Liber Isaiae prophetae

Glossa ordinaria on Isaiah, Anselm of Laon and his school. PL 113 cols 1231B–1316C, 35 chunks, 32,307 Latin words, translated in seven Opus stints on 2026-09-07. Merged from seven stint files. The concatenation is the easy half; what follows is the reconciliation, which is the required step.

Apparatus as shipped: 103 [var:] · 6 [sic:] · 3 [cj:] · 1 [d:] · 0 [ed:]. Every marker stands on a column read at Migne's plate (plate-gate.mjs 8956 green). The whole book was read at the plate — 53 leaves, f616–f658, printed corner numbers checked FIRST on every one — and the second printing was consulted wherever a letter was at stake. 36 TEI patches were filed and applied; all are digitization losses or a corrupted anchor, none is Migne's defect, so none carries a marker.

⭐ The foot of every column in this book is a CHECKED ZERO, on both printings. No numbered Forte conjecture layer, no asterisk cross-reference layer, anywhere in cols 1231–1316. That is a measurement of this book, not an inheritance from a sibling.

Work-wide conventions — what was reconciled at the merge, and what was left alone

⛔ 1. conversatio — swept, and swept to match the book shipping beside it

Four sites, all in the final stint, all rendered "conversation". Swept to "manner of life" at every one.

Two considerations, and they point the same way. First, the reader: conversatio is manner of life, and "conversation" is a false friend a general reader takes as talk — the 11551 epistylium class. Second, and decisive here: 8961 Leviticus made exactly this sweep at its own merge (29 of 30 sites) and ships in the same deploy as this book. Two volumes of one commentary, going out together, must not render one recurring term two ways. 8961 now carries 38 "manner of life"; 8956 carries none of the false friend.

⚑ Two of the four are conversatio in coelis, where "our conversation is in heaven" is the familiar Douay/AV wording of Phil 3:20. That is precisely the hazard the 11632 findings named — Douay wording arriving over the plate — and it is why the familiar phrase loses here: we translate what Migne prints, for a reader who does not already know the verse.

✅ 2. usque ad — no drift, checked rather than assumed

The abridgment formula is rendered "as far as" in all 33 chunks that carry it. Not one stint diverged. ("down to" occurs in five chunks and is ordinary prose in every one — checked individually, not counted.)

✅ 3. The lemma layer — a fact all seven stints established independently

This work sets no guillemets at all (0 «, 0 »): every lemma is italic, the Luke-9000 shape rather than the Genesis-8950 shape. Each stint counted its own spans against its brief and matched exactly — 293 · 303 · 278 · 327 · 253 · 352 · 248 = 2,054, the master inventory's own total.

✅ 4. Non-Latin type — one convention, settled mid-run rather than at the merge

Full detail in data/briefs/8956-CONVENTIONS.md. Two stints handled dropped Hebrew two defensible ways; the ruling was to file the patch and not insert into the English alone, because hebrew-recovery.md patches both twins in one operation. The pair is now converged UPWARD: צדק and (סגים) at 1235D, (דבר) at 1248D, (שחין) at 1280C, (מסכן) at 1282A stand in both twins. ⚑ The orphaned double and triple spaces our file carried at each site are the scars the loss left, and they are what made the recovery checkable.

⚑ 5. What the run established about the WITNESSES, and it cuts against the standing note

PL113-WITNESS-ENCODING.md says the archive copy is 1-bit JBIG2 and may never be a sole witness for a letter. That has been read as Gallica is the good copy. This book says otherwise, twice over. In cols 1291–1305 Gallica is the WORSE witness — five worn sorts in fourteen columns (tribuatus, lauaem, funiculu, Matih., irruguus), each a real non-word on Gallica and in our TEI, each clean on the archive. Earlier, at 1265A, the traffic ran the other way (potet/potest). Neither copy is privileged. Nothing stands on one witness, in either direction.

⭐⭐ Seven [sic:] candidates were withdrawn at the second printing in this book alone — four in cols 1291–1305, three more in the last stint (Chrito, fatres, and a meridia nam the archive prints hyphenated). Seven false public accusations against Migne's type, stopped by one HTTP request each. Open the other copy BEFORE firing, not after.

⚑ 6. A defect class that is ours and wears Migne's face

Five sites, one mechanism: a hyphen at a line or column break. cruceis, crucis is cru-|cis, ONE word — and from our file alone it satisfies Pattern 11a exactly, so it would have carried a [d:] on a doubling Migne never set. Also exallabitur (exal-|tabitur, across the very column break our own anchor marks), supereum, me moriter (= memoriter), sub coelo 1304est. All five are patched. When a candidate defect sits at or near a break, the break is the first hypothesis, not the last.

⚑ 7. Left open on purpose — two 7a sites, and neither was repaired

1311D prints In conturbatione, in dolore mentis, sed in gaudio et laetitia — no negation where its own sed demands one, confirmed on both printings. 1314D's Quae secundum apostolorum non solum hominum has no governed noun at all. Both are rendered as printed and the conjecture lives in the crux. A repair hidden inside the translation leaves no trace and passes every automated check.

⚑ 8. The comparand is not clean either

sources/vulgate/clementine-flat.txt prints the non-word cornonabit at Is 22:18 (upstream in Tweedale 2005; the post-2005 correction diffs are unapplied). A stint met it, could not collate, and declined to fire rather than reporting Migne as divergent — which is right: a [var:] fired against a corrupt Clementine is a false public claim about Migne's text. Now recorded in sources/vulgate/README.md and printed into every lemma brief.

⚑ 9. One patch candidate deliberately NOT applied

1302D, where the lemma's italics swallow the (HIER) attribution. Promoting it to a note changes noteCount on the Latin side alone and breaks twin parity; it wants a paired edit, not a patch. Recorded here so it is not lost and not re-proposed as new.

⭐⭐ The blind polarity read — 35 chunks, four readers, 17 sites, ALL OURS, and NOT ONE A MISSING NEGATION

Step 4a, run after the merge. Four readers over the whole book, none shown this file until its own findings were written down — the answer-key effect is measured (2026-07-28: every agent given the cruces first returned "all already logged" and found nothing).

Every one of the seventeen survived verify-english, the two-printing plate collation, the 36-patch pass, the 103-marker Vulgate check and the merge above. That is what this step is for.

⛔ translation-style.md 7a is written around a dropped non/nec/nisi. On this book it caught nothing, because in a lemma-and-gloss commentary the live seam is AGENCY — who does what to whom. 8961 Leviticus said the same on its own read (7 sites, 6 ours, zero negations); Isaiah is the second book to say it, on four times the readers. The rulebook should be widened, and the classes below are the evidence.

The classes, with the sharpest instance of each

A sentence reversed against its own neighbour. 1272A — Gregory's gloss is built on Non suum, sed alienum est, and the English made alienum ab ipso read "his strange work is from himself": the exact reverse, one line after "not his own but strange", with the enim clause then explaining the opposite of what stood. Repaired to "that work of his is alien to him."

A lemma silently conformed to the Vulgate. 1299C — Migne prints montem sanctum tuum; the English printed "my holy mountain", the Vulgate's reading, unmarked. The plate now stands, with a [var:] naming the divergence. ⚑ Two more of the same shape: His fire is in Sion (1275B, Migne prints no possessive) and Behold a king shall reign (1275C — Douay's Is 32:1 verbatim, where the plate prints no rex). This is the 11632 hazard exactly: Douay wording arriving over the plate, here inside the lemmata, where it is least visible and most load-bearing.

A verb or clause supplied to make a truncated incipit read. 1297A — Migne's lemma stops dead at Ecce accola and the English supplied "shall come" from the Vulgate. 1237C — Migne's comma orphans the accusative Judaeos and the English resolved it twice, once as the participle's object and again as "in place of the Jews", a phrase the plate does not print in any form.

Identifications swapped, in sentences whose only job is to identify. 1265B Domum David vocat domum Domini had the gloss and the lemma change places. 1249A Ordo: Reges sunt mei principes reversed the very word order the gloss exists to supply — against its own next clause, multos enim reges secum adduxit, quos subjugaverat.

An agent invented, or deleted. 1282C read the dative Illi as a demonstrative and deleted God from a sentence about what is numbered to him. 1291A supplied a "he" that made God call a burning to repentance. 1238B bound two instrumental relatives (ab eo quoque quo) to an unnamed personal crusher.

Voice flipped mid-sentence. 1293D — nubes levari and solis splendorem abscondi are both passive in one acc.+inf.; the English kept the first and made the second active, so the clouds became the agent and the sentence half-anticipated its own next clause. 1314C — the second, active dicere became an impersonal passive, so a claim its claimants make about themselves turned into a report about them.

Two words flattened into one, or one word split into two. 1302C — occidisti (of a light, going down) and cecidisti (falling) were both "fall", in a solar context (Surge, illuminare; sicut sol ortus) where the distinction is the point. 1301D — regulus, qui et basiliscus is a gloss whose whole business is that one creature has two names, and the English asserted the identity and then contradicted it three lines later.

A name normalised. 1245C — our Latin prints Rasim and the English printed the familiar Rasin, which our own Latin prints two columns later. So at the one place the twins differ, the English quietly conformed. ⚑ No marker fired: a single letter cannot stand on one witness, and this one was not put to the second printing.

What the readers rejected, and why that matters

Each reader named its clean chunks and its rejected candidates with reasons — the check that tells a real pass from a tidy "nothing new". Three rejections are worth keeping: one reader withdrew its own finding on Domini gentium after this file answered it; another withdrew three normalisation flags (tribuatus, Chrito, fatres) on finding them recorded here as Gallica-copy broken sorts that print clean on the archive; and a third confirmed positively that the two deliberately unrepaired 7a sites — 1311D, which prints no negation where its own sed demands one, and 1314D, whose governed noun is absent — were not quietly mended. That last is the check that had to come back clean, and it did.

Stint 0000

Stint file; the orchestrator merges it with the peers'. Every column in the range was read at the plate before these entries were written — Gallica ark:/12148/bpt6k5505319w, leaves f616–f621, f = (column + 1) / 2, calibrated against the printed corner numbers on all six leaves (f616 = 1231/1232, f617 = 1233/1234, f618 = 1235/1236, f619 = 1237/1238, f620 = 1239/1240, f621 = 1241/1242) before any collation was attempted. Reads recorded in data/briefs/8956-PLATE-READS-0000.json; the archive.org jbig2 copy was not consulted, and the plate-reads file says why it was not needed at any site.

Span count reconciled: 293 marked spans in the Latin of chunks 0000–0004 (293 italic …, counted paragraph-bounded; 0 guillemet spans — this work sets no « » at all, like Luke 9000 and unlike Genesis 8950), against the brief's 293. Exact agreement.

Plate vs. our TEI — four sites, all patched forward

1234A — Affectu, not Affecta. The plate reads Affectu piissimi patris, broken Affect-/u at the right edge of the line; our TEI has Affecta, which agrees with nothing and leaves the phrase without a construction. Read at native resolution and again at 900px: the final glyph has two stems joined at the foot. Corroborated by the work's own Affectu pii patris at 1236B. English renders the plate; patch filed.

