Liber Proverbiorum

Anselm of Laon and his school, on Proverbs. Translated 2026-08-25 by three translators working in parallel — 0000–0004 (Prov. I–VIII, cols 1079C–1091D), 0005–0009 (Prov. IX–XXIII, 1092A–1107B) and 0010–0013 (Prov. XXIV–XXXI, 1107C–1116D) — and merged here. 14 chunks, 13,510 Latin words. verify-english.mjs clean over the whole work.

Apparatus fired, work-wide: 33 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [cn:]. The work asserts nothing whatever about Migne's plate, and that is a deliberate result rather than an absence of candidates: every non-word met was rendered for its evident sense and logged, because a [sic:] is a public claim and the class it would blame Migne for has been our digitization's every time it has been tested.

⭐ This book is APHORISM, and the briefs warned all three stints that the danger moves with the genre: in a narrative a dropped negative leaves a sentence that will not construe, but in a maxim it leaves a maxim — one that reads perfectly and says the opposite. Each stint accordingly named the single Latin word every printed negative attaches to rather than merely counting them. The 2a trigger population came to 11 sites in one stint alone.

§M · THE MERGE

Three translators. The second and third read their predecessors' cruces off disk and inherited from them, which is the seam working: the second changed three of its own renderings to the first's table (doctrina → doctrine, never "learning" or "teaching") and the third followed both. doctores → teachers, charitas/dilectio held apart, scientia/sapientia held apart, Ecclesia mirrored word for word (every occurrence in the middle range is capitalized and so is its English), Roman-numeral chapter heads per the Ruth pilot.

⚖ mulier fortis — RULED: "a strong woman", and it stays

The third stint held fortis → "strong" everywhere, including the Prov. XXXI acrostic, and declared it a deliberate refusal of the Douay phrase, offering to move all nine sites if ruled against. It is not ruled against. Fortis means strong; "valiant woman" is another edition's translational choice, and reaching for it precisely because it is the familiar English is conformation — the defect this whole apparatus exists to catch. Measured across the work: five "strong woman", zero "valiant". Consistent, and it stays.

⭐ And the acrostic itself came back clean, which is worth recording because it was the chapter the brief told that stint to fear: Prov. XXXI agrees with the Clementine at every lemma. Its only two departures are punctuation — a period for 31:10's ? and a colon for 31:29's ; — both carried as printed. The divergences all sit in XXV–XXX, not in the famous chapter. Familiarity was the predicted risk and it did not fire; the risk was real, and naming it is what made the reading careful.

The correptio / correctio drift — one note, not two

Migne's text wavers between correptio (rebuke) and correctio (correction), and it drew a [var:] in Part I (1082A, the lemma Ad correctionem against Prov. 1:23's ad correptionem meam) and another in Part III (1110B, Prov. 27:5). The third stint asked whether the pair deserves one work-level note rather than two isolated markers. Both markers stand — each records a real divergence at its own verse — and the connection between them is recorded here rather than in a third marker, which would assert nothing new.

⚑ The Part I site is the sharper of the two: the lemma has drifted off the word its own gloss is built on, since the gloss below it reads corripit.

Checked and sound

• usque ad censused per occurrence in all three stints, never by rule: 98 of 98 the formula in the first range (recorded as a checked negative), 118 in the second, and 96 tokens with 93 formula in the third, including an anchor-split usque [1116C] ad mirrored in the English.
• Verse numbers out of register reproduced and never renumbered — and in two places the offset does not persist, so each was checked rather than a run being assumed.
• VERS. 1. printed with no -- at 1110B, reproduced as printed.

§P · WHAT THE BRIEF GOT WRONG ABOUT THIS BOOK, recorded so it is not repeated

The launch brief warned the first stint at length about Jerome's letter and double prologue under this work's opening head — continuous epistolary prose in Jerome's own voice, which must not be forced into the gloss register.

⛔ There is no such prose. Under that head Migne prints exactly one thing — the cross-reference note (Vide inter ejus Opera, tom. I, III et VIII.) — and goes straight to CAPUT PRIMUM. The warning had nothing to bite on.

⚑ Recorded for two reasons: so that a later reader does not conclude our text lost the prologue, and because the head is genuinely misleading — it promises a letter the volume does not reprint here.

§Q · THE UNDECODED ENTITY AT 1107B — a defect class of exactly one

The second stint found VERS. 35.emdash; in its Latin twin and did the right thing twice: it printed the house form VERS. 35.-- in the English under Pattern 9's section-label ruling, and flagged the Latin rather than editing it.

Measured: the only occurrence in all of src/latin, and the only one in all 5,276 PL TEI files. Not a letter substitution, not an unjoined line break, not a normalization — the entity reference &emdash; lost its ampersand upstream of us, so our Latin column was printing seven literal characters that are not a word, not Migne's, and not a defect of his that this edition exists to show. It was our own pipeline leaking XML at a reader.

Read at the plate before patching, per the standard the other patch files set: Gallica f554 prints VERS. 35. — Verberaverunt me. with the ordinary dash, as every verse address in the Glossa is set. So the repaired form is the corpus's own attested convention and not a conjecture. Patched at the TEI (data/tei-patches/8966.json) and re-chunked; the ten already-translated chunks still verified, and the timing was deliberate — no agent held this book and its last stint had not yet read its Latin.

§R · WHERE THE EVIDENCE POINTED AT US, AND THE MARKER WAS DECLINED

⭐ Worth naming as a class, because the discipline usually runs the other way. At 1103C the second stint declined a divergence on the ground that Migne's own next clause prints the word correctly, which makes the dropped letter likelier ours than his. At 1109D the third declined to call (Eccl. X) a misprint, because John 9:39 is already correctly cited four words earlier — so the second note is a cross-reference, and Eccles. 10:3 is the exact parallel to Prov. 26:12.

⚑ And at Prov. 23:4 Migne is the TIDIER witness: he prints Pone modum against the Clementine's own oddity ponde. Marking it would have misdescribed which text departs from which.

Seven misprinted citations were found across the three stints, every one identified from the quoted words and never from the numeral, all left exactly as printed and referred to data/citation-corrections.json. ⬜ They are not yet entered there — see the resume note.

Glossa ordinaria, Liber Proverbiorum (PL 113). Translated 2026-08-25, one translator, 5 chunks, 4,403 Latin words (frontmatter counts: 1,063 + 411 + 951 + 898 + 1,080). verify-english.mjs reports no failure over this range (the only failures it reports are the peers' still-missing 0007–0013).

Apparatus fired: 4 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [cn:] · 37 [n:] (exact parity with the Latin twins). Zero guillemets in either language. Marker parity, anchor parity, italic-span parity and paragraph parity are 1:1 in all five chunks; punctuation was compared band by band (colons, semicolons, question marks, exclamation marks), not by total, and all 51 bands match with zero mismatched bands after two repairs made by that test (see §5).

§V · WORK-WIDE VOCABULARY FIXED IN THIS STINT — for the merge

Stated before comparison, so a later merge can reconcile drift against the English files rather than against this table.

Head forms: CAPUT PRIMUM. → CHAPTER ONE.; CAPUT II. → CHAPTER II., and so on with the Roman numeral kept. ⚠ This follows Ruth (8968), the pilot, and 8954, and the register rule that heads keep their Roman numeral. 8965 spells its numerals out (CHAPTER THREE for CAPUT III.) and therefore diverges from the pilot; I did not follow it, and I flag the divergence rather than quietly siding with one. The head at 1079C, B. HIERONYMI EPISTOLA ET PROLOGUS DUPLEX DE LIBRIS SALOMONIS., is rendered THE LETTER OF BLESSED JEROME AND THE DOUBLE PROLOGUE ON THE BOOKS OF SOLOMON.

⛔ THE JEROME LETTER AND DOUBLE PROLOGUE ARE NOT IN THE TEXT — the head is a stub

My launch brief warned that chunk 0000 opens with continuous epistolary prose in Jerome's own voice which must not be forced into the gloss register. It does not. Under that head Migne prints one thing and nothing else: the cross-reference note [n: (Vide inter ejus Opera, tom. I , III et VIII .)] — "see among his Works, vols. I, III and VIII." He heads the section, refers the reader to Jerome's own works for the letter and the two prologues, and goes straight on to CAPUT PRIMUM. So there is no epistolary prose in this work at all, and the warning had nothing to bite on. Recorded because a later reader meeting that head will otherwise look for the letter and think our text has lost it: it is not lost, it was never printed here.

