Epistola II ad Corinthios

PL 114, cols 0551A–0570C · 10 chunks · 7,797 Latin words. Attributed by Migne to Walafrid Strabo; in fact Anselmus Laudunensis et schola. Translated 2026-08-30 by two Opus stints of five chunks each, neither of which could see the other.

This file is the MERGE LAYER, and it is not a copy of the range files. cruces-0000.md (chunks 0000–0004, 2 Cor 1–5) and cruces-0005.md (chunks 0005–0009, 2 Cor 6–13) stand beside it and carry the per-site detail — the full lemma collations, the negation tables, the unexpressed-subject rulings, the plate candidates. This file records only what was reconciled across the two ranges, which is the thing neither stint could do and the thing a later sweep must not churn.

⚠ Deliberately not the 8977/8944/9003 form. Those merges embed their range files, and because build-cruces.mjs concatenates every cruces*.md in the directory, their published pages carry the embedded text and the standalone file both — the apparatus appears twice. Verified 2026-08-30 on /cruces/pl/114/apocalypsis-b-joannis/. Keeping this file to the merge layer alone makes the published page read head → range A → range B, once each.

§0 · Apparatus, counted from the English

8 [var:] · 5 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [f:] — counted from src/english/8986/0*.md after the merge edits AND the plate gate, not summed from the two stints' reports. (Before the plate read the work stood at 7 [var:] and 0 [sic:]; the read confirmed all seven, fired an eighth, and produced the five [sic:] — see §5.)

verify-english.mjs 8986 → OK, 10 chunks, columns/notes/sections aligned, no duplicate paragraphs. scan-raw-markers.mjs → no marker rendered as literal bracket text. Word ratios 1.45–1.50×, no ratio warning.

✅ plate-gate.mjs 8986 PASSES (2026-08-30): every marker stands on a column that was read. Six pages read at 300 dpi — 282, 283, 285, 286, 289, 290 — recorded in data/plate-reads.json.

⭐ Both stints independently declined to fire a single [sic:], and both gave the same reason without having seen the other say it: a [sic:] accuses Migne's type, our Latin is Corpus Corporum's transcription and not the plate, and no column of this work has been read. Every suspicious form was carried as printed with a crux and a plate request instead. Given that the 2026-08-29b sweep found PL 113's [sic:] 93% false and PL 114's 65% false, that is the correct instinct and it should be read as evidence the stated marker policy transfers.

§1 · What was reconciled across the ranges — two sites, both applied

The runbook's merge step exists because two agents will diverge on terminology and neither can see the other. Two divergences were found; both are recorded here with the losing reading named, because a later sweep that cannot see the argument will otherwise churn them back.

1a · Bare pseudo → "the false apostles" (was "the false ones" once, at 0552A)

Migne prints the clipped noun pseudo eight times and the full pseudoapostol- four times. cruces-0005.md §1 rendered every bare pseudo as "the false apostles"; cruces-0000.md rendered its one instance (0552A, ut pseudo faciebant) as "the false ones".

Settled for "the false apostles", and the proof is Migne's own antithesis at 0566A: Hoc non ad apostolos pertinet … sed ad pseudo, qui componebant verba. The clipped noun is carrying "apostles" there — the sentence contrasts the true apostles with the false ones as apostles, and "the false ones" would dissolve the antithesis into a vague slur. 0565C (non sumus similes pseudo, qui usurpant sibi potestatem) and 0564C (qui pseudo sequebantur) read the same way.

⚑ Resolved by expanding the stump, not by contracting its twin — the settled rule from GLOSSA-NT-AGENT-BRIEF.md, applied here to a clipped noun in gloss prose rather than to a lemma. Recording that extension explicitly, because the rule as written speaks of lemmata and someone will otherwise re-derive it. One edit, 0000.md.

1b · consol- → console / consolation throughout (was "comfort" in four places)

Consolatio is the theme word of this epistle and it straddles the split: chapters 1–2 are in range A, chapter 7 in range B, and they are the same argument. A rendered the root console/consolation/consolatory in its gloss prose (nine sites) but "comfort" in one lemma; B rendered all five of its sites "comfort"/"comforted".

Settled for console/consolation, on three grounds:

1. The gloss at 0551C argues on the word. Non solum ut consolaretur, sed etiam exhortaretur: et ita nostra consolatio valet etiam ad alios exhortari. The passage turns on consolatio against exhortatio, and "console / exhort" keeps the pair distinct in English where "comfort / exhort" blurs which of the two the argument is about.
2. One Latin root, one English root, across a work where a reader tracking the theme from chapter 1 to chapter 7 is doing exactly what the gloss asks.
3. ⭐ "Comfort" is the Douay and AV word here, and conforming the lemma to the familiar English Bible is the CONFORMATION defect this brief exists to prevent. Standardizing on consolation moves away from conformation, not toward it. This is the one direction of the two that no later sweep can call a Bible-conforming drift.

Four edits: 0001.md (lemma Donetis et consolemini → "You should forgive and console"), 0005.md (Repletus sum consolatione → "I am filled with consolation"; Qua consolatus est in vobis → "Wherewith he was consoled in you"; and its gloss, "Because I saw Titus consoled in you, I also have been consoled").

§2 · Vocabulary compared across the ranges and found ALREADY in agreement

⚑ Naming the control's class: the comparison below is of Latin roots that occur in both halves — those are the only ones that could have drifted. A term confined to one half cannot diverge and is not evidence of anything either way. Seven roots qualified:

⭐ A rejected candidate, recorded because it looks like drift and is not. "Boasting" appears once, at 0568A in range B, against "glory/glorying" everywhere else. It renders jactantia (et jactantia vitatur), not gloriari — a different Latin word, correctly given a different English one. The two stints agree on gloriari at every site. Do not "fix" this.

§3 · What the two ranges did the same way without being told to

Recorded because agreement reached independently is worth more than agreement imposed, and because a later reader will otherwise wonder whether these were coordinated. They were not.

• VERS. n.-- passed through verbatim, untranslated (convention 1) at every occurrence in both halves.
• Verse numbers out of sequence reproduced as printed, never reordered — A found three (VERS. 2 for 1:3; VERS. 23 printed twice; VER. 1.-- at 0554B), B found the Baruch citation at 0561A standing one quotation early. Both logged, neither touched.
• Run-together non-words rendered for sense with a crux and a plate request, never a [sic:] — A's quamcredere (0554C) and duasepist. (in a note), B's abjicicimus, cum-tua and ibenter (0563–0564). Both stints cited the same measured ground: that class runs 10-for-10 as our corruption, and a [sic:] would misattribute it to Migne.
• Migne's stray and missing punctuation preserved — A carried two sentences ending with no stop and one missing mid-sentence stop; B preserved the stray full stop in in prima. Epistola at 0562A.
• Band-by-band punctuation comparison run by both, and it paid both times: A found 10 mismatches that had very nearly cancelled on totals; B found 8 (one deleted colon at 0569B restored, seven supplied marks removed). Zero mismatched bands in all 10 chunks at filing. ⭐ That is 18 punctuation defects in one short work that a totals comparison would have passed. The band test is not optional and this work is the second measurement saying so.

§4 · Open against this work

1. ✅ The plate gate — CLOSED 2026-08-30 (§5). 7 stood, 0 false.
2. ✅ The polarity read — DONE 2026-08-30 (§6). 10 sites, 9 ours. Recorded in data/polarity/8986.json.
3. ✅ Both held candidates settled at the plate (§5), and both against the stint's own guess: Transformantur was neither Migne's nor the Vulgate's, and Et Deus pacis erit vobiscum was Migne's own short lemma, not the TEI loss it was taken for.
4. ⚠ Our PL 114 scan clips the right margin on some pages in this stretch (p.265, p.281; p.327 clean). Any right-column crux above needs the second witness in data/volumes.json — never "illegible" from this PDF alone.

§5 · THE PLATE GATE — six pages, 2026-08-30

Read directly by the session rather than by fenced agents, deliberately: briefing several readers on the same suspicion produces convergence, not corroboration, and that failure is recorded three times in this project already. page = (column + 11) / 2, re-derived here by fitting the six PL 114 rows of the 2026-08-29 sweep ledger before it was compared with 9006's identical map. Printed corner numbers read FIRST on every page.

⭐ 7 STAND · 0 REVISED · 0 FALSE · 0 UNRESOLVED — and an eighth marker fired

⛔ Against a 65% false rate for PL 114 in the sweep three days earlier. The variable is the marker policy, not the volume: this work fires no letter-level accusation against Migne's type at all, and both stints reached that policy independently, without either seeing the other.

⭐⭐ NINE DEFECTS IN OUR LATIN THAT NO CHECK WE OWN COULD HAVE FOUND

Every one read correctly in our file, which is precisely why nothing caught them. All restored in data/tei-patches/8986.json; the five that are Migne's own type each now carry a [sic:].

