the second half of the sentence (et de Cadesbarne usque ad campestria Moab, et illo spatio undecim dierum scripto voluit commendare) has two locatives and one verb. Rendered as closely as the printed words allow, keeping both. Nothing supplied.

§4 · 0002 and 0003 — readings kept as printed

0452C · sin autem non est ipsa Cades — rendered literally, conjecture declined

The passage distinguishes the desert of Sin at Cades (Num. 20:1) from the desert of Sin between Elim and Sinai (Exod. 16:1), and closes sed alia prorsus, et aliis apud Hebraeos scripta litteris — "written with other letters among the Hebrews", which is the whole point. But what they passed between Elim and Sinai was Sin, not Cades, so the printed Cades here is suspect (Sin is the obvious candidate). Rendered as printed — "but Sin is not that same Cades which … they passed through between Elim and Sinai" — and the conjecture left here. No marker: Cades is a real name in its own grammatical slot, and the column is unread.

0456A · Quia omnis persona … traduntur victi in manus sanctorum — an anacoluthon

The long sentence opening Quia omnis persona cujuslibet dignitatis … confidens never gets a singular verb; it runs through cum persecutoribus … usque ad haereticorum conciliabula and ends in the plural traduntur victi. Rendered as it stands, with the subject left singular and the verb plural, which is what the page does. Not marked (no English exponent would show it any better than the parallel column does).

0456D · convenienter dux et princeps in terram promissionis introduxit — no terminal stop

The chunk, and Migne's sentence, simply stop. Pattern 8: nothing supplied. The English stops where he stops.

0451A · the denarius pun — "denary"

qui denarii perfectione vineae denarium exspectantes, ejusdem denarii multos praeparant participes: one word doing duty for the number ten and for the coin of Matt. 20. Rendered "denary" throughout, the one English word that carries both, so the pun survives; "penny" would have kept the coin and destroyed the number, which is what the gloss on the decani is built from.

0453B · Nox peccatoris, dies vita justi

Verbless in the first half. Rendered "Night is the sinner's, day the life of the just" — the ellipsis kept, nothing filled in.

§5 · Negation audit (Pattern 7a), and the supplied-pronoun list

Every negation particle in the five Latin chunks was located and matched against the English, one site at a time (the count is a census of the sites, not a substitute for reading them). 125 tokens: non 83 · nec 19 · ne (final) 6 · nisi 3 · nemo/nemini/neminem 5 · nullus/nullum/nulla/nulli 6 · nihil/nihilo 2 · neque 1. Per chunk: 0000 8 · 0001 20 · 0002 26 · 0003 32 · 0004 39. None was dropped and none was added, and no word was rendered as its opposite to make a sentence run. The sites where carrying one made the English stiff, and it was carried anyway: 0001 @0450A (praeceptum … non dicitur, kept as "is not spoken of" rather than smoothed to "no command is recorded"); 0001 @0450C (§3 above); 0003 @0456B (Honestum etiam esse non potest, quod turpe est → "Nor, moreover, can that be honourable which is base"); 0003 @0455C's four-fold non fieret nisi … nec resisteret nisi, kept in full.

Pattern 17. The range's only second persons are in quoted scripture and in the Decalogue exposition at 0004 @0458A–0458C, where the Latin is grammatically singular throughout (colas, suscipias, credas, operaberis, sanctifices, honora, non maechaberis) and the English is thou. The plural addetis (0004 @0457A) and venissetis / dixistis (0002 @0452A) take you. No addressee's number was altered.

Supplied pronouns and words, each one a place where the Latin has none and English requires one; listed so a later reader can see them without collation: "it" as subject of the Deuteronomy book, 0000 @0447B–C passim · "he" for the unexpressed Moses, 0001 @0449B (incipiens per Seir … venit) and 0001 @0449D · "them" at 0001 @0449D (trahere vel provocare) · "him" at 0003 @0455B (persequitur) · "us" at 0003 @0455C (conterit Deus Satan sub pedibus nostris is explicit; concidunt at 0002 @0453D took a supplied "us") · "it" for the people at 0003 @0454D (Israel, populum Christianum, an elliptical apposition rendered as one).

§6 · What is owed before this work ships

1. Plate reads on cols 0445D, 0452A, 0454C (§2) — or the three markers withdrawn.
2. A page render of col. 0449 to recover the Hebrew word behind ( ) (§3), which would retire the [ed:] and produce data/hebrew-patches/8945.json; the sibling ( ) in 0020.md belongs to the same pass.
3. Nothing else. The remaining entries above are recorded readings, not open questions.

Stint covering src/latin/8945/0005.md–0009.md (cols 0459D–0470C), translated 2026-08-31. All five chunks were read line by line against the Latin twin. Chunks 0006 and 0007 returned nothing beyond the punctuation/paragraph notes below; 0005, 0008 and 0009 carry the substantive entries.

⚠ No plate read exists for this work (data/plate-reads.json records nothing for 8945). Per CLAUDE.md rule 8a, no [sic:] and no [var:] was fired anywhere in this stint, including at the three columns below where a divergence from the Vulgate is visible in our Latin. Those are logged here and the markers are owed a plate read before they can be placed.