§1 · THE LEMMA COLLATION — every italic span in cols 1079C–1091D, walked

All 340 italic spans in my five chunks (67 + 44 + 72 + 71 + 86) were walked against sources/vulgate/clementine-flat.txt, cited by book chapter:verse. The agreements are listed as well as the divergences, because a list of findings alone cannot be told from a report by a reader who never looked.

1a · The four divergences that fired a [var:]

1b · Divergences found, logged, and DECLINED a marker — with the reason

None of these fires a [var:], because Pattern 14's bar is a divergence that CHANGES THE SENSE, and each of these is a form, an orthography, or a syntactic integration. They are listed because a blind reader will find them and needs to know they were seen.

1c · Checked and found to AGREE with the Clementine EXACTLY

Listed by verse, because this half of the collation is what distinguishes a report from a guess.

• Prov. I (1079C–1083A): 1:1 Parabolae Salomonis · 1:2 Ad sciendam sapientiam et disciplinam · 1:3 Ad intelligenda verba prudentiae · 1:5 Audiens sapiens, sapientior (×2) and Et intelligens · 1:7 Timor Domini · 1:8 Audi, fili mi, disciplinam · 1:9 Addatur gratia and Et torques collo tuo · 1:10 Fili mi · 1:11 Insidiemur and Insontem frustra · 1:12 Deglutiamus eum, sicut infernus viventem · 1:13 Omnem pretiosam substantiam reperiemus · 1:14 Sortem mitte nobiscum · 1:16 Ad malum currunt · 1:17 Frustra · 1:20 Sapientia, foris, In plateis dat vocem suam · 1:21 In capite turbarum clamitat, In foribus portarum urbis · 1:22 Usquequo · 1:23 Convertimini · 1:26 Ego quoque · 1:27 Cum irruerit repentina calamitas, Quando venerit super vos · 1:28 Mane consurgent et non invenient me, Invocabunt · 1:31 Comedent igitur, Suisque consiliis saturabuntur · 1:32 Aversio, Parvulorum · 1:33 Qui autem me audierit, absque terrore requiescet.
• Prov. II (1083A–1084A): 2:1 Fili mi · 2:2 Audiat sapientiam, Inclina cor · 2:3 Si enim sapientiam invocaveris (×2) · 2:4 Et sicut thesauros · 2:5 Tunc intelliges · 2:6 Quia Dominus dat, Ex ore · 2:7 Custodiet · 2:12 A via mala · 2:13 Qui relinquunt iter · 2:14 Qui laetantur · 2:16 Ut eruaris a muliere · 2:17 Et relinquit ducem · 2:18 Pacti Dei sui, Inclinata est enim · 2:19 Ad eam non revertentur · 2:21 Qui enim · 2:22 Impii.
• Prov. III (1084A–1086C): 3:1 Fili mi, ne · 3:2 Pacem · 3:3 Misericordia et veritas, Circumda eas gutturi tuo · 3:4 Et invenies gratiam · 3:5 Habe fiduciam in Domino · 3:7 Ne sis sapiens, Time Deum et recede a malo · 3:8 Sanitas quippe erit · 3:9 Honora Dominum de tua substantia, Et de primitiis omnium · 3:10 Implebuntur horrea tua saturitate, Torcularia redundabunt · 3:11 Disciplinam · 3:12 Quem enim diligit Dominus corripit · 3:13 Beatus homo, Et qui affluit prudentia · 3:14 Melior est, Primi et purissimi · 3:16 Longitudo dierum in dextera ejus et in sinistra · 3:17 Viae ejus · 3:18 Lignum vitae · 3:19 Dominus sapientia, stabilivit coelos prudentia · 3:20 Sapientia illius eruperunt abyssi · 3:21 Fili mi, ne effluant · 3:22 Et erit vita animae tuae, Et gratia faucibus tuis · 3:23 Tunc ambulabis fiducialiter, Et pes tuus non impinget · 3:24 Si dormieris, non timebis, quiesces, et suavis · 3:27 Noli prohibere benefacere eum qui, Noli, prohibere · 3:28 Ne dicas amico tuo… · 3:32 Quia abominatio Domini est omnis illusor, illusor, Et cum simplicibus sermocinatio ejus · 3:33 In domo impii · 3:34 Illusores.
• Prov. IV–V (1086C–1089A): 4:1 Audite, filii · 4:3 Nam et ego, Unigenitus coram matre · 4:4 Suscipiat · 4:5 Posside · 4:9 Dabit capiti · 4:10 Audi, fili mi · 4:11 Per semitas aequitatis · 4:12 Et currens · 4:17 Comedunt panem impietatis · 4:18 Justorum autem semita · 4:19 Via impiorum · 4:20 Fili mi, ausculta sermones meos · 4:23 Omni custodia serva cor tuum · 4:24 Remove a te · 4:25 Et palpebrae · 4:26 Stabilientur · 4:27 Ne declines, Averte pedem tuum · 5:1 Fili mi, attende, Inclina aurem · 5:2 Labia tua · 5:3 Favus enim, Labia meretricis, Oleo guttur · 5:4 Novissima autem, biceps · 5:6 Vagi · 5:8 Longe fac · 5:9 Ne des · 5:11 Et gemas · 5:12 Cur detestatus · 5:14 In medio ecclesiae · 5:15 Bibe aquam · 5:16 Deriventur · 5:17 Habeto, Nec sint alieni, Solus · 5:18 Sit vena · 5:21 Respicit Dominus · 5:22 Iniquitates · 5:23 Ipse morietur.
• Prov. VI–VIII (1089A–1091D): 6:1 Fili mi · 6:3 Suscita · 6:4 Ne dederis · 6:5 Eruere · 6:6 Vade ad formicam · 6:9 Usquequo, piger, dormies, quando · 6:11 veniet tibi quasi viator, Si vero impiger · 6:12 Homo apostata · 6:16 Sex sunt, Detestatur anima · 6:17 Oculos · 6:21 Liga ea in corde · 6:22 Cum ambulaveris, Cum dormieris · 6:25 Non concupiscat · 6:26 Pretium enim scorti · 6:30 Non grandis est culpa · 6:34 Quia zelus et furor viri non · 7:1 Fili mi · 7:2 Et legem meam · 7:4 Dic sapientiae, soror · 7:6 De fenestra · 7:8 Juxta angulum, prope viam domus illius graditur · 7:9 In obscuro advesperascente · 7:11 Nec valens in domo · 7:13 Apprehensumque deosculatur juvenem, Et procaci vultu blanditur, dicens · 7:16 Intexui funibus lectulum · 7:17 Aspersi cubile meum · 7:18 Veni, inebriemur uberibus · 7:20 Sacculum pecuniae secum tulit · 7:26 Et fortissimi quique interfecti sunt ab ea · 7:27 Viae inferi domus ejus, penetrantes, Interiora mortis · 8:2 In summis · 8:4 O viri, ad vos · 8:5 Intelligite, parvuli · 8:7 Labia mea · 8:13 Arrogantiam · 8:14 Meum est consilium · 8:15 Per me reges · 8:17 Ego diligentes, Mane, vigilant · 8:18 Mecum sunt divitiae · 8:22 Dominus possedit me · 8:31 Ludens in orbe · 8:34 Ad fores.
• Non-Proverbs quotations checked verbatim and agreeing: Jo. 16:25 Haec in proverbiis locutus sum vobis (1079C) · Ps. 118:113 Iniquos odio habui, et legem tuam dilexi (1082C) · Is. 1:20 Os Domini locutum est (1083B) · Job 1:1 simplex et rectus (1083C) · Pr. 1:11 Veni nobiscum, insidiemur sanguini (1083C) · Lc. 22:30 Ut edatis (1084D) · Apc. 3:19 Ego quos amo, arguo et castigo (1084D) · Ps. 33:9 beatus vir qui sperat in eo (1085B) · Rom. 8:28 diligentibus Deum omnia cooperantur in bonum (1085C) · Jo. 15:14 Vos amici mei estis, si feceritis (1085D) · Jac. 4:6 Deus superbis resistit, humilibus autem dat gratiam (1086C) · 1 Cor. 15:58 Stabiles estote et immobiles, abundantes in opere Domini semper (1087C) · Ps. 138:12 et nox sicut dies illuminabitur (1088D) · Is. 1:14 Solemnitates vestras odivit anima mea (1089C) · 2 Cor. 2:15 Christi bonus odor sumus Deo (1090D) · Mt. 16:18 Portae inferi non praevalebunt adversus eam (1091A) · Jo. 15:5 Sine me nihil potestis facere (1091C) · Eph. 1:3 Qui benedixit nos in omni benedictione spirituali in coelestibus in Christo (1086B).