Corpus Corporum silently tidied Migne (5): nou→non (0555B) · Vebum→Verbum (0561A) · ete.→etc. (0562B) · Ostendlt→Ostendit (0567B) · VERS. 51.→VERS. 21. (0559C) — the last is the 9006 @0043A class exactly: CC corrected the verse address, correctly on the merits, and being right is what made it invisible.

Corpus Corporum broke what Migne set correctly (4): quamcredere · duasepist. · Transformantur · ego for ergo.

⭐⭐⭐ 0570B — ONE DROPPED LETTER MANUFACTURED THE WORK'S HARDEST CRUX

cruces-0005.md §5.2 calls Istam ego imitetur "the worst site in my range": a nominative ego governing a third-person imitetur. The stint reasoned it through, declined a [cj:], named imiter and imitemur, took neither, and rendered the sentence visibly broken rather than repair what would not construe.

The plate prints Istam ergo imitetur. Verified at 5×. Corpus Corporum dropped one r and invented the whole problem. ⚑ The discipline of rendering literally and logging the crux instead of emending is what kept a wrong conjecture out of the text: the crux had no answer because it had no question.

The two held candidates, both settled against the stint's own guess

• 0555D Transformantur — the plate prints Transformatur, third person singular, across the break Trans-/formatur. Three readings existed and ours was nobody's: Migne Transformatur, Clementine transformamur, our file Transformantur. The stint declined the [var:] and made it its top plate request, reasoning that -mur→-ntur is a shape CC manufactures — right to decline, wrong about the true reading. The English had followed the corruption ("They are transformed"); re-rendered singular, its subject taken from the preceding lemma Vir.
• 0570C Et Deus pacis erit vobiscum — flagged by the stint as a probable TEI loss wanting a patch, since the Clementine has et dilectionis and the gloss expounds dilectio. The plate shows the short lemma is Migne's own. Not a loss but a divergence; [var:] fired, and the gloss reaching for dilectio anyway is the interesting part.

Negatives recorded, because a checked negative is worth having

• 0555C Ubicunque autem est spiritus Domini — the plate DOES print est; the stint's suspicion of a supplied word is refuted.
• 0562A non tantum — present on the plate; nothing was lost there.
• 0568 non est peccatum — the s is damaged type in this copy; the reading is not in doubt and no marker depends on it.

⚠ p.290 clips the LEFT margin and its left column has lost its opening characters. No marker depends on that column. Any future crux at 0569 needs the second witness (raw/scans/pl202/copyB_patrologiaecurs114unkngoog.pdf, a PL 114 copy B filed under pl202).

§6 · THE BLIND POLARITY READ — three readers, 2026-08-30

Three Opus readers over 10 chunks, each fenced from the cruces, the TEI patches and the plate reads, and from each other. 11 sites raised, 10 acted on, 1 declined. sites: 10 · ours: 9.

⛔ Deliberately fenced from this file: on the 2026-07-28 sweep every agent handed the cruces returned "all already logged" and found nothing, while blind readers on the same works found four live defects. One reader here disclosed that a grep glob had leaked six lines of the cruces into its context, and withdrew the affected item from its own findings so it could not inflate recall. ⚑ Scope the glob to src/english/<idno>/0.md; .md matches the apparatus.

⭐ What the fourth test found that nothing else could

0557B — a reader predicted a missing word from the English alone, and the plate had it. Our Latin read Quia in ea sine manebitur; the English, "there shall be abiding without", has no complement. The reader reasoned that sine cannot govern nothing, proposed sine fine, and asked for one page. The plate prints sine fine — Corpus Corporum had dropped the word outright, the only clean text LOSS in this work. No parity, count, band or marker check can see a class like this: nothing was misaligned, and the English was faithful to a Latin that was wrong.

The ten, by direction

OURS (8 in the English): Vel passive ergo read as though ergo were the word being glossed, when ergo is the inferential particle and passive governs operatur — which Migne's own gloss then equates with the passive efficitur (0557A) · per hoc mis-bracketed so that "through this life" made vita the instrument rather than the subject its own gloss names (0557A) · two Pauline sentences attributed to Christ by a bare "he" standing next to Christus, where alio loco is this commentary's formula for another place in the Apostle (0556D) · the negative left on ab Evangelio instead of on solum, turning "not on those grounds alone" into a flat denial that the Gospel commends them at all — the sed etiam is the structural proof (0563C) · ut dicatur read as purpose, making the Apostle act in order that he be called negligent, when the lemma is Devitantes hoc, "avoiding this" (0563C) · the subject of appareant given to the Corinthians when it is the apostles (0570A), which inverted who is called reprobate at the hinge of the paragraph — settled by probati enim videntur, dum judicant peccatores and by the lemma's own first-person appareamus reprobi · Istam rendered "this one" where it is pax (0570B) · patitur, printed indicative, quietly made subjunctive to harmonize with habeat (0567A).

OURS (1 in the Latin): the dropped fine, above.

MIGNE'S (1): Ut inopia illi (0563A) — a reader took it for a clipped illius and proposed repairing the lemma. The plate prints illi. So the English stays as printed and the divergence is now shown with a [var:] against Clementine 2 Cor 8:9 illius — which Migne's own next clause reads. Repairing it would have smoothed away the thing worth seeing.

The one declined, and why

0557C quo tamen nolumus mortem nobis tolli — a reader showed the sentence asserts the reverse of its own paragraph (2 Cor 5:4 is nolumus expoliari sed supervestiri throughout) and proposed reading morte for mortem. It rated itself "high on the defect, medium on the repair, needs the plate", which is the right way to raise it. The plate prints mortem. The difficulty is Migne's, our English renders it literally, and the proposed repair is declined because it would emend against the page. Logged, unmarked: a real, well-formed word that reads oddly is Migne's.

One further defect of Migne's, caught on the same pages

0563A abjicicimus — the plate prints abjici-/cimus across the line break, a doubled syllable for abjicimus. A reader saw the malformed word in our file and declined to call it Migne's without a plate, on the ground that our Latin is Corpus Corporum's transcription first. That was correct, and the plate then made it Migne's. Now carries a [sic:].

Rejected candidates worth keeping

Between them the readers rejected some two dozen, each with the loser named. The ones a later sweep is most likely to re-raise: Peregrinamur non fide, sed specie reads backwards against 2 Cor 5:7 until the preceding sentence fixes it as a claim about sight · sanctus mundi est, non Dei is Migne's own paradox, not a defect · super invocationem animae is odd and is what the plate prints · minor verbis is the deponent "I threaten", not the comparative · ministri justitiae aeternae takes aeternae with justitiae, since it cannot be predicate to masculine ministri · and VERS. 6's Hoc non ad apostolos pertinet, quia non erant eloquentes, which reads backwards at first sight and is right, because the comparison Paul disclaims is with the pseudo precisely on the ground that the apostles were not eloquent either.

Range: PL 114, cols 0551A–0559C. Anselmus Laudunensis et schola (Migne: Walafrid Strabo). Chunks 0005–0009 are covered by a second agent in its own file; nothing here touches that range.

Conventions applied without re-derivation: GLOSSA-NT-AGENT-BRIEF.md §"The Glossa conventions" (1 VERS. n.-- verbatim · 2 lemma Englished · 3 fragment rendered as the phrase it stands for · 4 sigla pass through as [n:]), translation-style.md 7, 7a, 7a″, 7a⁗, 8, 8b, 9, 10, 12, 14, 15, 17, 18. Format model: src/english/8968/ (Ruth); register anchor src/english/9741/0000.md.

0. Markers fired — 3, all [var:]

No [sic:], [d:], [ed:], [cj:], [nt:] fired. Reasoning for the two candidates that would have taken one is in §3.

1. Lemma collation against sources/vulgate/clementine-flat.txt (7a″)

Every lemma named in the launch message was grepped against the Clementine. 144 lemmata checked; 125 agree exactly with the Clementine on the words present; 19 diverge. The agreements are listed because a list of findings alone cannot be told apart from a report by an agent that never looked.

0000 (2 Cor 1) — 31 lemmata, 26 exact

Exact: Paulus apostolus. (1:1) · Paulum · apostolum · Paulus, (gloss-internal re-quotations of the same lemma) · Benedictus. · Deus. · Pater Domini. · Pater misericordiarum. · Et Deus. · totius … consolationis: (split by the glossator's perfectae; both fragments Clementine) · Qua exhortamur. (1:4) · Ipsi. (1:9) · Et eruit. (1:10) · Nam gloria nostra. (1:12) · Testimonium. ×2 (1:12) · Sinceritate Dei. (1:12) · Spero autem. (1:13) · Hac confidentia. (1:15) · Prius, (1:15, pickup) · Ut sit apud me, est et non. (1:17; Migne lowercases the Clementine's Est et Non and closes with a stop, not the Clementine's ? — orthography and Migne's own pointing, not a textual divergence) · Fidelis autem Deus, quia sermo noster, qui fuit apud vos non est, etc. (1:18) · Ad gloriam nostram. (1:20) · Vobiscum in Christo. (1:21) · Ego autem, etc. (1:23) · Non quia dominamur. (1:24) · parcens vobis non veni, subdit: Non quia dominamur, (1:23; the italic span runs over the glossator's own subdit: — preserved as one span in the English).

