Evangelium secundum Lucam

Merged from the eight translation stints' files (cruces-0000, -0006, -0011, -0017, -0021, -0026, -0031, -0036). The stints' own sections are preserved below, verbatim and unabridged, because their reasoning is the record of how each reading was reached. Nothing below §0 has been silently altered; every change made to a stint's text is named in §0.

⛔ THE PLATE HAS NOT BEEN READ FOR THIS WORK. Unlike 8999 and 9004, whose merged files record a completed plate read, everything here is the state before that step. The declined-oddity lists in §0.5 are what the plate reader works from. No attribution in this file is settled.

This is a nineteenth-century abridged recension of the marginal gloss: a VERS. n.-- address, the Vulgate words being glossed in italics, then the comment, with attribution sigla (BEDA., AMBR., AUG., HIER., GREG., ID.) arriving as ordinary notes. Migne's attributions in this recension are known to be unreliable, and passing a siglum through is not endorsing it.

Luke is printed inside the Glossa proper — the verso running head reads GLOSSA ORDINARIA. — EVANG. LUC., confirmed at the plate on pp. 144 and 164 during the chunking spot check. Like Mark and John, and unlike 9003 and 9004 in the same volume, whose bylines were corrected off the Glossa author. There is no attribution question here.

§0. THE MERGE — what was reconciled, and on what evidence

Eight stints, 41 chunks, 44,075 Latin words. Each was handed Mark's and John's six settled axes before it wrote a line, which is why this section is short: the merge cost 3 edits, against 15 on John, 100+ on Mark and 147 on 9004.

§0.1 ⭐ The axis that actually drifted was the one no brief had ruled on: when may a [var:] rest on a single word?

Three stints stated three different thresholds, each before its findings, which is the only reason the disagreement is visible at all:

None of the three is right, because word count is not what makes a [var:] true. A [var:] asserts that Migne's page diverges from the Vulgate — but our Latin is Corpus Corporum's transcription of that page, so a one-word difference has two possible owners, and the marker names only one of them. That is the same asymmetry the brief already states for [sic:], arriving on the other marker class.

⭐ THE RULE SETTLED HERE — a [var:] may rest on a single word only where the divergence has a witness of its own. Either:

• (a) Migne's own gloss argues from the divergent word. The commentator's exposition is then a witness that the word is what he read — 0250D puer (Jesus frequenter puer dicitur… unde Propheta: Ecce puer meus, where the Vulgate has servus); 0307B confusus in a gloss whose whole subject is blushing; 0330A in excelsis at Luke 2:14, which is the reading of Luke 19:38, the verse the gloss expounds two lines later.
• (b) The divergent form is one letter-level corruption cannot manufacture — a different word (dilectio for caritas, gratia for gloria, die for hora), a transliteration for a translation (sabaoth for exercituum), or a conflation of two verses.

Withdrawn under it: a dropped word, an inflectional ending, or a one-letter swap with no argument resting on it — because Corpus Corporum demonstrably does all three (this project has measured it dropping nos at a column break, and rejoining hyphenated words wrongly).

§0.2 The three edits

1. 0263B — [var:] WITHDRAWN. Migne prints Cor sapientis ubi est tristitia where the Vulgate has Cor sapientium. An inflectional ending, and the gloss (docens stultitiam ridentibus, prudentiam flentibus ascribendam) argues from neither number. Moved to the declined list.
2. 0264B–0267C, Lc 6:31 — [var:] WITHDRAWN. Migne prints et vos faciatis illis where the Vulgate has et vos facite illis similiter: a mood inflection plus a dropped word. The gloss argues from the wish against the deed (cum votis actus aequatur), which is untouched by either difference. Moved to the declined list.
3. 0258B — [var:] STANDS, its payload CORRECTED. ⚑ This one was nearly withdrawn on the strength of its own apparatus, and the Latin overturned it. The marker read "Vulg. 1 Jo 4:18 reads perfecta caritas" — which names the Vulgate and never says what diverged, so it reads like a dropped perfecta. The plate's own text settles it: Migne prints Perfecta enim dilectio foras mittit timorem, so the divergence is dilectio for caritas, an unrelated noun that no letter-swap produces — criterion (b) at full strength, and a known patristic variant besides. The payload now names it. A marker that quotes only the witness and not the divergence cannot be audited later.**

33 [var:] stand. 0 [sic:], 0 [cj:], 1 [d:].

§0.3 ⭐⭐ Zero [sic:] in 44,075 words, and it was a policy, not an accident

Every stint was told what John measured — that there, the 23 leaf-reads spent on fired markers produced no corrections while the nineteen declined oddities produced all seventeen — and every stint complied. About 140 suspect words are logged unmarked in §0.5's lists, each with its column, most with the plate leaf that would settle them, and none carrying an attribution claim. Five stints ordered their own list by which leaf to spend first; two noticed that several of their oddities share one leaf.

§0.4 The inherited axes, measured across the whole work — no drift, one exception

Counted mechanically over all 41 chunk pairs after the merge:

Axis 4 holds, and two stints found words the brief does not rule on and flagged them rather than deciding silently: gentilis and gentilitas. Every one of the work's 17 "nation(s)" was checked against its Latin: each is a singular gens of the Jewish people (gentem nostram at Lc 7:5, tollent locum nostrum et gentem at 0323C) or the plainly generic plural (in omnes gentes, Lc 24:47). Nothing was flattened. ⛔ Carry gentilis → Gentile and gentilitas → the Gentile world into the Matthew brief; they are the same axis and the brief should say so.

One real exception to usque ad → "as far as", taken with the Latin that forced it: at 0316A the phrase is Luke 16:16 itself — Lex et prophetae usque ad Joannem, the lemma and not the abridgment's splice formula — and renders "until John". Two temporal uses at 0350B (a vespera usque ad vesperam) take "until", and a fine usque ad finem is "from end to end". The axis has a class boundary, not just a preferred wording: it governs the splice formula, not the words wherever they occur.

§0.5 ⚑ Instrument traps — five stints hit them and five caught them before reporting drift

Every one of these produces a number that reads exactly like convention drift:

• \bturba matches the verb turbare — perturbatos, turbatam, turbine (a whirlwind). Fired on three separate stints.
• natio matches damnatio-, inclinatio-, examinatio-. One stint's grep returned nine hits in a range containing no natio at all.
• a raw usque ad grep overcounts the italic splice formula by catching unitalicised prose.
• \bdaemoni (the merge's own first attempt) misses daemones, daemonum, daemone, and reported 23 Latin against 41 English before the regex was fixed. The merge is not exempt.
• English italic-span parity is NOT 1:1 and never was. 23 chunks carry more English spans than Latin, +55 in all: 48 are [var:] payloads quoting Migne's Latin in italics, and the rest are non-words carried in italics under Pattern 7. ⭐ The control that settles it: John runs +61 across 28 chunks and Mark +26 across 24 — both shipped. A parity check defined against 1:1 would fire on two live works.

§0.6 A defect in the CONTROL, not in Migne

A stint refused to cite 1 Thess 5:3 in a marker because our sources/vulgate/clementine-flat.txt prints repentius; the Clementine reads repentinus, and the error is upstream in the 2005 Quasimodo release, so it sits in the faithful clementine/1Thes.lat too. ⛔ A marker citing that verse against a Migne lemma reading repentinus would have accused Migne of diverging from the Vulgate using our own typo as the standard. Recorded as a named erratum in sources/vulgate/README.md; neither file is edited, because their being faithful is what makes them a witness. The proof was inside the same file — Pr 29:1 carries the identical phrase spelled correctly.

§0.7 ⭐ The ✓ in the lemma briefs failed six times, and a gospel commentary is why

lemma-inventory.mjs clears a span if its words occur anywhere in the Clementine. In a gospel commentary the harmony's own parallels defeat that:

• Quid cogitatis mala in cordibus vestris cleared off Mt 9:4, inside a commentary on Lc 5:22.
• Gaudete et exsultate cleared off Mt 5:12; Lc 6:23 reads Gaudete in illa die, et exsultate.
• Gloria in excelsis Deo at Lc 2:14 cleared off Lc 19:38 — the entry song the gloss expounds two lines later.
• Ecce puer meus cleared, against Isaiah's servus, in a gloss arguing from puer.
• In illa die cleared, against Lc 17:31's hora.
• Supra terram cleared although the phrase occurs nowhere in the file (Lc 6:49 has super).

⚑ And a stint found a third mechanism the brief did not know: the checker searches the Clementine as one joined string, so a span can score ✓ by matching across a verse boundary — demonstrated at 0340D and 0353A, where the recension splices two consecutive verses. ✓ is not clearance, and in this genre it is weakest exactly where the work is densest.

Preserved verbatim from cruces-0000.md. Where it states a [var:] threshold of its own, §0.1 governs.

Chunks 0000–0005; 4,758 Latin words; Jerome's prologue head, the Proemium, and Caput I–IV (Luke 1:1 – 4:42). Written against data/briefs/9000-launch-brief.md, translation-style.md, the 9741 anchor pair, and §1 of src/english/8999/cruces.md and src/english/9001/cruces.md.

(a) THE POLICY I APPLIED — stated before the findings, so it can be argued with

1. No [sic:]. None fired, and none was close. The brief's measured result on John — 23 leaf-reads on fired markers produced zero corrections while nineteen declined oddities produced all seventeen — is the whole procedure here. Every suspect word in this range is in §(e) below, unmarked, with its column, and with no attribution claim. I do not know whose any of them is. Where a form looks like it could be Migne's type or Corpus Corporum's transcription, I name both possibilities and assert neither, and where I can see a mechanism (a hyphen-rejoin shape, an assimilated medieval spelling) I say what the shape is and stop there.

2. [var:] only on a quoted CLAUSE, only against sources/vulgate/clementine-flat.txt, cited book chapter:verse, and only where the divergence changes the sense. Three markers fired (§(c)). Every other divergence in the lemma inventory — and there are twelve — is listed with the reason it was declined. Declines fall into four classes and I applied them mechanically: (i) truncation (the abridgment cuts the lemma mid-phrase, so the sequence is absent from the Clementine because it is a fragment, not because it differs); (ii) orthography (Elizabeth/Elisabeth, Enoch/Henoch, nunquam/numquam, his/iis); (iii) word order with no change of assertion; (iv) a one- or two-word difference, which per the brief is suspicion of our file, not evidence about Migne, and therefore goes in the declined list rather than into a marker.

3. Pattern 7 without exception. Nothing is smoothed. Migne's printed punctuation stands where he sets it, including the semicolon that cuts a subject from its verb at 0248A and the question mark that breaks a sentence in half at 0252B. Both are logged in §(d); neither was repaired.

4. Negation fidelity (7a), counted by hand chunk by chunk. Every non, nec, neque, nisi, nullus, nihil, nunquam, ne the Latin prints stands in the English, and none was added. Two sites needed a ruling and got one in the running text, not in an apparatus: 0252C In hac vita omnis homo Christum videre non potuit is rendered "not every man could see Christ" (the contrast is with the day of judgment, when the elect and the reprobate see him) rather than the flatter "no man could"; and 0254A's nihil pro diaboli arbitrio, nec declarandae virtutis consideratione faciendum carries both particles.

5. Mechanism 5 was censused over ALL SIX chunks, not the first two. §(d) names the sites and says what fixed each binding — case and gender, never the easier sense. The brief's warning that John's stint apparatus named three sites all inside its first two chunks is the reason the census below runs to 0005.

6. The six inherited axes carried in from the first line, not reconciled afterwards. Measured across my whole range: usque ad → as far as, 62 italic spans, 62/62, no other form used; mysterium → mystery 7/7; diabolus/satanas → devil 13/13; daemon- → demon 4/4; natio (nationum, 0255B) → nation; gentes → Gentiles at all five sites, every one inside the Judaei/gentes typology (the widow of Sarepta, Naaman, the desert of the Gentiles, the calling of the Gentiles, the appearing after the resurrection). No exception was forced and none is claimed. multitudo and sacramentum do not occur in this range. ⚠ turba does not occur as a noun here either — the four turb- hits are the verb turbare (turbatus, turbati, turbari, turbantur), the exact false signal the brief names; I checked the instrument before reporting anything. One further word is not on this axis: turmis (0256A, a credentium turmis) is turma, a fourth noun, rendered companies so that crowd / multitude / throng stay free.

7. Gloss voice not archaised. dicit/ait → says; there is no saith in my range outside an italic span. The one saith I did write is inside the italic scripture at 0250A (Non est pax impiis, dicit Dominus), which is the italic-membership test working as intended. Second person follows Pattern 17: thou where the Latin addresses a singular (Zacharias at 0245A; the Father at 0253A), you where it is plural (quaerebatis, 0251D).

8. Ratio. verify-english.mjs passes my six chunks with two warnings, 0002 at 1.64 and 0005 at 1.62 against the 1.5× guide. Both are real expansions of the abridgment's own shorthand — truncated lemmata that need an English verb, and etc., usque ad splices — not padding. No clause was added.

(b) SPAN COUNT AGAINST THE BRIEF — I RAN IT, AND IT AGREES

data/briefs/9000-lemmata-0000.txt says 247 spans for chunks 0000–0005.

I counted the … spans in the body of my own six Latin files (frontmatter excluded, since the incipit: field can carry italics and inflate a naive file-level count):

247 = 247. The brief's total is right for this range. The English carries 62 as far as spans (61 plain plus as far [0245B] as, which has the column anchor inside the span exactly where Migne breaks the column) and 185 others, and every chunk's span count matches its Latin twin once [var:] content is stripped.

⚠ Two prose usque ad were deliberately not treated as the formula, per John's merge rule that only the italic splice is on the axis: a sabbato usque ad sabbatum (0245A) and A baptizato Dei Filio usque ad Deum Patrem (0253B). Both read naturally as "from … to" / "as far as" in ordinary prose and neither is italic.