§2 · FOUR MISPRINTED CITATIONS — every one identified from the QUOTED WORDS, never the numeral

All four are inside [n:] notes, which pass through verbatim; the text keeps exactly what Migne prints and only the index should be corrected, through data/citation-corrections.json. None was read at the plate, so each is a candidate and not a claim — but each was identified the settled way, from the words rather than from the number.

⚠ No Ibid. in this range depends on any of the four, so correcting them upstream cannot poison an anaphora chain (runbook false positive 2 checked, not assumed).

§3 · THREE VERSE NUMBERS THAT DO NOT MATCH THE VERSE GLOSSED — reproduced, never renumbered

⭐ The 6:26/6:27 pair is the interesting one, because the offset does not persist: the very next address, VERS. 30.-- Non grandis est culpa, is Pr. 6:30 exactly, and VERS. 34.-- Quia zelus et furor viri non is Pr. 6:34 exactly. So this is not a chapter numbered on a different system; it is two addresses out of register inside an otherwise correct sequence. All three are Migne's and all three are reproduced as printed.

§4 · usque ad — CENSUSED PER OCCURRENCE, NOT BY RULE

98 occurrences of usque ad in the body of my five chunks, and every one of them is the abridgment formula — 91 bare usque ad, and 7 that carry their resumption tail inside the same italic span: usque ad: in platea vocem ejus. (1082A) · usque ad et bene tibi erit. (1082D) · usque ad: et ipsi sunt jacula. (1083D) · usque ad: beatus vir qui sperat in eo. (1085B) · usque ad, (1088D, the comma inside the italics) · usque ad: et nox sicut dies illuminabitur. (1088D) · usque ad: gradiantur tecum mandata Dei, ut te instruant. (1089C). All render as far as, inside the italic span, with no verb ever supplied to complete the formula (8970 convention), and the seven tails were parsed as strictly as ordinary prose.

Six look-alikes, decided per occurrence from what stands on either side:

• 1080C VERS. 22.-- Usquequo. Jam — Pr. 1:22's own word, "how long".
• 1089B VERS. 9.-- Usquequo, piger, dormies, quando. and, in the gloss, Usquequo, piger, dormies in vitiis — Pr. 6:9's own word, twice.
• 1081D Hucusque persecutiones, quas Christo Judaei intulere and 1091C Hucusque locutus est parabolice ad Filium — "thus far", the glossator's own transition adverb, not a formula.
• 1088C juxta uniuscujusque qualitatem — the letters inside another word.

⚠ The trap the brief warned of did not fire in this range: no lemma in cols 1079C–1091D has usque ad among the verse's own words, so the formula never had a chance to swallow one. Recorded as a checked negative, since 8954's first stint met that trap 14 times in its 94.

⚑ One placement worth naming: at 1087B the column anchor falls between the etc., and the formula (Dicit de his, etc., [1087B] usque ad quia currens…), which is the one place a careless reading would attach the formula to the wrong sentence. The English anchor sits in the same place.

§5 · THE NEGATION CENSUS — every printed negative, with the WORD it attaches to

Proverbs is aphorism, so a lost negative leaves a maxim that reads perfectly and says the opposite. Every negative in all five chunks was walked and its host named; the list below gives the ones where the host was not the obvious verb, or where a correlative or quantifier stood nearby (the 2a trigger population). 100 negative tokens stand in the five chunks (43 non · 23 ne · 9 nec · 7 nihil · 4 nunquam · 4 nullus/nulla/nullum/nullo · 2 noli · one each of nisi, nil, neque, nemo, neminem, nequeunt, nequeat, noluerunt). The ~88 not tabulated below rest on their own finite verb with no correlative or quantifier in reach, and were checked one at a time.

The 2a trigger population, host proved from the sentence's own structure:

Contradictions carried, not repaired: none arose in this range — no printed negative in cols 1079C–1091D makes its own sentence contradict its neighbour, and I looked for the 0639C shape (a negative Migne supplies that the received text lacks) at every scripture quotation in §1c. That is a checked zero, and it is recorded as one: the one place a supplied negative could hide, 1082C's Mane consurgent et non invenient me, is Pr. 1:28's own non, printed by the Clementine too.

Two repairs the band-by-band punctuation test made, both mine: at 1079C I had rendered Migne's colon before ita in eis as a semicolon and moved a comma; at 1082A I had rendered his comma before unde as a semicolon. Both restored. Totals would have cancelled the first pair exactly (one colon lost, one semicolon gained), which is the failure the band test exists for.

§6 · PATTERN 17 — thou/you, with the Latin that fixes the number

The book addresses one hearer (fili mi) and takes thou throughout. One passage is plural and takes "you", and it is plural in Migne's own morphology:

• 1082A, VERS. 22 — Christum in carne vidistis… quem sprevistis… diligite… quod peccastis — four second-person plurals, addressed to the persecuting Jews, inside a paragraph whose neighbours are all singular. Rendered "you have seen… you have despised… love… you have sinned".
• 1082C, VERS. 27 — the lemma Quando venerit super vos, plural, likewise.
• Everything else — serva cor tuum, ne dederis, bibe aquam, keep thou thyself, thou art joined — is singular and takes thou.

§7 · PLATE CANDIDATES — raised, none patched, none marked

No [sic:] and no [cj:] was fired in this range, so these five chunks assert nothing whatever about Migne's plate (ship test B). Each item below is a crux and a request, not a claim.

7a · palma for plasma @ 1084C — a REAL WORD, so rendered as printed

VERS. 9.-- Honora Dominum de tua substantia. Ut scilicet homines, qui ejus palma sunt, qui ad ejus imaginem facti sunt recreentur.

palma is a perfectly good Latin word, and the sentence with it in reads "men, who are his palm". The word the argument wants is plasma — "who are his moulding, his handiwork" — which pairs exactly with the following qui ad ejus imaginem facti sunt, is the standard word for the creature in this register, and differs from palma by two letters.

⛔ No marker. The addendum's rule is measured and unambiguous: a real, well-formed word that reads oddly is Migne's until a plate says otherwise — refuted 5 of 5. The English therefore reads "who are his palm", the conjecture lives here, and a plate read at PL 113 col. 1084 band C would settle it in one request. Nor does it reach Pattern 18's bar, which is narrow: palma does not make the sentence assert something the author is arguing against; it makes it fail to construe, and 18a declines exactly that class.

7b · paraemias @ 1079C — the only near-non-word in the range

Graece παροιμιας paraemias, id est, proverbia dicit

The standard Latin transliteration of παροιμίας is paroemias; paraemias has ae for oe. It sits in italics beside the Greek it transliterates, which is where a compositor is most likely to err and least likely to be corrected — but it is a legible form of a real word, not a hole, so it is rendered as printed (carried in italics, untranslated, as a mentioned Greek word), with no [sic:]. ⚠ It is also exactly the shape of brief §4's normalization class running the other way, and I cannot tell from the file whether the ae is Migne's or ours.

7c · clàmitat @ 1082A — a grave accent on a finite verb

VERS. 21.-- In capite turbarum clàmitat, etc.

Migne prints the word with a grave accent. The Clementine has clamitat, and at 1091A the same verb's cognate is printed unaccented. An accent on à is ordinary in this corpus as a disambiguator on the preposition; on a third-person verb it is not. English has no exponent for a Latin accent (the Pattern 9 family), so the English reads "crieth out" with nothing marked, and this is logged so a later reader does not take the accent for our own.