0005

• 0461A — damnant, and it is the one marker fired in this stint. Migne prints Sic in unaquaque anima agitur, ut maxima lucra conferant, aliquando minima damnant. The sentence has no subject for conferant and the indicative damnant breaks out of the ut clause, so it does not construe. Gregory's sense (the source of the whole preceding paragraph, note [n: (RAB., GREG.)]) is that the smallest losses sometimes confer the greatest gains — i.e. minima damna (nom. pl.) as the subject of conferant. Rendered as printed and glossed additively under Pattern 18: [cj: damnant; read damna, "the smallest losses"]. Nothing was removed; the printed reading keeps its English.
• 0459D–0460A — Cassian's vice-names kept in italic Latin: gastrimargia, philargyria, acedia, cenodoxia. These are the technical Greek terms of the eight-logismoi scheme, on the same footing as the register's precaria / emphyteusis rule; the plain-Latin members of the same list (ira, tristitia, superbia, fornicatio) are translated. Fixed for the whole work: do not vary. Note that pigritia (under tristitia) is rendered "laziness" and acedia is left in Latin, so the two are not collapsed.
• 0460B — the lemma and its gloss spell the name differently. The lemma prints Haetheum et Gergezaeum, the Jerome extract immediately after prints Hethaeum. Both spellings are preserved ("the Haethite" / "the Hethite"): the variation is on the plate and is exactly the kind of datum a normalized English would destroy.
• 0461A — Ἀπὸ τοῦ ἀναθήναι. Reproduced verbatim and untranslated (the following id est, a sursum ponendo is Migne's own gloss and is translated). The Greek infinitive as printed is not a form of ἀνατίθημι that a grammar will own (ἀνατιθέναι expected); whether this is Migne's type or Corpus Corporum's is not decidable from our files, so nothing is marked.
• 0460A — Cappadocesve partem Hevaeorum has no verb of its own; deleverunt carries over from the preceding clause. Rendered with the elision preserved ("or the Cappadocians a part of the Hevites"), no supply.

0006

• 0462B, 0462D (end of CAPUT X), 0463B (end of CAPUT XI) — three glosses end with no terminal stop on the plate (…in montis vertice · …tendit in ardua virtus · …maledictioni subjacebunt). Pattern 8: none was supplied, so the English stops unpunctuated too. This is not an omission.
• 0463D — double negation carried. meritis tuis non assignabis … non viribus tuis, sed Deo: both negatives stand in the English ("thou shalt not assign to thy own merits … not to thy own strength, but to God"), though English would idiomatically drop the second.

0007

• 0465D — In lacte, mediocribus disciplinis, sed austerioribus. The sed has no negative to answer: "by middling disciplines, but by more austere ones" is self-contradictory as printed. The expected shape is non mediocribus … sed austerioribus. No negative was supplied (Pattern 7a) and no marker fired — the printed words are all real words in their slots, and the missing particle is as likely our digitization's loss as Migne's; a plate read would settle it.
• 0464B–0466A — the choerogyllus (Per choerogyllum et leporem) is kept in italic Latin: the animal has no settled English name and the gloss turns on its pairing with the hare, not on its identification.
• 0464C — sed quae significando exprimunt, mores, scilicet hominum: the relative has no expressed antecedent; rendered as the loose apposition it is ("but what they express by signifying are the manners, namely, of men").

0008

• 0466C — VERS. 3-- is printed without the period after the numeral that every other verse label in the work carries (VERS. 1.--, VERS. 6.--). Reproduced exactly as printed.
• 0467B — the verse labels run out of order: VERS. 19 (De primogenitis) and Non operaberis are followed by VERS. 16 (Sin autem habuerit), which belongs before them. The sequence is Migne's; nothing was reordered.
• 0466C — 1 John 3:17 is printed as a question (quomodo manet in eo charitas Dei?) and keeps its question mark.

0009

• 0469D–0470A — VERS. 81.-- Si difficile. The lemma is Deut. 17:8, so the label should read VERS. 8; 81 is not a verse of this chapter, and the label that follows it is VERS. 11. Carried as printed. No [sic:] — this is a structural label rather than prose, our Latin is not the plate, and the gate of rule 8a stands; a plate read would decide whether the extra digit is Migne's.
• 0468C — ad alteram diem explorationis hebdomadae septimae (Lev. 23:16). The received text reads expletionis, "of the completion" — which is what the passage needs, since the fifty days are being counted to their end, not to a searching-out. Rendered as printed ("of the searching out"). [var:] withheld pending a plate read (rule 8a); it is owed here.
• 0468D — Quia si terrena domus hujus habitationis dissolvatur, habebit domum non manufactam in coelis, cited [n: (I Cor. V)]. Two divergences at once: (a) the citation is 2 Cor. 5:1, not 1 Cor. 5 — printed reference kept verbatim per the refDisplay rule, no correction in the text; (b) the received text has habemus ("we have"), Migne prints habebit ("he shall have"), which the gloss's own third-person frame (Qui … utitur … potest) actually supports. Rendered as printed. [var:] withheld pending a plate read.
• 0470A — Supra cathedram Moysi sederunt scribae et pharisaei (Matt. 23:2) for the received Super cathedram. Rendered as printed ("Upon the chair"); the divergence is not material to the gloss, and is logged only so the absence of a marker is not read as agreement.
• 0470B — intelligendum est, non debebis. Rendered "it is to be understood, thou oughtest not" — the point of Augustine's remark is that the future indicative non poteris has deontic force, so the English keeps the bare future of the lemma and the "ought" of the gloss distinct.
• 0468A–0468B — three separate glosses cite 1 Cor. 5; each [n: (I Cor. V)] is reproduced at its own position, and the two quotations of Epulemur differ from one another as printed (in azymis, non in fermento veteri at VERS. 3; non in fermento veteri, neque in fermento malitiae et nequitiae at VERS. 4). Neither was conformed to the other or to the Vulgate — 7a″.

Standing note for the merge

Terms fixed in this stint and needing one voice across the whole work: phase (kept, not "passover"), "the pasch" for pascha, "chew the cud" for ruminare, "divide the hoof" for ungulam findere/dividere, "manner of life" for conversatio, "solemnity" for solemnitas, "retribution" for retributio, and the four Latin vice-names above.

Stint covering src/latin/8945/0010.md–0014.md, cols 0470D–0481D (CAPUT XVIII – CAPUT XXV), translated 2026-08-31.

All five chunks were read line by line against the Latin twin. Chunks 0010, 0011, 0012 and 0014 carry substantive entries below. Chunk 0013 was read and is clean — it carries no carried type, no dropped or intruded negation, and no lemma conformed to a familiar English version; its only notes are the two punctuation/format observations in §F, which are not defects.

Marker parity, reconciled per chunk (Latin body ‖ English body):

Apparatus fired in this stint: 4 [sic:] · 1 [cj:] · 1 [nt:] · 0 [d:] · 0 [var:] · 0 [ed:] · 0 [f:] · 0 [cn:]. Zero guillemets in the Latin, zero in the English.