Marker parity, checked mechanically before each file was closed: column anchors 1:1 and in order in all six chunks (1 / 15 / 3 / 12 / 6 / 10); [n: …] sequences byte-identical to the Latin in all six (3 / 35 / 5 / 25 / 16 / 22), matching every frontmatter noteCount.

⛑ §6.2 of the brief is mine and I did not report it. Chunk 0000 opens the work at 0243D below the full-width rule on plate p. 127, with Mark's gloss ending across the top of both columns. The band letters do not run down one column here, that is the plate, and no anchor was touched.

(c) THE LEMMA DIVERGENCE LIST — every span walked, three markers fired, eleven declined

I walked all 247 spans against the Clementine, ✓ and ⚠ alike, per 7a″. ✓ was treated as nothing at all: for every ✓ span I asked whether it matches the verse being glossed, not whether the words occur somewhere. That is what caught 0250D, which the pre-scan marks ✓.

The three [var:] fired

All three markers name the witness first, are English, and quote only Vulgate wording that does not stand in the Latin twin (checked, so the [ed:]/[var:] mirror test holds).

The twelve divergences I DECLINED, each with its reason

Inventory arithmetic, checked against the brief file itself: 247 lines = 201 ✓ + 16 ⚠ + 30 single-word. Of the 16 ⚠, three became the markers in the table above, twelve are declined in the table here, and one (Qui habet duas., 0252B) is a plain truncation whose separate typographic anomaly is #14 in §(e). Of the 201 ✓, one — 0250D Ecce puer meus — was a real divergence the mechanical test could not see.

The rest of the inventory, accounted for

• 62 italic usque ad — the abridgment's splice formula, not scripture. All rendered as far as. Not divergences.
• 30 single-word spans (Omnia, Theophile, Zacharias, Elizabeth, Sorte, Convertat, Unde, Occultabat, Gabriel, Jesum, Exsultavit ×2, Benedictus, Visitavit, Prima, Nazareth, Puerum, Gladius, Publicani, Herodes, Nathan, Mathusale, Seth, Esuriit, Mitte, Angelis, Docebat, Spiritus, Medice, Viduam) — each checked in place against its own verse. All are the Vulgate's word at that verse except the two orthographies already listed (Elizabeth, Mathusale). (The Qui habet duas. (BEDA) span at 0252B is carried in the inventory as a ⚠, not as a single word; its anomaly is #14 in §(e).)
• The remaining ✓ spans were each read word by word against their own verse. Beyond the table above I found no wording that diverges. Where the recension stops a lemma mid-phrase — and it does so constantly, which is what most of the ⚠ marks are — the English stops there too (And there was not unto them., And mercy., As was made., And being ended.), rather than completing the verse from memory. That is the single most frequent 7a″ temptation in this range, and refusing it is the reason so many of my lemma renderings read as fragments.

(d) THIS RANGE'S CRUCES, BY COLUMN

Printed punctuation preserved against the grain

• 0248A (0001) — Qui aeterna toto studio desiderant; saturabuntur, cum Christus apparuerit in gloria. Migne sets a semicolon between the subject clause and its verb. Pattern 8 forbids deleting a printed mark, so the English reads "They who desire eternal things with all zeal; shall be filled…". It is not a construction I would write and it is what the page prints.
• 0252B (0004) — Qui fidem et bona opera praedicat, quid aliud quam venienti Domino? ad corda audientium viam parat… A question mark inside the sentence, after the elliptical quid aliud quam venienti Domino. Preserved in place; the following clause is carried as a new English sentence, which is what Migne's mark makes it. Not repaired, not merged.
• 0246C (0001) — the sentence …ita sexto millenario missus est Christus, ends on a comma. Left ending on a comma (Pattern 8: never supply a terminal stop).

The mechanism-5 census — all six chunks

The brief's dominant defect class. Every site where a pronoun, relative or participle could bind to either of two nouns; each decided on case and gender, never on which reading was easier.

Constructions rendered as printed, where the printed Latin resists

• 0243D (0000) — et propriis quibusdam libri singuli mysteriorum gestorumque miracula eminerent: eminere is intransitive and miracula is accusative. Rendered so that the marvels are what stand out and propriis quibusdam is the means, which is the only reading the words allow. No emendation, no marker. §(e).
• 0252A (0004) — praecurretur suo baptismate quo peccata solvi non possunt: a future passive of praecurro, an intransitive verb, inside an etc., usque ad splice. Rendered impersonally, "there will be a going before by his own baptism". §(e).
• 0252B (0004) — Tunica plus sui necessaria quam pallium: a genitive where a dative or ablative of respect is wanted. There is no English exponent for the case, so the English reads "more necessary of itself"; nothing was emended. §(e).
• 0254A (0005) — Docet magis doctrina quam miraculis pugnare, humilitati quam potentia: a dative in a row of ablatives. Pattern 9: a case mismatch has no English exponent, so the English reads "with humility rather than with power" and the crux carries the observation. §(e).
• 0254D (0005) — qua doctrina per miracula praecedentia magnificat: magnificat with no expressed object. Rendered without one. §(e).
• 0247D (0001) — VERS. 47 carries two identical notes back to back inside one lemma span, Et exsultavit. [n: (ID.)] [n: (ID.)]. Notes are verbatim, so both are reproduced, in place, untranslated. Whether the doubling is Migne's or the transcription's is not something our file can settle. §(e).

Migne's citations — reproduced, never corrected

Per the brief and CLAUDE.md, a printed reference is refDisplay and is not ours to fix. Two in this range point elsewhere than their clause, and both stand as printed:

• 0256A (0005) — si cognovissent Dominum gloriae, nunquam crucifixissent is 1 Cor 2:8, and the note reads [n: (Matth. I)].
• 0250D (0003) — [n: (Isa. XLII)] closes the sentence after Filius hominis non venit ministrari, sed ministrare, which is Mt 20:28; Isaiah 42 belongs to Ecce puer meus, the span before it. The note is positioned after the second quotation, and I did not move it.

Two lemma renderings that are deliberately unfamiliar

• 0247C (0001) — Mater Domini. Migne cuts before mei; the English reads "The mother of the Lord", not the Douay's "the mother of my Lord". Conforming it would have supplied a word the plate does not print.
• 0248A (0001) — Et misericordia. cut before ejus: "And mercy", not "And his mercy".

(e) ⭐ THE DECLINED-ODDITY LIST — every suspect word I did NOT mark, in one place

No marker stands on any of these. I make no claim about whose any of them is. On John this list produced every correction the work received; it is what the plate reader works from. Ordered by column.

What I looked for and did NOT find, said plainly

• No <LACUNA>, no bracketed Forte, no ( sic ), no asterisk-keyed note, no foot-of-page apparatus anywhere in my six chunks. Consistent with §6.7 of the brief; recorded as a negative so a later reader knows it was checked and not merely unmentioned.
• No dittography — nothing in this range doubles a run, so no [d:] was owed.
• No split or fused word-division defect of the Pattern 10 shape. The hyphen-rejoin class the brief warns of (John's scie-/bat, im-/pleretur) did not appear here in its visible form; ⚠ but #12, #19 and #21 above are each a single-letter or single-syllable difference of exactly the kind that class manufactures, and none of them can be told from an authorial form in our text. That is the whole reason they are in this list.
• One Greek word exists in this work and it is in chunk 0024, not mine.

Preserved verbatim from cruces-0006.md. Where it states a [var:] threshold of its own, §0.1 governs.

Evangelium secundum Lucam, Glossa ordinaria. Chunks 0006–0010, 4,770 Latin words, CAPUT V (from VERS. 1) through CAPUT VI (to VERS. 49). Written under data/briefs/9000-launch-brief.md; §1's inherited axes were carried in from the first line, not reconciled afterwards.

§a. THE POLICY I APPLIED — stated before the findings, so it can be argued with

1. No [sic:]. None fired, and none was close. The brief's measured record governs: on John 23 leaf-reads spent on fired markers produced zero corrections and the nineteen DECLINED oddities produced all seventeen. Every suspect word in this range is in §e, unmarked, with its column, and with no claim about whose it is. Where a word is a non-word, Pattern 7's ruling is obeyed instead of a marker: it is carried into the English untranslated, in italics (ejunus, jejumare), so the reader meets the defect rather than a repair.

2. [var:] fires on a SUBSTITUTED OR ADDED WORD, never on order, orthography, or abridgment. This is the line I drew before reading, and I held it in both directions — it cost me four ⚠ spans I would otherwise have marked, and it forced two markers on spans the pre-scan had passed with ✓. Explicitly:

3. A one-word suspicion in a lemma is our-file territory, not Migne's. So Praebe illis (0263D) and Supra terram (0267C) — both of which are genuine divergences from Luke 6 — took no marker, and went to §e instead. The brief's asymmetry is the reason: an OCR letter-swap can manufacture a broken word, and cannot invent a coherent divergent clause.

4. Every [var:] is citable book chapter:verse against sources/vulgate/clementine-flat.txt, and none rests on the pre-scan or on recollection. Every one was re-grepped in the file.

5. Mechanism 5 was censused across all five chunks, not the first two. §d.2 lists every site where a relative, participle or pronoun could bind two ways, including the three in chunk 0009 and 0010 at the far end of the range.

6. The inherited axes were measured after writing, mechanically (§c.2). All six hold at 1:1; none needed an exception, and Luke forced none in this range.

§b. SPAN COUNT — 151 vs 151, counted

data/briefs/9000-lemmata-0006.txt prints 151 spans for chunks 0006–0010.

I counted the italic spans in my own Latin: 302 asterisks in the five chunk bodies (0006: 126 · 0007: 2 · 0008: 102 · 0009: 68 · 0010: 4), = 151 spans. The frontmatter heads[] of these five chunks carry no asterisks, so body and file totals agree.

151 = 151. The brief's split is correct for this range, and 151 lines parsed cleanly out of the brief file with none unmatched.

Marker parity, counted before each file was written and again after:

node scripts/verify-english.mjs 9000 reports no error and no warning on any of the five.

§c. THE LEMMA DIVERGENCE LIST — all 151 spans

§c.1 Summary

⭐ The three false ✓ are the finding of this section. The brief warns that ✓ means the sequence occurs somewhere; here it demonstrably came from another verse twice (Quid cogitatis mala from Mt 9:4 inside a commentary on Lc 5:22; Gaudete et exsultate from Mt 5:12 where Lc 6:23 reads Gaudete in illa die, et exsultate), and once from nothing at all: supra terram occurs nowhere in clementine-flat.txt, case-insensitively, yet the pre-scan marks span 148 ✓. One of the three carries a fired marker; the other two are declined on their own merits, and the pre-scan's verdict was wrong in all three.

⭐ Two ⚠ classes in this range are pure orthography and should be discounted wholesale by later stints: Nunquid (0 occurrences in the Clementine file, which spells Numquid) and charitas (0 occurrences, which spells caritas). Three of my nineteen ⚠ spans are nothing but that.

§c.2 The inherited axes, measured against the Latin after writing

⚑ The turba measuring trap fired here too, as John's cruces predicted. A \bturb grep over this range returns 11 hits, one of which is perturbatos (0265B, the verb) — a different word. The noun count is 10, and it is 10 in the English. Anyone re-measuring this range must exclude the verb before reporting drift.

⚑ One phrase settles mysterium / sacramentum mechanically inside this range: 0259D prints novorum mysteriorum sacramenta — both words in one noun phrase, "the sacraments of the new mysteries." They cannot share an English word here.

§c.3 The list — every span, in order

§d. THIS RANGE'S CRUCES, BY COLUMN

§d.1 Sites where the printed Latin does not construe, rendered as printed

0256B — quorum corda … provenit, a singular verb on a plural subject. Migne prints quorum corda a fluctibus saeculi ad futurae vitae tranquillitatem, quasi ad littus misericorditer vivendo, provenit. provenit is intransitive, so corda cannot be its object, and it is singular where corda is plural. The obvious conjecture is provehit ("he brings forward"), which would take corda as object and give the sentence the transitive sense the image wants. I did not make it. The English keeps the printed number — "whose hearts comes forth from the waves of the world…" — because English verbs do carry number and a smoothed "come" would erase the only trace of the defect. Pattern 9's "no English exponent" licence covers gender, not number. ⛔ I do not know whether provenit is Migne's or ours. provenit is a normal word (22 occurrences in our Latin corpus), so no frequency test bears on it at all.

0263A — cibumque perfidiae quae in praesenti eructant, a relative agreeing with nothing. The clause is the object of scient: cibumque perfidiae … causam tanti supplicii esse scient. cibum is masculine accusative and would take quem; perfidiae is feminine genitive and would take quam in this clause. quae is neither. Rendered literally, "the food of faithlessness which they belch forth in the present," binding the relative to the nearer noun without asserting that the printed form licenses it. Logged rather than repaired.

0266D — bonos auditores verbi sibi ipsis … assimilat. The whole sentence that follows (sicut Christus … unam Ecclesiam construit … sic bonus auditor … aedificat) requires the singular, i.e. Christ likens the good hearers to himself. Migne prints the dative plural ipsis, which cannot mean that and agrees with nothing in the clause. The English reads "to himself," because that is what the following sentence construes, and the printed plural is recorded here rather than parked in a marker. ⚑ This is the shape of the defect class the brief calls mechanism 5, met from the grammatical side.

0267B — quando Ecclesia utraque est consummata. utraque agrees with Ecclesia, so the literal sense is "when each Church has been consummated," which is what the English prints; the reading the argument seems to want is "when the Church in both its parts is consummated." No emendation made; the difference is a matter of what utraque qualifies and nothing in the type is defective.

0266B — Spina et rubus, haereticis. A bare dative where the parallel sentences use the nominative (Uva et ficus, dulcedo…, spinae et rubi, curae saeculi). Rendered "The thorn and the bramble, to the heretics," keeping the case. Possibly an ellipsis of conveniunt; possibly a defect. Not marked.