1234D — a colon our TEI drops. The plate prints …legem esse servandam : audiant hoc eorum socii…; our twin has no mark, so the two parallel imperatives run together. Two dots, vertically ranged, confirmed at 900px. English carries the colon (Pattern 8); patch filed.

1235D — Hebrew type our TEI drops, twice. The plate prints in Hebraeo sedek צדק and Hebraice siggim (סגים); our twin has the transliterations alone. Migne's own transliteration stands beside each Hebrew word, which is what licenses the reading. Restored silently in the English under Pattern 13a clause 1 (words lost mid-sentence, plate certain); patches filed. Note the second one leaves a visible scar in our twin — the stranded space in siggim : quod is where the parenthesis was.

Refuted at the plate, and worth recording as refuted. 1233C looked like clarem scientiae at column-half scale; the line-broken run cla- + vem at native resolution is clavem, as our TEI has it. 1235D looked like Ooliban; it is Oolibam. Both were candidate findings that a wider crop killed.

Migne's readings — markers fired

1237A — constabunt for the Vulgate's conflabunt (Is 2:4). [var:] fired; [sic:] deliberately NOT fired. The plate prints Et constabunt gladios suos, confirmed on a dedicated crop (s-t, not f-l), and our twin agrees. Consto is intransitive and cannot govern gladios, so the printed clause does not construe, and the English is correspondingly awkward: "And they shall stand fast their swords." The gloss that follows — Omne studium bellandi vertetur in pacem — is the swords-into-ploughshares argument, so the Vulgate's verb is certainly the sense. I decline the [sic:] on the class question, not the reading question: a [sic:] asserts that the type is defective, and this could as easily be Migne's copy-text as his compositor. [var:] states both readings and claims neither class. Recorded so a later reader can overturn it with evidence rather than with confidence.

1238A — allii for alii. [sic:] fired. The plate prints alii impiorum montes, allii colles, with alii set correctly six words earlier on the same sentence — so the plate itself supplies the norm. Allii is a real word (genitive of allium, garlic) with no slot in the clause, which is Pattern 12's test (1); the sense is rendered beside the marker per the 2026-09-05 amendment. Confirmed on a dedicated crop.

1240A — malitiam for militiam. [cj:] fired. The plate prints Quibus signamur ad malitiam Christi, read across the line break on two separate crops (mali- + tiam Christi); it is Migne's, not our transcription. The word is a perfectly good Latin form, so Pattern 12 has nothing to wrap — and the faithful English then says that the ring seals us unto the malice of Christ, which is not merely odd but the opposite of what the gloss is arguing. Militia Christi is the stock phrase, the sentence continues quem Pater signavit (Jn 6:27, the seal of enlistment), and the prodigal's ring follows in the next clause. That is Pattern 18's bar as Wilson set it: the printed word makes the sentence say something the author is arguing against.

1241C — Vox sanguis for Vox sanguinis (Gn 4:10). [sic:] fired on sanguis alone. The plate prints Vox sanguis fratris tui Abel clamat ad me de terra. A nominative in apposition to vox has no slot, and English cannot show a case error, so under Pattern 12's test (2) rendering it silently would be a repair; the marker carries the type and the sense stands beside it. The marker sits inside an already-italic lemma and therefore takes no second italic pair, per Pattern 10. The plate also adds Abel, which the Clementine does not have; that takes no marker — the referent is unambiguous and no sense turns on it.

The Greek

1235A — Ανθρωποσπαθῶς. Our twin reproduces the plate exactly, sigma-pi and all. The form is not a Greek word: the adverb wanted is ἀνθρωποπαθῶς, "after the manner of human feeling," which is precisely what the next clause glosses (Non quod animam habeat, sed nostro loquitur more). It is Migne's setting, not our source's loss, so no patch. Carried verbatim and untranslated, per the Greek rule, and no [sic:] fired — the following Latin supplies the sense in the author's own voice, so nothing is quarantined and nothing is hidden. Note also that Migne prints it without the rough breathing.

1237C — the Greek φ. Present on the plate and in our twin, in pro ea φ Graeco utitur. Reproduced. No issue; recorded because it is the only other Greek in the range.

Vulgate collation — every divergence found, marked and unmarked

Collated word by word against sources/vulgate/clementine-flat.txt. Marked (see above and the report): Lc 17:10 @1232D · Dt 22:10 @1233B · Is 1:26–27 @1236C · Is 2:4 @1237A · Sir 7:6 @1239B · Jo 4:34 @1240D. Found and deliberately left unmarked, with the reason in each case:

• 1232D, Lc 17:10 — Servus inutilis sum; quod debui facere, feci against the Clementine plural Servi inutiles sumus… fecimus. [var:] FIRED. The singular is the whole point of the gloss (it is Isaiah's own humility being expounded), and a reader who knows the verse would otherwise never learn that Migne's text differs.
• 1233B, Dt 22:10 — in vitulo for in bove. [var:] FIRED, and this is the most load-bearing of the six: the lemma under exposition is Is 1:3 Cognovit bos possessorem suum, and the Deuteronomy text is adduced for the ox. Migne's calf takes the proof-text's own keyword out from under the argument — exactly the 7a″ shape, running the other way.
• 1236C, Is 1:26–27 — the verse division. [var:] FIRED. The plate runs Civitas justi, urbs fidelis Sion, in judicio redimetur as one sentence, making Sion an appositive of urbs; the Clementine closes v. 26 at urbs fidelis and opens v. 27 with Sion in judicio redimetur, where Sion is the subject. Punctuation, but it changes what the sentence asserts about Sion, so it is Pattern 14 and not Pattern 8 alone.
• 1237A, Is 2:4 — constabunt for conflabunt. [var:] FIRED. See above.
• 1239B, Sir 7:6 — Ne quaeras judex fieri, ne forte non possis offerre iniquitates against the Clementine Noli quaerere fieri judex, nisi valeas virtute irrumpere iniquitates. [var:] FIRED. Offerre will not carry the Vulgate's sense at all (one does not "offer" iniquities), and the English renders it literally rather than borrowing irrumpere's meaning. Declined as a [cj:] under 18a: it is a divergence inside a quoted scripture, which is Pattern 14's business and not Pattern 18's, and it fails to construe rather than asserting a falsehood.
• 1240D, Jo 4:34 — voluntatem Patris mei for voluntatem ejus qui misit me. [var:] FIRED, because the gloss's argument is built on the word: mulierum panis Pater est, sicut et Christi, qui ait: Meus cibus est ut faciam voluntatem Patris mei. The Vulgate at that verse does not contain Patris at all.
• 1233B, Mt 11:28 — Venite ad me, qui laboratis drops the Vulgate's omnes. Truncation of a cited verse, no sense change. Unmarked.
• 1233C, Zach 5:7–8 — Iniquitas sedet super talentum plumbi compresses two verses into a clause that stands in neither. A compression, not a variant reading. Unmarked.
• 1233C, Ps 37:5 — Quasi onus grave for the Clementine sicut onus grave. Synonym. Unmarked.
• 1233D, Mt 13:55 — Nonne hic est Filius fabri? for Nonne hic est fabri filius? Word order only. Unmarked.
• 1233D, Phil 3:13 — Ad priora se extendens, posteriorum oblitus is a participial compression of quae quidem retro sunt obliviscens, ad ea vero quae sunt priora extendens meipsum. The author's own paraphrase, introduced as unde Apostolus, with no word for a [var:] to name. Unmarked.
• 1234A, Jer 51:9 — Curavimus Babylonem et non est curata for the Clementine sanata. Unmarked because the words are synonyms — but recorded, because the divergence is what makes the gloss work: the lemma being expounded is Is 1:6 nec curata medicamine, and Migne's verb is the one that echoes it.
• 1234A, Jn 14 (cited) — Omnes enim natura filii Dei sumus is not in John 14 or anywhere in the Clementine. A reference problem, not a text problem; not this stint's to correct.
• 1234B, Ps 49:12 — Mea enim est terra et plenitudo ejus for meus est enim orbis terrae et plenitudo ejus, with the person shifted to first for the speaker. Sense unchanged. Unmarked.
• 1234D, Ps 80:4 — Buccinate Domino in neomenia tuba adds Domino, which the Clementine lacks. Unmarked.
• 1235B, Mt 28:19 — Ite, docete omnes gentes for Euntes ergo docete omnes gentes. Unmarked.
• 1235A, Is 1:15 — the lemma prints plenae sunt sanguine where the Clementine has sanguine plenae sunt. Word order. Unmarked.
• 1235D, Is 1:22 (Symmachus) — mixtum for the Clementine's mistum. Orthography. Unmarked.
• 1236B, Is 1:24 — the lemma Heu ego is in no Vulgate text; the Clementine reads Heu! consolabor. A two-word lemma tag on which no sense turns, and the LXX rendering Migne quotes immediately before it (Vae fortes Israel) is correctly labelled LXX. Unmarked, recorded.
• 1236D, Ps 124:2 — Montes in circumitu ejus for circuitu. Circumitus is the regular formation from circum + eo and is attested; Pattern 9. Unmarked.
• 1237B, Is 2:5 — the gloss quotes its own lemma back as ambulemus in luce Domini where both the Clementine and Migne's own lemma four lines above read in lumine Domini. Recorded because it is a self-inconsistency on one column; sense unchanged, so unmarked.
• 1237B, Jn 3:20 — Qui male agit, odit lucem drops Omnis enim. Truncation. Unmarked.
• 1237D, Mt 24:30 — veniet cum potestate magna et majestate against venientem in nubibus caeli cum virtute multa et majestate. Paraphrase. Unmarked.
• 1238A, Is 2:11 — Exaltabitur autem solus Dominus for exaltabitur autem Dominus solus. Word order. Unmarked.
• 1238B, Ps 106:23 — Qui descendunt mare in navi for in navibus, in a gloss that is about naves bonae, plural. Recorded; the number clash is visible but no argument rests on it. Unmarked.
• 1238D, Ps 86:5 — Et ipse fundavit eam Altissimus is quoted exactly, but cited (Psal. CXII, LXXXIX) where the second reference should be LXXXVI. A reference, not a reading.
• 1239C, Wis 2:12 / Is 3:10 — Alligemus justum, quoniam inutilis est nobis, cited (Sap. I). Migne labels the whole quotation LXX, and it is the LXX of Isaiah rather than the Vulgate of Wisdom, so there is nothing for [var:] to correct. Unmarked.
• 1239D, Is 5:7 — Vinea Domini Sabaoth domus Israel est for Domini exercituum. A Hebraism for the same title. Unmarked.
• 1239C, Jn 14:6 — Ego sum via, veritas, et vita drops the first et. Unmarked.
• 1241A, Is 6:4 — Domus impleta est fumo for repleta, and cited (Exod. XIII). Synonym plus a reference problem. Unmarked.
• 1241A, Is 5:2 — the lemma Vineam electam supplies vineam, which the Clementine's plantavit eam electam does not have (the exact phrase stands at Jer 2:21, which is what the brief's ⚑ was seeing). The gloss immediately names the alia editio reading vineam Sorec, so Migne is knowingly working across editions here. Unmarked.
• 1242A, Is 5:20 printed as VERS. 19 — Migne's verse number is one short of the Clementine's for Vae qui dicitis. His numbering, reproduced.