7d · The clipped left margin — where it would show in this range, and it does NOT

Our archive.org copy of PL 113 clips the left margin on some pages, and 8965 found its split runs sitting at left-column heads because of it. I looked for that shape here and found none: there is no split run, no doubled syllable, and no truncated word anywhere in cols 1079C–1091D. ⭐ Recorded as a checked negative, because the same photograph produced positives two books away and a reader comparing the two needs to know this range was examined and came back clean.

7e · A misplaced semicolon @ 1086A — carried, unreadable, and logged

Illusor est et ille, qui promissa ejus, quasi prava, despicit, ac iram districtionis, quasi tolerabilem; spernit.

The semicolon falls between the object and its verb. Pattern 8 is explicit that a printed mark rides even where it makes the English unreadable, and that the crux, not the text, adjudicates — so the English reads "…and the anger of his strictness, as though tolerable; he spurns." The supplied "he" is the only addition; spernit's subject (ille) is four clauses back. ⚠ I cannot tell whether the semicolon is Migne's or our digitization's, and that is precisely why nothing was moved.

§8 · THE UNEXPRESSED SUBJECT — every pronoun I supplied that the Latin does not print

Named individually, because 12 of 25 sites in 11613's blind read were this class and both readings usually construe.

§9 · THE FOURTH TEST — read back for what the English ASSERTS

Run last, with counts already clean, on every paragraph. Three passages were re-read and changed; the rest stand.

1. 1082A Ad correctionem — first drafted with the gloss's corripit rendered "corrects", which made the lemma and the gloss agree and hid the divergence §1a records. Restored to "rebukes", which is what corripit says, and the [var:] now has something to be read against.
2. 1085D Non contendas adversus hominem — first drafted with "in vain" attached to the wrong clause ("this is not done in vain, which a sure necessity compels" reordered so that frustra floated onto cogit). Repaired: the frustra is on fit, and it must be, or the marker in §1a is arguing against our own sentence.
3. 1091D ne quis eum cum creaturis et temporibus esse prohiberet — the smooth reading ("lest anyone should deny that he began with the creatures") is the one the argument seems to want and is not what the words say. Rendered literally, "lest anyone should forbid him to be with the creatures and the times", and left rough. ⚠ A sentence that reads smoothly where the Latin is rough is a suspect, not a success; this one is logged rather than fixed.

Sections read clean under this test and needing nothing: the whole of CAPUT II (1083A–1084A), CAPUT IV (1086C–1087C), and CAPUT VII (1090A–1091A).

§10 · DECLINED, WITH THE REASON

• [cj:] at 1084C (palma) — declined: real word, no false assertion, and 18a's bar was not reopened. See §7a.
• [sic:] anywhere — declined throughout: there is no non-word, no split run and no doubled run in this range. A [sic:] is a public claim and seven false ones were withdrawn from an earlier batch.
• A [var:] on any of the nineteen divergences in §1b — declined individually, each with its reason in the table. The bar is a divergence that changes the sense; orthography, verb form and syntactic integration do not.
• A [var:] on the four misprinted citations in §2 — declined: the words agree with their true sources, so there is no textual variant to record. The defect is in the numeral, which is [n:] matter and belongs in data/citation-corrections.json, not in the English.
• Any tei-patch — declined: nothing in this range fails to construe from a transcription defect.
• Renumbering the three verse addresses in §3 — declined absolutely; they are Migne's.
• [cn:] — none, and none was expected: the plate spot check for this work is a checked zero.

Glossa ordinaria, Liber Proverbiorum (PL 113), CAPUT IX–XXIII. Translated 2026-08-25, one translator, 5 chunks, 5,713 Latin words (frontmatter counts: 1,139 + 1,049 + 1,189 + 1,156 + 1,133; body counts 1,146 + 1,058 + 1,200 + 1,167 + 1,142). verify-english.mjs reports no failure over this range — at the time of filing the only failures it reports are the peer's still-missing 0010–0013. EN/LA ratio 1.56 with [var:] content stripped.

Apparatus fired: 19 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [cn:] · 20 [n:] (5 + 3 + 5 + 1 + 6, exact parity with the Latin twins, contents verbatim and untranslated, Migne's own internal spacing kept — [n: (HIER., de morte Fabiolae. )] and [n: (Matth. X.)]). Zero guillemets in either language. The plate spot check for this work is a checked zero: Migne prints no numbered Forte apparatus here, there is no [cn:] in the Latin and none in the English.

Parity, measured not assumed: column anchors 12 + 12 + 13 + 13 + 12 = 62, verbatim and in order, and each was checked against the WORD it stands beside in the Latin, not merely for presence (test 2b). Italic spans 410 in the Latin, 410 in the English once [var:] contents are stripped (87 · 85 · 80 · 75 · 83). VERS. n.-- addresses pass through verbatim, untranslated. Paragraph structure 1:1. Punctuation was compared band by band (colons, semicolons, question marks, exclamation marks) across all 62 bands — see §5, which is where this stint's real repairs came from.

§V · VOCABULARY — INHERITED FROM cruces-0000.md, NOT RE-DERIVED

src/english/8966/cruces-0000.md was on disk before I filed, and I read it and followed its table rather than inventing my own. Where it decided a word, I used its word. The places where that changed what I had first written are named, because a merge needs to know a convention was obeyed on purpose and not by luck:

Head forms: CAPUT IX. → CHAPTER IX., and so on with the Roman numeral kept — Ruth's shipped form, 8954's, and the form chunk 0000 chose for this book. My heads are CHAPTER IX, X, XI, XII, XIII, XIV, XV, XVI, XVII, XVIII, XIX, XX, XXI, XXII, XXIII. (8965 spells its numerals out and therefore diverges from the pilot; chunk 0000 flagged that and did not follow it, and neither did I.)

Words this stint had to fix that chunk 0000's table does not reach, stated here so the merge can reconcile them against the peers' English rather than against this table:

• derisor → scorner (1097A ×2, 1105C), held apart from illusor → mocker, which chunk 0000 fixed and which does not occur in my range. Also illudere does not occur here.
• dissimulare → to dissemble (1095B Qui autem dissimulat, 1103A ligari dissimulat) and simulator → the dissembler (1094A) — both from the same idea, and Migne uses them of different people, so they are kept in one English family. versipellis → the turncoat (1097C), deliberately not "the dissembler", because Migne glosses it of the devil who changes his skin and the two words stand nine lines apart.
• fatuus → the fool · stultus → the fool · insipiens → the unwise · imprudentia → imprudence. ⚠ fatuus and stultus both come out "fool" because English has one word where the Latin has two, and 1097B prints them in the same sentence (Fatuitas stultorum → "The folly of fools"). Flagged rather than solved: inventing a second English word here would be our distinction, not his.
• pascha does not occur; typice does not occur; meretrix → harlot (1107A); parvuli → little ones (1092B).
• Proper names, form for form as Migne prints them: Porphyrius → Porphyry · Julianus → Julian · Caiphas → Caiphas · Sareptanam → of Sarepta · Origenes → Origen · Pelagianus → the Pelagian · Leonidis → Leonides · Alexandrum magnum → Alexander the Great · Herodes → Herod · Joannem → John · Judaea → Judea · Anna → Anna · Judae → Judas. None normalized toward a familiar English Bible spelling.

§1 · THE LEMMA COLLATION — every italic span in 1092A–1107B, walked

All 410 italic spans in my five chunks were walked, in order, against sources/vulgate/clementine-flat.txt, cited by book chapter:verse. The agreements are listed as well as the divergences (§1c), because a list of findings alone cannot be told apart from a report by a reader who never looked. Of the 410, 118 are the usque ad abridgment span (§2), which is not a scripture lemma at all, leaving 292 real spans.

1a · The nineteen divergences that fired a [var:]

Each names the book, chapter and verse explicitly, which is 8965's form and the house form.

1b · Divergences found, LOGGED, and declined a marker — each with its reason

None of these fires a [var:], because Pattern 14's bar is a divergence that CHANGES THE SENSE, and because a [var:] is a public claim. They are listed because a blind reader will find them and needs to know they were seen and weighed.