⛔ PLATE READ OWED — every [sic:] below is provisional

data/plate-reads.json records nothing for 8945, and data/plate-notes/coverage.json has no entry for this work at all (its sister book 8947 is recorded, 8945 is not) — so the CLAUDE.md 2026-08-19 spot check is also owed. Every judgment below about Migne's type rests on Corpus Corporum's transcription, not on the plate. The four [sic:] markers stand on cols 0471C, 0476C, 0481A and 0481C and must be plate-verified before this work is marked ours; node scripts/plate-gate.mjs 8945 will name them.

PL 113's on-disk archive.org copy (patrologiaecurs04migngoog, PDF page = (column + 11) / 2) clips the left margin on some pages, and the second witness is Gallica ark:/12148/bpt6k5505319w at f = (column + 1) / 2 — calibrate against the printed corner numbers before either read.

⚠ In-work marker policy is NOT uniform across the five parallel stints, and the merge must settle it. Stints 0015/0016/0020 fired [sic:] on carried type with the same "plate read owed" caveat (this stint follows them, as the majority convention). The 0005–0009 stint fired none, on the ground that rule 8a forbids a claim against Migne's type without a plate read, and logged its candidates instead. Both are defensible; they cannot both ship.

1. Carried type — the four [sic:]

Each is a non-word or a form with no grammatical slot, carried whole per Patterns 7/10 and marked per Pattern 12. 7a⁗ was checked on all four: none of them swallowed a verb, a negation, an interrogative or a conjunction, except where noted.

• @0471C — coversans for conversans. Sicque religiose coversans in mundo. Not a Latin word. Carried: And thus [sic: coversans] religiously in the world…. 7a⁗ check: the clause's finite verb is exspectat ("he looks for"), which is outside the marker, so the sentence still predicates. The conjecture is conversans, "conducting himself / living," near-certain from religiose + in mundo and from the parallel conversationem usage elsewhere in the work.
• @0476C — magnificam for magnificant. Migne prints, inside a quotation of Matt. 23:5, Dilatant phylacteria sua et magnificam fimbrias. magnificam is a real form (acc. fem. sing. of magnificus) but has no grammatical slot: the clause needs a finite verb governing fimbrias, and an adjective in the accusative cannot supply one — Pattern 12's first "real word that still fails the test" case. ⚠ 7a⁗ question 1 applies: the marked run was the second clause's only predicate, so the sense is supplied in plain words beside the marker — and they magnify [sic: magnificam] their fringes. Bare inside the italic run, per Pattern 10 (no nested italics). Conjecture: magnificant, the Vulgate's reading, which the gloss's own dilatabant fimbrias two lines above supports.
• @0481A — Catceati for Calceati. In the quotation Catceati pedes in praeparatione Evangelii pacis. Not a word; a broken l. Carried bare inside the italic lemma. The lemma-and-gloss coherence is the evidence: the gloss immediately following reads Doctorum enim pedes calceati sunt, with calceati set correctly, and the whole passage turns on being shod — so the defect is one sort, not the author's word.
• @0481C — pusillium for pusillum, and quo que for quoque. Two defects on one column. (a) Pondus magnum et pusillium — pusillium is not a word; carried bare inside the italic quotation, conjecture pusillum, "small," which the answering pondus magnum requires. The clause keeps its verb (sunt → "are"). (b) Spiritualiter quo que studendum est — a split run under Pattern 10, carried whole with its spacing: Spiritually [sic: quo que] we must take care…. The run is quoque, "also"; it is an adverb, not a verb, negation or conjunction, so 7a⁗ does not require the force to be supplied beside the marker, and it is recorded here instead.

2. The one [cj:] — @0475A, infructuosam

Migne prints: Non est de die in diem differenda poenitentia, sed statim agenda, ut per poenitentiam infructuosam ab oculis Dei abscondantur delicta.

infructuosam is a perfectly good word, so Pattern 12 has nothing to wrap — and the faithful English then says that unfruitful penance hides offences from God's eyes, in a sentence whose whole point is that penance must be done at once so that it works. The contradicting neighbour is on the same plate and in the same chunk: eight lines above, at 0472B, the gloss says Qui fraternum odium retinet in corde, non potest fructuosam poenitentiam agere. The privative prefix is the word that went wrong, which puts it inside Pattern 18's negation class (18a's ✅ extension, ruled 2026-08-15).

Rendered additively, nothing removed: through unfruitful [cj: infructuosam; read fructuosam, "fruitful"] penance. The printed word is verbatim in the Latin twin, satisfying the guard.

3. The one [nt:] — @0475A, a note that is prose

Migne's note (Adam namque in ligno praevaricationis corruit: ideo sibi et posteris suis maledictionem meruit, de qua dicit: Maledictus omnis qui pendet in ligno.) is a statement, not a locator — a four-clause explanation with a scripture quotation inside it. Converted to [nt:] and translated in place, at the same marker index, per Pattern 15. Its parentheses are kept as printed. Every other note in all five chunks is a citation and stays [n:], verbatim and unanglicized, including the ones whose reference is wrong (§5).

4. Declined conjectures — rendered as printed, logged only

Each of these is a real word occupying its slot; under Pattern 7 the English is ours and correct, and under Pattern 18's bar (as narrowed by 18a) none of them is a negation or a correlative, so none was glossed. Do not read this list as a to-do.