Readings rendered literally, no marker, conjecture recorded here only

1237A — Propheta for Prophetam. Dominus quoque in templo docuit quod non oportuit Propheta mori extra Jerusalem: a nominative as the subject of an infinitive. Declined as a marker under Pattern 9 — a nominative subject after quod/oportet is ordinary in twelfth-century school Latin, and English shows no case on "a Prophet" either way, so the marker would buy nothing and cost noise. (Also open, and left open: quod here can be read "because" or "that". The English takes "because," which fits the argument for why the law goes out from Sion.)

1238C — Aedificavit Salomon rem in Jerusalem. A real word in a real slot, so it renders in place — "Solomon built a thing in Jerusalem" — and takes no marker under Pattern 7. Fanum or aram is the obvious conjecture (3 Kings 11:7 aedificavit Salomon fanum Chamos), recorded here and not in the text. Declined as [cj:] because a vague word is not a false assertion.

1237C — cum caeteris delata est turpitudo haec. Delata (from defero) construes and is rendered "was carried away together with the rest"; deleta ("was destroyed") would be smoother and is what the sentence seems to want. The plate reads delata. Recorded, not emended.

1240C — Metonomicos. A transliterated Greek adverb (μετωνυμικῶς) set with an o for the y. Rendered "Metonymically" — it occupies its slot as an adverb and renders in place, so under Pattern 12's operational test the word on the page is ours and English, and takes no marker. The spelling is recorded here.

1234B — sic Omnipotens, Filius Omnipotentis est. Migne's comma falls between the predicate and its subject, which English cannot hold without a resumptive. Rendered "so the Almighty, [he] is the Son of the Almighty," with the supplied pronoun bracketed in the corpus's usual way, so that the printed mark survives (Pattern 8) and no word of Migne's is altered.

1238D — Hebraeum tamen excelsum significat, non excelsus. The distinction is between two Latin forms and has no English exponent at all, so both are kept in italic Latin and the author's own next sentence carries the sense ("For he is height itself, not merely high"). Not a defect; recorded so that a later reader does not try to translate the pair.

Stint 0005

Stint file; the orchestrator merges it with the peers'. Every column in the range was read at the plate before these entries were written — Gallica ark:/12148/bpt6k5505319w, leaves f622–f628, f = (column + 1) / 2, calibrated against the printed top corners of f622 (1243 / 1244) before any collation was done. The archive.org copy (patrologiaecurs04migngoog) was not consulted: every leaf was legible at native resolution and no candidate turned on a letter Gallica left open. Reads recorded in data/briefs/8956-PLATE-READS-0005.json; the foot of every column in the range is blank, so this work carries no numbered and no asterisk conjecture layer in cols 1243–1255.

Plate vs. our TEI

1245D — auctoritatatem, an extra syllable in our file. The plate reads tanquam auctoritatem habens; our TEI prints auctoritatatem, which is not a word. The English renders the plate ("as having authority"); patch filed in data/briefs/8956-PENDING-TEI-PATCHES-0005.md. No [sic:] — the defect is not Migne's.

1248D — Hebrew type our source dropped. The plate sets Verbum Hebraicum (דבר) Dabar; our TEI carries the transliteration only. Unpointed, read at 4×; the classic Corpus Corporum scar of hebrew-recovery.md. Not inserted into the English on its own, because that recipe patches both twins in one commit and a translating stint cannot edit the Latin; patch filed.

Migne's own type — markers fired

1250C — apotolos. The plate prints reprehendentes apo-/tolos, broken over a line, with no s; our TEI has the same. A non-word, and the plate licenses the accusation, so [sic: apotolos] stands and the sense is rendered beside it ("them that reproved the apostles"), per Pattern 7's 2026-09-05 amendment. Conjecture: apostolos, which the same sentence's argument requires.

1254A — quas for quasi. The plate prints impii quasi plumbum merguntur in aquis vehementibus, et descendunt in profundum quas lapis — read on a tight native crop, with a clear word space before lapis on the following line. Quas is a real Latin word (accusative feminine plural relative) but has no grammatical slot in its clause, which is Pattern 12's first test, so it takes [sic: quas] and the sense is supplied beside it ("as a stone"). Inside an italic lemma, so the marked run takes no second italic marker (Pattern 10). The conjecture quasi is what merguntur quasi plumbum, nine words earlier in the same sentence, already prints — which proves an error exists, not whose it is; the plate settles that.

Migne's own type — marker DECLINED

1252B — Et requiscere. Migne prints the lemma Et requiscere, confirmed on a tight crop, and our TEI agrees. The shape invites [sic:] — the ordinary form is requiescere, and quiescere faciet arrogantiam daemonum prints eleven lines later at 1252C. No marker was fired, and the reason is a witness outside this plate: our own Clementine flat file prints et quiscere faciam superbiam infidelium at Is 13:11, the very verse this lemma quotes. A syncopated -quiscere stem therefore has a life beyond this compositor, and Pattern 9 (attested spellings are not Pattern 7 defects) governs. The English renders it plainly, "And to rest". Recorded here so a later reader meets the argument and not just the silence.

Vulgate divergences — the collation

Every lemma in the range was walked word by word against sources/vulgate/clementine-flat.txt. Divergences in wording took a [var:]; divergences that are only Migne's abridgment of a verse, or that English cannot show, are recorded here without a marker.

Marked [var:] (eleven lemmata, twelve markers):

The two that matter most to the argument on the page are 1248B and 1247B. Puer natus est nobis is the christological lemma of the chapter, quoted once as the address and once inside the exposition, and the gloss builds on nobis ("Non Judaeis, nobis"), not on the noun — so the substitution of Puer for Parvulus passes unnoticed by every check but this one. At 1247B ex Deo nati replaces the Vulgate's whole relative clause with a phrase out of John 1:13; the gloss that follows ("The other Prophets and their disciples") is indifferent to either reading, which is exactly why it survives.

Not marked — abridgment, not divergence. Migne routinely quotes a verse short, and an omission is not a variant reading:

• 1248C Is 7:11 Pete tibi signum a Deo for a Domino Deo tuo.
• 1252A Ps 67:27 Benedicite Domino de fontibus Israel for benedicite Deo Domino.
• 1253B Mt 6:12 Dimitte nobis debita for debita nostra, cited (Matth. V) for Mt 6.
• 1251B Is 11:10 Qui stat in signum for in signum populorum; 1255C Is 14:31 Ab aquilone enim fumus for fumus veniet.

Not marked — no English exponent. A one-letter mood or spelling variant that renders identically in English is recorded, not marked, because a [var:] a reader cannot see in the text beside it is noise:

• 1255B Is 14:29 Ne laeteris against Clementine Ne laetaris (subjunctive for indicative; both "Rejoice not").
• 1255C Is 14:31 qui effugiat against qui effugiet.

Not marked — conflation, wording exact. 1244A–B quotes Et regem Dominum exercituum vidi oculis meis: et volavit ad me unus as one lemma. That is the tail of Is 6:5 followed immediately by the head of Is 6:6, each verbatim; the join is Migne's, the words are the Vulgate's. 1250D Ego lux et vita, ascribed to Christ, is likewise a fusion — of Jn 8:12 Ego sum lux mundi and Jn 11:25 / 14:6 ego sum ... vita — and belongs to no single verse to be collated against.

Readings carried through under Pattern 7

1254C — the double negative nisi ... non. Migne prints quod adhuc mansisset, nisi superbia tua non fuisset, which says the opposite of the sentence's own argument (the kingdom fell because of the pride). Both particles are carried: "which would still have remained, unless thy pride had not been." Confirmed on the plate; conjecture, nisi superbia tua fuisset, is not in the text by design.

1254B — Invitus Christus; descendit voluntarius. The plate's semicolon falls after Christus, so the printed sentence reads "Christ unwilling; he came down willing," which attributes the unwillingness to Christ in a gloss whose whole point is the contrast with Lucifer, who is drawn down against his will. Punctuation follows the plate (Pattern 8); the English keeps the printed break. Conjecture: Invitus; Christus descendit voluntarius — "[thou art drawn down] unwilling; Christ came down willing."

1253D — an infinitive with no expressed subject. Nos pro imbecillitate sensuum putabamus non posse resistere potentiae Dei, et te solum in altitudine tua permanere. The second accusative-and-infinitive has te; the first has no subject at all, and supplying one decides who could not resist whom. The English keeps it impersonal — "thought that it was not possible to resist the power of God" — rather than choosing.

1251B — the seized asp. Id est, divinam potentiam in comprehensum exinde aspidem vel regulum, id est, diabolum, tradet captivum. The in + accusative will not attach to tradet, and divinam potentiam is left without a governing verb. Rendered as closely as the words allow, with the agency the Latin does fix: the subject that delivers up is God, and what is delivered up captive is the asp. No word supplied.

1246D — nullius est enim. Ad Spiritum sanctum ... quae Graece neutri generis pneuma, Hebraice feminini Ruach, Latine masculini generis est, et ideo diversis positionibus sortitur, nullius est enim. The last clause is elliptical past construing — the sense wanted is that the Spirit belongs to no one gender — and sortitur has no object. Rendered literally.

1244D–1245A — Et pro, id est, adhuc. Et adhuc. Et pro, id est, adhuc ex ea gente decimatio. Pro stands with no complement; the gloss is telling the reader what the lemma's Et stands in place of. Rendered "And in place of, that is, still out of that nation a tithing," which keeps the ellipsis visible.

Not defects — recorded so a later reader does not report them

Migne's 3 glyph. Throughout this range the VERS. numeral 3 prints in a shape that reads as a 5 at native resolution (VERS. 3 Et clamabant at 1243C, VERS. 3 Derelictus est at 1245A, VERS. 3 Ad prophetissam at 1246D). The verse numbers in our TEI are correct against Isaias throughout; nothing here is a divergence.

Genezareth / Genesareth. Both spellings stand within four lines at 1247C–D, and the plate prints them exactly so. Not normalized.

Siole at 1247A. Migne's own spelling of Siloe, confirmed on a top-strip crop of the column. Covered by the [var:] on that lemma; no [sic:] — a metathesized proper name is not the same claim as broken type.

δορκὰς at 1252C. Read at native resolution and again at 3×: grave accent over the alpha, exactly as our TEI has it. Reproduced verbatim, untranslated.

Span count

The brief data/briefs/8956-lemmata-0005.txt declares 303 marked spans for chunks 0005–0009. Counted paragraph-bounded in my own Latin: 47 + 55 + 57 + 72 + 72 = 303, and the unbounded count is the same number, because this work sets no guillemets at all (0 « and 0 » across all five chunks) and therefore has no unmatched-delimiter failure mode in either direction. The two counts agree for a reason, not by cancellation.