1c · Checked and found to AGREE with the Clementine exactly

Listed by verse, because this half of the collation is what distinguishes a report from a guess. Fragments are counted as agreeing when every word Migne prints is the verse's own word in the verse's own order.

• Prov. IX (1092A–1092C): 9:1 Sapientia aedificavit, Columnas septem · 9:5 Venite, comedite panem meum, and the gloss's pane / vino · 9:6 Relinquite · 9:7 Qui erudit · 9:8 Noli arguere · 9:12 Si sapiens, sapiens fueris · 9:13 Mulier stulta (×2), Mulier · 9:16 Qui est parvulus · 9:17 Aquae furtivae · 9:18 Et ignoravit quod ibi.
• Prov. X (1092C–1093D): 10:1 Parabolae Salomonis, Filius sapiens · 10:2 Nil proderunt · 10:3 Non affliget Dominus · 10:4 Egestatem operata · 10:5 Qui congregat · 10:6 Os autem impiorum, os · 10:7 Memoria · 10:8 Stultus ceditur labiis (see §3 on the ae/e) · 10:9 Qui ambulat · 10:13 Virga in dorso ejus · 10:14 Sapientes abscondunt · 10:15 Substantia divitis · 10:23 Quasi per risum stultus operatur scelus · 10:24 Quod timet impius, veniet super · 10:25 Quasi tempestas · 10:26 Sicut acetum dentibus · 10:28 Spes autem. Also Psal. 118:2 Domine, libera animam meam a labiis iniquis (Migne's [n: (Psal. CXIX)] numbering is the Vulgate's own, and correct).
• Prov. XI (1093D–1094D): 11:1 Statera dolosa · 11:2 Ubi fuerit superbia · 11:7 Mortuo homine impio, Exspectatio sollicitorum · 11:9 Simulator · 11:12 Qui despicit · 11:14 Ubi non est gubernator, populus corruet, salus autem ubi multa consilia · 11:15 Affligetur, Qui autem cavet laqueos securus erit · 11:20 Abominabile Domino cor pravum · 11:21 Manus in manu · 11:22 Circulus aureus in naribus · 11:27 Bene consurgit diluculo qui quaerit bona · 11:28 Justi autem quasi · 11:29 Qui stultus est · 11:30 Qui suscipit · 11:31 Si justus.
• Prov. XII (1094D–1096A): 12:2 Qui bonus est · 12:4 Mulier · 12:5 Cogitationes · 12:7 Verte impios · 12:10 Novit, Viscera autem impiorum crudelia · 12:12 Desiderium, Radix autem justorum proficiet · 12:16 Fatuus, Qui autem dissimulat · 12:19 Labium veritatis · 12:23 Cor insipientium provocat · 12:24 Manus fortium, Quae autem remissa · 12:25 Maeror in corde · 12:26 Qui negligit damnum · 12:27 Non inveniet.
• Prov. XIII (1096A–1096D): 13:1 Filius sapiens · 13:4 Vult et non vult piger, Anima autem operantium, impinguabitur · 13:7 Est quasi dives · 13:11 Substantia · 13:12 Spes quae differtur · 13:13 Qui detrahit · 13:14 Lex sapientis · 13:16 Cum consilio · 13:20 Qui cum sapientibus, Amicus stultorum · 13:23 Multi cibi · 13:24 Qui parcit virgae odit filium suum. Also Psal. 62:6 Sicut adipe (Migne's [n: (Psal. XXVI)] does not match those words — §3), Luc. 16:19 purpura et bysso, Gal. 6:1 Si praeoccupatus fuerit homo in aliquo delicto.
• Prov. XIV (1096D–1097D): 14:1 Sapiens, Destruet · 14:3 In ore stulti · 14:4 Ubi non · 14:6 Quaerit derisor sapientiam · 14:7 Vade contra · 14:10 Extraneus · 14:16 Sapiens timet et declinat · 14:17 Impatiens · 14:19 Jacebunt mali ante bonos · 14:24 Fatuitas stultorum · 14:25 Liberat animas testis fidelis, Profert mendacia, Versipellis · 14:26 In timore Domini · 14:27 Timor Domini · 14:30 Vita carnium, Putredo ossium invidia. Also Psal. 110:10 Initium sapientiae timor Domini (Migne's [n: (Psal. IX)] does not match — §3). ⚠ Dominus impiorum (1097B) is treated separately in §3; it agrees with no verse.
• Prov. XV (1097D–1098D): 15:1 Responsio mollis frangit iram · 15:2 Ebullit stultitiam · 15:5 Stultus irridet · 15:7 Cor stultorum dissimile · 15:10 Doctrina mala · 15:15 Omnes dies · 15:17 Melius est vocari ad olera · 15:19 Pigrorum, Via justorum absque offendiculo · 15:23 Laetatur homo · 15:24 De inferno novissimo · 15:26 Firmabitur · 15:30 Lux oculorum laetificat animam, Lux oculorum. ⚠ Viduae Ecclesiae (15:25) and Videns bona oculus delectat corda bona are treated in §3 — the first fuses a gloss word into the lemma's italics, the second is announced by Migne himself as Alia translatio and is therefore not a Vulgate citation at all.
• Prov. XVI (1098D–1100B): 16:2 Omnes viae · 16:3 Revela Domino · 16:4 Universa propter semetipsum · 16:5 Abominatio est · 16:7 Cum placuerint Domino · 16:8 Melius est parum cum justitia · 16:10 Divinatio in labiis regis · 16:11 Pondus et statera · 16:13 Voluntas regum · 16:14 Nuntii mortis, Placabit · 16:15 In hilaritate vultus regis vita, Et clementia ejus quasi imber serotinus · 16:22 Fons vitae · 16:24 Favus mellis composita verba · 16:26 Anima laborantis laborat sibi · 16:29 Vir iniquus lactat amicum suum · 16:30 Qui attonitis oculis cogitat · 16:32 Melior est patiens viro forti · 16:33 Sortes mittuntur in sinum. Also Luc. 3:6 videbit omnis caro salutare Dei, Psal. 9:24 quoniam laudatur peccator in desideriis animae suae, et iniquus benedicitur, Jac. 3:6 inflammat rotam nativitatis nostrae inflammata a gehenna.
• Prov. XVII (1100B–1101A): 17:1 Melior est buccella · 17:6 Corona senum · 17:8 Gemma gratissima · 17:12 Expedit magis ursae · 17:14 Qui · 17:16 Quid prodest · 17:17 Omni tempore · 17:22 Animus gaudens · 17:24 In facie prudentis · 17:26 Non est bonum damnum inferre justo. Also Rom. 8:28 Quoniam diligentibus Deum omnia cooperantur in bonum, Prov. 26:10 Qui imponit stulto silentium, iras mitigat. ⚠ 17:18 Homo stultus plaudet manibus reverses the Clementine's Stultus homo — word order only, and it is listed here rather than in §1b only because the two words are the whole lemma.
• Prov. XVIII (1101B–1102A): 18:3 Impius cum in profundum venerit · 18:4 Aqua profunda, Et torrens redundans fons sapientiae · 18:9 Qui mollis et dissolutus · 18:11 Substantia divitis urbs roboris ejus · 18:13 Qui prius respondet quam audiat · 18:14 Spiritus viri sustentat imbecillitatem, Spiritum vero ad irascendum · 18:16 Donum hominis · 18:17 Justus prior · 18:19 Frater qui adjuvatur · 18:23 Cum obsecrationibus loquetur pauper · 18:24 Vir amabilis.
• Prov. XIX (1102A–1103B): 19:1 Melior est pauper · 19:2 Ubi non · 19:3 Stultitia · 19:4 Divitiae addunt amicos · 19:5 Testis falsus · 19:10 Non decent stultum · 19:11 Doctrina viri, Et gloria ejus est iniqua praetergredi · 19:13 Dolor patris, Tecta jugiter · 19:15 Pigredo immittit soporem (×2), Et anima dissoluta esuriet · 19:18 Erudi filium · 19:19 Qui impatiens est sustinebit damnum et cum rapuerit · 19:22 Melior est pauper · 19:24 Abscondit piger manum · 19:28 Devorat. Also Gen. 6:6–7 tactus dolore cordis intrinsecus: Delebo, inquit, hominem quem creavi a facie terrae.
• Prov. XX (1103B–1104B): 20:1 Luxuriosa res · 20:4 Aestate · 20:10 Pondus et pondus · 20:11 Ex studiis suis intelligitur · 20:12 Dominus fecit utrumque · 20:14 Malum est · 20:15 Est aurum et multitudo · 20:17 Suavis est homini panis mendacii · 20:21 Haereditas · 20:25 Ruina est homini devorare sanctos, et post vota retractare · 20:26 Dissipat impios rex sapiens, et incurvat super eos fornicem · 20:30 Livor vulneris. Also Rom. 8:18 non sunt condignae passiones hujus temporis.
• Prov. XXI (1104C–1105A): 21:1 Sicut divisiones · 21:8 Perversa via · 21:13 Qui obturat aurem · 21:15 Pavor operantibus iniquitatem · 21:18 Pro justo datur · 21:20 Thesaurus desiderabilis · 21:22 Civitatem fortium · 21:28 Testis mendax · 21:29 Vir impius. Also Psal. 94:4 in manu ejus sunt omnes fines terrae, Matt. 25:41 Discedite a me, maledicti, in ignem aeternum, qui paratus est diabolo et angelis ejus, Matt. 10:28 Nolite timere eos qui occidunt corpus.
• Prov. XXII (1105A–1106A): 22:1 Melius est, Nomen bonum, Gratia bona · 22:2 Dives et pauper · 22:3 Callidus vidit malum · 22:4 Finis modestiae, Divitiae et gloria · 22:6 Proverbium est · 22:8 Qui seminat iniquitatem, Et virga irae suae · 22:16 Qui calumniatur · 22:24 Noli esse amicus, Homini iracundo, neque ambules cum viro · 22:26 Noli esse cum his, qui defigunt manus · 22:28 Ne transgrediaris · 22:29 Vidisti virum velocem in opere suo? coram. Also Gal. 1:10 Si adhuc hominibus placerem, Christi servus non essem, Matt. 26:52 Omnes enim qui acceperint gladium, gladio peribunt.
• Prov. XXIII (1106A–1107B): 23:1 Quando sederis · 23:3 Ne desideres · 22:28 (quoted at 1106B) Ne transgrediaris terminos antiquos quos posuerunt patres · 23:6 Ne comedas · 23:7 Comede · 23:8 Cibos quos comederas · 23:9 In auribus · 23:10 Ne attingas · 23:17 Non aemuletur cor tuum peccatores · 23:21 Dantes symbola consumentur · 23:27 Fovea enim profunda est meretrix · 23:29 Cui rixae? cui foveae? Rixae, Cui sine causa vulnera? · 23:30 Nonne his · 23:33 Oculi tui videbunt · 23:34 Et eris sicut dormiens · 23:35 Verberaverunt me. Also Apoc. 2:10 Esto fidelis usque ad mortem, et dabo tibi coronam vitae, Joan. 8:44 Vos ex patre diabolo estis, Isa. 5:22 Vae qui potentes estis ad bibendum vinum, et viri fortes ad miscendam ebrietatem.