0263C — debet patientiam verberantis sustinere. Read as an objective genitive ("the patient bearing of him that beats"), which is the only reading that construes; a subjective reading would make charity endure the beater's own patience, which the paragraph nowhere supports. Recorded because the English deliberately turns the genitive rather than smoothing it.

0258C — Omnis quaestio ex lege quae tenax justitiae, non habet, etc. The abridgment cuts the sentence with etc. before its predicate is complete; non habet is left governing nothing. Rendered as printed, with the truncation visible. This is the recension's own splice, not a defect.

0259A — quia qui opera legis, etc. Same class: the usque ad abridgment cuts before the verb, leaving opera legis accusative with nothing to govern it. The English reproduces the cut ("because they who the works of the law, etc.") rather than supplying a verb; supplying one would be Pattern 8's "never supply" at the scale of a clause.

0264A — ut beneficiis aliis trahantur. aliis is printed, so the English reads "that they may be drawn by other kindnesses." The sense the sentence seems to want is "that others may be drawn by kindnesses," which would need nominative alii. Not emended.

§d.2 MECHANISM 5 CENSUS — every two-way binding in the range, all five chunks

The brief names re-binding to the contextually easier noun as the dominant defect class. These are the sites where a relative, participle or pronoun in my range could attach two ways. Case and gender decided each one, not sense.

No dropped and no added negation anywhere in the range. Every non, nec, neque, nisi, nunquam, nullus, nihil, and every ne-purpose clause the Latin prints stands in the English, including the double non modo non at 0263B ("not only not reproved, but moreover praised"), the triple ne … ne … ne at 0261C, and non nisi at 0261D ("not able to see Christ except in what is lowly").

§d.3 Pattern 8 — a printed mark I kept rather than improved

0265A — Nunquid potest. Migne sets a full stop on a lemma that is the opening of Luke 6:39's question (Numquid potest caecus caecum ducere?). The English reads "Can he." — the printed mark kept, the sentence left abrupt. §8b's colon device does not apply: this is not an indirect question refused by English grammar, it is Migne's own punctuation of a truncated lemma, and recording it is the point. verify-english initially flagged 4 EN vs 3 LA question marks on chunk 0009 and caught my first draft, which had supplied a mark the plate does not print. Parity is now 3 = 3.

§d.4 Pattern 17 — thou/you in this range

The range swings between the two repeatedly and the swings are all morphological. Singular → thou: 0257C Tu homo peccator … tibi peccata relaxo · 0259A ut … intelligas · 0260B (Sir 4:36) manus tua · 0265A–B the whole te punit / te salvat / te caecaverit / coronaberis sequence · 0265C–D Si quem vis reprehendere … ut sis bona arbor · the Lc 6:29–30 lemmata qui te percutit, Ne repetas, and Praebe illis · Lc 6:41–42 Quid autem vides, Hypocrita. Plural → you: 0257D–0258A cogitatis … vestras … Ex vobis intelligite · 0260B Interrogo vos · 0262D–0263B the vae vobis series · 0263B–0264A Diligite inimicos, Orate, prout vultis, Estote, Nolite judicare · 0264A–B vobis gradus professionis … amplectamini. Nothing was decided by who is addressed or how exalted; 0265A moves from plural Nolite judicare to singular tibi nocet within four lines and the English moves with it.

§d.5 Migne's citations passed through, not corrected

Three of his [n:] notes point somewhere other than where the words stand. All are reproduced verbatim and untranslated, per the note rule and §6.3 of the brief; none is queried in the text.

• 0264A [n: (I Cor. III)] on charitas est patiens, benigna est — the words are 1 Cor 13:4.
• 0259C [n: (Ephes. IV)] on Exuite veterem hominem cum actibus suis, et induite novum — the first half is Col 3:9, the second Eph 4:24. The [var:] states both places; the note is untouched.
• 0260B [n: (Eccles., IV)] on Non sit manus tua… — Ecclesiasticus (Sir) 4:36, i.e. his abbreviation is right and only looks like Ecclesiastes; not an error, recorded so a later reader does not "fix" it.

§d.6 Not defects — checked and left (Pattern 9)

• 0256D Fotinum for Photinus, beside Arium and Manichaeum. An attested Latin form of the name; translated normally as "Photinus," no italics, no crux beyond this line.
• 0265C sunt objurgationes adhibenda — neuter plural gerundive on a feminine plural noun. Gender has no English exponent. Left alone, and marked here so a later sweep does not churn it.
• 0260B Ebionitas, 0256D Manichaeum — normal forms, listed only because a heresiarch name is a common false positive.
• Column-band arithmetic. 0256A→0256B→…→0260A in chunk 0006, then chunk 0007 opening and closing at 0260A, then 0260B in chunk 0008; and 0267C→0267D across the 0009/0010 boundary. All normal per §6.1. No text in this range is broken mid-word or mid-construction across an anchor — checked at all 47 anchors.
• (cont.) on the heads of 0007, 0009 and 0010 — our chunk division, translated as the head text plus the untouched marker, per §5.

§e. DECLINED-ODDITY LIST — every suspect word I did NOT mark

This is the list the plate reader works from. Nothing here carries a [sic:], a [cj:], or an attribution. Where I have a conjecture I give it, and I say plainly that I do not know whose the defect is. Two entries are non-words carried into the English in italics under Pattern 7; the rest are real words that read oddly and are rendered as printed.

⚑ Two of these (1, 2) are non-words and twelve are well-formed Latin. Per the brief, that split predicts nothing about attribution: 9001 @0203A magnitudino is a genuine non-word that Migne prints, and the John reads went 4-of-4 Migne's in one range and 5-of-5 ours in another 200 columns away. I make no class claim on any of the fourteen.

⚑ If a leaf budget is limited, the order I would spend it in is 3, 4, 1, 2, 5 — the first two because they are lemma divergences that would each convert into a standing [var:] if the plate confirms them, the next two because they are the only non-words, and 0256B because a transitive/intransitive swap changes what the sentence asserts.

⚑ And a class this range cannot see at all: the brief's silent-repair mechanism, where Corpus Corporum quietly fixed Migne's broken type. Nothing in a clean-looking Latin can betray it, and no Vulgate collation reaches it. The fourteen above are the visible half only.

§f. Columns carrying a fired marker — for plate-gate.mjs

Nine [var:], on eight columns: 0257A, 0257D, 0258B, 0258D (two: the Gen 3 quotation at 0258D and the Rom 10 quotation at 0258D), 0259C, 0263B, 0264A, 0267D. Under the brief's map (leaf n carries columns 2n−11 and 2n−10) those eight columns fall on four leaves: n134 (cols 257/258 — five of the nine markers), n135 (259/260), n137 (263/264), n139 (267/268). No [sic:] and no [cj:] stands anywhere in this range.

Preserved verbatim from cruces-0011.md. Where it states a [var:] threshold of its own, §0.1 governs.

Chunks 0011, 0012, 0013, 0014, 0015, 0016 — capita VII–IX, the central miracle and parable sequence. 6,297 Latin words, 196 italic spans, 65 column anchors, 26 [n:] notes.

§1. THE POLICY THIS STINT APPLIED — stated before the findings, so it can be argued with

1.1 Markers

• [sic:] fired: ZERO. Deliberately, per the launch brief §2. Every suspect word — non-word, broken word, or real word that reads oddly — is in the declined-oddity list at §5, with its column, named and unattributed. I do not know whose any of them is, and I say so per site. On John the 23 leaf-reads spent on fired markers produced no corrections and the nineteen declined oddities produced all seventeen; I am doing what John's stints did.
• [cj:] fired: ZERO. Two sites tempted one (0280A dilationem, 0279A positionis). Both are real, well-formed Latin words, which is exactly the class the brief tells me to log and not mark.
• [var:] fired: THREE. Each rests on sources/vulgate/clementine-flat.txt, cited book chapter:verse, and each is a quoted clause, not a single word: 1. 0268D Matt 8:13 — Migne sanatus sit puer, Clementine Et sanatus est puer in illa hora. Mood, and the gloss's whole point is that "Matthew explains it more plainly", i.e. reports the healing. 2. 0270A Luke 7:22 — Migne renuntiate Joanni quae vidistis, Clementine quae audistis et vidistis. A dropped pair of words inside the clause, not at the etc. truncation. 3. 0280C 1 John 3:2 — Migne nondum apparuit quid erimus, sed tandem similes ei erimus, quando videbimus eum sicuti est, Clementine …Scimus quoniam cum apparuerit, similes ei erimus: quoniam videbimus eum sicuti est. A whole clause replaced.
• Materiality test for [var:] (Pattern 14, "in a way that changes the sense"): I declined it for orthography (coelo/caelo, quicunque/quicumque, fenum/foenum, Plusquam/plus quam), for a dropped connective at the head of a quotation begun mid-verse (enim, autem), for abridgment at either end of a span, for word order inside a truncated address, and for lexical doublets that do not move the sense (dilectio/caritas, vestimentis/vestibus, Cupio/desiderium habens). All 22 such declines are listed with reasons in §3.

1.2 Non-words

I render the sense in the running English and log the printed form, its column and both candidate owners in §5; I do not wrap it, because wrapping it is [sic:] by another name and the brief forbids that here. This keeps 7a⁗ (the force stands in the running text) while leaving the plate reader a clean list. One site only: 0274C jutius.

1.3 The five inherited axes — applied from the first line, measured after

Counts are mechanical, Latin family against English family, over all six chunks.

⚑ I CHECKED THE INSTRUMENT BEFORE REPORTING, TWICE, AND BOTH SIGNALS WERE FALSE.

1. A nation grep over my Latin returns four hits — incarnationis ×3 (0272A, 0275B/0276B, 0279A) and dignatione (0268D). natio does not occur in this range at all. The single English "nation" renders Gentem nostram at 0268C, singular gens of the Jewish people, which is the class convention 4 expressly permits ("their own nation").
2. A raw usque ad grep returns eleven, not ten. The eleventh, 0277C et usque ad illos praedicatio pervenit, is ordinary prose and unitalicised — the same class as 8999's unitalicised "down to". I rendered it "as far as them" anyway, so no drift signal exists either way, but the count is 10 formula + 1 prose and a later sweep should not read 11 as 11.
3. Related: the one "throng" in my English (0268B, "is thronged about with the soldiery of the virtues") renders the verb stipatur, not turba. This is 9001's documented near-miss; do not sweep it.

⭐ No inherited convention had to be broken. No exception was forced by Luke in this range.

• gentes took Gentiles throughout because every instance in capita VII–IX is inside the Judaei/gentes typology: the centurion as gentilis populus (0268A–0269A), plebs gentium (0276C), the Gerasene demoniac read as gentes quas … Dominus per praedicatores visitavit (0275A), and Paul's scandal in gentibus (0273A). Even gentium leges at 0275B, which looks generic, sits inside that same allegory and is the laws of the Gentile peoples whom the demoniac figures.
• gentilitas (0268C errore gentilitatis, 0272A, 0276A) is a sixth word on this axis and the brief does not rule it. I rendered it "the Gentile world" (and errore gentilitatis as "the crimes of Gentile error"), never "heathenism", so it stays inside the gentes family and is distinguishable at the merge. ethnicus does not occur here, so "heathen" appears zero times and the fourth word is untouched.

1.4 Register

Pattern 7 literal throughout; Pattern 8 punctuation follows the plate (three sites in §4 where that looks like an error and is not); Pattern 17 thou for the singular (0269A to the centurion, 0271C cognoveris/confitere/expande), you for the plural (0270A vidistis, 0274B vos, 0282B Vos qui … adhaeretis, 0283C estis). No em-dashes in any of the six chunks (grep: 0). No [f:]: this is a commentary that quotes and expounds a lemma, which false positive 5 excludes.

§2. THE SPAN COUNT — I RAN IT, AND IT AGREES

data/briefs/9000-lemmata-0011.txt declares 196 spans for chunks 0011–0016.

Counting … runs in the body of my own six Latin files (frontmatter excluded):

196 = 196. The brief's split is correct for this range, at both ends. Nothing is missing at the head (my first span, [0267D] Cum autem implesset., is the first line of the brief) and nothing runs past the tail ([0283B] Renuntiare his. is the brief's last line and my last span). Frontmatter counts also agree: 65 column anchors and 26 [n:] notes in the Latin, 65 and 26 reproduced verbatim and in order in the English (checked mechanically, string-for-string).

§3. THE LEMMA DIVERGENCE LIST — all 196 spans walked, with the reason for every decline

Verdicts: = agrees with the Clementine at the verse being glossed (or is a clean truncation of it) · ≠ diverges, [var:] fired · ≈ diverges, [var:] declined, reason given · tr truncated address, nothing to collate · gl the glossator's own word, not scripture.