Stint 0010

Stint file; the orchestrator merges it with the peers'. Every column in the range was read at the plate before these entries were written — Gallica ark:/12148/bpt6k5505319w f628–f634, f = (column + 1) / 2, calibrated against the printed corner numbers on all seven leaves; the archive copy patrologiaecurs04migngoog (PDF page = (column + 11) / 2, calibrated on p.636 and p.638) was brought in as second witness at cols 1258, 1262 and 1265. Reads recorded in data/briefs/8956-PLATE-READS-0010.json.

Span count: 278 marked spans in the Latin of chunks 0010–0014, against the brief's 278. Counted both whole-file and paragraph-bounded; zero paragraphs carry an odd number of asterisks and the range contains no guillemets at all, so the two counts cannot have cancelled.

Plate vs. our TEI

1260C — destrueret, not destruerat. Patch filed (data/briefs/8956-PENDING-TEI-PATCHES-0010.md §1). The English renders the plate's subjunctive.

1262D — ingruat ei, not ingruat et. Confirmed on both printings. Patch filed (§2). The English gives the verb its indirect object: "what may fall upon him after the table."

1267A — an intrusive full stop, mercedes. ejus. The plate breaks the word across the line (merce- / des) and prints no stop. Patch filed (§3).

1265A — potet / potest: the two printings differ, and NO marker is fired. Gallica sets stare non potet (read at 3.6×, terminal -et unmistakable) and our TEI agrees; the archive copy sets potest clean. A word defective on one printing only does not license a [sic:] against Migne's type, so the English reads "cannot stand" and nothing is patched. Logged as a press variant, same class as 8961 @0317D and 8950's Mititte/Mittite.

1262A — expugnat / expugnant. The archive copy sets expugnant; Gallica and our TEI set expugnat. Press variant or jbig2 artifact on the lossy copy; nothing marked, nothing patched.

Migne's foot-of-page apparatus

None in this range. The feet of all fourteen columns (f628–f634, both columns of each leaf) were rendered to the bottom edge of the leaf and are blank of any numbered, asterisked or lettered note. Col 1258 carries the signature 40 and nothing else. This is a checked zero for cols 1255A–1268D on the Gallica copy; it is not a claim about the rest of the work.

Scripture: Migne against the Clementine

The whole divergence list for the range, in column order. [var:] was fired where the divergence bears on the sense or on the gloss built upon it; the rest are recorded here only.

Fired [var:]

Recorded, no marker

• 1256C — Aqua Dibon (Is 15:9) for Clementine aquae Dibon: singular for plural, and the gloss glosses the city name, not the water.
• 1258A — Auferetur laetitia et exsultatio de Carmelo (Is 16:10) for Et auferetur…: only the connective dropped. But the citation is (III Reg. XVIII), which is Elijah on Carmel and not the verse quoted; reproduced verbatim per the refDisplay rule.
• 1258C — Mercenari. (Is 16:14, quasi anni mercenarii). The clipped form stands on both printings, so it is Migne's own type and not a digitization loss. Read as an abbreviated lemma — this commentary clips lemmata constantly (Ecce iste venit vir, Cingulum ultra tibi, Transite, ululate, qui habitatis) — and rendered "Of a hireling." No [sic:]: the reading is Migne's abbreviation, not broken type, and the Latin twin keeps the clip visible.
• 1258D — Emarcescet (Is 17:4) for marcescet: prefixed. The gloss's own closing verb is marcescet, unprefixed.
• 1259A — Messis quidem multa operarii, autem pauci (Mt 9:37): the same words as the Clementine, differently pointed (multa, operarii autem pauci). Punctuation follows the plate.
• 1259A — duarum aut trium (Is 17:6) for duarum vel trium.
• 1260A — super eum hyemabunt (Is 18:6) for super illum hiemabunt.
• 1261A — Principes Thaneos (Is 19:11) for principes Taneos: an aspirated form of Tanis. Recorded because this range is dense in proper names and Migne's spellings of them are exactly where a divergence hides; the sense is untouched, so no marker.
• 1263A — Ecce iste venit vir (Is 21:9): Migne clips ascensor out of Ecce iste venit ascensor vir bigae equitum. An abridged lemma; the gloss then expounds super bigam sedentem.
• 1263C — Occurentes (Is 21:14) for Occurrentes: one r. Attested as a spelling; no marker (Pattern 9).
• 1264A — Quidnam tibi quoque (Is 22:1) for Quidnam quoque tibi est: word order and the copula dropped.
• 1264B — Elam (Is 22:6) for Aelam, twice, where the same chapter's LXX gloss at 1262C spells it Aelam.
• 1264D — Comedamus et bibamus, cras enim moriemur is the Clementine's Is 22:13 word for word, but is cited (Jerem. XXXIV). Reproduced verbatim.
• 1265A — Coronans coronabit te (Is 22:18). No claim is made in either direction: our own sources/vulgate/clementine-flat.txt reads Coronas cornonabit te at that verse, which is not Latin, so the file is corrupt at exactly the place a collation would want it. The English renders Migne; a later pass with a sound Clementine should revisit this line.
• 1266A — Transite, ululate, qui habitatis (Is 23:6): Migne clips maria and in insula.
• 1266A — Nunquid non haec vestra est (Is 23:7) for Numquid non vestra haec est: order.
• 1266B — Contaminatione contaminabit eum — a Hebraism quoted from the leprosy law; no single Clementine verse carries the phrase, so no received reading can be named.
• 1266B — dum recordaremur Sion (Ps 136:1) for cum recordaremur.
• 1266B — Cingulum ultra tibi (Is 23:10): the lemma clips the Vulgate's non est. An abridgment, not a dropped negation — the gloss expounds the girdle, not its absence.

Readings and constructions worth recording

• 1256B — Expediti Moab ululabunt, glossed Ab uxoribus. The gloss reads the Vulgate's expediti ("the ready men") as free of wives, and then proves it from Abraham's vernaculos… expeditos. Rendered "Free of wives, brave warriors" so that the gloss's own inference stands; the ablative is one of separation, not of agent.
• 1257D — Domini ergo gentium apostoli et apostolici viri vineae Sabama exciderunt flagella. Domini gentium is the nominative plural subject lifted straight out of Is 16:8 (vineam Sabama domini gentium exciderunt), in apposition with apostoli: the lords of the nations, that is the apostles, cut off the shoots. Not a genitive depending on apostoli.
• 1258B — et Jacobus, cui una deest, desunt omnes. James 2:10 compressed: the dative cui governs both clauses, so "to whom one is lacking, to him all are lacking." Rendered with the dative carried through rather than smoothed into a nominative.
• 1258B — Ad sanctam, scilicet Ecclesiae, quae conabitur sua facere, non praevalebit. The construction does not resolve cleanly: ad sanctam is left hanging, and the subject of praevalebit is Moab (feminine) while quae has sanctam nearest. Rendered as literally as the printed words allow, negation intact; both printings agree, nothing is emended.
• 1259C — Non recipiens Patrem, quia non recipis Filium; qui enim credit in Patrem, credit in Filium. The plate moves from participle to second person singular within the clause (recipiens … recipis). Both printings agree. Rendered as printed, with thou for the singular address (Pattern 17).
• 1264D — unde ei Dominus haec mala minatur: the chunk's last sentence carries no terminal stop on the plate. The English leaves it unstopped (Pattern 8).
• 1265B — id est crucem possumus intelligere. Migne italicizes only crucem possumus, which cuts across the clause; the italic boundary is reproduced where he sets it rather than tidied to the phrase.

Stint 0015

Stint-scoped. The orchestrator merges these into the work's cruces.md.

All seven leaves in this range (Gallica ark:/12148/bpt6k5505319w f634–f640, f = (column + 1) / 2, calibrated against the printed corner numbers before any reading) were read at the plate and the running text collated word by word against our Latin. The foot of every column in the range is blank — this book sets neither the numbered nor the asterisk conjecture layer here. Three digitization losses were found and are filed in data/briefs/8956-PENDING-TEI-PATCHES-0015.md.

The work sets no guillemets at all in these five chunks (0 «, 0 »), as the peer stint reported for its own range: every lemma is italic. Span count reconciles exactly — 327 marked spans in the Latin against the brief's 327, paragraph-bounded (52 + 76 + 67 + 74 + 58).

Plate defects marked in the English

1280B — velmors, a fused run. The plate sets Non infernus velmors. as a single token with no word space, verified at native resolution; our TEI has it fused too, so the defect is Migne's and not ours. Marked [sic: velmors] under Patterns 10 and 12 (the whole run is carried, including the real word vel). Per 7a⁗ the marker does not swallow what the run governs: the disjunction and the noun are both said in plain words beside it — "Not hell nor death [sic: velmors]" — since the clause has no other predicate and the gloss is glossing Vivens vivens against infernus and mors together. Conjecture: vel mors.

Migne's own citations, left as printed

refDisplay keeps what he prints; none of these is corrected, and none takes a marker.

• 1267C — the Lazarus parable is cited (Luc. XIX); it is Lc 16.
• 1268D — Mt 25:46 is cited (Matth. XIII); the verse is Mt 25.
• 1269C — Omnis qui se exaltat humiliabitur is cited (Luc. I); it is Lc 14:11 and 18:14.
• 1274B — Quia justi fulgebunt sicut sol is cited (Matth. XIII); correct, Mt 13:43, but the opening word is Migne's (see below).
• 1276C — Habitabis in loco isto amodo et usque in sempiternum is cited (Zach. XI); the words stand nowhere in Zacharias, and the nearest scriptural home is Jer 7:7.
• 1277C — Erit Saron sicut desertum is cited (Isa. XXXII); it is Is 33:9.

Vulgate divergences too small to fire [var:]

Every one collated against sources/vulgate/clementine-flat.txt and confirmed on the plate.

• 1267B — Is 24:1 quoted as Ecce Dominus nudabit; the Clementine reads Ecce Dominus dissipabit terram : et nudabit eam. An abridged incipit, not a variant reading.
• 1268A — Is 24:16 printed A finibus terrae, laudes audivimus; the Clementine has no comma after terrae. Punctuation only, kept as printed (Pattern 8).
• 1268D — Mt 13:30 quoted Alligate ea per fasciculos ad comburendum; Clementine in fasciculos.
• 1268D — Eph 6:12 quoted as Rectores tenebrarum, contra spiritualia nequitiae in coelestibus, abridged from adversus mundi rectores tenebrarum harum.
• 1269D — Gal 5:17 quoted without enim (Caro concupiscit adversus spiritum).
• 1271C — 1 Cor 3:2 quoted Lac dedi vobis potum, non escam; Clementine lac vobis potum dedi. Word order only.
• 1271D — Ps 117:22 quoted Lapidem quem aedificantes reprobaverunt; Clementine quem reprobaverunt aedificantes. Word order only; the verse is set inside the Is 28:16 lemma with no citation of its own, which is worth noting as a conflation even though the words are otherwise exact.
• 1272B — Is 29:3 printed in circumitu tuo for in circuitu tuo. Our TEI and the plate agree; an attested spelling, exempt under Pattern 9.
• 1273D — Is 30:21 printed verbum monentis post tergum; Clementine verbum post tergum monentis. Word order only, and the gloss turns on post tergum, which both have.
• 1274A — Is 30:24 printed Commixtum migma; Clementine commistum migma. Spelling.
• 1274B — Mt 13:43 quoted Quia justi fulgebunt sicut sol; Clementine Tunc justi fulgebunt. The connective has been adapted to the host sentence.
• 1274C — Ex 15:1 quoted Cantemus Domino, gloriose enim, and cut there, as the plate cuts it.
• 1275A — Ora pro nobis Dominum quid agere debeamus? is set in italic as though quoted, but the words are a paraphrase of Jer 42:2–3 and stand nowhere verbatim. Rendered as printed.
• 1275C — Is 32:10 quoted Post dies et annum, dropping the Clementine's enim.
• 1279B — Is 38:1 printed audivisset for the Clementine's syncopated audisset; and the plate really does set the comma inside Et factum, est cum audivisset. Both kept (Patterns 7 and 8).
• 1280D — Mt 19:12 quoted qui se castraverunt; Clementine qui seipsos castraverunt.
• 1280D — Ex 32:32 quoted Dimitte eis hanc noxam, aut dele me de libro quem scripsisti, dropping tuo and the conditional aut si non facis.