1d · Clipped lemmata — BOTH classes listed, so the split is not misread later as inconsistency

Ruth's convention 3 renders a fragmentary lemma as the phrase it stands for; the 2026-08-19 addendum adds that a complete one-word lemma has nothing elided and is rendered as it stands. Both classes occur here and the split is deliberate.

Expanded (the governed word is missing and English will not carry the stump): 1092A Sapientia aedificavit → "Wisdom hath built herself a house" · 1092A Relinquite → "Forsake childishness" · 1092A Noli arguere → "Rebuke not a scorner" · 1092B Si sapiens → "If thou be wise" · 1092C Et ignoravit quod ibi → "And she knew not that there are giants there" · 1092C Nil proderunt → "The treasures of wickedness shall profit nothing" · 1092D Non affliget Dominus → "The Lord will not afflict with famine" · 1092D Egestatem operata → "The slack hand hath wrought want" · 1093B Quod timet impius, veniet super → "…shall come upon him" · 1093D Spes autem → "But the hope of the wicked" · 1094A Affligetur → "He shall be afflicted with evil" · 1094D Qui suscipit → "He that receiveth souls" · 1094D Si justus → "If the just man receive on earth" · 1096C Qui cum sapientibus → "He that walketh with the wise" · 1096D Sapiens, etc. → "A wise woman," etc. · 1097A Ubi non, etc. → "Where there are no oxen" (Prov. 14:4) · 1098A Stultus irridet → "The fool laugheth at the discipline of his father" · 1100C Qui. → "He that letteth out water" (see §3) · 1102B Ubi non, etc. → "Where there is no knowledge of the soul" (Prov. 19:2) · 1103B Devorat → "Devoureth iniquity" · 1103B Luxuriosa res → "Wine is a luxurious thing" · 1102C Non decent stultum → "Delicacies are not seemly for a fool".

⚠ The two Ubi non lemmata are the control for this whole class: Migne prints the identical stump at 1097A and at 1102B for two different verses, and each was expanded from its own verse (oxen at Prov. 14:4, the knowledge of the soul at Prov. 19:2). Expanding them alike, or leaving them alike, would have been the error.

Left as they stand (nothing is elided; the lemma is a complete English phrase): Columnas septem · Mulier stulta · Aquae furtivae · Memoria · Substantia divitis · Quasi tempestas · Statera dolosa · Simulator · Manus in manu · Mulier · Cogitationes · Desiderium · Fatuus · Substantia · Extraneus · Impatiens · Versipellis · Aqua profunda · Stultitia · Haereditas · Aestate · Thesaurus desiderabilis · Nomen bonum · Gratia bona · Pone modum · Comede · Fili mi.

§2 · usque ad — CENSUSED PER OCCURRENCE, NEVER BY RULE

126 tokens of usque in my five Latin chunks. 118 are the abridgment formula (25 · 24 · 22 · 23 · 24), each inside its own italic span and each answered by exactly one italic as far as in the English — verified 1:1 per chunk, not in total. Eight are not the formula:

• Six sit inside another word and are invisible to a careless sweep: Vultusque (1098A), utriusque (1101B, 1101C), ejusque (1096B, 1105A), Hucusque (1105D).
• 1099D — Serotinus imber in Judaea usque hodie … venire consuevit: ordinary "even to this day". Rendered so.
• ⭐ 1106D — Esto fidelis usque ad mortem, et dabo tibi coronam vitae [n: (Apoc. II)]. This is the trap in my range and it is a real one: the words usque ad stand in italics, inside a quotation, immediately after an abridgment formula three words earlier in the same paragraph. They are Apoc. 2:10's own words — "faithful unto death" — and rendering them as the formula would have cut the crown out of the promise. Decided from what stands either side, as the brief requires.

Two further spans carry the formula with Migne's own punctuation inside the italics and are reproduced that way rather than normalized: usque ad, discedite a me operarii iniquitatis. (1098B, where the resumption is fused into the same italic run) and usque ad, (1100C).

§3 · THIS RANGE'S CRUCES — including everything declined, with the reason

⭐ 3a · 1107B · VERS. 35.emdash; — a broken HTML entity in OUR TEI. The English prints VERS. 35.--

The Latin chunk reads VERS. 35.emdash; Verberaverunt me.. Every other one of the 201 verse addresses in this work reads VERS. n.-- . emdash; is the tail of an undecoded &emdash; character entity, and it occurs exactly once in the whole of src/latin/ — a corpus-wide grep returns this one line and nothing else. It is not type Migne set; it is our source's decoding failure.

What I did, and why: the English prints VERS. 35.-- , which is what Migne prints. This follows Pattern 9's ruling on section labels — "the English head translates a structural label and there is no English exponent for a Latin typo, and the Latin side of the parallel page preserves the misprint verbatim, so nothing is concealed from the reader." A verse address is exactly that kind of structural label.