Divergent (all declined):

1. 0551D · in nobis, — Clem. 1:9 in nobismetipsis. One word, intensive for simple; assertion unchanged. Declined.
2. 0552A · Non in carnali sapientia. — Clem. 1:12 et non in sapientia carnali. Word order only; English has no exponent for it. Declined.
3. 0552B · Quia gloria vestra sumus. — Clem. 1:14 quod gloria vestra sumus. Quia for quod; synonymous in medieval Latin. Declined.
4. 0552B · ut secundam gratiam, — Clem. 1:15 ut secundum gratiam haberetis. ⚠ This is a real divergence of construction (adjective secundam agreeing with gratiam, "a second grace," vs the preposition secundum). Declined, and this is the reason: it is a one-letter difference in a verb-adjacent inflection, which is exactly the shape CLAUDE.md 8a records the digitization manufacturing (judicabit→judicavit), and our Latin is not the plate. Rendered as printed ("that a second grace"). Plate candidate.

0001 (2 Cor 2) — 19 lemmata, 15 exact

Exact: Statui autem. (2:1) · ne iterum venirem: (2:1) · Si enim ego contristo. (2:2) · Non ut contristemini. (2:4) · Si quis autem contristavit. (2:5) · Sed ex parte, ut non onerem, etc. (2:5) · Objurgatio haec, quae fit a pluribus ita ut e contrario magis. (2:6–7, spanning the verse break with no mark, as Migne prints) · Donetis et consolemini. (2:7) · Ideo enim et scripsi. (2:9) · Ut non circumveniamur. (2:11) · Cum venissem autem Troadem, etc. (2:12) · Eo quod non. (2:13) · Odorem notitiae, etc. (2:14) · Bonus odor. (2:15) · Odor mortis. (2:16) · Sicut plurimi. (2:17).

Divergent:

1. 0553A · sed statui hoc idem: — Clem. 2:1 hoc ipsum. idem for ipsum, near-synonyms; the opening sed is the glossator's own connective inside the italics. Declined.
2. 0554A · Nomen Domini per vos blasphematur inter gentes. — Clem. Rom 2:24 Nomen enim Dei per vos blasphematur inter gentes. Domini for Dei, enim dropped. A divine-name substitution, but the assertion is unchanged and the gloss does not turn on it. Declined — noted here because a divine name is the class 7a″ warns about, and it was checked, not assumed.
3. 0554A · Caupones tui vino aquam miscent. — [var:] fired, see §0.

0002 (2 Cor 3) — 30 lemmata, 25 exact

Exact: Incipimus iterum. (3:1) · Scripta non. (3:3) · Non in tabulis. (3:3) · Carnalibus. (3:3) · Fiduciam. (3:4) · Non quod sufficientes. (3:5) · Sed sufficientia. (3:5) · Qui et idoneos. (3:6) · Multo magis abundat. (3:9) · Nam nec. (3:10) · Quod claruit. (3:10) · hac / parte, (3:10, split round the glossator's praedicta) · Habentes. (3:12) · Multa fiducia. (3:12) · Idipsum. (3:14) · Sed usque, etc. (3:15) · cum, · legitur, · Moyses. (3:15) · Cum autem conversus fuerit. (3:16) · Auferetur velamen. (3:16) · Dominus autem spiritus est. (3:17) · ibi libertas: (3:17) · Nos revelata facie. (3:18) · In eamdem imaginem. (3:18).

Divergent:

1. 0555B · Quod in Christo evacuatur. — Clem. 3:14 (quoniam in Christo evacuatur). Quod for quoniam; probably attracted from the quod evacuatur of 3:13 standing just above. Conjunction swap, assertion unchanged. Declined.
2. 0555B · sed super cor eorum est velamen, — Clem. 3:15 velamen positum est super cor eorum. Word order rearranged and positum dropped. Multi-word, but the assertion is identical ("a veil is on their heart"). Declined under the stated bar (multi-word and changes the assertion — the second condition fails).
3. 0555C · Ubicunque autem est spiritus Domini, — Clem. 3:17 ubi autem Spiritus Domini (no copula). Two divergences: Ubicunque for ubi, and a supplied est. ⚠ This was my strongest declined candidate and I want the reason on record. Ubicunque is a whole distinct lexeme with a quantifying force and is not OCR-producible from ubi, so it clears the "whole distinct word" half of the bar cleanly. It fails the other half: in a gnomic statement ubi is already generalizing, and the gloss below (et ideo Judaei qui non habent spiritum, non possunt libere intelligere ut nos) does not rest on universality. Declined to keep the marker set to claims I can stand behind. The supplied est is separately the named "digitization silently normalizes" class (brief §4: supplying an est Migne prints bare). Plate candidate for the copula.
4. 0555D · Vir quidem imago Dei est, et gloria Dei. — Clem. 1 Cor 11:7 Vir quidem non debet velare caput suum : quoniam imago et gloria Dei est. Migne drops the first clause and splits the single predicate into two (imago Dei est, et gloria Dei). The gloss needs exactly the two predications the Clementine also makes. Assertion unchanged. Declined.
5. 0555D · Transformantur. — Clem. 3:18 transformamur (1st pl.). ⚠ The most interesting site in the range and it is declined on principle, not on merit. The divergence is material — third person against first — and the gloss's own next words are first person (Transimus de forma in formam), which means the printed lemma and the printed gloss disagree. But -mur → -ntur is a verb-ending change, the precise shape CLAUDE.md 8a names as the class where the digitization manufactures a Vulgate divergence that a collation then dutifully "finds." A [var:] here would be a public claim about Migne's page resting on a two-letter difference in a file that is not his page. Rendered as printed ("They are transformed"), so the disagreement with Transimus is visible to a reader. Plate candidate, highest priority in this range.

0003 (2 Cor 4) — 17 lemmata, 12 exact

Exact: Commendantes. (4:2) · Coram Deo. (4:2) · In quibus. (4:4) · Deus hujus saeculi. (4:4) · Non enim. (4:5) · Lucem splendescere. (4:6) · Scientiae claritatis. (4:6) · Habemus. (4:7) · Thesaurum. (4:7) · Semper mortificationem, etc. (4:10) · Manifestetur. (4:10) · Ergo mors, etc. (4:12) · Vobiscum. (4:14).

Divergent:

1. 0556A · "Ut luceat lux vestra coram hominibus" — Clem. Mt 5:16 Sic luceat. Not italicized by Migne, so not a lemma; one-word connective. Declined.
2. 0556A · "Non ut videantur ab hominibus" — a negated construction built by the glossator out of Mt 23:5 / 6:1, not a quotation of either. Not italicized. Declined.
3. *0556B · fuistis aliquando tenebrae, nunc autem lux in Domino [n: (Ephes. V)] — Clem. Eph 5:8 Eratis enim* aliquando tenebrae. fuistis (perfect) for eratis (imperfect), enim dropped. English has no exponent that changes the assertion. Declined.
4. 0556B · In faciem Jesu. — Clem. 4:6 in facie Christi Jesu. Accusative for ablative (one letter) and Christi dropped. Neither changes the assertion. Declined; the faciem/facie half is the manufactured-inflection class again.
5. 0556D · Cum infirmor, tunc potens sum, — Clem. 2 Cor 12:10 cum enim infirmor. Connective dropped. Declined.
6. 0556D · Per tribulationes, etc. — Clem. Act 14:21 per multas tribulationes. One word dropped inside a clip that already ends in etc. Declined.

0004 (2 Cor 5) — 47 lemmata, 40 exact

Exact: Scimus enim quoniam si terrestris domus nostra hujus habitationis dissolvatur quod aedificationem, etc. (5:1) · Aeternam. (5:1) · Nam. (5:2) · in hoc, (5:2) · ingemiscimus. (5:2) · De coelo. (5:2) · Nam et qui, etc. (5:4) · Si tamen vestiti / non nudi. (5:3) · Gravati eo (5:4) · ingemiscimus (5:4) · Eo. (5:4) · Ut absorbeatur, etc. (5:4) · Audentes. (5:6) · Peregrinamur a Domino. (5:6) · Ut referat unusquisque. (5:10) · Suademus. (5:11) · Deo autem manifesti sumus, (5:11) · Sive enim mente. (5:13) · Sive excedimus, (5:13) · Sive sobrii: (5:13) · Sive enim mente excedimus. (5:13) · Sive enim sobrii sumus. (5:13) · Vobis. (5:13) · Charitas enim Christi, (5:14; ch- is orthography) · Si unus pro omnibus. (5:14) · Qui vivunt (5:15) · Neminem novimus. (5:16) · Secundum carnem. (5:16) · Et si cognovimus secundum carnem. (5:16) · Vetera. ×2 (5:17) · Nova. (5:17) · Deus erat. (5:19) · Reconciliamini. (5:20) · Pro nobis peccatum fecit. (5:21) · "Absorpta est enim mors in victoria" (1 Cor 15:54, unitalicized, enim is the glossator's) · "Videbitis coelum apertum…" (Jo 1:51, unitalicized; drops Dei, super for supra) · "irriguum superius et irriguum inferius" (Jdc 1:15, unitalicized, verbatim — note it matches Judges, not the Josue 15:19 doublet, which reads superius et inferius).