3.1 The 25 spans that are NOT plain agreements

3.2 The remaining 158 spans — agreements and truncated addresses

Walked line by line against the Clementine at the verse actually being glossed, not merely against the brief's ✓. They fall into three classes and none carries a divergence:

• Verse-address lemmata, complete (52): 0267D Cum autem implesset · 0268A Intravit Capharnaum. Centurionis · 0268A Misit ad eum seniores · 0268C Gentem nostram · 0268C Jesus autem ibat · 0268D Non sum dignus · 0268D Nam et ego · 0268D Nec in Israel · 0268D Invenerunt servum · 0268D Et factum est · 0268D Quae vocatur Naim · 0268D Cum autem appropinquaret · 0269C Ecce defunctus efferebatur · 0269C Et turba · 0269C Et resedit · 0269C Et nuntiaverunt · 0271B Similes sunt pueris · 0271B Venit Filius hominis manducans et bibens · 0271B Justificata est sapientia · 0271C In domo · 0271C Coepit rigare · 0271D Is cui plus · 0271D Recte judicasti · 0271D Intravi in domum · 0271D Oleo caput meum non unxisti · 0271D Remittuntur ei peccata · 0271D Et coeperunt qui simul · 0271D Fides tua te salvam fecit · 0272B Et factum est · 0272B Maria quae vocatur Magdalene · 0272D Et Joanna · 0272D Procuratoris Herodis · 0272D De facultatibus · 0272D Cum autem turba · 0272D Exiit qui seminat · 0272D Secus viam · 0273C Haec parabola · 0273C Ad tempus credunt · 0273C In terram bonam · 0273C Nemo autem · 0273C Non enim · 0274C Qui enim habet · 0274C Et cessavit, et facta est tranquillitas · 0274C Mirati sunt ad invicem · 0274C Ad regionem Gerasenorum · 0274C In monumentis · 0274C Jesu Fili Dei · 0274C In principe daemoniorum ejicit daemonia · 0274C Vinciebatur catenis · 0275C Et rogabant · 0275C Grex porcorum · 0275C Fugerunt et nuntiaverunt in civitatem, et in villas · 0275D Exierunt autem videre quod factum est, et venerunt ad Jesum, et invenerunt hominem · 0275D Et rogaverunt · 0275D Redi in domum · 0275D In medicos · 0276B Tetigit fimbriam · 0276B Et confestim · 0276B Quis est · 0276B Videns autem mulier · 0276B Flebant autem · 0276B Nolite flere · 0276B Non est mortua · 0276B Tenens manum · 0276B Puella, surge · 0276B Quibus praecepit · 0277A Nihil tuleritis · 0277C Duas tunicas · 0277C Ibi manete · 0277C Pulverem pedum vestrorum · 0277C Joannes surrexit · 0277C Et assumptis · 0277C Secutae sunt illum · 0277C Dies autem coeperat · 0277C Nisi forte · 0277C Quinque millia · 0277C Distribuit discipulis suis · 0277C Et saturati sunt · 0278D Cophini duodecim · 0278D Et factum est · 0279A Quem me dicunt esse turbae? · 0279A Vos autem quem me esse dicitis? · 0279A Joannem Baptistam · 0279A Vos autem · 0279A Christum Dei · 0279A Praecepit ne · 0279A Dicebat autem · 0279A Abneget semetipsum · 0279A Nam qui · 0279A Hunc filius hominis · 0279A Aliqui hic stantes · 0279A Donec videant regnum · 0279A Post haec verba · 0279A Assumpsit Petrum · 0280B Et factus est (see §3.1) · 0280C Et vestitus ejus · 0280C Ecce duo viri · 0280D Erant autem Moyses, et Elias visi · 0280D Dicebant excessum · 0280D Gravati erant somno · 0281A/0281B Bonum est nos hic esse (twice) · 0281B Tria tabernacula · 0281B Unum Moysi · 0281B Facta est nubes, et obumbravit eos · 0281C Hic est Filius meus · 0281C Et timuerunt · 0281C Et vox · 0281D Inventus est Jesus solus · 0281D Factum est autem in sequenti die, descendentibus illis de monte · 0281D Occurrit [0282A] illis turba · 0281D Et ecce spiritus apprehendit · 0281D Et cum accederet · 0281D Et increpavit · 0281D Ponite vos · 0281D Futurum est ut tradatur · 0282C At illi ignorabant · 0282C Intravit autem cogitatio · 0282C Me recipit · 0282C Joannes dixit · 0282C Nolite prohibere. Qui enim · 0283B Domine, vis · 0283B Nescitis cujus spiritus estis · 0283B Dixit quidam · 0283B Vulpes foveas habent · 0283B Ait autem ad alterum · 0283B Sine ut mortui · 0283B Renuntiare his. Each was read word by word against its own verse. Where a span stops short, it stops at a word boundary of the Clementine and nothing inside it differs.
• Single-word addresses (16): fluctus · commotio · Visitavit · Beatus · Quid · Justificata · Rogabat · Attulit · Maria · Centuplum · Videntes · Suffocantur · Deridebant · Narraverunt · Sanabat · Qui autem / Dico autem / Qui non / Cui ergo / Arundinem vento (short two-word addresses read the same way). Checked in place: each is the Clementine's own word at its verse. ⚑ The two exceptions are gl, not scripture: fluctus and commotio at 0268D are the glossator's Hebrew etymology of Naim, not a quotation, and must not be collated against anything. They are italicised because they are glossed words.
• The usque ad splice formula (10, at 0268D, 0269C ×2, 0270A, 0271C, 0283B ×5): Migne's own abridgment marker, not scripture. Rendered as far as in all ten, per the settled convention.

⚑ One brief-file caution, recorded because it would mislead a later reader. The brief prints the [0269C] usque ad spans and the [n:]-bearing spans with a ✓, and both are meaningless verdicts: usque ad occurs in the Clementine somewhere (in Genesis, in Kings), and a span whose last token is a note marker was searched with the marker text stripped. Neither ✓ was clearance and neither was treated as one.

§4. CRUCES BY COLUMN — Migne's own, rendered as printed, no marker

0269B — Credente Domino sanatur servus, quia potest meritum Domini etiam servis suffragari. Migne capitalises Domino/Domini, and the easy reading makes this Christ. It cannot be. The sentence's argument is that a master's merit avails for his servants ("not only by the merit of faith, but also by zeal for discipline"), and the master in this pericope is the centurion, who is the one who believed. Rendered "The master believing, the servant is healed … the merit of the master." A mechanism-5 site, decided on the argument's own structure rather than on the capital letter. No emendation is involved: dominus carries both senses and only English must choose.

0270D — Prophetico ergo exemplo ad virtutem subeundae virtutis hortatur. virtutem … virtutis in four words, and the second occurrence has no work to do. Rendered as printed ("he exhorts to the virtue of a virtue to be undergone"). I do not know whether this is a compositor's repeat of a word from the line above, a scribal one in the recension, or the abridger's own clumsiness, and I assert none of the three.

0270D — Ecce quae in veste pretiosa. Lc 7:25 reads Ecce qui in veste pretiosa sunt, and Migne prints quae. ⚑ This span is NOT italicised in the Latin, so it is not in the brief's inventory and no lemma test reaches it — but it is a quotation, and it is the only unitalicised one in the range. A relative's gender has no English exponent (Pattern 9), so nothing changes in the English; recorded so the divergence is not invisible.

0270D — exteros coelestis regni … consenescunt. exteros is accusative plural with no verb to govern it; the clause's own verb is consenescunt, intransitive. Rendered as an apposition ("strangers to the heavenly kingdom, grow old…"), which is the closest the printed case allows.

0271B — Venit sicut tunc, ita et nunc, utramque viam salutis respuitis. A second-person plural verb dropped into a third-person sentence, with no address. Kept ("you spurn both ways of salvation"); the next sentence (Lamentavimus … cantavimus) is first-person plural, so the passage is already changing person and the shift may be the abridger's compression of a homiletic original.

0271C — osculare, id est, nihil nisi sapientiam loqueris. Five imperatives (recurre, inquire, confitere, expande, osculare) and then a future indicative. Rendered "thou shalt speak", which is what is printed; loquaris would give the jussive the series wants, and I do not assert it.

0272B — Sanata aegra, de salute ejus aliis aegrotant. aliis is dative/ablative and cannot be the subject aegrotant requires. The obvious reading is alii, "others sicken over her salvation" — which is exactly what the verse says (coeperunt qui simul accumbebant dicere intra se) — and I declined to adopt it, because adopting it is the silent repair Pattern 7a exists to forbid. Rendered "they sicken over her salvation, to others", which is deliberately awkward and true to the page.

0272D — Joanna, Dominus gratia ejus, vel Dominus misericors. cujus est omne quod vivimus. A full stop mid-sentence, followed by a lowercase cujus. Preserved exactly (Pattern 8, and 9001 @0202D is the precedent: the stop there was our source's, at Migne's line break). ⚑ This is the site in my range I most want read at the plate, because a stop that appears at a line turn and a stop Migne set are indistinguishable in our file and only one of them is a defect.

0273D — vel per adversa terrendo, vel prospera blandiendo retardant. retardant is active with no expressed subject: grammatically it must be the adversa/prospera, but the two gerunds name the means, not the agent. Rendered with an unspecified "they hold them back". A mechanism-5 shape that has no safe resolution; recorded rather than resolved.

0274A — Quia voluptates et divitiae suis importunis cogitationibus guttur mentis, ne bonum desiderium ad cor possit intrare quasi aditum vitalis strangulant flatus. The word order splits vitalis … flatus across the verb and puts the purpose clause inside the object phrase. Every word construes; only the order is violent. Rendered in the order English needs, with nothing added or dropped, and logged because a later reader will meet the Latin and think something is missing.

0274B — Et bona terra cum patientia fructum reddit … et dum humiliter flagella suspiciunt, post flagella in requiem suscipiuntur. Singular terra governs the first two verbs and plural verbs govern the last two. Kept as printed; the sentence has slid from the ground to the hearers it figures.

0275A — unde Gerasa sive Gergesi colonum ejiciens, id est diabolum a quo prius incolebatur vel advena propinquans, qui longe erat in tempore. Two Hebrew etymologies run together with no main verb: Gerasa = "casting out the settler", Gergesa = "a stranger drawing near". Rendered as printed, with the two glosses left in apposition. Gergesi is a form I cannot vouch for (§5.2).

0276A — non capientes verbum Dei, et in firma mente pondus sapientiae non sustinentes. Printed as two words. It construes as printed ("not sustaining the weight of wisdom in a firm mind") and I rendered that, but infirma is equally available and would give "in an unsteady mind", which suits Conscii fragilitatis better. A word-division ambiguity is the shape our source manufactures most often (the hyphen-rejoin class), so this is a §5 entry and not a marker.

0277D — …Jesum resurrexisse negabant, multa mirabilia opera viderant, in cujus passione multa miranda contigerant. The middle clause has no connective and no relative, and then in cujus takes an antecedent the sentence never supplies in that position. Rendered as printed, asyndeton and all.

0278B — Nondum intellexerant apostoli fidem populi credentis non esse venalem, sed noverant nos potius esse redimendos per suas escas gratuitas. Nondum intellexerant … sed noverant is self-contradictory as it stands. ⛔ I did not supply a negative, which is precisely the repair 7a forbids, and I did not soften sed. Rendered "had not yet understood … but they had known". If a negative was lost here it was lost before our file, and the plate is the only thing that can say so.

0279A — incomprehensibilia paternae positionis arcana. positio is a real word and gives "the Father's position", which is not what the sentence wants; dispositionis would give "the Father's disposing". A [cj:] was available and is declined per §1.1. Rendered as printed.

0280A — visuros dilationem Ecclesiae antequam per mortem recedant de mundo. Read as dilatio = "delay", the sentence is nonsense; read as the attested medieval dilatio = dilatatio, "spreading", it says exactly what the following clause says (in praesenti contra mundi gloriam erecti, certius credantur in coelo regnaturi). I took the attested medieval sense (Pattern 9: the test is attestation, not classicism) and rendered "the spreading of the Church". This is a rendering decision, not an emendation — the printed letters are unchanged — but it is a decision, and it is logged so it can be reversed.

0280C — ita et oculis mentis tuae jam divinarum sensus albescit lectionem. divinarum is genitive plural feminine and lectionem is accusative singular: they cannot construe together, and the sentence needs lectionum. I rendered the genitive plural that the adjective demands ("the sense of the divine readings"), and I say plainly that I am rendering against one printed letter. No marker**: one letter is suspicion of our file as much as of his.

0281C — cujus perfectus ostenditur audita voce dicentis. perfectus as a bare masculine nominative has no referent; perfectio would give "whose perfection is shown". Rendered "whose perfect one is shown", which is what the case allows.

0283B — presbyteros Baal morte affecit. 3 Kings 18 has Elijah killing the prophetas Baal, and presbyteros is a real word in the wrong place. Not a lemma, so no [var:] is even available; rendered "the elders of Baal" and logged.

0283D — sed eum cujus patrem sciebat mortuum, illum patrem de quo dicitur: Obliviscere domum patris tui, etc. The sentence never reaches a main verb: sed eum … illum patrem … dangles after elegit innocentem. Rendered as printed, with the anacoluthon intact.

0284A — quid fiet illis qui nulla utilitate saepe visitat domos suorum quas dereliquerunt. Plural relative subject, singular verb, and a question that Migne closes with a full stop. Both kept: "what will become of those who often, with no profit, visits the houses of their own people which they had forsaken." The period is Pattern 8, not 8b — 8b covers an indirect question English cannot mark, and this is a direct one whose mark Migne simply does not print.

4a. MECHANISM-5 CENSUS — ALL SIX CHUNKS, not the first two

Every place in the range where a pronoun, relative or participle could bind to either of two nouns. Decided on case and gender, then checked against the sense; the easier reading is named where it differs. Sites 12–19 are in chunks 0015 and 0016, which is the half of a range that John's census never reached.

4b. NEGATION — carried, including where it hurts

All of non/nec/neque/nisi/haud/nunquam/nullus/nihil in the range appear in the English and none was added. The sites where carrying it produces awkward or contradictory English, and where I carried it anyway: 0270C (nec … nec … nec … nec, four limbs on John the reed, all four kept) · 0271B (nec psalmis … nec threnis, both kept, and the second nec governs a long parenthesis that could easily swallow it) · 0273C (nulla fide, nullo intellectu, nulla saltem — three, kept) · 0275C (nisi quis more porci vixerit … non accipit … non autem ad perdendum) · 0277B (ut nihil nisi virgam ferrent, et ut nec virgam ferrent — the gloss's point is that the two commands contradict, so flattening either kills the paragraph) · 0278B (Nondum intellexerant … sed noverant, §4: the contradiction is on the page and stays there) · 0282A (qui nec fidem confitentur, nec ipsius fidei audire volunt sermonem) · 0283A (non sacramenta … et adversum nos non sunt … et Dominum non sequuntur — four negatives in one sentence with two different scopes).