Forms rendered as printed, no marker

• 1270B — distingunt for distinguunt (Migne, Alii distingunt sic). A real form that occupies its slot and renders in place: "Others punctuate thus." Pattern 9 / the fuges–munde boundary.
• 1270B — the catchword Spiritum: salute, repeats the lemma peperimus spiritum, salutem non fecimus two lines above, but with the ablative salute for the accusative salutem. A bare catchword has no syntax to violate and Latin case has no English exponent, so the English reads "Spirit: salvation," with no marker; the divergence is recorded here. Both readings are on the plate.
• 1275D — sicut In Pentecoste, with a capital I mid-sentence. The plate sets the capital; kept.
• 1274A — the italic run is split by Migne himself as usque et ad, two spans with a roman et between them. Not our markup and not a defect: the plate sets exactly that, and the English mirrors it as "as far and as".
• 1274D — the chapter head prints CAPUT XXXI with no full stop, unlike every other head in the range. Confirmed on the plate; the frontmatter's heads array already carries it unstopped.
• 1278D — the paragraph on VERS. 33 ends quanta de imminentibus cura with no terminal stop. Confirmed on the plate. The English ends unstopped (Pattern 8: never supply).

Column-band and structural notes

Nothing in this range breaks mid-word or mid-construction across an anchor. The band sequence in chunk 0018 runs 1275A → 1276D → 1277A → 1278A with the usual positional irregularity, which is not a gap (known false positive 1).

Stint 0020

Stint file; the orchestrator merges it with the peers'. Every column in the range was read at the plate before these entries were written — Gallica ark:/12148/bpt6k5505319w f641–f646, f = (column + 1) / 2, calibrated against the printed corner numbers on all six leaves (f641 = 1281/1282 … f646 = 1291/1292) before any read was recorded; the archive copy patrologiaecurs04migngoog (PDF page = (column + 11) / 2, calibrated on p.646 and p.651) was brought in as second witness at cols 1282, 1283, 1284, 1286, 1287, 1290 and 1291. Reads recorded in data/briefs/8956-PLATE-READS-0020.json.

Span count: 253 marked spans in the Latin of chunks 0020–0024, against the brief's 253 (50 · 57 · 44 · 58 · 44). Counted both whole-file and paragraph-bounded; no paragraph in the range carries an odd number of asterisks and the range contains no guillemets at all, so the two counts cannot have cancelled. This confirms for cols 1281A–1292A what two peer stints established: this work sets no guillemets, and every lemma is italic.

Plate vs. our TEI

1282A — Migne sets HEBREW type that our TEI drops. The plate reads Hebraice mesuchan (מסכן); our twin reads Hebraice mesuchan , quia, the orphaned space being the scar. Confirmed on both printings. Patch filed (data/briefs/8956-PENDING-TEI-PATCHES-0020.md §1). The Hebrew was not inserted into the English alone.

1281B — VERS. 4-- for the plate's VERS. 4. —. A lost point in a verse address. Patch filed (§2); the English prints VERS. 4.--.

1290A — (I Cor.. X) for the plate's (I Cor. x). A doubled point inside a citation, confirmed on both printings. Patch filed (§3). The note is reproduced verbatim as our twin has it until the patch lands.

1291A — docem for decem. Both printings read decem: ten tribes. Patch filed (§4); the English renders the plate.

1291C — plerosquo for plerosque. Both printings read plerosque, the accusative subject of tentari. Patch filed (§5).

The two printings differ — and no marker is fired

Five sites. In each our TEI agrees with Gallica and the archive copy is clean; a word defective on one printing only does not license a [sic:] against Migne's type (the 8956 @1265A potet/potest precedent, 8961 @0317D, 8950's Mititte/Mittite).

1283A — tunduntur.I / tunduntur. Gallica sets a stray capital-I-shaped sort after the full stop that closes quae salsis et amaris fluctibus saeculi tunduntur.; archive p.647 prints the stop clean and nothing after it. Read at 4× on Gallica: the mark has serifs top and bottom and sits on the baseline, so it is a sort, not an ink blot — but it is on one copy only. The English ends the sentence and carries nothing.

1284D — vestustate / vetustate. Gallica sets In novitate Spiritus, non in vestustate litterae; archive p.647 sets vetustate, the Rom 7:6 word. English: "not in the oldness of the letter."

1286B — nonhabet / non habet. Gallica fuses the run in rem autem nonhabet; archive p.648 sets the space. This is the dangerous shape (7a⁗-b): the fused lexeme carries the negation. Since no marker is fired, nothing quarantines it, and the English says the negation in plain words: "but does not have the thing itself."

1290A — ob oriente / ab oriente. Gallica sets et quem Magi ob oriente venientes adoraverunt; archive p.650 sets ab. English: "and him whom the Magi, coming from the east, adored."

1290C — Curavivimus / Curavimus. Gallica hyphenates the Jer 51:9 lemma as Curavi- / vimus Babylonem; archive p.650 sets Curavi- / mus. Before recording this the archive column's left margin was measured against the three neighbouring lines (volenti, Babyloni, et quasi), all flush with mus, so the archive really lacks the syllable rather than having it cropped. English: "We have healed Babylon, and she is not healed."

Carried type — the one [sic:] in the range

1287B — Super aedificati, confirmed on both printings. The Ephesians lemma is set with the word divided mid-line, Super aedificati super fundamentum apostolorum et prophetarum, where Eph 2:20 reads superaedificati. Gallica f644 and archive p.649 agree, so the defect is Migne's and the marker stands. Per Pattern 10 the whole run is carried, including the real word Super, and per the 2026-09-05 amendment the sense is rendered beside it; the marker takes no second italic marker because it sits inside an already-italic lemma.

Vulgate divergences — the deliverable

Seventeen [var:] fired. Collated word by word against sources/vulgate/clementine-flat.txt; where a lemma is quoted twice, both occurrences were collated.

1281A — Zach. 8:7–8. Migne: Ego congregabo populum orientis, et erunt mihi in populum, et ego ero illis in Deum. Clementine: Ecce ego salvabo populum meum de terra orientis … et ego ero eis in Deum. The verb of the proof-text is changed (congregabo for salvabo) in a gloss adduced for the gathering of the Gentiles. [var:] fired.

1281A — Gen. 34:3, cited (Gen. XXIV). Migne: Locutus est Sichen ad cor ejus, et consolatus est eam. The Clementine has no such sentence anywhere; at Gen. 34:3 it reads tristemque delinivit blanditiis. The gloss needs the words ad cor because it is expounding Loquimini ad cor, so the quotation is doing real work in a form the Vulgate does not carry. [var:] fired. The citation (Gen. XXIV) for Gen. 34 is Migne's and is reproduced verbatim — the third mis-citation class a peer already recorded for this work.

1281D — the VERS. 13 lemma is a conflation, and it loses its agent. Migne prints Spiritum Domini super me, eo quod unxerit me ad annuntiandum mansuetis misit me: Et requiescet super eum Spiritus Domini under the address for Is 40:13. The first half is Is 61:1 without Dominus (eo quod unxerit Dominus me), the second is Is 11:2. Dropping Dominus removes the anointer from a first-person servant-song verse. [var:] fired naming the Vulgate's agent.

1282B — Amos 3:6. Migne: Si est malum in civitate quod non faciat Dominus. Clementine: si erit malum in civitate, quod Dominus non fecerit? Tense and mood differ and Migne prints a full stop where the Vulgate has the interrogative; the negation is in the same place in both, and the sense of the proof (God makes the malum that is affliction) is unchanged. Crux, no marker.

1282B — Is 40:25. Migne: Cui assimilastis me, adaequastis me, dicit Sanctus. Clementine: Et cui assimilastis me, et adaequastis? dicit Sanctus. Migne supplies the object to the second verb and drops both et, and closes with a stop where the Vulgate has a question mark. The gloss answers it as a question anyway (Quasi diceret: Quare…?). Crux, no marker.

1282D — Rom. 11:33. Migne: Incomprehensibilia judicia ejus, et investigabiles viae ejus. Clementine: quam incomprehensibilia sunt judicia ejus. An abridgment that turns an exclamation into a pair of adjectives. Crux, no marker.

1283B — 1 Cor. 15:31. Migne: Quotidie morior propter vestram gloriam, fratres. Clementine: Quotidie morior per vestram gloriam. propter makes the Corinthians' glory the cause of Paul's dying, per makes it the ground of his oath; the gloss adduces the verse for the apostles as worms who die daily. [var:] fired.

1283A — Is 41:1. Migne's lemma Taceant insulae drops the Vulgate's ad me, so the islands are silent at large rather than silent before God. Crux, no marker — the gloss expounds insulae, not the addressee.

1283B — Ps. 21:7. Migne quotes Ego sum vermis et non homo without autem. Crux.

1285A — Is 42:11, the lemma quoted twice, and only the second diverges. The verse address is De vertice montium (plural, with the Vulgate); the gloss then re-quotes it three words later as De vertice montis, singular, immediately after Super montem excelsum ascende (Is 40:9). The gloss's own argument — the height of virtue and contemplation, one summit — runs on the singular. [var:] fired on the second occurrence.

1284C — Jo. 9:39. Migne: In judicium ego veni in hunc mundum. Clementine: In judicium ego in hunc mundum veni. Word order only. Crux.

1285A — Rom. 11:25. Migne: Cum plenitudo gentium intraverit. Clementine: donec plenitudo gentium intraret. Crux.

1285D — Matt. 8:11. Migne: Multi venient ab Oriente et Occidente, et recumbent cum Abraham, Isaac et Jacob, ab omnibus scilicet mundi partibus. Clementine: multi ab oriente et occidente venient, et recumbent cum Abraham, et Isaac, et Jacob in regno caelorum. Migne's italic run swallows the gloss's own explanatory tail in place of the Vulgate's in regno caelorum, so a reader collating the printed quotation meets a phrase the Gospel does not have. [var:] fired.