Divergent:

1. 0557B · Non manufactam, sed aeternam. — Clem. 5:1 domum non manufactam, aeternam in caelis (no sed). One connective added. Declined.
2. 0557D · Qui sumus in tabernaculo. — Clem. 5:4 qui sumus in hoc tabernaculo. One demonstrative dropped from a phrase Migne has already quoted once as Nam et qui, etc. Declined.
3. 0557D · Quod nolumus exspoliari, — Clem. 5:4 expoliari. Attested orthographic doublet; Pattern 9 exempts it. Declined.
4. 0557D · Deus autem hoc effecit. / Deus tamen effecit — Clem. 5:5 Qui autem efficit nos in hoc ipsum, Deus. Multi-word recast: God fronted, present → perfect, nos in hoc ipsum → hoc. Declined because it is paraphrastic compression that leaves the agent and the act identical (God is the one who brings this about, in both), and the second form (tamen for autem) is plainly the glossator re-quoting himself to head a second reading, not a second citation.
5. 0558A · In ipso vivimus, movemur et sumus? [n: (Act. XVII.)] — Clem. Act 17:28 In ipso enim vivimus, et movemur, et sumus. Two connectives dropped. Declined. (The ? is Migne's, closing the question Quomodo? two clauses earlier; preserved.)
6. 0558A · prout gessit in corpore, — Clem. 5:10 prout gessit (the "of the body" sits earlier, as propria corporis). Migne moves it into the clause. The gloss does lean on it (non ut gesturus erat si viveret), but the Clementine already predicates the body of the deeds, so the assertion is unchanged. Declined.
7. 0558C · Deo est: — Clem. 5:13 mente excedimus Deo (no copula). A supplied est again — the named normalization class. Declined; plate candidate.
8. 0558D · Cum exaltaveritis Filium hominis… [n: (Joan. VIII)] — Clem. Jo 8:28 exaltaveris (2nd sg.). ⚠ Worth naming: here it is very likely the Clementine that is the odd witness — its own following verb cognoscetis is plural, and the received Vulgate text elsewhere reads exaltaveritis. sources/vulgate/README.md warns that later upstream corrections are not applied and that a call turning on a couple of letters should say so rather than trust this copy. Declined, and it would have been the wrong marker to fire.
9. 0559A · Neminem non secundum carnem. — no Clementine counterpart at all. 2 Cor 5:16 reads neminem novimus secundum carnem … sed nunc jam non novimus. This italic run is the glossator's own reconstitution of the verse's sense, not a quotation; declined as not a scripture-citation divergence. Its parsing is at §2.4.
10. 0559C · De peccato delevit vel damnavit peccatum, — Clem. Rom 8:3 de peccato damnavit peccatum in carne. The delevit vel is the glossator offering an alternative verb inside the italics (his own vel, not a variant reading), and in carne is clipped. Declined.
11. 0559C · quia misit Deus Filium suum in similitudinem carnis peccati. — Clem. Rom 8:3 Deus Filium suum mittens in similitudinem carnis peccati. Finite verb for participle, quia supplied to attach it. Assertion unchanged. Declined.
12. 0559C · Maledictus omnis… and 0559C · terra es… — [var:] fired, see §0.

2. Cruces — passages read, and what was decided

### 2.1 Verse numbering (reproduced as printed, never reordered)

• 0551B, VERS. 2.-- Benedictus. — Benedictus Deus et Pater Domini is 2 Cor 1:3, not 1:2. Migne's address is one short. Reproduced verbatim.
• 0552D–0553A, VERS. 23 printed twice. The second (Non quia dominamur) is 1:24. Migne's ordering; reproduced, not renumbered.
• 0554B, VER. 1.-- (not VERS.) opens CAPUT III. Convention 1 makes the address verbatim, so the English prints VER. 1.-- too. Not normalized.

### 2.2 Defective and doubtful type

• 0554C · quamcredere in Quis non videat prius esse cogitare quamcredere? — a run-together non-word (quam credere), Pattern 10's class. Rendered for its evident sense ("that thinking is before believing") and NO [sic:] fired. The reason, stated so it is auditable: Addendum A measures this class 10-for-10 as our corruption, and a [sic:] accuses Migne's type. CLAUDE.md 8a records live works shipping false [sic:] against Migne on exactly this reasoning. Plate candidate: one look converts this crux into a marker or closes it. The Latin twin carries the broken run unmarked and visible, so nothing is concealed.
• 0554C · contra duasepist. inside [n: (AUG., lib. II, contra duasepist. Pelag., c. 8.)] — a second run-together (duas epist.), inside a note. Notes pass verbatim and take no marker of ours. Logged. Plate candidate.
• 0557B · Quia in ea sine manebitur. — ⛔ sine is a preposition with no object. Almost certainly sine fine ("because in it there will be abiding without end"), which is what the lemma Aeternam. is being glossed to mean. The conjecture is NOT in the text: the English reads "Because in it there shall be abiding without," and the hole is visible to the reader. No [cj:] — Pattern 18 requires Migne's printed word to be a real word whose faithful English asserts something false, and here nothing is asserted at all; the defect is a missing word, which is [ed:]'s territory only if source loss can be demonstrated, and it cannot be from our files. Plate candidate, second priority.
• 0557C · cupimus super invocationem animae, quam jam accepimus. — invocationem is a real, well-formed word that reads oddly; innovationem ("the renewal of the soul, which we have already received") construes the sentence perfectly and matches the passage's whole argument (the body's glory still awaited, the soul's already had). Rendered as printed, per Addendum A: a real word that reads oddly is Migne's until a plate says otherwise, and a corpus-frequency argument, however good, cannot settle whose the reading is. No marker. Plate candidate.
• 0557D · quo tamen nolumus mortem nobis tolli, sed superindui. — mortem (acc.) as subject of the passive tolli gives "we do not wish death to be taken from us," which is not what the passage wants; morte (abl., "to be taken from us by death") construes with the whole figure of being stripped rather than over-clothed, and with non ut superveniat four words later. Rendered as printed, one-letter inflection, no marker (§1.4's reasoning). Logged.
• 0558C · Isti sunt angeli de quibus Dominus ait in Evangelio… — ⚠ a non sequitur: no angels have been mentioned anywhere in this chapter, and the sentence introduces them with a back-reference (Isti) that has no antecedent, then runs on into the Jo 1:51 and Jacob's-ladder material and stops. This looks like an intrusion into the recension, not a defect of type. Rendered exactly as printed; nothing supplied, nothing marked. Flagged here because a blind reader will stop at it and should know it was seen.
• 0551D · Deus praesidia sua non negat suis in necessitate positis and 0558B · …dies Domini in dubium venit — both end with no terminal stop. Pattern 8: never supply. The English sentences stop without a stop too.
• 0552B · …debetis orare pro nobis Vel ideo orationes vestrae… — Migne prints no stop between nobis and Vel. Carried through as printed; the English reads "…ought to pray for us Or therefore your prayers…".

• 0556B · post accedente luce quod factum est; melius redditur. — Migne's semicolon falls inside the construction. Two parses: (a) quod factum est is parenthetical ("afterwards, the light being added, which was done; it is rendered better"), which respects the printed mark; or (b) quod factum est melius redditur ("what was made is rendered better"), which requires the semicolon to be misplaced. (a) was taken, because Pattern 8 makes the mark evidence and (b) can only be reached by overriding it. The English is rough at that point; so is the Latin.
• 0557A · per vos multos abundet gratiarum actione — Clem. 4:15 reads per multos in gratiarum actione; Migne inserts vos, leaving vos multos ("you many"). Rendered "through you many it may abound in thanksgiving." Suspiciously the shape of an intruded word; logged, no marker. Plate candidate.

### 2.3 Punctuation (Pattern 8 / band test) Bands compared 1:1 across all five twins for : ; ? « » after stripping English-only markers: zero mismatched bands in all five chunks. Nine supplied semicolons and one softened one were found by the band test and repaired before filing — they had cancelled to near-parity on totals (0000: 6/6, 0002: 5/5), which is the failure the band test exists to catch. Question marks: 0002 has one (quamcredere?), 0004 four (Quomodo? · …et sumus? · maledictus? · omnis?); all four are direct questions, so 8b's colon recast was not needed anywhere in this range and was not used.