⚑ 7a⁗ is not at issue anywhere in this range, because nothing is quarantined: zero [sic:], zero [cj:], and the three [var:] sit beside a complete rendering and carry no force of their own. Every verb, negation and conjunction is in the running English.

⚑ And per the brief's closing warning, I re-read every ruling above against the English I actually shipped. The four that were most at risk of the John @0409B failure (a crux that says the right thing over an English that does not) are 0269B master, 0272B aliis, 0278B sed noverant, and 0280A dilationem: all four read in the shipped chunks exactly as this file says they do.

§5. THE DECLINED-ODDITY LIST — every suspect word I did NOT mark, in one place

⛔ I do not know whose any of these is. Each could be Migne's type or Corpus Corporum's transcription of it, and my guess carries no evidential weight — the brief gives the numbers. They are ordered by column so a plate reader can work down a page.

5.1 Non-words and broken words — the highest-value entries

5.2 Single wrong letters and wrong endings — real forms, wrong place

5.3 Real words that merely read oddly, and proper names I cannot vouch for

5.4 What I checked and did NOT log

Reported so the list can be trusted as complete rather than merely long: I did not log band-arithmetic (false positive 1 — my anchors run 0268A→0271D, 0272A→0272B, 0272C→0276B, 0276C→0277A, 0277B→0281A, 0281B→0284A, all continuous, and no text breaks mid-word or mid-construction across any of the 65 anchors); I did not log the recurring (cont.) heads at 0012, 0014 and 0016 (false positive: our chunk division); I did not log Migne's verse addresses where they sit a verse off the words they quote (0269C VERS. 13.-- Et turba quotes Lc 7:11 or 7:12, and 0272D VERS. 6.-- Supra petram takes 8:13's wording under 8:6 — Migne's numbering is his and an Ibid.-class reference is not ours to correct); and I found no ( sic ), no bracketed Forte, no <LACUNA>, no asterisk-keyed note and no foot-of-page apparatus anywhere in the six chunks, which agrees with the brief's false-positive 7.

Preserved verbatim from cruces-0017.md. Where it states a [var:] threshold of its own, §0.1 governs.

Chunks 0017, 0018, 0019, 0020 — capita X, XI (and cont.), XII. 5,748 Latin words. Written against data/briefs/9000-launch-brief.md, translation-style.md, the 9741 anchor pair, and §1 of src/english/8999/cruces.md and src/english/9001/cruces.md.

(a) THE POLICY I APPLIED — stated before the findings, so it can be argued with

a.1 Markers

I fired no [sic:]. Not one, in 5,748 words. Every suspect word in my range — non-words included — is rendered as printed and logged unmarked in §(e) for the plate reader. This is John's result carried forward deliberately, on the brief's numbers: on 8999 the 23 leaf-reads spent on fired markers produced no corrections and the nineteen declined oddities produced all seventeen. I have no evidence about the plate, and I say so at every site rather than guessing a class.

I fired no [cj:]. Two sites tempted one (0287C cadit, 0295A mansuros); both are argued and declined in §(d). Where a [cj:] would have been fired, the running English would still have had to carry the force outside the bracket — the 8999 @0415A rule — and at both sites the printed reading construes well enough that quarantining it would have been the riskier act.

I fired seven [var:], under one stated test:

⭐ Fire [var:] only where the divergence from the Clementine is larger than a single word — a substituted or omitted phrase, or two texts joined. Log every single-word lemma divergence unmarked.

That test is the brief's own asymmetry made operational: "an OCR letter-swap manufactures broken words and cannot invent a semantically coherent divergent clause." A one-word divergence is inside the class our transcription can manufacture; a missing prepositional clause or a conflation of two books is not. Three candidates were withdrawn by this test after being drafted — 0288A habet for habebit, 0293B Qui for quia, 0297A Vestiebat for vestiebatur — and each now stands in §(e) instead. Every fired marker cites sources/vulgate/clementine-flat.txt by book chapter:verse.

Final marker counts for the range: [var:] 7 · [sic:] 0 · [cj:] 0 · [d:] 0 · [ed:] 0 · [nt:] 0.

a.2 The inherited axes (§1 of the brief), applied from the first line

⚑ The one place I extended an axis rather than applied it, said plainly. gentilis is a fifth word on the gentes axis and the brief does not rule on it. I render it Gentiles, because it is the adjective of gentes and both my sites sit in the same Judaei/gentes opposition (0285B sunt pejores gentilibus, after post descriptae legis contemptum; 0289B propter derisionem gentilium, of the pagan god-name). I did not reach for "heathen", which §1.4 reserves for ethnicus — a word my range does not contain. If a later stint has ethnicus and gentilis in one sentence, this needs re-deciding; I do not.

⚑ Non-italic usque ad was left as ordinary prose, per the 8999 merge note that only the italic formula was normalised: 0285A usque adeo … ut → "to such a degree … that"; 0290C usque ad adventum Filii Dei → "until the coming of the Son of God"; 0294B usque ad arcana … penetrare → "as far as the hidden things". The one scripture lemma that opens with it — 0293B Usque ad sanguinem Zachariae, Lc 11:51 — is rendered "As far as the blood of Zacharias" and its gloss's Cur usque ad hunc follows suit, so lemma and exposition agree.

⚠ The measuring-instrument trap fired here too, and I checked before reporting. A \bturba grep over my four chunks returns three hits and only one is the noun: 0296D mentem perturbat is the verb turbare, and 0294B tanto intonante turbine is turbo, a whirlwind — rendered "at so great a thundering whirlwind", not a crowd. That is the 8999 false signal reproduced exactly in this range.

a.3 Register and structure

• Pattern 17 applied as a number marker: thou/thee/thy wherever the Latin address is grammatically singular (0285B Vae tibi; 0286B Cum audis … agnoscis; 0288C Excita ergo somnum tuum; 0292A ut aut peccare desistas; 0293B Miserere animae tuae; 0296C Quia longa tibi tempora … promittebas), you/your for every plural. No archaism is used for reverence: the singulars addressed to a sinner and to the reader take the same thou.
• Pattern 7 / 7a. Every non, nec, neque, nisi, ne, nullus, nihil, nemo the Latin prints is in the English, including the triple stack at 0297D (Non tantum non timeatis ne … desint) and the four-fold at 0288D. Nothing is smoothed, nothing supplied.
• Punctuation follows the plate (Pattern 8): Migne's commas after est at 0284A and after eorum at 0293D are kept; nothing is merged or split.
• Markers sacred. Anchors 15 / 15 / 11 / 16, in order, verbatim. Notes 3 / 5 / 2 / 9, verbatim and untranslated — no [n:] in my range is editorial prose, so no Pattern-15 [nt:] is owed. Frontmatter copied verbatim. ## heads 1:1, (cont.) untranslated. Question-mark parity 0/0, 1/1, 1/1, 2/2.
• Italic spans are 1:1 with the Latin in 0019 and 0020. 0017 and 0018 each carry one more than the Latin, both accounted for: the [var:] at 0285B quotes the Clementine in italics inside the marker (0017), and Aegyto is carried in italics under Pattern 7 (0018).
• Em-dashes: zero in running English across all four chunks. The two that remain are inside [var:] apparatus.
• ⚠ (Ephes, IV) at 0290D is not a [n:] marker but a parenthesis inside the italic run; reproduced verbatim, comma and all.

a.4 Re-reading my own rulings against the English I wrote

Per §4 of the brief (the 8999 @0409B failure: a crux that records the right decision over an English that carries the wrong one), I re-read every ruling below against the shipped text before closing. Three checks worth naming: 0290C's Cum … fortis … armatus … custodit … atrium suum renders each fragment in the order Migne splits it, so the reader can reassemble Lc 11:21; 0295C's in Filio hominis really does read "in the Son of man" and not "against"; 0297A's Vestiebat really is active in the running text ("Was clothing as one"), matching the decline recorded at §(e).

(b) SPAN COUNT AGAINST THE BRIEF'S 187

I ran the count. It agrees exactly.

data/briefs/9000-lemmata-0017.txt declares 187 spans for chunks 0017–0020, bands 0284A–0297D. 187 = 187, no discrepancy. Counted with a regex over the four Latin chunk bodies (frontmatter excluded), independently of the brief file.

(c) THE LEMMA DIVERGENCE LIST — all 187 spans, with the reason for every decline

Column and pre-scan mark are the brief's; the verdict is mine, formed by reading the printed Latin word by word against sources/vulgate/clementine-flat.txt at the verse being glossed — not at wherever the ✓ matched.

Summary: 7 divergences marked [var:] · 22 divergences or near-divergences examined and declined · 158 spans agree.

The seven [var:], gathered

⚑ The three I drafted and withdrew (0288A habet, 0293B Qui estis, 0297A Vestiebat) are in §(e), not here. Each is a single word. Under the policy in §(a.1) a single word is a claim about our file, not about Migne's plate, and I decline to make either claim.

(d) THIS RANGE'S CRUCES, BY COLUMN

### 0284A — Sicut in apostolis est, forma episcoporum Migne's comma stands between est and forma, breaking a nominal predicate. Preserved (Pattern 8); the English reads "there is the form of bishops" and the comma is not reproduced where English cannot carry it without becoming unreadable. No emendation, no marker.

### 0285B — pejores gentilibus, naturale solummodo jus dissipantibus The participle is dative plural agreeing with gentilibus, not with the cities: it is the Gentiles who squander only the natural law, and that is the whole force of pejores. Bound accordingly. (Mechanism 5 candidate; checked, not a defect.)

### 0285B — cum illae civitates praedictae, Domino praesente credere noluerunt, Tyrus et Sidon olim fuerunt amicae The sentence has no main clause answering cum: Migne joins a cum-clause to a bare perfect. Rendered as it stands ("for whereas those foresaid cities … would not believe, Tyre and Sidon were once friends…"), with nothing supplied. No marker — the words are all real and the looseness is the recension's, not a type defect.

### 0286A — ut unde illi ceciderunt sublevati, isti ascendant humiliati ⛔ Mechanism 5, and the easy reading is the wrong one. illi are the fallen angels and sublevati attaches to them ("from whence they, being lifted up, fell"); isti are the disciples and humiliati attaches to them. Reading either participle across to the other subject inverts the whole antithesis. Bound by case and number, not by sense.

### 0286B — Cum audis omnia, agnoscis omnipotentem, non decolorem, non degenerem Patris Second person singular (Pattern 17) → thou hearest / thou acknowledgest. Both negatives carried. The genitive Patris governs degenerem, not omnipotentem: "not degenerate from the Father".

### 0286D — ejusdem gentis · 0290A — illius gentis Singular gens of a people → nation, per §1.4's third class. Neither is gentes, and neither is rendered "Gentiles".

### 0287A — descendit quidem lex per Moysen in mundum, et nullam sanitatem contulit hujusmodi hujusmodi is the wounded man of the parable, not the law and not the world. Rendered "conferred no health upon such a one". Mechanism-5 site; checked.

### 0287C — ⭐ Quanto humilius sedet, tanto amplius cadit — a declined [cj:] Printed: Quanto humilius sedet, tanto amplius cadit. Aqua confluit ad convallem de tumoribus collis. The expected sense is that the lower Mary sits the more she takes in, which is what capit would give, and capit → cadit is a one-letter difference.

Declined, and here is the argument for declining. cadit is a perfectly good verb and the sentence construes without emendation if the subject is the water of the very next sentence: "the more humbly she sits, the more abundantly it falls" — water running down from the swellings of the hill into the valley. Pattern 18 exists for a real word whose faithful English misleads; here the faithful English is the image the next sentence supplies. Rendered "The more humbly she sits, the more abundantly does it fall." I record the alternative so a reader is not left to invent it, and I fire nothing. Worth a glance at the plate at 0287C, but it is a reading question, not a type question.

### 0287C — haec convivium parat Domino, in convivio Domini illa jam jucundatur haec is Martha (named second, nearer) and illa is Mary. Rendered "the latter … the former" rather than "this one … that one", which in English inverts as often as not. The alternative binding would have Mary preparing the banquet, which the same sentence denies.

### 0288A — Quis vestrum habet amicum See §(e). Present for the Vulgate's future habebit. Rendered as printed; no marker.

### 0288B — Psal. XLIV for Ps CXLIV The note keyed to Prope est Dominus … voluntatem timentium se faciet reads (Psal. XLIV); the text is Ps 144:18–19, and Ps 44 contains none of it. [n:] contents are Migne's citations and pass through verbatim and untranslated — the brief forbids correcting or querying one, and §6.3 forbids proposing corrections to references. Logged here only, so the indexer's operator knows the refKey and the refDisplay will not agree at this site.

### 0288C — Excita ergo somnum tuum excitare somnum — literally "rouse thy sleep". Rendered "Shake off thy sleep", which is the idiom's force, not a repair of a defect: nothing here is defective. Recorded because the literal English would be nonsense and I want the choice visible.

### 0289C — et domus supra domum cadet cum habitatoribus unius domus ab alterius domus familia vel opprimuntur vel puniuntur Two things at once: a singular cadet followed by plural opprimuntur/puniuntur, and a cum-clause with no conjunction to hang the plurals on. ⛔ Mechanism 5 checked: the ablative agent ab alterius domus familia is the other house's household, and the ones oppressed are habitatoribus unius domus. Getting that backwards is the easy reading and the wrong one. Rendered as printed, number mismatch included; no marker — every word is real.

### 0290A — qui sibi conscii erant nihil malarum artium a Domino didicisse qui is the apostles (the filii Judaeorum of the lemma), not the Jews who accuse them. The English binds it to the apostles. Mechanism-5 site; checked.