⚠ This is the one band where my punctuation comparison does not match, and the mismatch is this and only this: the semicolon of emdash; counts as a semicolon in band 1107B of the Latin and has no counterpart in the English. It is not page punctuation.

⛔ It should be fixed by a data/tei-patches/8966.json entry, not by anything in the English — find VERS. 35.emdash;, replace VERS. 35.-- , which corrects the markup and not one word of Migne. I have not written the patch, because a patch requires re-chunking and confirming chunk boundaries are unchanged, and that is the merge's business, not a translating stint's. It is the cheapest open item in this range.

3b · Plate candidates — raised, never patched, and NO MARKER FIRED ON ANY OF THEM

Ship test B: an unread candidate is safe if and only if it fires no marker. This range asserts nothing about the plate. Our archive.org copy of PL 113 (patrologiaecurs04migngoog) is known to clip the left margin on some pages, and data/volumes.json lists a second witness for PL 113 — Gallica ark:/12148/bpt6k5505319w, IIIF, one HTTP request per read, map f = (column + 1) / 2 (NOT the archive map). Every candidate below is one fetch away.

⭐ And the negatives, recorded because a checked negative is worth having: I looked for the "suspiciously tidy" normalization class (brief §4 / addendum C) in every line I examined closely and found no candidate. The passages here are rough in the ordinary way — ellipsis, anacoluthon, abridgment tails — and nowhere does a form strike me as tidier than its neighbours.

3c · Migne's own irregularities, reproduced and not tidied

• Citations whose numerals do not match the words they carry — THREE, all identified from the QUOTED TEXT and never from the numeral, all left exactly as printed. They are the index's business (data/citation-corrections.json), not the text's, and I have not touched that file:
• 1093A · [n: (Isa. LXVI)] on mors et vita in manibus linguae — the words are Prov. 18:21. ⚑ Not a random miss: Isa. 66:24 is the source of ignis eorum non exstinguetur five lines above, so the compositor's eye had Isaiah 66 on the page.
• 1096B · [n: (Psal. XXVI)] on Sicut adipe — the words are Psal. 62:6 (Sicut adipe et pinguedine repleatur anima mea). Psalm 26 has no such words.
• 1097B · [n: (Psal. IX)] on Initium sapientiae timor Domini — the words are Psal. 110:10 (also Sir. 1:16). Psalm 9 has no such words.
• Verse numbers that do not match the verse glossed — FOUR, reproduced as printed, never renumbered: 1093B VERS. 18 over a lemma from 10:14 (§1a) · 1100C VERS. 13 over a gloss on 17:14 (Aquam dimittere) · 1101B VERS. 1 over 18:3 · 1103B VERS. 2 over 20:1 (Luxuriosa res). ⚑ At 1100C the lemma is the bare stump Qui., so the verse number and the gloss disagree about which verse it opens. I expanded it from Migne's own next words (Aquam dimittere), i.e. from Prov. 17:14, and not from the printed numeral — the expansion is licensed by his gloss, not by my guess. Both readings are on the page for a reader to check.
• 1098C · Viduae Ecclesiae — the italic span takes in Ecclesiae, which is not in Prov. 15:25 (firmos faciet terminos viduae); Migne has drawn his own gloss word inside the lemma's italics. Reproduced with the italics where he set them: "Of the widow the Church,". No [var:], because the lemma is not claiming to be the verse.
• 1098D · Alia translatio: Videns bona oculus delectat corda bona — Migne says himself that this is another version. Pattern 14 is for scripture quoted as the Vulgate; a labelled alternate version is not a divergence, it is a second witness he names. No marker.
• 1105B · quia opus sunt divinum — plural subject, singular predicate, of the rich man and the poor man together. Rendered as printed ("because they are a divine work"); English carries the disagreement visibly, so there is nothing to mark.
• 1105D–1106A · praecipue Judaeo, qui … furebant — singular antecedent, plural verb. English has no exponent (Pattern 9); rendered "who raged", crux, no marker.
• 1095C · et aliis, quae sequantur, demonstrant — the sentence ends with no terminal stop. Pattern 8: the English stops where he stops, with no full stop supplied. This is deliberate and is not a truncation of ours.
• 1106B · ita haereticus, quae non intelligit in Scripturis, ut libet interpretatur? — a question mark closing what is plainly a statement. Preserved (Pattern 8; this is HIS oddity and recording it is the point). It is not the 8b indirect-question case: the English construction takes the mark without recasting.
• 1105A · Christi servus non essem! — Migne's exclamation mark on Gal. 1:10. Preserved.
• All four question marks at 1107A and the one inside the lemma at 1106A (Vidisti virum velocem in opere suo? coram.) are Migne's and stand where he sets them, including inside the italic spans.

3d · Passages that will not construe cleanly, rendered as closely as the printed words allow

Named, because a blind reader will stop at each of them and needs to know they were read, not skated over.

• 1092C, Nil proderunt gloss: usque ad quod pro vita sunt dati, conferunt. The abridgment has taken the subject. Rendered "because they were given for life, they profit"; the subject supplied is the thesauri of the lemma. Both the quod-clause's force ("because"/"in that") and the bare conferunt are elliptical in the Latin and are elliptical in the English.
• 1096A, Filius sapiens: ut hic … perveniat: ille, ne, cum arguitur, audiat. The second limb has no verb of its own; ne … audiat hangs on the ut of the first. Rendered "the other, that he may not hear when he is reproved," which keeps the ellipsis visible.
• 1104D, Thesaurus desiderabilis: Reprobi non tantum virtutes non assequi, etc. An accusative-and-infinitive with its governing verb cut away by the abridgment. ⚑ Two negatives, and both are carried on their own hosts: rendered "The reprobate not only do not attain the virtues, etc." A finite verb is supplied where Latin has the infinitive, because English cannot carry a bare infinitive here; nothing else is added.
• 1106A, Vidisti virum velocem: usque ad quorum cavit errorem esse ponendum. Rendered "he took heed that their error was to be set down." I can construe the words and I am not confident of the sense; the resumption tail is short and the antecedent of quorum is inside the elided matter.
• 1099C, Pondus et statera: Lapides justi fortes. justi is a genitive singular or a nominative plural and both construe. Rendered "The stones of the just are strong"; the alternative ("Just stones are strong ones") is named here rather than chosen silently.
• 1094C, Circulus aureus in naribus: Si suis naribus circulum auri fixeris. suis is either the reflexive possessive ("in his own nostrils") or the genitive of sus, a swine. ⭐ I chose the swine, and the reason is grammatical rather than traditional: the sentence's only person is the second-person fixeris, so a reflexive suis would have no antecedent at all, while Prov. 11:22's in naribus suis is the swine's snout on every reading of the verse. Rendered "in a swine's nostrils"; the rival is here.

§4 · SUPPLIED PRONOUNS AND SECOND PERSON

4a · Every "he/she/it/they" the Latin does not print, named, with what names it

The 11613 measurement makes this the dominant defect class. Where two readings both construe I say so rather than choosing silently.

• 1092C, Et ignoravit quod ibi → "she knew not". Prov. 9:18's subject is not expressed; Migne's own next words name her — Nescit mulier adultera. Not my inference.
• 1092D, Non affliget Dominus tail → "he turns it back by justly avenging". The Lord, from the lemma directly above.
• 1093B, Quod timet impius → "He speaks of that wicked man"; 1093C, Quasi tempestas → "He speaks of the persecutors". The subject of dicit throughout Migne's Proverbs glosses is the writer of the proverb; the same unexpressed subject governs subdit, subditur, adjungit, praemissum est everywhere in this range, and is rendered "he" uniformly.
• 1093C, Quasi tempestas: Sed illa … collocata → "that house". illa is feminine and the only feminine noun in the sentence is domum fidei.
• 1094D, Qui stultus est: ille damnatur, iste accipit bravium → "the former … the latter". ille is the impius of the preceding clause (who is condemned), iste the pius (who receives the prize) — fixed by which of them the prize belongs to, not by the demonstratives alone.
• 1095D, Cor insipientium provocat → "they take away that which they seemed to have". The plural of the lemma (insipientium), carried through appetunt · habent · tenent · nequeunt.
• 1096A, Filius sapiens: hic … ille→ "the one … the other", mapped in the printed order to sapientem and stultum. ⚠ Migne uses hic of the FIRST-named, which inverts the classical habit; Prov. 13:1's own second half decides it (qui autem illusor est non audit cum arguitur — the fool is the one who will not hear), and that is the limb ille carries.
• 1096D, Sapiens, etc. / Destruet → "She shall destroy". Prov. 14:1's subject for destruet is the insipiens woman, against the Sapiens mulier of the first half — the lemma Migne prints.
• 1098D, Videns bona oculus → "it beholds them … it is delighted". The eye of the lemma, which is the sentence's only available subject for contemplatur and delectatur.
• 1105C, Dicet piger tail: aut, sicut alios justos, consumat tormentis → "or, as it did other just men, may consume him". The lion of Prov. 22:13 (Leo est foris), which is the verse the lemma opens.
• 1104C, Sicut divisiones → "he inclines", "he bestows", "he renders" — the Dominus of the lemma and of in manu Domini.