### 2.4 Negation — the sites where the host was proved from structure, not from what reads well Every printed negative was mapped to its host word and the host checked in the English. The test-2a population (a negative on a verb with a quantifier / correlative / pivot adjacent) is enumerated here in full, with the structural proof:

Non-quantifier negatives (verb-hosted, no correlative) were checked one by one and are correct: non ait · nullus praesumat · non est judicandus · nec est liber · nesciens · non esse verum · in nullo alio · non implet · nil nisi · nunquam-class absent · non condolet · non venit · non profecit · non poterat · non egere · non est ex nobis, nec credere · non datur · non celamus · ut non intelligeretur · non intellecto Christo · non habent spiritum, non possunt · non immisit · nollent · non praebuit · non est inutilis · non faceremus, nisi · ne illi de se minus sentiant · ne … nudi inveniantur · non est curandum, cum non sit tabernaculum nisi ad tempus · nusquam + Non infra, non supra, non intra, non extra · non tamen hic videtur · si non per se · vel non crediderunt · non secundum quod gesturi erant · ut non alta de nobis dicamus · non vobis, qui non potestis capere · non est, sed apparet · nec hoc debet facere aliquis · non quin … sed quod non · non jam … nunc ita esse credit · nulla diversitas · non ficta. Count: every negative printed in the Latin is present in the English, and none is present that the Latin does not print.

### 2.5 Unexpressed subjects — supplied pronouns, and what decided each

• 0551B Per quem est et nobis. — the elided predicate is Pater, from the lemma directly above (Pater Domini); quem is Christ. "Through whom he is the Father of us also."
• 0551C non solum ut consolaretur, sed etiam exhortaretur — both verbs are deponent actives ("console", "exhort"), not passives, and the following clause (nostra consolatio valet etiam ad alios exhortari) is active in the same way. Subject: the one who has received the exhortatio. Rendered "that he might console… but also exhort". The passive reading ("be consoled… be exhorted") also construes; it lost because the deponent morphology and the parallel active infinitive both point the other way.
• 0555B non aliter modo intelligunt — subject is the Jews, from 3:14's obtusi sunt sensus eorum, not from anything in the sentence.
• 0555C–0555D, the Moses/veil stretch: Moyses, Dominus, spiritus and the filii Israel all compete. Resolved as: cum … legitur … Moyses → Moses is what is read (the gloss supplies exponitur, "expounded to them"); Cum autem conversus fuerit → the subject is eorum animus, which the gloss itself prints; ad quem convertuntur → the Jews, same antecedent. No pronoun in this stretch is supplied without a printed antecedent.
• 0555A illa nec dicenda est gloria — illa is the old glory / the ministration of death, against ejus (the new) in the same clause. Rendered "that other," not "she" or "it."
• 0559A Vitam futuram ita certa spe tenet — third singular with no printed subject. Taken as the man born again new in Christ, because the nearest printed third-singular subject is si quis novus … et novam habet in the immediately preceding sentence. The rival is the Apostle (whose Neminem novimus is the lemma); it lost on nearness and on the fact that the intervening sentence changed the subject to quis. Both construe — recorded rather than chosen silently.
• 0559B dum infidelis … intellexit — Paul, self-evidently (dum infidelis, and 5:16 is his own before-and-after). Named, not assumed.
• 0556A quod non facit malitia, sed justitia — subjects printed; no supply needed.

### 2.6 Fragmentary lemmata — what class each belongs to before it was expanded (⚑) The controls here are of two different classes and were kept apart:

• Complete one-word lemmata (nothing elided): Testimonium. · Habemus. · Thesaurum. · Fiduciam. · Carnalibus. · Idipsum. · Habentes. · Aeternam. · Audentes. · Suademus. · Vobis. · Vobiscum. · Nova. · Vetera. ×2 · Reconciliamini. · Manifestetur. · Commendantes. · Moyses. — rendered as they stand.
• Clipped lemmata whose governed words are elided — convention 3, rendered as the phrase they stand for, and only where the continuation is not itself picked up elsewhere in the gloss:
• Nam nec. → "For neither was it glorified" (3:10; the continuation glorificatum est is picked up nowhere, and the gloss turns on the nec).
• Eo quod non. → "Because I found not Titus my brother" (2:13; the Ruth Non te. precedent).
• Non enim. → "For we preach not ourselves" (4:5; the gloss glosses the whole clause).
• Deus erat. → "God was in Christ" (5:19; the gloss's whole argument is Pater esse in Filio, which needs in Christo).
• Word-pickups inside the glossator's own sentence, where the continuation is supplied by the gloss's further italic pickups, so expansion would duplicate: Prius, · hac / parte, · cum, / legitur, · Nam. (continued by in hoc, and ingemiscimus.) · Eo. (continued by the next lemma Quod nolumus exspoliari) · Sive excedimus, · totius / consolationis:. Rendered as the words they are. This is a third class, not an inconsistency with either of the two above; naming it is the point of the ⚑ rule.

### 2.7 usque ad and its look-alikes (Addendum D) Three occurrences in the range, none of them the abridgment formula, all decided from what stands on either side:

• 0555B et hoc usque in hodiernum tempus — the verse's own words (3:15).
• 0555B Sed usque, etc. — the lemma for 3:15 sed usque in hodiernum diem; its own words.
• 0558D etsi Christus usque ad mortem fuit infirmus — ordinary idiom, "even unto death." No resumption tail in this range, and no verb supplied anywhere to complete one.

### 2.8 Vocabulary settled from the corpus

• 0552A · bare pseudo (ut pseudo faciebant) — the Glossa's substantival use. The corpus already renders it "the false ones" (8990's cruces, propter pseudo qui vobis se ingerunt); taken as a hint, checked against this work's own pseudoapostolos two paragraphs above, and used.
• 0558B · praviloquio — rare but regularly formed; rendered "perverse speech." Real word, no marker (Pattern 9).

3. Read clean / candidates rejected

Read clean and nothing found: the whole ARGUMENTUM (0551A); 0551A–0551D (CAPUT I opening, Benedictus through Et eruit); 0553B–0553C (CAPUT II, Si enim ego contristo through Donetis); 0554B–0554D (CAPUT III opening); 0555A (Quod claruit); 0556B–0556D (CAPUT IV); 0558A–0558C (Audentes through Sive enim sobrii sumus); 0559A–0559B (Neminem novimus through Et si cognovimus).

Rejected with reasons (each was raised against itself and lost):

1. A [sic:] on quamcredere — rejected: the class is measured 10-for-10 as the digitization's, and the marker accuses Migne. Crux + plate request instead. §2.2.
2. A [var:] on Transformantur — rejected: verb-ending change, the manufactured-divergence shape. §1.12. This is the one I would most like a plate read on.
3. A [var:] on Ubicunque autem est spiritus Domini — rejected on the second half of the bar (the assertion does not change); the lexical half was clean. §1.10.
4. A [var:] on secundam gratiam — rejected: one-letter inflection. §1.4.
5. A [var:] on Cum exaltaveritis — rejected, and it is probably the Clementine that diverges here, not Migne. §1.26.
6. A [cj: sine; read sine fine] at 0557B — rejected: Pattern 18 needs a printed real word whose faithful English asserts something false; a stranded preposition asserts nothing. §2.2.
7. A [cj: invocationem; read innovationem] at 0557C — rejected under Addendum A and 18a (the census is closed and the bar is assertion-falsity, which this does not reach). §2.2.
8. A [cj: mortem; read morte] at 0557D — rejected, same class. §2.2.
9. Reordering VERS. 23 / VERS. 2 — rejected outright: Migne's numbering, reproduced. §2.1.
10. Normalizing VER. 1.-- to VERS. 1.-- — rejected: convention 1 makes the address verbatim.
11. Column-band arithmetic (0552D→0553A, 0554A→0554B, etc.) — not raised; runbook false positive 1. No text is broken mid-word or mid-construction across any anchor in this range.
12. Pattern 4 [f:] tags — not applied; runbook false positive 4, this is a lemma-and-gloss commentary.

## 4. Plate-read requests, in priority order

1. 0555D · Transformantur (against Clem. transformamur) — decides a [var:].
2. 0557B · sine in Quia in ea sine manebitur — decides whether fine is Migne's loss or ours.
3. 0554C · quamcredere — decides a [sic:].
4. 0557C · invocationem (v. innovationem).
5. 0555C · the copula in Ubicunque autem est spiritus Domini and 0558C · Deo est* — both the "digitization supplies an est* Migne prints bare" class; one look at each settles both.
6. 0552B · secundam (v. secundum) and 0557D · mortem (v. morte). ⚠ Cols 0551–0559 fall in the stretch where our PL 114 scan clips the right margin on some pages (p.265, p.281); a truncated right column is not evidence of anything, and a second witness from data/volumes.json is required for any right-column reading.

Columns 0559D–0570C (CAPUT VI–XIII). Chunks 0000–0004 were translated concurrently by a peer agent and are not covered here; that half's cruces are in cruces-0000.md.