1286A — Is 43:14. Migne's lemma: Haec dicit Dominus Salvator. Clementine: Haec dicit Dominus, redemptor vester, Sanctus Israel. The same author prints Redemptor correctly at 1287B for Is 44:24, which is what makes this one visible. [var:] fired.

1286C — Is 5:4. Migne: Quid ultra debui facere tibi, et non feci? Clementine: Quid est quod debui ultra facere vineae meae, et non feci ei? The addressee has changed from the vineyard to a second-person thee, which is precisely what the gloss on Reduc me needs. [var:] fired.

1286C — Gen. 15:8. Migne: In quo sciam quia possidebo eam? Clementine: unde scire possum quod possessurus sim eam? An entirely different wording of Abraham's question. [var:] fired; no claim is made about which recension it comes from.

1288B — Is. 5:6. Migne: Mandabo nubibus meis ne pluant super eam imbrem. Clementine: et nubibus mandabo ne pluant super eam imbrem. Migne adds meis, and the gloss's whole point is that the clouds are God's own prophets. [var:] fired.

1288C — Is. 45:9, quoted as a proof of itself. Migne: Vae qui contradicit factori suo; Clementine: Vae qui contradicit fictori suo — fictor, the potter who moulds, not factor, the maker at large; the verse continues testa de samiis terrae and Numquid dicet lutum figulo suo. [var:] fired. The citation (Psal. XXXV) is Migne's and is reproduced verbatim: the quotation is Isaiah, and Ps. 35:6 is the Veritas tua usque ad nubes quoted a few lines earlier at 1288B, so the reference has slid down the column. No correction is proposed.

1288D — Rom. 5:20. Migne: ut ubi abundavit delictum, superabundet et gratia. Clementine: Ubi autem abundavit delictum, superabundavit gratia. Indicative to subjunctive, plus an intrusive et: Migne's form makes the abundance of grace the purpose of the abounding of sin, which is the sense the gloss on the monks of Egypt is reaching for. [var:] fired.

1289A — Exod. 20:3. Migne: Non erunt tibi dii alieni. Clementine Exod. 20:3: Non habebis deos alienos coram me. [var:] fired.

1289B — Is. 45:21, and the divergence is one of PERSON. Migne: Praedixi illud — first person, God speaking. Clementine: ex tunc praedixit illud — third. The gloss then continues in the first person (quando locutus sum Abrahae, vel quando legem dedi), so Migne's person is what the exposition rests on. [var:] fired.

1289B — the verse address at In Domino dicet is printed VERS. 25 and the words are Is 45:24. Migne's numbering; reproduced as printed. Crux.

1290A — 1 Cor. 10:11. Migne: Nos sumus in quos fines saeculorum devenerunt. Clementine: scripta sunt autem ad correptionem nostram, in quos fines saeculorum devenerunt. A recast that supplies a subject the verse does not have. Crux.

1290D — Matt. 2:1–2. Migne: Ab oriente venerunt magi dicentes: Vidimus stellam ejus in oriente. Clementine: ecce Magi ab oriente venerunt Jerosolymam, dicentes: … vidimus enim stellam ejus in oriente. A conflation of the two verses with enim dropped. Crux.

1291B — Act. 7:51. Migne: dura cervice et incircumcisi cordibus. Clementine: Dura cervice, et incircumcisis cordibus et auribus. Migne's incircumcisi is nominative where the Vulgate has the ablative of respect, and et auribus is gone; the gloss is expounding Nervus ferreus, the neck, so the case shift matters to the construction. [var:] fired. The citation (Act. XVII) for Acts VII is Migne's and is reproduced verbatim.

1291C — Is. 48:9. Migne prints the lemma longe faciam; feci: where the Vulgate has longe faciam furorem meum. The gloss then supplies the missing object itself (scilicet furorem meum), which is the tell that something is absent from the printed lemma; feci has no counterpart in the verse. [var:] fired.

1292A — Jo. 19:34. Migne: Unus militum lancea latus ejus aperuit, et continuo exivit sanguis nostrae redemptionis, et aqua baptismi. Clementine: et continuo exivit sanguis et aqua. The gloss's interpretation has been drawn inside the italic quotation, so the page prints as John's words a sentence that names redemption and baptism. [var:] fired.

Migne's own citations, reproduced and not corrected

Beyond the three named above, one more in the range:

1284B — the (Phil. II) / (Luc. I) pair. The plate attaches (Phil. II) to the quotation Spiritus sanctus superveniet in te (Lc 1:35) and (Luc. I) to the sentence about the dove at the baptism. Read at native resolution on Gallica f642: the plate prints them exactly where our TEI has them, so the apparent swap is Migne's, not the digitization's. Reproduced verbatim.

Migne's foot-of-page apparatus

None in this range. The feet of all twelve columns (f641–f646, both columns of each leaf) were inspected: no numbered notes, no asterisk conjectures. Col 1289 carries the running signature PATROL. CXIII. and col 1290 the sheet signature 41; both are press furniture and are neither translated nor marked.

Register notes

Second person. Pattern 17 was applied throughout: tu-addresses to Cyrus (1287D–1288A), to Jacob (1282C, 1286C), to Babylon as a woman (1290C–D) and to Jerusalem's citizen render thou; vos-addresses to the apostles (1284D, 1286A), to the Gentiles (1282B) and to the hardened Israel of ch. XLVIII (1291A–B) render you. The range switches between the two inside single chapters, so this was checked paragraph by paragraph rather than per chapter.

Agency. The range is thick with God speaking in the first person, and the readings above were formed by asking of every clause who acts on whom. Two places where the printed Latin fixes an agency an English reflex would invert, and where the English follows the Latin: 1290C Et non resistet, sicut, Moyses Deo volenti delere filios Israel — the resister is Moses, and God is the one willing to destroy, so the English reads "as Moses did to God when he willed to blot out the sons of Israel"; and 1288D Nec ipse aedificavit civitatem vel templum, sed permisit aedificari — Cyrus does not build, he permits, and both negations stand.

Etymological glosses kept in Latin. pugillus / palmus (1281D) and inter partes for interpretes (1286D) are the words the gloss is defining; an English equivalent would make the gloss explain nothing, so they stand in italic Latin beside their renderings.

Stint 0025

Stint file; the orchestrator merges it with the peers'. Every column in the range was read at the plate before these entries were written — Gallica ark:/12148/bpt6k5505319w, leaves f646–f653, f = (column + 1) / 2, calibrated against the printed corner numbers on all eight leaves (f646 = 1291/1292, f647 = 1293/1294, f648 = 1295/1296, f649 = 1297/1298, f650 = 1299/1300, f651 = 1301/1302, f652 = 1303/1304, f653 = 1305/1306) before any collation was attempted. Reads are recorded in data/briefs/8956-PLATE-READS-0025.json. The band letters print in the centre gutter and mark the RIGHT column on every leaf in this range, as the peers report.

The archive.org second printing (patrologiaecurs04migngoog, PDF page = (column + 11) / 2) was consulted at five sites, on pp. 652, 653, 654, 656 and 658. It is the better witness in this range, and it withdrew four defect claims against Migne — see "Four [sic:] withdrawn" below.

Span count reconciled: 352 marked spans in the Latin of chunks 0025–0029, counted paragraph-bounded (0025 = 87, 0026 = 61, 0027 = 69, 0028 = 59, 0029 = 76), against the brief's 352. Exact agreement. ⚑ 0 guillemet spans: this work sets no « or » at all, in this range as in the peers' — Luke-9000 shape, not Genesis-8950 shape. Confirmed by counting, not assumed.

Markers fired: 16 [var:], and nothing else. No [sic:], no [d:], no [ed:], no [cj:].

⭐ The finding of the stint: our own line-break fusions read as Migne's defects

Five sites in this range are our digitization mishandling a hyphen at a line or column break, and three of them presented as textbook plate defects. They are itemized in data/briefs/8956-PENDING-TEI-PATCHES-0025.md; the point worth carrying forward is what they looked like from our file alone:

The 1295D doubling is the dangerous one. Read from the twin it satisfies Pattern 11a exactly — two copies of a run, differing in one word, the first ending in the neighbouring phrase's shape — and 11a instructs marking the intrusion. There is nothing internal to our file that could have told that from Migne's own eye-skip. Only the hyphen on the plate could, and it says Migne set the word once.

Four [sic:] candidates withdrawn at the second printing

Each was a real non-word in the Gallica copy AND in our TEI, which is exactly the corroboration reference_plate-read-triage warns proves an error exists and never whose it is. Each was clean on the archive printing, so no marker stands.

• 1297B — tribuatus. Gallica f649 breaks tribu- / atus; archive p.654 breaks tribu- / latus. The English reads "lest the just man, being afflicted, fail."
• 1305B — lauaem. Gallica f653 prints the Is 61:11 lemma Et lauaem, an italic a where the d belongs — and laudis prints cleanly two lines below on the same page, which is precisely the shape that licenses a broken-sort claim (the 8955 @0629D ulit/Tulit precedent). Archive p.658 prints Et laudem.
• 1296D–1297A — funiculu. Archive p.653 prints funiculus; the Gallica final s is worn to a dot at 3×.
• 1293B — Matih.. The Mt 16:19 note prints (Matih. XVI) on Gallica, (Matth. XVI) on archive. Our twin has Matth. and is right; a note's content rides verbatim in any case.

⚑ Together with 1301A irruguus/irriguus, that is five places in fourteen columns where the Gallica copy carries a worn or broken sort the archive copy does not. This inverts the PL 114 finding, where the archive copy was the worse witness by twelve to one in Matthew's opening range. It is a fact about these two copies of PL 113 in this range, not a general ranking, and it is the reason no [sic:] in this stint could stand on Gallica alone.

The column anchor [1240D] at 1294D

Our twin prints [1240D] where the sequence runs 1294C → 1295A. The plate settles it as a digit corruption and not a misplacement: the D band on f647 stands in the centre gutter beside VERS. 9. — Consurge in the right-hand (1294) column, exactly where our twin sets the anchor. So the position is right and only the digits are wrong (9 → 4). The English reproduces [1240D] verbatim — anchors are sacred and verify-english checks them 1:1 — and the patch must move both twins in one operation.

⚠ Deliberately not reported as defects, per the standing false-positive list: the [1296D] that stands before [1296A] at the head of CAPUT LIII (a recurring column number and a D-band followed by an A-band are both normal), and the Ibid.-style [n: (ID.)] chains throughout.

Migne's own mis-citations — reproduced verbatim, six of them

All confirmed on the plate; none corrected, per the standing rule.

• 1295C [n: (Psal. XCVII)] on In brachio meo gentes sperabunt, which is Is 51:5.
• 1295C [n: (Matth. XVIII)] on Ite, docete omnes gentes, which is Mt 28:19.
• 1293D [n: (Joan. VIII)] on Qui sitit, veniat et bibat, which is Jo 7:37.
• 1294B [n: (Ezech. XVIII)] on nolo enim mortem peccatoris, which is nearest Ez 33:11.
• 1299C [n: (Isa. XL)] on Non sicut viae meae sunt viae vestrae, which is Is 55:8. ⭐ This one is doubly interesting: the quotation itself diverges (fired [var:]), and the chapter number is wrong, in a work whose own book is Isaiah.
• 1301B [n: (Psal. LXXXVI)] on Fundamenta ejus in montibus sanctis — correct, but the note numbers the Psalm by the Vulgate reckoning while the neighbouring [n: (Psal. XCVIII)] at 1293A does likewise; recorded so a later index pass does not read the pair as inconsistent.