### 0290B — the split lemma Profecto pervenit … regnum … Dei: … Regnum Migne breaks Lc 11:20 into four italic fragments and glosses between them. All four reproduced in place and in order, with the colon after Dei preserved.

### 0290B — Hunc digitum magi confessi sunt in Aegyto See §(e). Carried in italics, untranslated, unmarked.

### 0290C — the six-fragment lemma Cum … fortis … armatus … custodit … atrium suum The gloss dismembers Lc 11:21 and glosses each word. All six fragments kept italic in the printed order with the gloss words between them, so the verse reassembles on the page.

### 0292A — quae in Ninivitis per poenitentiam … in regina austri studium percipiendae sapientiae … congregatur quae is the Church (from Ecclesiae mysterium), and studium is a bare nominative in apposition with no verb of its own. Rendered as an apposition inside dashes-free parenthesis; nothing supplied. Mechanism-5 site: quae could be pulled to poenitentiam, and is not.

### 0292B — sed Judaei veram Dei sapientiam cum ipsis conversantem, non audiunt sed blasphemant ⛔ conversantem is accusative and agrees with sapientiam: it is Wisdom that dwells among them. Binding it to Judaei would make the Jews the ones dwelling, and lose the point.

### 0293C — ne fidei vos studiosos faciat operum negligentes Both studiosos and negligentes agree with vos. Rendered "lest it make you, zealous for faith, negligent of works."

### 0293B — Qui estis ut monumenta · Repondens autem See §(e). Both rendered as printed; no markers.

### 0294B — Hanc legisperiti, nec ipsi lege et prophetis intelligunt A bare ablative lege et prophetis where in lege et prophetis is expected. Rendered "neither themselves understand by the law and the prophets". Real words; a possible lost in. Logged in §(e).

### 0294B — Intrare clavem, est non esse contentum superficie litterae intrare with an accusative object clavem ("to enter the key"). Real words, and the sense is recoverable — to enter by the key. Rendered literally ("To enter the key is not to be content with the surface of the letter"), nothing supplied. Logged in §(e).

### 0295A — ⭐ vel mansuros in denuntiatione supplicii — the second declined [cj:] Printed: Totus hic locus ad subeundam pro confessione Domini passionem vel contemptu mortis, vel spe praemii, vel mansuros in denuntiatione supplicii instruit. The first two limbs are ablatives of means (contemptu, spe); the third is a future participle in the accusative plural with nothing to agree with. It does not construe.

Declined. Every word is real and I can see at least two readings (mansuros as a second object of instruit — "instructs also those who will abide"; or a corruption of an ablative in the series). Choosing one in a [cj:] would be exactly the confident wrong call the brief prices. Rendered as literally as the printed words allow — "either by contempt of death, or by hope of reward, or those who will abide, in the denunciation of punishment" — which is deliberately as awkward in English as it is in Latin. Plate read owed at 0295A.

### 0295C — Et omnis qui dicit verbum in Filio hominis remittetur ⛔ The most consequential single word in my range. Lc 12:10 reads in Filium hominis — a word spoken against the Son of man. The ablative makes it a word spoken in him. The gloss's own surrounding argument (Confitetur Filius apud Patrem illum…, and the Ei autem that follows about the unforgivable sin) works either way, so the context does not decide it. Rendered as printed and not marked; see §(e).

### 0295D — Commendanti gratiam supernae pacis et unitatis, vult iste ingerere molestiam The dative participle is the Lord who commends; iste is the man of Lc 12:13. Bound accordingly. Mechanism-5 site; checked.

### 0295D — Inter fratres, patrimonium, non judex medius, sed pietas debet sequestra dividere, quamvis immortalitatis patrimonium non sit pecunia hominibus petendum sequestra is a feminine noun in apposition to pietas ("affection as arbitress"); the final clause has patrimonium (neuter) with petendum but pecunia (feminine) as predicate, so it does not agree cleanly. Rendered literally, gender mismatch and all — Pattern 9: a Latin gender mismatch has no English exponent, there is nothing to emend and nothing to mark.

### 0296D — nec communes fructus aliquo sibi dominatu specialius vindicantibus The participle agrees with volatilibus, the birds, not with the providence of God. Checked.

### 0297A — Vestiebat sicut unum See §(e). Rendered active, as printed.

### 0297C — quibus si haec subtrahuntur, ad probationem est, si dantur ad gratiarum actionem quibus is the faithful of the preceding clause. Note Migne prints no comma before ad gratiarum actionem; not supplied.

### 0298B — the chunk ends mid-argument at aliis exemplum praebeatis Chunk 0020's last column band is 0298B and the sentence closes there. Not a defect; the chunk division is ours.

(e) DECLINED-ODDITY LIST — every suspect word I did NOT mark

⛔ This is the list the plate reader works from. On 8999 the nineteen entries of this list produced all seventeen corrections the work received, while the fired markers produced none. Each entry says what is odd, names both possible owners, and asserts neither.

⚑ Priority for a plate reader, if the range gets fewer leaves than columns: entries 4 (0295C, sense-bearing), 1 (0290B, non-word with a clean control one column away), 2 (0293B, non-word), 9 (0295A, does not construe), then 3, 5, 6.

⚑ What this list cannot reach. The brief's second mechanism class — Corpus Corporum silently repairing Migne's broken type — leaves no trace in our file by construction. Nothing in my range can be offered against it, and a clean-looking Latin here is not evidence of a clean plate.

(f) Open items handed forward

1. gentilis → Gentiles is my extension of §1.4, not the brief's ruling (§a.2). Ratify or overturn it at the merge; two sites, 0285B and 0289B.
2. Seven [var:] stand on six unread columns — 0285A/B, 0289C, 0290C–D, 0292D, 0296A, 0297B, 0297C. plate-gate.mjs will hold the work until they are read.
3. Nine columns owed a plate read from §(e): 0288A, 0290B, 0293B, 0295A, 0295C, 0297A, and (low priority) 0284D, 0287C, 0294B.
4. Ps CXLIV cited as Ps XLIV at 0288B — an indexing note, not a text correction.

Preserved verbatim from cruces-0021.md. Where it states a [var:] threshold of its own, §0.1 governs.

Stint range: src/latin/9000/0021.md … 0025.md · cols 0298B–0310C · 4,988 Latin words (5,255 by wc -w including frontmatter) · CAPUT XII (cont.), XIII, XIII (cont.), XIV, XIV (cont.).

Apparatus fired in this range: 4 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:]. Column anchors reproduced 1:1 (13 · 16 · 4 · 15 · 1 = 49, diffed against the Latin, all in order). Notes reproduced 1:1 and verbatim (1 · 2 · 0 · 9 · 0 = 12); none is editorial prose, so no [nt:] is owed.

(a) THE POLICY THIS STINT APPLIED — stated before the findings, so it can be argued with

1. No [sic:], none, on any ground. The brief's measured finding governs: on John, 29,727 words produced zero fired [sic:], the 23 leaf-reads spent on fired markers produced no corrections, and the nineteen DECLINED oddities produced all seventeen. Every suspect word in this range — real word reading oddly, non-word, fused or split run alike — is rendered as printed, logged in §(e), and left unmarked, with no claim about whose defect it is. I do not know whether any of them is Migne's type or Corpus Corporum's transcription, and I say so site by site rather than in a preamble only.
2. No [cj:] either. Pattern 18's bar is that the printed word must make the sentence assert something the author is arguing against, and 18a's extension reaches negations and correlatives only. Four sites in this range are strong conjecture candidates (§e.1, e.4, e.11, e.12) and none of them is a negation or a correlative; each is a word that leaves the sentence obscure rather than false. All four are declined and logged with the conjecture named, which is where 18a says an uncertain reading belongs.
3. [var:] is the one marker I fire, and only where a word or clause is REPLACED by different matter that changes what the verse says. Each of the four rests on sources/vulgate/clementine-flat.txt, cited book chapter:verse, and each sits on a quoted clause rather than a single word — the class the brief's table shows standing 88 of 88 across three Glossa gospels. Declined, explicitly: abridgment (a lemma truncated, or closing in etc.), orthography (sepes/saepes, Usquedum/usque dum), preposition variants (super/supra, twice), and every single-word substitution — ei for et, esse for est, an added se. A one-word [var:] on an unread column is the trap 9003's collation exposed, where the digitization manufactured the divergence a collation then dutifully found.
4. Migne's wrong citation ADDRESSES take no marker. Three notes in this range point at chapters that cannot be right (Rom. XXXV, Rom. VIII for Rom 13:13, II Cor. IV for 1 Cor 4:9). Notes are reproduced verbatim and untranslated; the words Migne quotes agree with the Clementine in two of the three, so there is no divergence to mark — only an address to log (9001's precedent at 0192B, where a wrong address alone took no [var:]).
5. Pattern 7 literalism, including where it hurts. nihil mundamini, petierunt, domum intellectus, cognitione conjunctos, operatur stranded at a sentence end, affectum for affectuum — all rendered as the printed words allow, none repaired, none smoothed by a substituted English word. Where the printed sentence will not construe cleanly the English is correspondingly hard, and the crux says so.
6. Punctuation follows the plate (Pattern 8). Migne's ? closing the fragment Qua hora non putatis Filius hominis veniet? rides as printed on what is a statement. The paragraph at 0304A ending …noxiae inclinationis erigi has no terminal stop in the Latin and none was supplied in the English.
7. §1 conventions carried from Mark and John, from the first line. turba → crowd (5 sites), multitudo → multitude (1 site, 0298B, and it is the only one); daemonium → demon (0305C, twice), diabolus → devil (0301B, twice); gentes → Gentiles in the Judaei/gentes opposition (0304B, 0304A, 0307A, 0309B), singular gens of the Jewish people → nation (0302C infructuosa gente Judaeorum, 0306B tulerunt locum et gentem); no natio occurs anywhere in this range — checked with word boundaries, because a bare grep for natio matches damnatio, inclinatio and examinatio, which is the instrument error the brief warns about, and it fired here nine times out of eleven. mysterium → mystery, sacramentum → sacrament, 1:1. usque ad → "as far as", every occurrence: two splice formulas (0299C in the AMBR. extract, 0308D in the AUG. extract) and two ordinary uses (0300A usque ad finem saeculi, 0300B usque ad perfectionem).
8. The gloss's own voice is not archaised. dicit/ait → says/said, no saith. Every -eth form in the range was audited: all of them sit inside an italic scripture lemma (cometh, knocketh, possesseth). One -eth in the gloss voice was written and then removed — "it behoveth thee" at 0300C is now "thou oughtest", which keeps Pattern 17's required -est on a singular address without archaising a third-person gloss verb. The -est forms that remain (goest, wouldest, mayest, purposest) are all Pattern 17 singulars.
9. Pattern 17 in the second person. The gloss switches number inside single paragraphs and the English follows the morphology, not the addressee: thou at 0300C (posthabeas), 0301B (vadis, praeparas, tuae carni), 0307A–0308A (invitatus fueris, te junxeris, extollas, incipias, distribueris); you at 0301C (vivite … damnemini) and 0309B (Voca pauperes context, plural imperatives).
10. The Greek is reproduced in Greek. Ὕδωρ stands at the head of 0024 exactly as printed, space before the comma included, not transliterated and not translated away (§d, 0306C).

(b) SPAN COUNT AGAINST THE BRIEF — 142 vs 142, COUNTED

data/briefs/9000-lemmata-0021.txt declares 142 spans for chunks 0021–0025. I counted the … runs in the body of my own five Latin chunks (frontmatter excluded):

142 = 142. The brief's split is correct for this range, and every line of it is walked below, in the brief's own order, agreements included — because a report of findings alone cannot be told apart from a report by an agent that never looked.

(c) THE LEMMA DIVERGENCE LIST — all 142 spans, with the reason for every decline

Chunk 0021 — spans 1–37 (bands 0298B–0301C)

1. [0298B] VERS. 36.-- Exspectantibus. — single word, check in place → AGREES Lc 12:36.

1. [0298B] Quando revertatur. ✓ Clementine verbatim → AGREES Lc 12:36.

1. [0298B] Ut cum venerit, confestim aperiant ei. ⚠ NOT in Clementine verbatim — CHECK → DIVERGES — [var:] FIRED. Vulg Lc 12:36 ut, cum venerit et pulsaverit, confestim aperiant ei. Migne's lemma drops et pulsaverit from the middle of the run, and the gloss immediately after it expounds precisely the knocking (Non vult aperire judici pulsanti). An internal omission, not a truncation, and the exposition rests on the words that are gone: fired, cited Lc 12:36.

1. [0298B] Beati servi illi, quos cum [0298C] venerit, ✓ Clementine verbatim → AGREES Lc 12:37. Cue truncated, closing in etc.

1. [0298B] VERS. 37.-- Vigilantes. — single word, check in place → AGREES Lc 12:37.

1. [0298B] Amen dico vobis. ✓ Clementine verbatim → AGREES Lc 12:37.

1. [0298B] Et si venerit in secunda, ✓ Clementine verbatim → AGREES Lc 12:38. Truncated before vigilia.

1. [0298D] Hoc autem scitote quoniam si sciret paterfamilias. ✓ Clementine verbatim → AGREES Lc 12:39.

1. [0298D] VERS. 40.-- Qua hora non putatis Filius hominis veniet? ✓ Clementine verbatim → AGREES Lc 12:40 (veniet is the Clementine's own reading). Migne closes the fragment with ? on what is a statement; the mark rides as printed (Pattern 8).

1. [0299A] VERS. 41.-- Ait autem ei Petrus. ⚠ NOT in Clementine verbatim — CHECK → DIVERGES, one word — DECLINED. Vulg Lc 12:41 Ait autem et Petrus; Migne prints ei. A one-word divergence on an unread column is the class 9001's cruces names as the trap the manufactured judicabit/judicavit set: it is as likely to be our transcription as his type, and [var:] would make it a public claim about the plate. Logged, not marked; the English reads "And Peter said to him."