• @0471A — in successum projici for in secessum (Matt. 15:17's word, "the draught"). Rendered "is cast forth into the outcome." successus is a real noun in the accusative and construes; the sense the gloss needs ("passes out of the body") is the conjecture, not the reading.
• @0471A — prodigorum for prodigiorum. The Augustine excerpt ends sic illae inspectiones prodigorum a significationibus miraculorum discernendae sunt, and its own lemma four words earlier is portentorum inspectores — so the sense is "of portents." prodigorum (gen. pl. of prodigus, "of the wasteful") is a real genitive with a real governor; rendered "of the prodigal."
• @0472D — ut ea, scilicet. sunt ei utilia aut nociva. The stray full stop after scilicet is Migne's and is preserved under Pattern 8 ("never delete"); the indicative sunt where an ut clause wants sint is rendered as printed. The English is correspondingly broken: "Working: that those things, namely. are useful or hurtful to him which he perverts in another." Conjecture: ut ea scilicet sint. Nothing supplied.
• @0476B — et profectionibus diversi generis for professionibus. Rendered "and with departures of divers kinds." A real ablative plural; the following examples (sanctimonialis / uxor / virgo) are professions, which is the evidence for the conjecture and not for an emendation.
• @0479B — non praecipitandum … incurras judicium. The gerundive stands unattached beside a second-person subjunctive. Rendered as closely as the printed words allow, with the non left where it governs incurras: "and mayest not, by that which must be hurried, incur the judgment of the reaped harvest." Conjecture: non praecipitando, "not by hurrying." No negation was moved, added or dropped.
• @0479A — Et pecunia, quae ad usuram dari prohibetur, quia… has no main verb; the answering Altera est quam ad usuram dare debemus two clauses later shows a Una est… has fallen out or was never set. Rendered with the anacoluthon intact ("And money, which is forbidden to be given for usury, because…").
• @0478D — VERS. 37 in a chapter of 25 verses. The head VERS. 37.-- Non erit meretrix stands between VERS. 15 and VERS. 19 and glosses Deut. 23:17. The numeral is Migne's and is reproduced; 17 is the obvious reading (a turned digit), but a verse number is not a word that asserts anything false about the argument, so it takes no marker.
• @0481A — VERS. 5 printed twice (Quando and Mortuus fuerit), consecutively. Reproduced as printed; the second gloss does belong to Deut. 25:5.

5. Migne's citations that point at the wrong place — verbatim, per the [n:] rule

Reproduced exactly as printed. refDisplay keeps what Migne prints; resolution is the indexer's business, not the translator's. Listed so a later pass need not re-derive them:

• @0470C [n: (I Cor. IX.)] on Dignus est enim operarius cibo suo — Matt. 10:10.
• @0470C [n: (Num. XVIII)] on Qui altario serviunt cum altario participant — 1 Cor. 9:13.
• @0472D [n: (Gen. XXIX)] on Dentes ejus lacte … candidiores — Gen. 49:12.
• @0474A [n: (I Cor. XV)] is right for Corrumpunt bonos mores colloquia mala (15:33); @0475A [n: (II Tim. V)] on Peccantes coram omnibus argue is 1 Tim. 5:20.
• @0475D [n: (I Cor. XIII)] on the woman not permitted to teach — 1 Tim. 2:12.
• @0478C [n: (Job. II)] on Balteum regum dissolvit — Job 12:18.
• @0479A [n: (Proverb. X)] on Ruina est homini post vota tractare — Prov. 20:25.
• @0481A [n: (Ezech. VI)] on Calceati pedes in praeparatione Evangelii pacis — Eph. 6:15.
• @0477A, @0480C [n: (Psal. CIII)] on Non nobis, Domine (Ps. 113B:1) and on vinum quod laetificat cor hominis (Ps. 103:15) — the second is right, the first is not.

6. Scripture divergences noticed and NOT marked

Per Pattern 14 the [var:] bar is a divergence that changes the sense, and all of these are rendered as Migne prints them. None was material enough to fire, and ⛔ none could honestly fire in any case while the plate is unread (rule 8a: a [var:] asserts that MIGNE'S text diverges, and our Latin is not the plate — the 9003 collation found Corpus Corporum manufacturing exactly this class of divergence). Recorded for a later plate pass:

• @0470C Gal. 6:6 in omni opere bono for the Vulgate's in omnibus bonis.
• @0474D Luke 3:7 quis docebit vos fugere for quis ostendit vobis fugere.
• @0473A Ps. 34:25 devorabimus eum (future) for devoravimus eum.
• @0471A 1 Cor. 6:13 et hunc et hanc (singular) for et hunc et has. Rendered "both the one and the other"; Latin grammatical gender has no English exponent (Pattern 9), so there is nothing here to mark either way.
• @0478C Job 12:18 praecinxit (perfect) for praecingit.
• @0481C Prov. 20:10 as printed, Pondus magnum et pusillium, et mensurae duplices, immundae sunt apud Deum, is a long way from the Vulgate's Pondus et pondus, mensura et mensura — but that is a loose quotation, not a defect, and the pusillium inside it is handled at §1.
• @0477C Deut. 23:4: the gloss says the nations met Israel cum pane tantum ("with bread only"), where Deuteronomy says they did not meet them with bread and water. This is the author's own paraphrase and it is rendered as printed. No negation was added to rescue it — the noluerunt of his own lemma four words earlier stands untouched.

7. Format and vocabulary held across the stint — for the merge

• usque ad → as far as, italic, matching 8947/8976. Every truncated patristic excerpt keeps its etc. and stops exactly where Migne stops.
• doctor → teacher (never "doctors"), inherited from 8976 §V via 8947. ⚠ This stint was drafted with "doctor" and converted; the parallel stints are not yet uniform and the merge must sweep the whole work. 8 occurrences converted here (0010 ×1, 0011 ×2, 0012 ×6 — counting doctus/indoctus glosses separately — 0014 ×1).
• VERS. N.-- kept unanglicized and ## CAPUT N. → ## CHAPTER N., final stop preserved because this work's heads carry one (8947's CAPUT XXIV does not; the plate decides, per work).
• LXX kept as printed, never expanded to "the Septuagint" in the text.
• Etymological glosses keep their Latin, because the gloss is the etymology and an English word sharing no root would explain nothing (the stylus / murenulae precedent): @0471B avigeria / avigaria, with gerantur and garritus carried in parentheses beside the English; @0481D Amalec lingens sanguinem rendered as the name-gloss it is. Mamzer (@0477B) likewise stands.
• Hebrew place- and tribe-names as Migne spells them (@0472A: Cades, Nephtali, Sichem, Ephraim, Cariatharbe, Hebron, Juda, Bosor, Ruben, Ramoth in Galaad, Gad, Gaulon in Basan, Manasse). Not normalized to their familiar English shapes.
• (Josue XX) at 0472A is an inline italic locator, not an [n:] note, and it is the only one in these five chunks. It is reproduced verbatim and untagged. Pattern 4's typographic test would admit an [f:] here, but a single occurrence in a work that otherwise cites through [n:] is not the inline-tail genre the pattern was written for, and tagging it in one stint of five would put an unmatched [f:] into the index. ⬜ Flagged for merge-time reconciliation: tag it in all stints or none.