Vulgate divergences — the 16 fired

All collated word by word against sources/vulgate/clementine-flat.txt, and every Clementine reading quoted below was checked to be sound Latin before the marker went in.

Chunk 0025 (6).

1. 1293A, Ps 98:5 — Adorabunt scabellum pedum ejus for Clementine adorate scabellum pedum ejus. ⭐ The richest shape on this book, and here it is exact: the lemma under exposition is Is 49:23 Adorabunt, and the proof-text adduced to gloss it has been conformed to it, imperative into future. The argument's keyword has been written into its own proof-text.
2. 1293B, Mt 10:14 — Excutite pulverem pedum vestrorum for excutite pulverem de pedibus vestris. The same shape again, on the same page: the lemma is Pulverem pedum tuorum (Is 49:23), and the proof-text now shares its exact genitive construction.
3. 1293C, Rom 6:12 — Non regnet peccatum in corpore vestro for Non ergo regnet peccatum in vestro mortali corpore. mortali is dropped.
4. 1293D, Jo 7:37 — Qui sitit, veniat et bibat for Si quis sitit, veniat ad me et bibat. ad me is dropped, which takes Christ out of a sentence quoted to show that he calls.
5. 1294B, Ez 33:11 / 18:32 — nolo enim mortem peccatoris, sed magis ut convertatur et vivat. The Clementine has nolo mortem impii, sed ut convertatur impius a via sua, et vivat (33:11) and nolo mortem morientis (18:32); Migne's peccatoris is in neither, and the note cites XVIII. This is the liturgical form of the verse, not the Vulgate's.
6. 1294B, Mt 25:41 — Ite, maledicti, in ignem aeternum for Discedite a me maledicti in ignem aeternum.

Chunk 0026 (5).

1. 1295D, Is 52:13 — Elevabitur et exaltabitur for exaltabitur et elevabitur. The two verbs are transposed in the hinge verse of the servant song, and this is Migne's own book. ⚑ Fired on the plate reading; our twin's exallabitur is our defect, patched separately.
2. 1296B, Jo 1:29 — qui tollit peccata mundi for qui tollit peccatum mundi. Singular into plural, in the sentence that names the Lamb.
3. 1296C, Lc 8:5 — Exiit qui seminavit semen suum for Exiit qui seminat, seminare semen suum. Present participle into perfect, and the infinitive of purpose dropped. The gloss is arguing that the Gospel word does not grow old, so the tense is not idle.
4. 1296C, 1 Pet 2:24 — Peccata enim nostra portavit in corpore suo for qui peccata nostra ipse pertulit in corpore suo super lignum. A different verb, ipse dropped, and the tree dropped — in a chapter on Is 53.
5. 1296C, 1 Cor 9:22 — Omnibus omnia factus sum, ut omnes lucrificarem for ut omnes facerem salvos. lucrifacerem is the verb of 9:19 and 9:22a; the two halves of Paul's sentence have been spliced.

Chunk 0027 (3).

1. 1297D, Eccl 11:1 — Mitte panem tuum super aquas, quia in multitudine dierum invenies eum for super transeuntes aquas, quia post tempora multa invenies illum. Two divergences in one verse; the gloss then expounds panem doctrinae tuae.
2. 1298C, Rom 12:1 — Exhibeatis corpora vestra hostiam Deo viventem for hostiam viventem, sanctam, Deo placentem. ⭐ The gloss immediately distinguishes the holocaust (offered whole to God) from the victim (part given to the priests), so Deo viventem — "a living victim to God" — is precisely the phrase the argument turns on, and it is not the Vulgate's.
3. 1299C, Is 55:8 — Non sicut viae meae sunt viae vestrae for Non enim cogitationes meae cogitationes vestrae, neque viae vestrae viae meae. Migne recasts the verse into the shape his gloss needs (many ways of worldly men against one way of truth) and cites the wrong chapter.

Chunk 0028 (2).

1. 1301A, Is 58:11 — Et erit quasi hortus irriguus for et eris quasi hortus irriguus. ⭐ The person changes, and the gloss's own next words are a vocative (o fidelis vel Ecclesia) that only the second person answers; two lines later Assimilaberis Christo is second singular again. Migne's third person stands in the English, with the marker beside it.
2. 1302C, Is 11:2 — Requiescit super eum Spiritus Domini for Et requiescet super eum spiritus Domini. Future into present, adduced to prove that the Spirit rests on Christ now.

Divergences examined and DECLINED as immaterial

Recorded so a later reader can see they were weighed, not missed.

• 1298B, Jo 15:19 — Si de mundo essetis for fuissetis; and idcirco odit vos mundus for propterea. Mood and synonym; the sense is untouched.
• 1298B, Rom 10:12 — Sed non est distinctio for Non enim est distinctio. Connective only.
• 1299C, Jo 14:6 — Ego sum via, veritas et vita for Ego sum via, et veritas, et vita.
• 1301B, 1 Cor 3:11 — Fundamentum aliud nemo potest ponere drops enim.
• 1301A, Jo 12:26 — Ubi ego sum for ubi sum ego. Word order.
• 1303A, Jo 4:35 — quia jam albae sunt ad messem for quia albae sunt jam. Word order.
• 1303A, Eph 2:13 — Vos qui eratis longe drops aliquando.
• 1303C, Rom 10:10 — Corde creditur drops enim.
• 1304D, Lc 21:20 — cum videritis ab exercitu circumdari Jerusalem for Cum autem videritis circumdari ab exercitu Jerusalem. Word order and connective.
• 1293C, Jo 8:34 — servus peccati est for servus est peccati. Word order.

Pattern 8 — three sentences that stop without a stop, and one that keeps a question

Confirmed at the plate in each case; the English carries the plate's punctuation exactly and supplies nothing.

• 1297D — the column ends quia in novissimo praemia recipies with no terminal stop on Gallica f649. The English ends "thou shalt receive rewards", unstopped.
• 1300A — opere complevi, unstopped on f650. So in the English.
• 1302B — audita praedicatione apostolorum, unstopped on f651.
• 1295B — ut etiam in terra sua opprimat illum, et calumnietur ei? Migne closes an usque ad abridgment tail with a question mark, and the English keeps it as a direct question, which the tail already is (Pattern 8b is not needed here).
• 1302C — chunk 0029's Latin ends mid-sentence on a comma (Spiritus meus, qui est in te,). That is our chunk boundary, not the plate's; the English ends on the same comma.

Odd readings rendered as printed, with no marker

• 1298B — Ne sit, scilicet de sex diebus quibus factus est mundus. The gloss on Custodiens sabbatum is elliptical past construing; Ne sit has no expressed subject and the de phrase hangs. Rendered literally ("Let it not be, that is, of the six days in which the world was made") and confirmed word for word on f649. Nothing is defective — it is simply compressed — so no [cj:]; the bar is that the English asserts something false, and an ellipsis asserts nothing.
• 1297A — victis perditionis, victoribus causa est praemii. Two datives against one predicate nominative; the plate reads exactly so on f649. Rendered "the cause of perdition to the conquered, of reward to the conquerors."
• 1300C — cui charitatem Deo invidebat. Two datives again, and the sense turns on which is the person grudged. Rendered "to whom he grudged charity towards God" — Cain is the actor, his brother the person, the charity the thing. Recorded because this is exactly the agency seam that a smooth English can silently re-assign.
• 1304C — Cum miserit Filius hominis angelos, et colligent… A cum clause answered by et + future indicative rather than a main clause. Rendered as printed, conjunction kept.
• 1302B — Et salvavit sibi. Pater, scilicet per Filium gentes. The gloss supplies subject, agent and object as three bare appositions. Rendered in that order without recasting: "The Father, namely, through the Son, the Gentiles."

Second person — Pattern 17 as applied here

The range moves between a singular and a plural addressee repeatedly, and the English marks it. Singular → thou: the Church at 1302D–1303A (Quos in fide genuisti, nata es), the soul at 1301A–1301B (Assimilaberis Christo, capiti tuo; carnem tuam; domesticos seminis tui ne despexeris), Jacob at 1293A (fortitudinem Jacob luctandi), the idolater at 1299B (idola adorasti). Plural → you: the Jews addressed throughout CAPUT LV (Quare appenditis argentum, Qui solam speciem litterae custoditis, Ne putetis), and CAPUT LIX's Sed iniquitates vestrae. No instance in the range required a judgment call: the morphology decided every one.

Stint 0030

The last stint of the work: Isa. LXII–LXVI, to the end of the book. Stint-scoped; the orchestrator merges this into cruces.md.

Span count: the brief lists 248 spans for chunks 0030–0034; my own Latin gives 248 whole-file and 248 paragraph-bounded, with 0 « and 0 » — so the two counts agree for the reason 8956-CONVENTIONS.md §2 gives (this work sets no guillemets at all), not by two errors cancelling.

Plates: all six leaves read, printed corner numbers checked first on every one — Gallica f653 (1305/1306), f654 (1307/1308), f655 (1309/1310), f656 (1311/1312), f657 (1313/1314), f658 (1315/1316). Second witness archive.org pp. 658, 659, 661, 662, 663. Details in data/briefs/8956-PLATE-READS-0030.json; our-side losses in data/briefs/8956-PENDING-TEI-PATCHES-0030.md.

1. Vulgate divergences — the collation, in column order

Every lemma and every quoted proof-text in the range was collated word by word against sources/vulgate/clementine-flat.txt. Twenty-six [var:] markers fired, over twenty-four distinct sites (Lc 12:49 fires at both of its occurrences, and Is 66:1 is noted at 1307D against its clean occurrence at 1312A). The Clementine was sound Latin at every site touched here; the cornonabit defect at Is 22:18 does not fall in this range.