1. [0299A] VERS. 42.-- Dixit autem Dominus. ✓ Clementine verbatim → AGREES Lc 12:42.

1. [0299A] Ignem veni mittere in terram. Quis putas esse fidelis dispensator? ⚠ NOT in Clementine verbatim — CHECK → DIVERGES, one word — DECLINED. Vulg Lc 12:42 Quis, putas, est fidelis dispensator; Migne prints esse, and the English renders the infinitive it is ("Who dost thou think to be the faithful steward?"). The span also carries the head of 12:49 before it: that is the recension's own splice of the two verses it is about to expound together, not a divergence in either.

1. [0299B] Et prudens quem constituit Dominus super familiam suam, ut, ⚠ NOT in Clementine verbatim — CHECK → DIVERGES, one word — DECLINED. Vulg Lc 12:42 supra familiam suam; Migne super. A preposition variant with no change of sense; the two alternate freely in medieval copies (Pattern 9).

1. [0299B] VERS. 43.-- Beatus. — single word, check in place → AGREES Lc 12:43.

1. [0299C] VERS. 44.-- Super omnia quae possidet. ⚠ NOT in Clementine verbatim — CHECK → DIVERGES, one word — DECLINED. Vulg Lc 12:44 supra omnia quae possidet; Migne super. Same class as the preceding and the same decline. ⚑ Two instances in one column is itself worth the plate reader's eye.

1. [0299C] VERS. 45.-- Quod si dixerit servus. ✓ Clementine verbatim → AGREES Lc 12:45. Truncated before ille.

1. [0299C] Et edere. ✓ Clementine verbatim → AGREES Lc 12:45.

1. [0299C] VERS. 46.-- Partemque. — single word, check in place → AGREES Lc 12:46.

1. [0299C] VERS. 48.-- Paucis. — single word, check in place → AGREES Lc 12:48.

1. [0299C] usque ad ✓ Clementine verbatim → Not a lemma: the abridgment's splice formula inside the [n: (AMBR.)] extract, rendered as far as (§1.6).

1. [0300A] Omni autem. ✓ Clementine verbatim → AGREES Lc 12:48.

1. [0300A] VERS. 49.-- Ignem, — single word, check in place → AGREES Lc 12:49.

1. [0300A] Baptismo autem habeo, ✓ Clementine verbatim → AGREES Lc 12:50.

1. [0300B] Et quomodo. ✓ Clementine verbatim → AGREES Lc 12:50.

1. [0300B] Usque dum perficiatur. ✓ Clementine verbatim → AGREES Lc 12:50.

1. [0300C] VERS. 51.-- Putatis. — single word, check in place → AGREES Lc 12:51.

1. [0300C] VERS. 52.-- In domo una. ✓ Clementine verbatim → AGREES Lc 12:52.

1. [0300C] Divisi. — single word, check in place → AGREES Lc 12:52.

1. [0300C] Tres in duos. ✓ Clementine verbatim → AGREES Lc 12:52.

1. [0300C] VERS. 54.-- Dicebat autem. ✓ Clementine verbatim → AGREES Lc 12:54. Truncated before et ad turbas.

1. [0300C] Cum videritis. ✓ Clementine verbatim → AGREES Lc 12:54.

1. [0300C] Nimbus venit. ✓ Clementine verbatim → AGREES Lc 12:54.

1. [0300C] VERS. 56.-- Hypocritae, — single word, check in place → AGREES Lc 12:56.

1. [0300C] VERS. 58.-- Cum autem vadis. ✓ Clementine verbatim → AGREES Lc 12:58.

1. [0300C] Cum adversario. ✓ Clementine verbatim → AGREES Lc 12:58.

1. [0301C] Da operam, ✓ Clementine verbatim → AGREES Lc 12:58.

1. [0301C] VERS. 59.-- Reddas. — single word, check in place → AGREES Lc 12:59.

Chunk 0022 — spans 38–74 (bands 0301C–0305C)

1. [0301C] VERS. 1.-- Aderant autem. ✓ Clementine verbatim → AGREES Lc 13:1.

1. [0301D] Nuntiantes. — single word, check in place → AGREES Lc 13:1.

1. [0301D] De Galilaeis. ✓ Clementine verbatim → AGREES Lc 13:1.

1. [0301D] VERS. 3.-- Omnes, — single word, check in place → AGREES Lc 13:3.

1. [0301D] VERS. 4.-- Sicut illi decem. ✓ Clementine verbatim → AGREES Lc 13:4.

1. [0302B] missus, — single word, check in place → AGREES Jo 9:7 (Siloe, quod interpretatur Missus). The single word is the standing interpretation of the name, not a Luke lemma.

1. [0302B] VERS. 6.-- Arborem fici. ✓ Clementine verbatim → AGREES Lc 13:6.

1. [0302B] Qui sedebant sub umbra mortis lumen ortum est eis [n: (Isa. IX)] ⚠ NOT in Clementine verbatim — CHECK → DIVERGES — [var:] FIRED. Vulg Is 9:2 habitantibus in regione umbrae mortis, lux orta est eis; the qui sedeba(n)t form belongs to Mt 4:16 (sedentibus in regione umbrae mortis), and neither witness reads sub umbra or lumen. Three words replaced in a clause quoted as a proof-text under an explicit [n: (Isa. IX)]. Fired, cited Is 9:2, with the Matthew parallel named inside the marker.

1. [0302B] VERS. 7.-- Quaerens fructum. Vinea Domini sabaoh, domus Israel est, ⚠ NOT in Clementine verbatim — CHECK → Lc 13:6–7 (Quaerens fructum) AGREES. The second sentence is Is 5:7, where the Vulgate reads Vinea enim Domini exercituum domus Israel est: Migne keeps the Hebrew word where the Vulgate translates it. [var:] DECLINED — sabaoth and exercituum are the same thing in two languages, so nothing a reader takes away changes; and the form as printed, sabaoh, is a non-word, which is a question about type and not about the received text (§e.6). Rendered as printed inside the italic lemma.

1. [0302B] Dixit autem ad cultorem vineae. ✓ Clementine verbatim → AGREES Lc 13:7.

1. [0302B] Ut quid etiam terram, ✓ Clementine verbatim → AGREES Lc 13:7.

1. [0302D] Vae vobis, Scribae et Pharisaei, ✓ Clementine verbatim → AGREES Mt 23:13 (and the repeated formula through Mt 23).

1. [0302D] VERS. 8.-- Usquedum fodiam, ⚠ NOT in Clementine verbatim — CHECK → AGREES Lc 13:8. The ⚠ is word division only: the Clementine sets usque dum, Migne Usquedum.

1. [0303A] Mittam stercora. ✓ Clementine verbatim → AGREES Lc 13:8.

1. [0303A] VERS. 9.-- Et si quidem. ✓ Clementine verbatim → AGREES Lc 13:9 (Clementine et siquidem; spacing only).

1. [0303A] Sin autem. ✓ Clementine verbatim → AGREES Lc 13:9.

1. [0303A] VERS. 10.-- Erat autem docens. ✓ Clementine verbatim → AGREES Lc 13:10.

1. [0303A] Sabbatis. — single word, check in place → AGREES Lc 13:10.

1. [0303C] VERS. 11.-- Ecce mulier, ✓ Clementine verbatim → AGREES Lc 13:11.

1. [0303C] Annis, — single word, check in place → AGREES Lc 13:11.

1. [0303D] VERS. 14.-- Quia Sabbato. ✓ Clementine verbatim → AGREES Lc 13:14.

1. [0303D] VERS. 15.-- Hypocritae. — single word, check in place → AGREES Lc 13:15.

1. [0303D] Non solvit bovem. ✓ Clementine verbatim → AGREES Lc 13:15. Truncated before suum.

1. [0303D] VERS. 16.-- Hanc autem, ✓ Clementine verbatim → AGREES Lc 13:16.

1. [0303D] VERS. 17.-- Erubescebant omnes. ⚠ NOT in Clementine verbatim — CHECK → ⚠ RESOLVED AGAINST OUR CONTROL FILE, NOT AGAINST MIGNE. clementine-flat.txt prints erubescant at Lc 13:17, a subjunctive in a narrative clause; Migne's Erubescebant is the ordinary reading of the verse. The pre-scan's ⚠ is a fact about the control, so no divergence is asserted against Migne and no marker is fired. Worth a line in any future collation: the control has an oddity here.

1. [0304B] Adversarii. — single word, check in place → AGREES Lc 13:17.

1. [0304B] Omnis populus gaudebat. ✓ Clementine verbatim → AGREES Lc 13:17.

1. [0304B] VERS. 18.-- Dicebat ergo. ✓ Clementine verbatim → AGREES Lc 13:18.

1. [0304C] VERS. 19.-- Simile est, ✓ Clementine verbatim → AGREES Lc 13:19.

1. [0304D] VERS. 21.-- Quod acceptum, ✓ Clementine verbatim → AGREES Lc 13:21.

1. [0304D] VERS. 22.-- Et ibat, ✓ Clementine verbatim → AGREES Lc 13:22.

1. [0304D] VERS. 24.-- Contendite intrare, ✓ Clementine verbatim → AGREES Lc 13:24.

1. [0305A] VERS. 25.-- Cum autem, ✓ Clementine verbatim → AGREES Lc 13:25.

1. [0305A] Incipietis foris stare. ✓ Clementine verbatim → AGREES Lc 13:25.

1. [0305A] Nescio vos. ✓ Clementine verbatim → AGREES Lc 13:25.

1. [0305B] VERS. 26.-- Manducavimus, — single word, check in place → AGREES Lc 13:26.

1. [0305B] In plateis, ✓ Clementine verbatim → AGREES Lc 13:26.

Chunk 0023 — spans 75–88 (bands 0305C–0306C)

1. [0305C] VERS. 28.-- Ibi erit fletus, ✓ Clementine verbatim → AGREES Lc 13:28.

1. [0305C] VERS. 29.-- Et venient. ✓ Clementine verbatim → AGREES Lc 13:29.

1. [0305C] VERS. 32.-- Dicite vulpi, ✓ Clementine verbatim → AGREES Lc 13:32. Truncated before illi.

1. [0305C] Ecce ejicio daemonia, ✓ Clementine verbatim → AGREES Lc 13:32.

1. [0305C] VERS. 33.-- Verumtamen oportet. ✓ Clementine verbatim → AGREES Lc 13:33.

1. [0305C] Hodie et cras, ✓ Clementine verbatim → AGREES Lc 13:33.

1. [0305C] VERS. 34.-- Jerusalem, — single word, check in place → AGREES Lc 13:34.

1. [0305C] Quoties volui congregare. ✓ Clementine verbatim → AGREES Lc 13:34.

1. [0306B] Filios tuos. ✓ Clementine verbatim → AGREES Lc 13:34.

1. [0306B] Quoties volui congregare filios tuos, ✓ Clementine verbatim → AGREES Lc 13:34.

1. [0306B] Quemadmodum avis, ✓ Clementine verbatim → AGREES Lc 13:34.

1. [0306B] VERS. 35.-- Domus vestra. ✓ Clementine verbatim → AGREES Lc 13:35.

1. [0306B] Non videbitis me, ✓ Clementine verbatim → AGREES Lc 13:35.

1. [0306B] Cum dicetis, ✓ Clementine verbatim → AGREES Lc 13:35.

Chunk 0024 — spans 89–139 (bands 0306C–0310B)

1. [0306C] Et factum est, ✓ Clementine verbatim → AGREES Lc 14:1.

1. [0306C] VERS. 3.-- Si licet, ✓ Clementine verbatim → AGREES Lc 14:3.

1. [0306C] VERS. 4.-- At illi tacuerunt. ✓ Clementine verbatim → AGREES Lc 14:4.

1. [0306C] Ipse vero. ✓ Clementine verbatim → AGREES Lc 14:4.

1. [0306C] Omne opus servile non facietis in eo [n: (Lev. XXIII)] ✓ Clementine verbatim → AGREES Lv 23:7 verbatim (and 23:21, 23:25, 23:35, 23:36).

1. [0306C] VERS. 5.-- Cujus vestrum. ✓ Clementine verbatim → AGREES Lc 14:5.

1. [0307A] Bos in puteum. ✓ Clementine verbatim → AGREES Lc 14:5.

1. [0307A] Omnes enim peccaverunt, et egent gratia Dei [n: (Rom. XXXV)] ⚠ NOT in Clementine verbatim — CHECK → DIVERGES — [var:] FIRED. Vulg Rom 3:23 omnes enim peccaverunt, et egent gloria Dei; Migne prints gratia. A whole word replaced in a clause adduced as a proof-text, and the substituted word is doctrinally live in its own right. Fired, cited Rom 3:23. ⚑ The note's address, (Rom. XXXV), names a chapter Romans does not have — reproduced verbatim, no marker (§e.15).

1. [0307A] VERS. 7.-- Dicebat autem. ✓ Clementine verbatim → AGREES Lc 14:7.

1. [0307A] VERS. 8.-- Cum invitatus fueris: ✓ Clementine verbatim → AGREES Lc 14:8.

1. [0307A] Ne forte honoratior. ✓ Clementine verbatim → AGREES Lc 14:8.

1. [0307A] Exaltatus autem humiliatus sum et confusus [n: (Psal. LXXXVII)] ⚠ NOT in Clementine verbatim — CHECK → DIVERGES — [var:] FIRED. Vulg Ps 87:16 exaltatus autem, humiliatus sum et conturbatus; Migne prints confusus. A word replaced in a verse quoted with its address, and it is the word the exposition leans on: the gloss around it is about shame (cum rubore, erubescens). Fired, cited Ps 87:16.