0. Marker policy actually applied in this range

• Fired: 4 [var: …], all substitutions or additions inside a quoted scripture.
• Fired: 0 [sic:], 0 [d:], 0 [ed:], 0 [cj:], 0 [nt:].
• ⛔ No [sic:] was fired anywhere in this range, deliberately. data/plate-reads.json records no read for any column of this work, and CLAUDE.md hard rule 8a forbids a [sic:] or [var:] on an unread column: a [sic:] accuses Migne's type, and our Latin is Corpus Corporum's TEI, which departs from Migne at roughly 3 sites per page. The three non-word / fused-run candidates below (§4) are therefore rendered for their evident sense and logged here as plate candidates, not marked. If the plate is read and confirms them, each is a one-line [sic:] addition.
• The [var:] bar I used, stated so it can be checked: fire only where the divergence from sources/vulgate/clementine-flat.txt is (a) a substitution or an addition of a whole distinct word, and (b) something the sentence or the gloss actually turns on. Declined for every OMISSION, however multi-word, because a dropped word is precisely what a digitization loses and what a [var:] would then blame on Migne. Declined for word order, for orthography (charitate/caritate, coelum/caelum), and for lemma truncation, which is Migne's citing practice, not a divergence.
• ⚠ All four [var:] stand on unread columns (0561A/B, 0562A, 0567D, 0568A) and plate-gate.mjs will name them before this work can be marked ours. They rest on the Clementine collation, not on the plate; each is stated below with the exact printed reading so a plate reader can settle it in one look.

1. Work-vocabulary conventions (⚠ MERGE NOTE — the peer agent cannot see this)

Fixed deliberately, because chunks 0000–0004 touch the same vocabulary and the two halves must be reconciled at assembly:

Lemma-expansion policy (convention 3, made checkable): a clipped lemma is expanded only as far as the printed words' own grammar forces an English complement — a transitive verb gets its object (Supportate. → "Bear with me."; Et non egerunt. → "And have not done penance."; Nec enim debent, → "For neither ought the children,"), a bare noun or a complete two-word clause is rendered as it stands (Confidentia. → "Confidence."; Est veritas. → "There is truth."; Insipientiae. → "Of foolishness."). No further words are imported from the verse. ⚑ This is why Est veritas. is NOT "The truth of Christ is in me" — and it is not an inconsistency with the expanded stumps, which belong to the other class.

Repeated lemmata, checked identical in both places (the prompt's three, resolved by expanding, never by contracting):

• Qui multum non abundavit. — 0563B ×2 → "He that had much did not abound." both times.
• Utinam sustineretis, — 0566A ×2 → "Would that you would bear with me," both times.
• Sed quasi in insipientia. — 0567A/B, lemma then re-quotation → "But as it were in foolishness." / "as it were in foolishness" — same words both times.
• Also checked: Non quasi imperans (0562D ×2), sive in corpore, sive (0568A/B ×2), In vobis (0570A ×2), Nam et nos (0569D–0570A ×2). All rendered identically.

2. Lemma ↔ Clementine collation — the full list

Every lemma named in the launch message was greped against sources/vulgate/clementine-flat.txt. 2 Cor 6:1–13:13 plus every non-Corinthian quotation. Result: every named lemma checked, plus nine non-Corinthian quotations · 8 divergent · 4 marked.

2a. AGREE EXACTLY with the Clementine (rendered from Migne, no marker, nothing to report)

Chapter VI–VII (0005): Ne in vacuum · Nunc tempus · Ut non vituperetur · Sicut Dei ministros · In multa patientia, in tribulationibus · In castitate · in scientia · In suavitate · in Spiritu sancto · In charitate non ficta (orthography only) · In virtute Dei · Per gloriam et ignobilitatem · Per gloriam · ut seductores · habiti seductores · ut · veraces · ut castigati · Quasi tristes · Omnia possidentes · Os nostrum · Nolite jugum · Quae autem conventio Christi · Cum idolis? (mark included) · Quoniam inhabitabo · Ab omni inquinamento carnis · Ad commoriendum · Repletus sum consolatione · Superabundo · Caro nostra · caro · Foris · intus · Qua consolatus est in vobis · Et si … poeniteret · Videns · Videns quod · Contristati estis · Poenitentiam in salutem · Saeculi autem · Ecce enim · Quantam in vobis · Exhibuistis · Qui fecit injuriam · Abundantius magis · Sed sicut · Et viscera ejus. Also Matth. XI, 28 Venite ad me, omnes qui laboratis et onerati estis, et ego reficiam vos — agrees word for word with Mt 11:28.

Chapter VIII–IX (0006): Notam autem vobis · Quod in multo experimento · Et altissima · Primum Domino · Non quasi imperans ×2 · Egenus factus est · cum dives esset (word order only; and the gloss's own point is fuisset vs esset, not the order) · Sed et velle · Sed ex aequalitate · Ut fiat aequalitas · Sicut scriptum · Qui multum non abundavit ×2 · In hanc gratiam · Devitantes hoc · Coram Deo · Confidentia · Sive · Ostensionem · Et vestra aemulatio · Erubescamus · Qui parce seminat · Non ex tristitia · Hilarem enim datorem · Abundare · Incrementa frugum · Quae operatur · Gratias Deo.

Chapter X (0007): Ipse autem ego Paulus · Paulus · humilis · Non secundum carnem militamus · Ad destructionem munitionum · Et in promptu habentes · Ulcisci · Quae secundum faciem · Non erubescam · Ut autem non existimer · Hoc cogitet · Tales · Non enim audemus · Nos autem · Sed secundum mensuram · Non in alienis laboribus (a clip that joins 10:15's Non to in alienis laboribus; both words are the verse's) · Qui autem gloriatur · Sed quem Deus.

Chapter XI (0008): Utinam sustineretis ×2 · Insipientiae · Aemulor · Vos uni · Timeo autem · Nam si is qui venit · Recte pateremini · Existimo · Nam etsi imperitus · Quoniam gratis Evangelium · Est veritas · Ut in quo gloriantur · Nam ejusmodi · Inveniantur sicut et nos · In quo · Transfigurantes se · Quorum finis · erit secundum opera ipsorum · Quod loquor · Sed quasi in insipientia ×2 · Caedit · Hebraei · Israelitae sunt, et ego · Ut minus sapiens · In laboribus · A Judaeis quinquies · Una minus · Periculis latronum · In fame · Sollicitudo · Non uror? (mark included) · Non mentior.

Chapter XII–XIII (0009): Scio hominem · Sive in corpore, sive · sive in corpore, sive extra · Raptum · Usque ad tertium coelum · Raptus est · Arcana verba · Supra id quod · Et ne magnitudo · Dixit mihi · Nam virtus in infirmitate perficitur · Factus sum insipiens · In omni patientia · Et virtutibus · Nec enim debent · Et superimpendar (Clementine sets super impendar as two words — spacing only) · Olim putatis · Ne iterum cum venero · Et non egerunt · Non parcam · An experimentum · Qui in vobis non infirmatur · Nam et nos infirmi · Nam et nos · infirmi in illo · vivemus cum eo · virtute Dei · in vobis · In vobis · Oramus autem · Reprobi simus · Gaudemus enim · Secundum potestatem · Non in destructionem · Gratia · Et charitas Dei.

2b. DIVERGENT — the eight, and what was done with each

2c. Non-Corinthian quotations checked, not marked

• 0560A · Mt 11:29–30 Tollite jugum meum super vos, et invenietis requiem animabus vestris. Jugum enim meum suave est, etc. — Migne runs 11:29 into 11:30, dropping et discite a me, quia mitis sum, et humilis corde. A mid-quotation abridgment closed by his own etc., not a divergent reading. No marker.
• 0561A · Habitabo inter illos et ambulabo (Baruch III). ⚠ The words are not in Baruch at all. They are 2 Cor 6:16's own inhabitabo in illis, et inambulabo inter eos, itself quoting Lv 26:12 Ambulabo inter vos. Baruch III is the reference for the NEXT quotation — Terris visus est, et cum hominibus conversatus est = Bar 3:38. So Migne's parenthetical locator sits one quotation early. Reproduced exactly where he prints it, inside the italics, untranslated. No marker and no reordering: a mis-bound reference is his (runbook false-positive 2 warns specifically against "fixing" one).
• 0561A · Bar 3:38. Migne prints Terris visus est, the Clementine Post haec in terris visus est. Opening truncation of a quotation, plus the dropped preposition; not marked.
• 0561A · Jo 1:14 Verbum caro factum est, et habitavit in nobis — agrees exactly.
• 0566A · Non te laudent labia tua, sed proximi tui (VERS. 1, uncited, and not in italics). Pr 27:2 reads Laudet te alienus, et non os tuum; extraneus, et non labia tua. A free reformulation from memory, and Migne does not present it as a quotation (no italics, no locator). Declined: Pattern 14 marks a citation that diverges, not an allusion.
• 0567D · Lc 12:31 / Mt 6:33, cited (Luc. XII.): Migne Primum quaerite regnum Dei, et haec omnia adjicientur vobis? against Lc 12:31 Verumtamen quaerite primum regnum Dei, et justitiam ejus: et haec omnia… Declined: an omission (et justitiam ejus) plus word order, and neither changes what the sentence asserts in the argument about food. The ? is Migne's own and is kept — it closes his direct question Ubi ergo promissio Dei: …?
• 0569A/B · Phlp 1:23 Cupio dissolvi, et esse cum Christo against desiderium habens dissolvi, et esse cum Christo. Declined: the traditional citation form, sense identical.
• 0570B · Act 5 (Saphira and Ananias) — an allusion, not a quotation.