Fired as [var:]

Collated, divergent, and NOT marked — crux only

• 1305D Eph 5:31 Propter hoc relinquet homo patrem et matrem, cited (Gen. II; Matth. XIX). The wording is Ephesians 5:31 exactly (minus suam); Gen 2:24 has relinquet but not Propter hoc, Mt 19:5 has Propter hoc but dimittet. Migne's citation stands verbatim.
• 1306B Jo 6:56 Caro mea vere est cibus for Caro enim mea… — a dropped particle.
• 1310A Rom 9:29 quasi Sodoma for sicut Sodoma.
• 1310B Rom 8:15 spiritum adoptionis filiorum Dei, cited (Rom. IX) for Rom 8.
• 1311A Jo 14:6 Ego sum via, veritas et vita for via, et veritas, et vita.
• 1311C Pr 3:18 qui apprehenderunt eam for apprehenderint.
• 1311C Ps 125:6 quia venient cum exsultatione for Venientes autem venient.
• 1312C Jo 9:22 conspiraverant (pluperfect) for conspiraverunt, and confiteretur Christum for eum confiteretur esse Christum.
• 1312C / 1314C 1Cor 5:7 quoted twice: at 1312C without Etenim, at 1314C with it, matching the Clementine. Only the first occurrence diverges, and nothing in either gloss turns on the particle.
• 1313A Ps 64:6 Spes omnium finium terrae et in mari longe — verbatim, cited (Psal. LXXIV).
• 1313A Rom 8:35 Quis nos separabit for Quis ergo nos separabit.
• 1314D Mt 28:19 Ite, docete omnes gentes for euntes ergo docete.
• 1315B 1Cor 3:17 Templum Dei sanctum est for Templum enim Dei sanctum est.
• 1311B / 1311B Is 65:20 quoted twice, the second time truncated by Gregory (Puer centum annorum. Et peccator centum annorum.) with the verbs left off; not a divergence.
• Migne's mis-citations reproduced verbatim, per the standing rule — six in this range: (Gen. II; Matth. XIX) for Eph 5, (Exod. XIV) for Ex 32:34, (Psal. XVIII) for Ps 28, (Rom. IX) for Rom 8:15, (Psal. LXXIV) for Ps 64, (II Thes. II) for 1Thes 2, (II Cor. VI) for 1Cor 3:16. None corrected.

2. Plate findings

2.1 Three [sic:] candidates opened on the archive BEFORE firing, and all three withdrawn

The brief's instruction paid for itself three times over. Each of these is a well-formed defect on the Gallica exemplar with our TEI agreeing — the exact shape that fires a marker — and each is clean on the second printing.

• 1312C Gallica a Chrito Christiani dicti; archive (p.661, 300 dpi) a Christo.
• 1313D Gallica Obsecro vos, fatres; archive (p.662, 300 dpi) fratres.
• 1315A Gallica breaks meridia / nam with no hyphen, which reads as Pattern 10 split type; archive (p.663, 300 dpi) prints meridia- with the hyphen. Ordinary line break.

No marker fired at any of the three. The English renders the sense at all three sites, and the pending-patches file records them so a later pass does not re-open them.

2.2 Readings confirmed as Migne's on both printings, and kept

• 1306C Videtur tamen AUGUSTUS dicere super Genesim. Not Augustinus, on Gallica f653 and archive p.658. A perfectly good Latin name, so nothing is broken and [sic:] has nothing to wrap — but the English then says that the emperor Augustus wrote on Genesis, while the sentence is contrasting him with Hieronymus two lines below and pointing at De Genesi ad litteram. Pattern 18: the printed word is rendered and [cj: Augustus; read Augustinus, "Augustine"] stands beside it.
• 1309C Omnis haereticus Deum ad iracundiam PROCAT. The lemma it glosses (Is 65:3) reads provocat. The plate hyphenates pro- / cat across a line, so no letters are hidden: the form really is procat, a real verb ("solicits, demands"). Rendered "solicits" with [cj: procat; read provocat, "provokes"].
• 1310B–C Ubi ACHOR filius Charmi de spoliis Jericho furatus… periit. Jos 7 names the man Achan and the valley Achor; Migne gives the man the valley's name, twice, and then derives the place-name from him. On the plate. Rendered "Achor" — not normalised.
• 1313A the dittography — see §2.3.
• Punctuation, four sites, each checked rather than assumed: 1312C …praeponentes Pars autem populi has no stop before the capital; 1314A scripsere naturis and per quem omnia operatur each end with no terminal stop; 1309B Allegorice. carries its stop. All four are on the plate and all four ride into the English unaltered (Pattern 8).

2.3 1313A — a doubling that varies, marked on the FIRST copy (Pattern 11a)

The plate prints, in italic, reddentis reddentes retributionem inimicis suis — confirmed on both printings, with no punctuation between the two words (the speck above the second is dirt, checked at 1900 px).

Position tells nothing here, so 11a governs: the intrusive copy is the one whose removal leaves the passage saying what the author is arguing. The frame is Vox etiam Domini audita est in Romanis exsultantibus, qui etiam nutu Dei venerunt, … scilicet Judaeis, per Romanos — a nominative plural subject, which reddentes agrees with and reddentis cannot. reddentis is the genitive singular of the Vulgate lemma itself (Is 66:6 vox Domini reddentis retributionem inimicis suis), i.e. the compositor's eye has gone back to the verse. So the marker sits on the first copy, and the English reads [d: rendering] rendering the recompense to his enemies.

2.4 Candidates refuted at the plate, no marker, no patch

• 1305C impe dio on Gallica reads impendio on archive p.658, as our TEI has it.
• 1312D in praesen i. on Gallica reads in praesenti on archive p.661.
• 1314C my- / teria is an ordinary line break: the next line begins steria veritatis, so the word is mysteria, as our TEI has it.
• 1307D (Jer. XXIII) — the archive prints (Jer. XVIII); Gallica is right and our TEI follows it. A jbig2 digit substitution, exactly the class PL113-WITNESS-ENCODING.md predicts.
• Band letters. Cols 1315 and 1316 carry A, B, C and no D, because the work stops two thirds down the page. Not a gap (known false positive 1).

2.5 The last page — what it carries, and what it does not

The work ends at 1316C, agentes gratias visa impiorum ineffabili poena. Checked for everything a last page can carry:

• No colophon, no subscription, no EXPLICIT, no editor's closing note for the Glossa.
• Below a full-measure rule the leaf carries the tome's own table of contents, ORDO RERUM QUAE IN HOC TOMO CONTINENTUR — which is also the running head already printed at the top of this leaf, so the head has changed over before the work has ended. It closes FINIS TOMI CENTESIMI DECIMI TERTII.
• ⭐ The two printings diverge inside that table. Gallica prints 0000 against Liber Ecclesiastes, Canticum canticorum and Liber Ecclesiastici, and 0090 against Liber Sapientiae — unset placeholders. The archive prints 1115, 1125, 1183 and 1167. Both print 1231 against Prophetia Isaiae, which is this work's true opening column. So the Gallica copy is from a state of the sheet in which four references had not yet been set, and the archive from the corrected state. (Recorded as a fact about the tome, not about our text.)
• The archive alone carries a printer's imprint below the FINIS line, Imprimerie générale de Châtillon-sur-Seine. — Jeanne Robert.; the Gallica copy's foot below FINIS is blank.
• Foot of both columns blank on both printings — no numbered Forte layer, no asterisk cross-reference layer. A checked zero on the final page as on the other five, consistent with 8956-CONVENTIONS.md §2.

3. Sense and construction cruces

• 1311D, Is 65:24, In conturbatione, in dolore mentis, sed in gaudio et laetitia. ⭐ The page contradicts itself and the English says so. The lemma being glossed is Is 65:23 neque generabunt in conturbatione, and the sed demands that the first two members be denied — but Migne prints no negation, on the plate, and it is the same on the archive. Rendered as printed ("In trouble, in grief of mind, but in joy and gladness"). Pattern 7a: the missing non is the evidence, and supplying it would destroy it. The conjecture is recorded here and is not in the text.
• 1309C, et in quadrum exstruere, et secti in latere pulveris roborare lineis. The tail of an usque ad abridgment of Jerome and it will not construe cleanly. Rendered as closely as the printed words allow ("to build four-square, and to strengthen with lines the dust cut upon the brick"), taking latere as the ablative of later, brick, which the lemma Is 65:3 sacrificant super lateres supports.
• 1314D, Quae secundum apostolorum non solum hominum, etc. The noun governed by apostolorum is simply absent; the sense wanted is 1Cor 13:1 linguis hominum… et angelorum. Rendered with the gap left open ("Which according to the apostles', not only of men, etc.") rather than filled. No marker: nothing is broken type, and there is no single word to conjecture.
• 1314A, multitudinem viventium ad fidem. venientium ("coming to the faith") is the obvious sense and viventium is a real participle that just about carries "living unto the faith". Rendered as printed; the conjecture is declined under 18a's bar, since the English asserts nothing false.
• 1309C sed per verba nostra; Dei intelligimus affectum erga nos — Migne's semicolon falls inside what English wants as one clause. Kept where he puts it (Pattern 8).
• 1313B creabo mihi… Ecclesiam unam quae aedificatur de vivis lapidibus: ends on a colon with nothing after it; the colon rides into the English.
• 1313B Ego qui omnes creavi homines in nihilo — in nihilo for ex nihilo is ordinary medieval usage (Pattern 9); translated normally, no marker.
• 1308D and 1316A — the same etymology, two different languages. At 1308D Migne writes Mene Hebraice, Luna Latine; at 1316A, of the same word, Mene Graece, Latine luna. Both are on the plate. Neither is emended and neither is harmonised to the other; the English keeps Mene and luna in Latin at both sites so the gloss still explains something.
• 1310B–C vallis effundens sanguinem / vallis fundens sanguinem — Migne gives the interpretation of Saronas twice, with and without the prefix. Both on the plate, both rendered ("pouring out" / "shedding").
• 1312A Caelum sedes mea — our file's one Caelum in a work that otherwise spells Coelum; the plate has Coelum. English unaffected; patch filed.
• Agency read back, per the 8961 warning. Six passives and ablative absolutes in the range were re-read for who acts: 1312B quod humana manus construxit (a human hand builds, not God); 1312D Romanis urbem irrumpentibus (the Romans burst in); 1314B quo purgabitur quid in bonis erit purgandum (the fire purges, the good are purged); 1315A Domino qui per illos figuratur (the Lord is figured by David and Solomon, not they by him); 1316A a sole… lumen recipit (the moon receives from the sun); 1316C boni videbunt eos… separatos (the good see, the wicked are separated). All six construe as rendered.

4. Chunks read and found clean

Reported so this file can be told from one written by a reader who did not read.

• 0030 (1305C–1308A) — 66 spans, 11 anchors, 28 notes. Every word collated against Gallica f653 and f654 and against archive pp. 658–659. Clean apart from the sites above.
• 0031 (1308B–1309B) — 32 spans, 5 anchors, 14 notes. Collated against f654 and f655 and against archive p.659. No plate/TEI divergence anywhere in the chunk; its only findings are the three Vulgate [var:] at 1308B and 1308C.
• 0032 (1309C–1312A) — 68 spans, 10 anchors, 24 notes. Collated against f655 and f656. Col 1309 and col 1310 are entirely clean at the plate — every word of both columns agrees with our TEI, including the three readings that most invited a marker (procat, Achor, the effundens/fundens doublet), all three of which are Migne's.
• 0033 (1312B–1316B) — 76 spans, 16 anchors, 29 notes. Collated against f656, f657, f658 and against archive pp. 661–663. Carries the two withdrawn [sic:] candidates, the dittography, and four of the seven our-side patches.
• 0034 (1316C) — 6 spans, 1 anchor, 3 notes. Collated against f658 and archive p.663; the Latin agrees with both printings word for word. Two [var:], no defects.

verify-english 8956 passes over all 35 chunks: columns, notes and sections aligned, no duplicate paragraphs. English/Latin word ratios: 1.46 · 1.52 · 1.46 · 1.50 · 1.45 (marker contents excluded), against the register's ~1.5× expectation.