8. Punctuation preserved that will look like our error

All Pattern 8 — the marks are Migne's and none was supplied, deleted or moved.

• @0470D Quid ergo deesse potest illi qui omnia habentem habet. — a question closed with a full stop. The stop stands; 8b's colon recast does not apply, because this is a direct question in Latin and English can hold the mark it was given, however odd.
• @0476D …prius in Deum, postea in virum, vir tantum in Deum — the sentence simply stops, with no terminal mark at all. The English stops the same way.
• @0480C Praeterea cum haec populo praecipiuntur; qui non indigent, admonentur — a semicolon breaking the subordinate clause from its main clause. Kept.
• @0472D the stray stop after scilicet — see §4.
• @0472D the italic run of Dentes ejus lacte … candidiores is interrupted by Migne's own roman gloss (id est lege) and by the column anchor. The English preserves both italic runs and the interruption in place rather than reflowing the verse into one quotation: His teeth than milk, that is, [0472D] than the law, are whiter.

Stint covering src/latin/8945/0015.md – 0019.md (cols 0482A–0492D). All five chunks were read line by line against the Latin. Chunk 0017 and chunk 0019 returned no defect of Migne's type; every other finding is listed below. Findings are grouped by column.

Plate reads performed for this stint

⚠ Our Latin is Corpus Corporum's transcription, not the plate, so no [sic:] or [var:] below stands on an unread column (CLAUDE.md rule 8a). Second witness used: Gallica ark:/12148/bpt6k5505319w (PL 113), IIIF, f = (column + 1) / 2 — calibrated at the start of this stint against the printed corner numbers on f242, which read 483 / 484. Nine columns read:

No foot-of-page apparatus of either layer (numbered Forte conjectures or the asterisk cross-references) appears on any of these nine columns. That is a sample, not a proof — it does not license a claim that the work carries none.

0482A — cartalo — no marker

Migne prints ponamus in cartalo conversationis nostrae; the Vulgate at Deut. 26:2 has cartallo. Single-l cartalum is an attested medieval spelling of cartallum, "basket," so this is Pattern 9, not a defect: rendered normally as "basket," no italics, no marker.

⚑ Incidentally noted at the same read: the plate prints a comma at de Aegypto, sic nos which our TEI does not carry. Recorded as a transcription loss, not a defect of Migne's; the English carries the comma, since Pattern 8 follows the plate.

0482C — hoc legis implesse Tobiam — no marker, rendered as printed

Migne prints, verbatim (plate-verified):

VERS. 12. — Quando compleveris: hoc legis implesse Tobiam, sed qui longe erant a templo tertio anno collectas trium annorum decimas dabant pauperibus…

An accusative-and-infinitive (implesse Tobiam) with no governing verb — almost certainly legimus has dropped out ("we read that Tobias fulfilled this of the law"), and hoc legis is "this [precept] of the law." The English renders the printed words as closely as they allow and leaves the clause hanging as the plate leaves it.

Declined for Pattern 18 deliberately: the defect is a missing main verb, not a false assertion, and 18a keeps the [cj:] bar to negations, correlatives, and readings that make the page assert what the author is arguing against. Nothing here asserts anything false; it simply fails to construe.

0483D — trangressione — [sic: trangressione]

Plate-verified. Col. 483 ends the line Quis enim potest sine trangres- and col. 484 opens sione omnibus sermonibus legis permanere — so the s of trans- is genuinely absent from Migne's type, across a column break. A non-word; carried untranslated in italics under Patterns 7, 10 and 12.

Conjecture: transgressione, certain — the sentence is Quis enim potest sine transgressione omnibus sermonibus legis permanere, "who can abide in all the words of the law without transgression," and the Galatians proof-text that follows (Quicunque ex operibus legis sunt, sub maledicto sunt) requires exactly that sense.

7a⁗ check: the quarantined run is a common noun in a prepositional phrase. The clause keeps its verb (permanere), carries no negation and no conjunction inside the marker, so nothing has to be supplied beside it.

0484A — Migne's ? on the Galatians quotation — preserved

cum et Paulus dicat: Quicunque ex operibus legis sunt, sub maledicto sunt? — Gal. 3:10 is a statement, and Migne closes it with a question mark. Preserved under Pattern 8 (a ? the plate prints on a statement rides as printed; that oddity is his, and recording it is the point). Not 8b: the question is already direct, so nothing had to be recast.

0484D — aectinaveris — [sic: aectinaveris], and a divergence from the plate

Migne's lemma at Deut. 28:14 is a non-word. Plate-verified: the plate does print a non-word here, beginning aec-, where declinaveris belongs.

⚠ But our twin and the plate do not agree on one letter. Corpus Corporum reads aectinaveris; at native IIIF resolution the plate looks like aeclinaveris — the fourth letter is a tall ascender with no visible crossbar, though at this resolution l and a t whose bar has failed cannot be told apart with confidence. The marker therefore wraps our twin's reading, as Pattern 12's verbatim-substring rule requires; the letter is recorded here as unresolved. Either way the word is Migne's own broken type and the [sic:] is licensed.

Conjecture: declinaveris, certain from the LXX gloss Migne prints immediately after (Non praeteribis ab omnibus verbis).

7a⁗ check: the marker swallowed the clause's only verb, so the sense is supplied in plain words beside it — "Thou shalt not turn aside [sic: aectinaveris]" — per the [sic: eccidit] model. The negation Non stays outside the marker and stays in the English.