3. Negation — every printed negative, and what it attaches to

Test 2 was run on all five chunks: each non · nec · neque · ne · nisi · nunquam · nemo · nihil · nullus in the Latin was named, its host word identified from the sentence's own structure, and the English checked to carry the negative ON that host. Counts came out equal in both directions in all five chunks and nothing was relocated. The 2a-population sites (a negative near a quantifier, a correlative, or a sed pivot) are these, with the host proved from the structure:

• 0562A · non tantum propter eum qui fecit injuriam, nec propter eum … sed ad manifestandam. Host of non is tantum, not scripsi — the nec … sed correlative proves it. English: "not only for his sake … nor for his sake … but to make manifest". ⚑ Note tantum is Migne's roman supplement inside the italic quotation, so it is the glossator's word, not the verse's; it is rendered as text, not as lemma.
• 0562A · non magis causa eorum … quam causa Ecclesiae. Host is magis, a comparative correlative, not scripsisse. English: "not more for the sake of … than".
• 0562B · non solum de correctis, sed etiam de aliis. Host is solum. English "not only … but also".
• 0562D · Non solum rogavi Titum, sed ego ipse dico. Host solum.
• 0563A · Nemo ergo se contemnat — nemo is the subject; English "Let no man … despise himself", the negative on the subject, not on the verb.
• 0563B · non plus exigitur quam retinere sibi debet. Host plus. English "no more is required than".
• 0563B · Non quin differant in claritate: sed quia…. A double negative (non quin) carried as "Not but that they differ".
• 0563C · non unum solum, sed et duos et tres — host is unum solum, English "not one only, but two and three"; and non ab Evangelio, neque ab electione solum, sed etiam quod…, where both negatives sit on the phrases, not on commendat.
• 0563D · fama nostra non pollui, sed pollere debet. Host is the infinitive pollui and the sed pollere correlative proves it: "our good name ought not to be defiled, but to be strong". (The Latin's own pollui/pollere wordplay is not reproducible; noted, not repaired.)
• 0564B · non quaeritur de quanto, sed quo animo detur. Host is the phrase de quanto.
• 0565B/C · ut … non existimer … tantum terrere verbis, et opere … non implere. Two negatives, two hosts: existimer and implere, with tantum on terrere. Rendered "that I may not … be thought by you only to terrify with words, and not to fulfil in deed".
• 0566A · ne in eo illis videatur inferior quam ipse quidem non curat. The non is on curat, and quam is the relative accusative picking up gloriationem, not a comparative — "which glorying indeed he himself does not care for".
• 0566A · Non te laudent labia tua, sed proximi tui. Host laudent; thy singular per Pattern 17.
• 0566A · Insipientiam dicit … ne … videatur inferior: ne on videatur.
• 0566B · Hic amoris est, non livoris and non ut ad laudem ejus proficiant, sed ad horum profectum: both on the phrases, per the non … sed pivots.
• 0566B · virginitate mentis (etsi non corporis) servare: the parenthetical negative stays inside the parenthesis and on corporis.
• 0569D · non solum se pro salute eorum impendere, sed et mori paratum: host solum.
• 0570A · Non destruuntur autem qui corriguntur, sed correcti aedificantur: host destruuntur.
• 0566A · Nam si is qui venit. Non missus. — the two-word gloss is itself the negative.
• 0570B · We rejoice if it be not necessary… (si non sit necesse): host necesse.

⭐ The one 7a site I want a later reader to look at, because it is where a negative would most naturally be smuggled IN:

• 0565C · Nos autem, etc. Sicut illi qui usurpando gloriantur ultra mensuram. The verse is 10:13 Nos autem NON in immensum gloriabimur, so the gloss's elliptical Sicut illi qui… plainly means "[not] as those who by usurping glory beyond measure". Migne prints no negative and I supplied none: the English reads "As those who by usurping glory beyond measure." It reads as an affirmative comparison and it is meant to be read against the lemma's own non. This is Pattern 7 obeyed, and it is exactly the shape 7a warns is easiest to "improve".

4. Plate candidates — three sites in 0006, all rendered, none marked

All three are in the Corpus Corporum TEI itself (sources/pl/tei/8986.xml), so they are not our chunker's; that does not make them Migne's.

1. 0562D/0563A · abjicicimus for abjicimus. A non-word (doubled syllable). Corpus control: our chunked Latin has abjicimus 3×, abjicicimus only here. Per the addendum-A rule a non-word is worth raising as a plate candidate and has been right every time — but "right" there has meant our corruption, so a [sic:] would publicly blame Migne for a defect that is probably the digitization's. Rendered for its evident sense ("we cast off"), which the clause requires (it is the only verb). ⬜ Plate read wanted.
2. 0563A · cum-tua for cum tua. The only hyphenated token in the entire work. The shape is a line-break hyphen surviving transcription — the single likeliest digitization artifact there is. Rendered "with thy beggary". ⬜ Plate read wanted; if Migne really sets it so, it is a Pattern 10 run and takes a [sic: cum-tua].
3. 0563D · ibenter for libenter. A non-word; corpus control 155 libenter to 1 ibenter, and the one is here. A dropped initial letter is the classic OCR loss. Rendered "gladly". ⬜ Plate read wanted.

⭐ And the opposite class (brief §4 / addendum C), recorded as a negative so the check is on the record: I looked for forms that read suspiciously tidy in rough surroundings and found none I would name. The passages where a normalization would be invisible are 0563A (the pauper in cella, dives in conscientia rhetorical run, which is polished but is polished in a polished paragraph) and 0570B–C (the Chrysostom peroration on peace, which is the smoothest prose in the range and sits immediately beside the broken Istam ego imitetur of §5.2 — anyone reading that plate for §5.2 should read the whole page while there).

5. Passages read closely, decided, and why

5.1 · 0562A — quos superius in prima. Epistola … dixit

Migne prints a full stop between prima and Epistola, breaking "in the first Epistle" in half. Preserved exactly (Pattern 8: never delete a mark he prints), so the English reads "…whom above in the first. Epistle he said had done wrongs and frauds to their brethren." It is unreadable as English and it is what the page says. ⬜ A plate read settles whether the stop is Migne's or our TEI's; either way nothing here is emended.

5.2 · 0570B — Istam ego imitetur. — the worst site in my range

Printed: Ista pax orbem servavit … ista homines angelos fecit. Istam ego imitetur. Nominative first-person ego stands with third-person singular imitetur. The sentence does not construe in any reading.

• Rendered "This one, I, let him imitate." — both printed persons kept visible, nothing smoothed. It reads as broken because it is broken.
• Two conjectures, and which I took. imiter ("let me imitate this one") keeps the printed ego and needs one letter; imitemur ("let us imitate") is the natural sense of an exhortation and matches the Chrysostom source this paragraph is drawn from, but it contradicts the printed ego, which is why it loses. If a conjecture is ever put into the text, it should be imiter.
• No [cj: …] fired. Pattern 18's bar is that the printed word makes the sentence assert something the author is arguing against, and 18a's extension covers negations and correlatives. This is a person mismatch that asserts nothing at all — the declined class named at 11064 @1154C (suscepit for suscepi). Declined for that reason, not overlooked.
• No [sic:]: imitetur is a real, well-formed word, and the column is unread.

5.3 · 0563A — Ut inopia illi.

Migne prints the lemma Ut inopia illi; the Clementine has ut illius inopia vos divites essetis, and the gloss itself expounds per illius paupertatem. Rendered "That by poverty for him." — illi is a dative and is rendered as a dative. The obvious reading "That by his poverty" was rejected: it renders a printed dative as a genitive, which is Pattern 7's silent repair, and convention 3 licenses expanding a fragment, not changing one of its words. The gloss's next sentence supplies the sense to any reader. No marker (real word, unread column).

5.4 · 0566B/C — the Jerome usque ad tail, and its unexpressed subject

Serpens Evam de Deo mentiendo seduxit, etc., usque ad penitus non esse, aut aeternam non esse affirmat contra auctoritatem omnium Scripturarum. The resumption tail was parsed as strictly as ordinary prose (the ferias warning): affirmat is the finite verb, the two esse clauses are its objects, and aeternam is a feminine accusative — so the thing denied is a feminine noun that the abridgment removed, almost certainly poenam or gehennam (Jerome on those who deny that hell exists or that it is eternal). English says "it", and this note is what tells a reader what "it" is. Nothing was supplied to complete the formula.