1. [0307A] VERS. 9.-- Et tunc incipias, ✓ Clementine verbatim → AGREES Lc 14:9.

1. [0307A] VERS. 10.-- Tunc erit tibi gloria, ✓ Clementine verbatim → AGREES Lc 14:10.

1. [0307A] Mihi autem nimis honorificati sunt amici tui, Deus [n: (Psal. CXXXVIII)] ✓ Clementine verbatim → AGREES Ps 138:17 verbatim; Migne stops before nimis confortatus est principatus eorum.

1. [0307A] VERS. 12.-- Dicebat autem. ✓ Clementine verbatim → AGREES Lc 14:12.

1. [0307D] Et ei qui se invitaverat. ⚠ NOT in Clementine verbatim — CHECK → DIVERGES, one word — DECLINED. Vulg Lc 14:12 Dicebat autem et ei, qui invitaverat; Migne adds a reflexive se. The addition only makes explicit the object the Vulgate leaves understood, and the gloss's own paraphrase (qui forsitan superbire poterat, quia tot ad convivium … vocaverat) reads the same either way.

1. [0307D] Noli vocare amicos. ✓ Clementine verbatim → AGREES Lc 14:12. Truncated before tuos.

1. [0307D] Non in comessationibus [0308A] et ebrietatibus [n: (Rom. VIII)] ✓ Clementine verbatim → AGREES Rom 13:13 verbatim — the pre-scan's ✓ is genuine at the right verse. ⚑ Migne's address (Rom. VIII) is wrong for Rom 13:13; reproduced verbatim as a note, no marker (§e.17).

1. [0308A] Ne forte. ✓ Clementine verbatim → AGREES Lc 14:12.

1. [0308A] VERS. 13.-- Voca pauperes. ✓ Clementine verbatim → AGREES Lc 14:13.

1. [0308A] VERS. 14.-- In resurrectione. ✓ Clementine verbatim → AGREES Lc 14:14.

1. [0308A] VERS. 15.-- Haec cum audisset, ✓ Clementine verbatim → AGREES Lc 14:15.

1. [0308A] Beatus, — single word, check in place → AGREES Lc 14:15.

1. [0308A] Si quis manducaverit ex hoc pane, vivet in aeternum [n: (Joan. VI)] ✓ Clementine verbatim → AGREES Jo 6:52 verbatim.

1. [0308A] VERS. 16.-- Homo quidam, ✓ Clementine verbatim → AGREES Lc 14:16.

1. [0308C] Coenam magnam. ✓ Clementine verbatim → AGREES Lc 14:16.

1. [0308C] VERS. 17.-- Et misit servum. ✓ Clementine verbatim → AGREES Lc 14:17. Truncated before suum.

1. [0308C] in quos fines saeculorum devenerunt [n: (I Cor. X)] ✓ Clementine verbatim → AGREES 1 Cor 10:11 verbatim.

1. [0308C] VERS. 18.-- Et coeperunt, ✓ Clementine verbatim → AGREES Lc 14:18.

1. [0308C] Villam emi. ✓ Clementine verbatim → AGREES Lc 14:18.

1. [0308C] usque ad ✓ Clementine verbatim → Not a lemma: the splice formula inside the [n: (AUG.)] extract, rendered as far as.

1. [0309A] Rogo te. ✓ Clementine verbatim → AGREES Lc 14:18.

1. [0309A] Rogo te, habe me excusatum, ✓ Clementine verbatim → AGREES Lc 14:18.

1. [0309A] VERS. 19.-- Juga boum. ✓ Clementine verbatim → AGREES Lc 14:19.

1. [0309A] Eo probare illa. ✓ Clementine verbatim → AGREES Lc 14:19.

1. [0309A] VERS. 20.-- Uxorem duxi. ✓ Clementine verbatim → AGREES Lc 14:20.

1. [0309A] VERS. 22.-- Factum est, ✓ Clementine verbatim → AGREES Lc 14:22.

1. [0309A] VERS. 23.-- Exi in vias et sepes, ⚠ NOT in Clementine verbatim — CHECK → AGREES Lc 14:23. The ⚠ is orthography only: Clementine saepes, Migne sepes, the ordinary medieval spelling (Pattern 9). No marker.

1. [0309A] VERS. 25.-- Ibant autem turbae, ✓ Clementine verbatim → AGREES Lc 14:25. Truncated before multae.

1. [0309A] VERS. 26.-- Si quis venit. ✓ Clementine verbatim → AGREES Lc 14:26.

1. [0309A] Adhuc. — single word, check in place → AGREES Lc 14:26.

1. [0309D] VERS. 27.-- Et qui non bajulat. ✓ Clementine verbatim → AGREES Lc 14:27.

1. [0309D] Et venit post me. ✓ Clementine verbatim → AGREES Lc 14:27.

1. [0309D] VERS. 28.-- Quis enim ex vobis. ✓ Clementine verbatim → AGREES Lc 14:28.

1. [0309D] Turrim aedificare. ✓ Clementine verbatim → AGREES Lc 14:28.

1. [0310A] Computat sumptus. ✓ Clementine verbatim → AGREES Lc 14:28.

1. [0310A] VERS. 29.-- Omnes qui vident. ✓ Clementine verbatim → AGREES Lc 14:29.

1. [0310A] Facti sumus spectaculum mundo, angelis et hominibus [n: (II Cor. IV)] ⚠ NOT in Clementine verbatim — CHECK → DIVERGES in order and connectives — DECLINED. Vulg 1 Cor 4:9 quia spectaculum facti sumus mundo, et angelis, et hominibus; Migne re-orders the verb and drops the two et. Nothing a reader of either takes away differs, and Pattern 14 fires on sense, not cadence. ⚑ The address (II Cor. IV) names the wrong epistle for a 1 Corinthians verse; verbatim, no marker (§e.16).

1. [0310A] VERS. 31.-- Aut quis, ✓ Clementine verbatim → AGREES Lc 14:31.

1. [0310B] Si possit. ✓ Clementine verbatim → AGREES Lc 14:31.

Chunk 0025 — spans 140–142 (band 0310B)

1. [0310B] VERS. 33.-- Sic ergo, ✓ Clementine verbatim → AGREES Lc 14:33.

1. [0310B] VERS. 34.-- Bonum. — single word, check in place → AGREES Lc 14:34. Truncated after the first word of Bonum est sal.

1. [0310B] VERS. 35.-- Neque in sterquilinium. ✓ Clementine verbatim → AGREES Lc 14:35. Migne quotes only the second limb of Neque in terram, neque in sterquilinium.

(d) THIS RANGE'S OWN CRUCES, BY COLUMN

### 0298B — nihil mundamini (0021) Sive vitatis mala sive facitis bona, nihil mundamini, sed solum adventum Redemptoris exspectetis. Rendered "you are cleansed in nothing". mundamini is a well-formed passive, but "whether you shun evils or do good, you are cleansed in nothing" argues against the sentence's own second limb. A conjecture that construes is nihil moremini, "delay not at all, but await only the coming of the Redeemer" — one minim's difference. No marker: the printed word is real, the sentence is obscure rather than false, and Pattern 18a keeps that class in the crux. Logged for the plate (§e.1).

### 0298B — the Lc 12:36 lemma, [var:] FIRED See §c.3. The gloss's judici pulsanti expounds words the printed lemma does not carry.

### 0298D — exteriorum damnatorum similitudinem (0021) Rendered "the likeness of outward condemned men". The simile actually introduced is the householder and the thief, i.e. outward losses, which is damnorum; damnatorum is a real word and is rendered as one. No marker (§e.2).

### 0299C — the anacoluthon at VERS. 43 (0021) Iste autem qui repertus fuerit fideliter annonam verbi ministrat, et in curam gregis vigilans, aeterna beatitudine glorificabitur. The relative clause takes an indicative ministrat where the construction wants a participle, so the sentence has two finite verbs and no clean subject. The English carries the join as printed ("he who shall have been found faithfully ministers … and, watching …, shall be glorified"). Nothing is emended (§e.3).

### 0300A, 0300B — usque ad used four ways in two columns Two are the recension's splice formula inside patristic extracts (0299C AMBR., 0308D AUG.); two are ordinary prepositional uses in the gloss's own prose (usque ad finem saeculi, usque ad perfectionem). All four render "as far as" per §1.6, which reads naturally in each. Recorded because a later sweep looking for the formula will find four hits in this range and only two are the formula.

### 0301B — ne contemptus te accuset (0021), a mechanism-5 site contemptus is nominative masculine and agrees with sermo Dei, not with te. The English reads "lest, being despised, it accuse thee" — the word of God accusing, not the reader being accused for contempt. The easier reading here is the wrong one, which is the class the brief names as dominant.

### 0301D — petierunt for perierunt (0022) nuntiatur ei de Galilaeis, quod quasi juste ideo petierunt, quod non bene se praeparaverunt. Rendered "who as it were justly sought this for the cause that…". The same sentence continues et illos vere fuisse peccatores et juste periisse, and the next clause alios peccatores perituros, so perierunt is what the argument wants — one letter away. No marker, no [cj:]: the printed word is real, and internal corroboration proves an error exists, never whose it is (§e.4).

### 0302A — ex iota et ita (0022) The Greek for eighteen is ιη´, iota + eta. Migne prints ita, which is also the ordinary medieval Latin NAME of the letter η (Byzantine ita), so this is very possibly correct as printed. Rendered as printed, "by iota and ita", and flagged only so a plate reader can confirm the sort (§e.5).

### 0302B — Vinea Domini sabaoh (0022) A non-word: sabaoh for sabaoth. It sits inside an italic lemma, where a second italic cannot nest, and it is carried as printed. It could be Migne's dropped final sort or a Corpus Corporum loss; I do not know which and assert neither. No [sic:] (§e.6). The Vulgate's own exercituum at Is 5:7 is discussed at §c, span 46, and took no [var:].

### 0302B — the Is 9:2 proof-text, [var:] FIRED See §c, span 45. Three words replaced in a clause carrying its own address.

### 0302C — in Synagoga crebris docet (0022) crebris is an ablative plural adjective with no noun; the sense wants crebris [sermonibus] or an adverb. Rendered "teaches often in the Synagogue", which is the least that the printed words allow without supplying a noun. Logged (§e.7).

### 0303C — …curvata fuit, quae magis terrena quam coelestia noverat, operatur. (0022) Two finite verbs, the second stranded at the end of the sentence with no subject of its own. Rendered as printed, with operatur left where Migne sets it: "…was bent away from the uprightness of her mind, she who had known earthly things rather than heavenly, works." The conjecture that construes is noverat operari ("knew how to work earthly things rather than heavenly"). No marker (§e.8).

### 0304A — a paragraph with no terminal stop (0022) VERS. 16 ends …quod est filiam Abrahae a vinculo noxiae inclinationis erigi with no full stop. Pattern 8: none supplied. The English paragraph ends the same way, mid-air.

### 0304B — Deum de sua erectione glorificantem (0022), mechanism-5 checked The participle is accusative feminine and agrees with filiam Abrahae, id est Ecclesiam — so it is the Church that glorifies God for her raising up, not the rejoicing bystanders, who are the easier antecedent. Rendered accordingly.

### 0306C — Ὕδωρ, the only Greek in the work (0024) Ὕδωρ , aqua, inde hydropis, morbus aquosus subcutaneus. The Greek is reproduced verbatim, in Greek, with the space Migne sets before the comma. hydropis is set as a nominative where it is the genitive of hydrops; the English renders it "the dropsy", the one English word that keeps the ὕδωρ derivation visible (dropsy < hydropsy < ὕδρωψ), so that the etymology the sentence exists to make still works in the English. The choice is recorded rather than silent (§e.10).

### 0307A — the Rom 3:23 proof-text, [var:] FIRED; and (Rom. XXXV) gratia for gloria — see §c, span 96. The note's address names a chapter Romans does not have; reproduced verbatim, unmarked (§e.15).

### 0307A/B — the Ps 87:16 proof-text, [var:] FIRED confusus for conturbatus — see §c, span 100.

### 0307B — Honoratiori post invitato locum dat, qui … praeitur (0024), mechanism-5 site qui is the man who gives place — the over-confident earlier guest — and praeitur is passive, so he is outstripped by the nimbleness of those who followed Christ. The easier reading binds qui to the more honourable newcomer and makes the verb active; that reading is wrong on both the case and the voice. Rendered per the morphology.

### 0309B — domum intellectus accipiunt (0024) Quidam autem vocantur et contemnunt, qui domum intellectus accipiunt, sed opere non implent. Rendered "who receive the house of understanding". The parallel three lines later — qui acceptam intelligentiam opere adimplent — wants donum intellectus, "the gift of understanding," and domum/donum differ by one sort. Real word, so no [sic:]; not a negation or correlative, so no [cj:]. Logged (§e.11).

### 0309C — cognitione conjunctos (0024) ut et cognitione conjunctos, et divino praecepto diligamus. Rendered "those joined to us by acquaintance". The passage is about hating father, mother, wife and children while loving them, so the expected word is cognatione, "by kinship". One letter. No marker (§e.12).

### 0310B — piorum affectum (0024) legationem lacrymarum, bonorum operum, piorum affectum — three genitives, and the third is an accusative singular where the series requires affectuum. English has no exponent for the case, so the rendering "of holy affection" is unavoidable and the defect is invisible in the English; that is exactly why it is logged here (§e.13).

### 0310B — the 1 Cor 4:9 quotation, [var:] DECLINED Order and connectives only; see §c, span 137. The address (II Cor. IV) is wrong for the verse.

### 0310C — spiritualia sapientia (0025) ad condienda corda auditorum spiritualia sapientia: a neuter plural adjective against an ablative singular noun. Rendered "with spiritual wisdom"; Pattern 9 governs — English carries no exponent for the disagreement, there is nothing to emend and nothing to mark (§e.14).