0484D–0485A — the doubled clause — [d: …]

Plate-verified. Col. 484 ends Et aere corrupto. Vita polluta ex carnali im- and col. 485 opens munditia. Carnalis vita polluta ex carnali immunditia. Carnalis vita per aerem exprimitur, quia aeris spiramine humanum corpus vivificatur. The doubling is Migne's, not our transcription's.

The two copies differ, so Pattern 11a governs and the marked copy is argued, not positional. The intrusive copy is the second (Carnalis vita polluta ex carnali immunditia.): its tail repeats the phrase immediately before it — the eye-skip's source — and its head, Carnalis vita, is a false start on the sentence that then actually runs, Carnalis vita per aerem exprimitur. Delete it and the passage reads Vita polluta ex carnali immunditia. Carnalis vita per aerem exprimitur, quia…, which is exactly the argument: corrupted air glosses a polluted carnal life, and carnal life is figured by air because breath quickens the body. Here position and argument agree — the second copy is both the later one and the intrusion — but the reasoning is recorded because 11a requires it whenever the copies are not identical.

0486A — statibilitatem — [sic: statibilitatem]

Plate-verified: Migne prints Fortitudinem, scilicet et statibilitatem auferat. A non-word; carried in italics.

Conjecture: stabilitatem, near-certain — paired with fortitudinem, and the gloss is on in genibus et in suris, knees and calves, which are precisely strength and stability.

7a⁗ check: the clause keeps its verb (auferat) and carries no negation or conjunction inside the marker; nothing supplied.

0488C — In fidelibus — [sic: In fidelibus], sense supplied beside it

Plate-verified: Migne's own type reads In fidelibus autem dicitur: Vos ex patre diabolo estis. This is a word-division defect of the split kind (Pattern 10), Infidelibus set as two words, and it is the dangerous shape 7a⁗-b names: the quarantine is a whole lexeme, and the lexeme carries the polarity.

Conjecture: Infidelibus, certain. The gloss is built as a contrast — Dedit eis potestatem filios Dei fieri, his qui credunt in nomine ejus (John 1:12, of believers), answered by John 8:44, Vos ex patre diabolo estis, which is addressed to unbelievers. Read as printed, "among the faithful it is said: you are of your father the devil" reverses the contrast the sentence exists to draw.

The whole broken run is carried per Pattern 10 (In stays inside it), and because the run carried the polarity, 7a⁗ requires the force to stay outside the quarantine: the English reads "But to the unbelieving [sic: In fidelibus] it is said." The marker says the type is Migne's; this note says what it was meant to be.

0490D–0491A — terna flamma — [cj: terna; read aeterna, "eternal"]

Plate-verified on both sides of the break, and the break is clean: col. 490 ends id est carnem, propter opera libidinis, — at a comma, with no hyphen and no lost syllable — and col. 491 opens terna flamma comburet;. So terna is what Migne prints, and nothing was lost by the digitization.

terna is a real word (nom. fem. sg. of ternus, "threefold") and it occupies its slot, agreeing with flamma — so Pattern 12 has nothing to wrap, and the faithful English is "a threefold flame shall burn." That is the founding Pattern 18 shape (11081 @0030B, munde for mundo): the page asserts something the author is not arguing. His subject two sentences earlier is aeternis suppliciis and ardebit usque ad inferos; a threefold flame appears nowhere in the passage and answers nothing in it. The gloss is additive — Migne's word keeps its English and the conjecture stands beside it.

⚠ Not Pattern 14: the divergence sits in the glossator's own prose, not inside the quoted scripture. The quotation (comedent terram et nascentia ejus) closes before id est carnem.

0491C — the Isaiah quotation — [var: …]

Plate-verified: Migne prints Audite verbum Domini, principes Sodomorum Gomorrhae (Isa. I). Isa. 1:10 reads audite verbum Domini principes Sodomorum percipite auribus legem Dei nostri populus Gomorrhae — Migne's line has collapsed the verse's two vocatives into one, losing populus, so his text addresses "princes of Sodom of Gomorrah" where the received text addresses the princes of Sodom and, separately, the people of Gomorrah. Rendered as printed, with the received reading noted beside it.

Cause not speculated on, per Pattern 14: compression by the glossator quoting from memory and a compositor's eye-skip are both available and telling them apart is a research question.

0491C — obtulerunt with no terminal stop — preserved

Plate-verified: …vinum enim cum felle mistum obtulerunt ends the paragraph with no point of any kind. Pattern 8 forbids supplying one; the English stops where the plate stops.

Same class, same chunk, both plate-verified and both preserved: 0491D Mea est ultio, etc, Paulus de hoc loco sumpsit — a comma where a period belongs — and 0491D In die ultionis visitabo et hoc peccatum eorum (Exod. XXXII), again closing with no stop.

⛔ 0492B — pre dexteram — OUR defect, not Migne's. No marker fired.

This is the finding of the stint that changes a file, and it goes the other way from all the rest. Our Latin twin (src/latin/8945/0019.md) reads:

Tollam in coelum manum meam, et jurabo pre dexteram meam…

pre is not a Latin word and the obvious move was [sic: pre]. The plate refutes it. Gallica f246, right column, band B, reads plainly:

Tollam in cœlum manum meam, et jurabo per dexteram meam, et dicam: Vivo ego in æternum.

So Migne prints per, and pre is Corpus Corporum's transcription error. The gloss on the same column confirms it independently — Migne's own next sentence re-quotes the phrase as Jurare per dexteram, id est per Filium.

Consequences, and they are the point of rule 8a:

1. The English renders per — "and I will swear by my right hand" — because Pattern 7 renders what Migne's plate prints, and the plate prints per.
2. No [sic:] was fired. It would have accused Migne's type of a defect that is ours. This is the class 8a exists to prevent and it turned up on the first work where the check was actually run against a second witness.
3. ⬜ Owed at merge: a TEI patch. data/tei-patches/8945.json should correct pre → per with this provenance, after which 0019 needs re-chunking. It was not done inside this stint because four sibling stints were writing chunks of the same work concurrently and a re-chunk rewrites them all. Until the patch lands, the Latin column on the reading page shows pre beside an English that says "by" — which is a visible inconsistency, and this note is the record of why.