5.5 · usque ad — all four occurrences decided one at a time

Resumption tails parsed strictly: 0565B's peccantibus is a dative governed by fortis existo ("toward sinners I stand forth strong"), not an ablative absolute.

5.6 · Unexpressed subjects supplied — every one, with what decides it

• 0559D–0560A · Hi non adulantur — "These" = the Dei ministros of the lemma above.
• 0560B · Ficta est, quae deserit in adversitate — the subject is charitas, from the lemma In charitate non ficta immediately above; English "That charity is feigned which…".
• 0560B · Vel hoc dicit and every later bare dicit / ait / ostendit / tangit / addit — the Apostle, named explicitly at 0561A (Apostolus vult eos separari) and again at 0562B, 0563C, 0567B. Rendered "he" throughout, and "the Apostle" where Migne writes it.
• 0561A · Idola prohibet coli — subject is the Apostle, not Christ: the sentence sits inside the exposition of Paul's Cum idolis? and is followed by His omnibus modis ostendit.
• 0561C · quod Epistola contristavit — the object vos is unexpressed; supplied from 7:8 epistola illa … vos contristavit. English "made you sorrowful".
• 0562A · cui consentiendo inquinati eratis — cui is the offender (illius qui uxorem patris habuit), not the negotium: "by consenting to whom you had been defiled". ⚠ Both readings construe; I took the person because consentire in this gloss governs persons and because the offender is the nearest masculine antecedent.
• 0562A · the 7:12 cast of characters — qui fecit injuriam = the incestuous man (Migne says so: Ut ille qui incestum fecit); qui passus est fraudem = the injured party; Titus is not in this sentence at all, though he is the subject two paragraphs later at 0562B.
• 0562B · quia se gaudebat Apostolus — the reflexive se with gaudebat is pleonastic; rendered "the Apostle rejoiced in himself" rather than dropping the word. Logged because dropping it would have been invisible.
• 0565D · qui ad omnes missus erat a Deo — "he who had been sent by God to all" = Paul, from Ad regulam autem Apostoli in the same sentence.
• 0567B/C · detrahebant illis — "them" = the uncircumcised Gentile believers, from the same clause's eo quod essent incircumcisi.
• 0567D · cum tamen nihil ferret — the subject is Paul, not the devil and not the robbers: the robbers are the ones who would want what he does not carry.
• 0568A · Quia fert mihi testimonium — the subject is God, from 11:31's own Deus et Pater … scit quod non mentior.
• 0570A · et sic reprobi appareant — ⚠ a real ambiguity, and it is not adjudicated silently. The plural must be either (a) the apostles, who "appear reprobate" when there is no one left to judge — which is 13:7's own nos autem ut reprobi simus, and which the next two sentences state twice (probati enim videntur, dum judicant peccatores; cessante auctoritate quasi reprobi videntur) — or (b) the Corinthians. I took (a), and the English keeps a bare "they" so the reader is not committed further than Migne is; (b) loses because it makes the following Reprobi simus, id est appareamus reprobi change subject mid-argument.
• 0570C · Sua tribuit congrua cuique — the subject is the Apostle distributing the three gifts of the salutation, from Totam Trinitatem optat benedicere eos just before.

5.7 · Smaller readings settled, with the loser named

• 0565B/C · quod minor verbis … quod minor verbis impleturus sum factis. minor is the deponent verb "I threaten", not the comparative "less". Minatur stands four words earlier in the same paragraph (quod absens verbo minatur*), and "what is less by words" yields nothing. Rendered "what I threaten with words" in both places.
• 0567B · Caedit. Id est, decipit. Migne glosses "he strikes" by "he deceives", which is odd but is what he prints, and his next sentence (In faciem caeditur is, in cujus os injuria irrogatur) shows he means a figurative blow. Rendered as printed; no marker.
• 0565A · Et ad haec devincendum — gerundive singular against plural haec. Pattern 9: a number agreement with no English exponent. Rendered normally, no marker, no churn.
• 0562C · quibus suadere non potest / quibus sua dare non prodest. Migne records the variant himself (ut alii codices habent), and the two readings are the same letters differently divided — suadere / sua dare. Both rendered as printed; the pun is not reproducible in English and is recorded here rather than forced into the text.
• 0565C · pseudo qui commendant — the Vulgate's qui seipsos commendant; Migne clips the reflexive. English supplies "themselves" because commend is transitive in English and the gloss's own next words are id est, usurpant potestatem. Recorded because it is a supply.
• 0570C · communicatio facti Spiritus sancti — facti is genitive of factum; rendered literally "the communication of the deed of the Holy Spirit". It reads strangely and is printed so. No marker (real words, unread column).
• 0568C · ne alter puter — alter rendered "other than I am", the small supply English needs; the alternative "lest another think" is wrong, because puter is first person.

6. What I read clean

• Column anchors: counted before and after writing each file — 11 · 9 · 6 · 8 · 9, in order, and each was checked against the word it stands beside in the Latin, not merely for presence (2b). Four sit inside a construction and stayed there: In multa patientia, in [0560A] tribulationibus and vos esse incontaminatos, a [0562A] negotio (both inside an italic lemma), Ab omni [0561B] inquinamento carnis, and penitus [0566C] non esse. Two were moved after a first draft placed them at the nearest sentence break instead of the nearest word — 0562D (jura auctoritatis [0562D] inclinant) and 0563C (pecuniarum collectarum [0563C] multitudo).
• The 0569C→0570A jump (no 0569D) is Migne's band spacing, false-positive 1. Not reported.
• Notes: 3 · 0 · 1 · 4 · 3, verbatim, same order, contents untranslated. All eleven are citations or sigla — (Matth. XI), (Ibid.), (Joan. I), (HIER.) ×3, (AMBR.), (Luc. XII.), (THEOPH.), (CHRYS.) — so no Pattern 15 [nt:] conversion arises; judged by content, not length. Two of them sit inside the italic lemma (Timeo autem. [n: (HIER.)], Est veritas. [n: (HIER.)]) and are reproduced there.
• Question marks: 1 · 1 · 1 · 2 · 3, all preserved, none supplied. Every one is a direct question in Latin, so Pattern 8b never fired: Cum idolis? · pro confidentia vestri ivit? · quare me potius quam pseudo sequamini? · …adjicientur vobis? · Non uror? · At cujus gratia … gloriaris? · Quis? · Sed nunquid mox ut nascitur? ⚠ The one indirect question in the range — 0568A/B Hoc ignoravit Apostolus, utrum … an … an … — Migne closes with a full stop, and it stays a full stop.
• Punctuation by band (test 3): colons, semicolons, question marks, guillemets and exclamation marks compared band by band across all 48 bands of the five chunks, with our English-only markers stripped first. Zero mismatched bands. Eight edits were made to get there, all of them ours: a supplied semicolon at 0560A, a supplied colon at 0563A, a supplied colon at 0565C, two supplied semicolons at 0566A/B, a deleted colon at 0569B (Jamdudum forsan putatis:) with a supplied semicolon beside it, and two supplied semicolons at 0570C. No comma→semicolon sweep was made in either direction.
• Guillemets: none in the Latin, none in the English.
• Migne's own ( sic ): none in this range, so nothing was stacked.
• Verse numbers: VERS. 1–14 (ch. VI–VII), 1–15/18–24 and 2–15 (VIII–IX), 1–18 (X), 1–31 (XI), 2–21 and 2–13 (XII–XIII) — all in ascending order, none out of sequence.
• Heads: CAPUT VI. → CHAPTER VI. and so on through CHAPTER XIII., 1:1, Roman numeral kept, matching the corpus's shipped forms. No (cont.) heads in this range.
• Word ratio: 1.46 · 1.47 · 1.47 · 1.49 · 1.48 — at the Abbo-pilot 1.5×.
• verify-english.mjs 8986 passes (all 10 chunks, this half and the peer's).

7. The fourth test

Every paragraph was read back against the Latin asking only does it assert what he asserts, after the counts were known clean. It caught three things, all repaired in the text:

1. 0570A · ut … corripienda non inveniat had been rendered "may find in them nothing to be corrected", which relocates the negative from inveniat onto the object. Now "may not find in them things to be corrected".
2. 0565C · Quod est, non ponimus nos, where our punctuation had turned Migne's comma into a colon and made a second lemma-like announcement out of a continuing clause.
3. 0562C–0563A, where three smooth English sentences sat over the three defective forms of §4 — the "reads smoothly where the Latin is rough" tell. The smoothness is now on the record here rather than passing as success.

⬜ Left for a plate reader, in priority order: 0563A cum-tua · 0562D/0563A abjicicimus · 0563D ibenter · 0570B Istam ego imitetur (read the whole page — the Chrysostom peroration is on it) · 0562A in prima. Epistola · 0570B et dilectionis · 0563C In hac parte · and the four [var:] columns 0561A/B, 0562A, 0567D, 0568A, which plate-gate.mjs will list.