Citations Migne prints that point elsewhere — recorded, not corrected

refDisplay keeps what Migne prints; none of these is a defect of the text and none takes a marker.

• 0487B Ubi venit plenitudo temporis, misit Deus Filium suum natum ex muliere is cited (Ephes. IV); it is Gal. 4:4. Migne's wording also gives natum where the Vulgate has factum. Logged without a [var:]: "born" and "made" of a woman name the same thing here, and no clause of the gloss turns on the difference.
• 0487D Non est speciosa laus in ore peccatoris is cited (Eccli. XIII); it is Ecclus. 15:9.
• 0491B Vermis eorum non morietur, et ignis eorum non exstinguetur is cited (Isa. LVI); it is Isa. 66:24.
• 0483B Usque in hanc horam esurimus is cited (I Cor. IV) — correct (4:11).

Read clean

Chunk 0017 (cols 0487B–0488A) and chunk 0019 (cols 0492C–0492D) were read against the Latin word by word and contain no defect of Migne's type, no doubling, and no scripture divergence material enough to mark. 0017's only entries above are two mis-pointed citations, which are apparatus, not text; 0019's only entry is the pre case at 0492B, which is our file's defect and not the plate's. Chunks 0015, 0016 and 0018 each carry findings, itemized above.

Lemma-and-gloss coherence — checked, and it is what caught two of these

Per 7a″, every scripture lemma in the five chunks was read word by word against the printed Latin before the English was trusted, and each gloss was read back against its own lemma. Two findings above were caught by that check and by nothing else: 0488C (In fidelibus, where the gloss's contrast identified which word had broken) and 0491A (terna, where the surrounding argument about eternal punishment identified the lost syllable). No lemma in these chunks was conformed to a familiar English version; the Deuteronomy lemmata are rendered from Migne's printed Latin throughout, including Non aectinaveris, which has no English at all.

Chunks 0020, 0021, 0022, 0023 (cols 0493A–0506C) — CAPUT XXXIII from VERS. 1 through the end of CAPUT XXXIV and the close of the book. 5,282 Latin words, 8,076 English (1.53×). All four chunks were read line by line against the Latin twin; every column band below was examined, and the chunks are otherwise clean.

Apparatus fired in this stint: 1 [sic:] · 1 [ed:] · 0 [d:] · 0 [var:] · 0 [cj:] · 0 [nt:] · 0 [cn:]. Notes at exact parity (27 · 29 · 26 · 5); column anchors verbatim and in order (16 · 16 · 14 · 8); guillemets zero on both sides, as in the Latin; six supplied em-dashes across four chunks.

⛔ PLATE READ OWED. Every judgment below about Migne's type rests on Corpus Corporum's transcription, not on the plate (CLAUDE.md 8a). The one [sic:] (§2) stands on col. 0496C and must be plate-verified before this work is marked ours; PL 113 has a second witness at Gallica ark:/12148/bpt6k5505319w, f = (column + 1) / 2. Where a defect could be rendered at all, I declined the marker and logged it here instead — the line taken throughout this stint is that a marker fires only where carrying the type is forced (a non-word that cannot be put into English), never merely where a reading looks wrong.

1. @0493B — the empty parentheses after the Hebrew names → [ed:]

Migne prints Sinai ( ) interpretatur amphora mea…, Seir ( ) pilosus…, Pharan ( ) ferocitas eorum… Three parentheses, all empty in our source. The type inside them has been dropped in the digitization: the same empty parenthesis appears in this work at chunk 0001, Qui Hebraice Choreb ( ), where what stands inside can only be the Hebrew form of the name. Rendered faithfully the English shows a hole that reads as our bug — the 11055 case exactly — so one [ed:] is placed at the first occurrence and covers all three, supplying nothing and naming what recovery requires (Pattern 13, and 13a clause 3: the loss is not recoverable from our source).

⚠ Merge note for the work-level cruces: this class is not confined to my chunks. Chunk 0001 carries it too, and whatever that stint did with it and this must agree at merge time. If the other stint rendered it silently, one of the two has to move.

2. @0496C — pli for pii → [sic: pli], the stint's only carried type

et quasi holocaustum ab ara pli cordis flamma devotionis in coelum subvolat. The sense is past doubt (ab ara pii cordis, "from the altar of a devout heart"), and pli is not a word, so there is nothing to render: carried untranslated in italics under Patterns 7 and 10, and marked under Pattern 12 so the defect is not read as ours. 7a⁗ check: the marked run is a single attributive adjective — the clause keeps its subject (flamma devotionis), its verb (subvolat), and carries no negation, interrogative or conjunction inside the marker. ⛔ Plate read owed before this ships (§ head).

3. @0497C — requiscet: rendered, deliberately NOT marked

Lemma Et inter humeros illius requiscet. — for requiescet, and the gloss that follows turns on the verb (Tanto autem libentius Christus in illo requievit). Loss of e in hiatus is an ordinary medieval orthographic reduction, i.e. Pattern 9's "attested spelling", not type unique to this plate; and a [sic:] here would swallow the lemma's only verb, the shape 7a⁗ was written against. Rendered "he shall rest". If a plate read later shows the sort itself is broken, the marker can be added without touching the English.

4. @0497D — homine delectantur: no nominative in the sentence

Temperie coeli, et rore fructus, et fontium et fluminum abundantia, homine delectantur, quae in abysso significantur. Every candidate subject is ablative and homine is singular ablative, so the sentence has no nominative to govern delectantur. Rendered as closely as the printed words allow ("…are delighted with man"). Conjecture: homines delectantur — "men are delighted by the mildness of the sky, by dew, by fruits, and by the abundance of springs and rivers" — which is what the surrounding fertility gloss is plainly saying. Declined for [cj:]: the English is incoherent rather than false, and the word that went wrong is neither a negation nor a correlative (Pattern 18a's scope, ruled 2026-08-18).

5. @0499D — passionem where the sense wants possessionem

Zabulon et Issachar in Galilaea passionem acceperunt, ubi Christus maxime docuit. The clause is about tribal territory (Galilee is where their lot fell, and therefore where Christ taught and chose the apostles), so possessionem is near-certain. Rendered "received their passion in Galilee" — Pattern 7, no silent repair. Declined for [cj:] on the same ground as §4: a substitution that leaves the sentence odd, not a false assertion of the class Pattern 18 admits.

6. Punctuation carried against the English grain (Pattern 8)

• @0494D, Isa. 1:19 — Migne prints Si volueritis et audieritis, me bona terrae comedetis, with the comma falling before me, where the Vulgate has audieritis me, bona terrae comedetis. Rendered with the pronoun left where the plate puts it ("…and shall hearken, me you shall eat the good things of the land"). The comma is not deleted and the pronoun is not moved.
• @0496C — pro vitulis pergentes mugiebant sed accepto, itinere gressus non deflectebant: the comma falls inside the ablative absolute. English syntax has no position for a mark there; rendered "but, the journey once taken up, they did not turn their steps aside", i.e. the mark is kept but re-sited by one word. Logged rather than silently smoothed.
• @0505B — quia non est perturbatus tempestate vitiorum, ordo discretionis eorum centum et viginti sunt anni legislatoris quo numero… is one run-on with a single comma and no full stop. Kept as a run-on; no stop or semicolon supplied.
• @0503D — Beatus es, Israel, etc ends the paragraph with no point after etc. The English ends the same way.
• @0494B — the plate prints VERS.--3.-- (dashes before the numeral, unlike every other verse label in the work). Reproduced as printed.

7. Scripture as Migne prints it — divergences logged, none marked [var:]

Every lemma in this stint was read word by word against the printed Latin before the English was written (7a″). The divergences found are all of the kind that a plate read could reassign to the digitization, so each is recorded here and none fires a marker.

⚠ 0503D is the one that would take a marker under Pattern 12's second test: prae + accusative has no English exponent, so "before the brightness" renders only by hiding the disagreement. It is declined here because a case ending is exactly the class our transcription is least trustworthy on, and a [sic:] is a public claim against Migne's type. Re-open it if the plate read confirms the accusative.

8. Citations in the notes: printed, not corrected

[n:] contents ride verbatim, so these stand as Migne set them; the true reference is the indexer's business (refDisplay vs refKey), not the translator's.

• @0496B Nisi quis reliquerit patrem et matrem… keyed [n: (Luc. XVIII)] — the verse is Luke 14:26.
• @0499D Caecitas contingit ex parte in Israel keyed [n: (Rom. II.)] — Rom. 11:25, and the same verse is correctly keyed [n: (Rom. XI.)] at 0493B in chunk 0020, so the plate disagrees with itself inside one chapter.
• @0506B the triple [n: (Hebr. X; Joan. I; Rom. X)] on one composite quotation is Migne's own; kept whole.

9. Two marker-class judgments, stated so a later reader can overturn them

• @0498B — [n: (unde: Mons domus Domini elevabitur super colles (Isa. II) )] stays [n:], untranslated. It is longer than a locator and it quotes a verse, but its content is a proof-text with its own citation, i.e. apparatus, not editorial prose making a statement (Pattern 15's test is content, and this is a citation carrying its quotation). It is the only note in the stint where the call was close.
• @0502D — Ferrum et aes cal. is carried untranslated, unmarked. The lemma cue truncates mid-word (calceamentum), and a truncation is not a defect: it is Migne abbreviating a lemma he expects the reader to complete — the gloss immediately below supplies it (Incorruptibile quoque praebuit calceamenti munimentum). Carried as printed in its italic run, on the L4 principle; no [sic:], because nothing is broken.
• @0503A — the inline (Psalmus XCI) sits inside the italic quotation as running text, not as an [n:] note. Kept verbatim and in place; it is not promoted to a note marker, which would change the note count against the twin.

10. Negation audit (Pattern 7a)

Every non · nec · neque · nisi · nullus · nihil · nunquam · nusquam in the four chunks was checked against its English, in both directions. Count of negative particles in the Latin body and of their English exponents agree at every site; nothing was dropped and nothing supplied. Three places where the printed negation makes the English harsher than a smooth reading would like, and all three stand:

• @0493B prohibere novit, non juvare: nec spiritu dilectionis liberat — "it knows how to forbid, not to help: nor does it set free by the spirit of love".
• @0504D nusquam sacrificia, nusquam libamina, nusquam sacerdos vel pontifex — all three nusquam carried ("nowhere… nowhere… nowhere"), not compressed into one English negative.
• @0505A nec profunditatem sapientiae Dei, vel secretum consilii alicui patere — the nec is carried onto both members ("neither the depth… nor the secret… lies open to anyone").

Also carried: Dedi eis praecepta non bona (@0504D, Ezech. 20:25), in eos qui non peccaverunt (@0506A), oculus non caligavit / nec dentes moti (@0505B), Non est ita legendum (@0499A), quia non de tua virtute (@0504A), Neque enim ex operibus legis justificabitur omnis caro (@0501C), and non necesse habetis ut aliquis doceat vos (@0501B).

11. Renderings fixed for this stint, for merge-time reconciliation

Ecclesia → the Church (capital, throughout) · gentes/gentiles → nations in scripture lemmata, "the gentiles" in the gloss where the contrast with Judaei is the point · primitiva Ecclesia → the primitive Church · charitas → charity (never "love", except inside a scripture lemma that prints dilectio) · doctrina → doctrine · sol justitiae → the sun of justice · interpretatur → "is interpreted" · Quasi: opening a gloss → "As if to say:" · usque ad → as far as (8978 precedent) · unde → "whence" · Isaias, Habacuc, Balaam, Juda, Ruben, Aser, Manasses, Zabulon, Esdras keep Migne's forms · second person singular renders thou (Pattern 17) — @0496D Suscipe, @0495D quam sis perfectus, @0505A si conjunxeris… si respicias.