4b · Pattern 17 — thou is the singular, you is the plural

Every archaic form in these five chunks answers a grammatically singular Latin address, and the Latin that fixes the number is named:

• 1092A Quasi interrogares → "As though thou shouldst ask" (2 sg.).
• 1092B Moneo te … quod tamen tibi prodest … Si enim sapiens fueris → "I admonish thee … profits thee … if thou be wise".
• 1093B si … desideras · conteges · proferas · tuum cor · os → "if thou desirest … thou shalt neither cover … nor utter … thy heart … thy mouth".
• 1094A Ne enim putares → "lest thou shouldst think"; 1094C si … fixeris → "if thou shouldst fix".
• 1102D–1103A conspicis · castiga · tuis verbis → "if thou seest … chastise … thy words"; 1103A labaris → "if thou shouldst slip".
• 1106A quo indutus videbaris? → "wherewith thou didst seem to be clothed?"; Non transgrediaris → "Do not thou pass over".
• 1106B–1106C sederis · desideres · loquaris · comedas · attingas · comederas · didiceras · deseras · sustineas · perdas · putaveras · auscultares · ingeras → all singular, all thou.
• ⭐ 1106C neque mittas margaritas → "neither cast thou pearls" — and this is where the rule earns its keep: Matthew's verb is the plural mittatis, Migne prints the singular, and the [var:] at that point and the thou in the text are saying the same thing from two directions. Migne's own Nolite four words earlier stays plural ("Give ye not"), so the shift is visible on the page exactly as he printed it.
• No plural vos is rendered "thou" and no singular tu is rendered "you" anywhere in the range.

§5 · THE TESTS, RUN, AND WHAT THEY RETURNED

Test 1 — a [sic:] must not swallow the word it marks. ⭐ Vacuously clean, and for the right reason: this range fires ZERO [sic:]. Nothing is quarantined, so nothing can be quarantined with its verb, its negation or its object inside. I also ran the mechanical form — stripped every marker from the English and read what was left — and found no stranded article, preposition or doubled mark; every sentence reads aloud without its brackets. The [var:] markers are all beside a complete rendering, never in place of one.

Test 2 and 2a — name the single Latin word each printed negative attaches to. Proverbs is aphorism, so this was run on every proverb, not on every paragraph. The non solum / non tantum / nisi / correlative population — 2a's own trigger condition — is nine sites, and each was proved from the sentence's structure and not from what reads well:

Every other printed negative — the non · nec · neque · ne · nisi · nunquam · nullus · nemo · nihil · nolite · noli population, ~90 tokens across the five chunks — was checked the same way, one host at a time. No negative was relocated, none was dropped, and none was invented. Two are worth naming because they read as contradictions and were carried anyway: 1093A et ignis eorum non exstinguetur inside a gloss on the memory of the just, and 1094D etsi enim impii non sumus, certe si peccatores nos esse negamus…, where three negations stack in one sentence and all three stand.

Test 2b — each anchor against the WORD it stands beside. All 62 were checked at the site, not for presence. Migne's band breaks fall mid-construction constantly in this book and the English follows them there: Denique [1093A] os → "In fine, [1093A] the mouth" · Stultus [1093C] est igitur → "A fool [1093C] therefore is he" · Hoc Origenes [1094A] nescivit → "This Origen [1094A] knew not" · Prudens autem [1094B] publice tacet → "But the prudent man [1094B] keeps silence in public" · glorificatur [1105B] Pater → "is glorified [1105B] the Father", where the English keeps Migne's order rather than the natural English one precisely so that the anchor lands where he puts it. ⚠ One anchor is approximate and is declared: 1103A, where Migne breaks inside the noun phrase cupiditatis infimae famae | sauciatur; my English breaks inside the same noun phrase ("the hunger of lowest [1103A] covetousness") but two words earlier than the Latin, because English word order puts sauciatur at the head of the clause. It is inside the same clause and the same phrase.

Test 3 — punctuation compared BY COLUMN BAND, never by total. This is where the repairs came from, and the peers' warning was exactly right. First pass: four mismatched bands, and all four were marks I had supplied, three of them semicolons:

Every one is the documented shape: an aphorism in two halves invites a semicolon, and the translator supplies one without noticing. A fifth, in 0008, was a different error the same pass caught — an [n: none] I had typed into a paragraph that carries no note at all, which broke note parity and would have shipped an empty marker.

After the repairs: 61 of 62 bands match exactly — colons, semicolons, question marks and exclamation marks. The one that does not is band 1107B and it is emdash; and nothing else (§3a). Totals were never used; had they been, 1092B's supplied semicolon and 1094C's supplied colon would have hidden inside a chunk total that happened to come out right.

The fourth test — does the paragraph ASSERT what he asserts? Run last, with the counts already clean, reading each finished paragraph against its Latin and asking that one question. Three paragraphs were rewritten by it, and none of them was missing anything:

1. 1103D, Malum est — Qui aeterna praemia … desiderat, malum est, quod in praesenti debet sustinere. My first English quietly repaired the anacoluthon into a well-formed sentence. Migne's sentence is not well-formed: he begins with a nominative Qui and finishes with an impersonal malum est, because he is playing on the buyer's cry of Prov. 20:14 (Malum est, malum est, dicit omnis emptor; et cum recesserit, tunc gloriabitur) — the recesserit and the glorietur are both in his sentence. Restored to the printed shape.
2. 1094D, Qui stultus est — Impius vivit pio. First rendered as though the wicked man merely lives alongside the godly. The dative is a dative of service and the verse is qui stultus est serviet sapienti: "The wicked man lives for the godly."
3. 1099A, Omnes viae — my first version put "he discerns" at the head of the sentence, which read better and moved the anchor. Migne's discernit is at the end and the anchor falls after bonorum; the English now follows him.

⚑ Two habits, applied: a sentence that reads smoothly where the Latin is rough was treated as a suspect (that is what caught 1 and 3), and the abridgment resumption tails were parsed as strictly as ordinary prose rather than excused as "abridged" — which is what produced §3d's list instead of a silent smoothing of five awkward tails.

§6 · WHAT I DECLINED, AND WHAT THE MERGE SHOULD KNOW

• Four markers that a strict reading of the patterns would allow, all declined, each named above with its reason: the [cj:] at 1097B (Dominus impiorum), the [sic:] at 1094A (damuante), the [sic:] at 1100B (eum charitate), and the [var:] at 1103C (Qui potest dicere). ⚑ The common reason is not timidity: in three of the four the evidence points at our file rather than at Migne's plate, and a marker would publicly blame him for a fault that is probably ours — which is precisely why seven false [sic:] were withdrawn from an earlier batch. The fourth (1097B) is a real-word case the brief rules is Migne's until a plate says otherwise.
• No tei-patch written, though §3a argues for one and states the exact find/replace. Patching requires re-chunking and a boundary check, which belongs to the merge.
• No data/citation-corrections.json entries written for the three misprinted citations of §3c, for the same reason: the printed forms stand verbatim in the text and the index is where they are resolved. All three are identified from the quoted text and never from the numeral.
• [cn:]: a checked zero for this range, consistent with the work's own spot check.
• For the merge, the two things most likely to need reconciling with the peers' halves: (1) the fatuus/stultus collision — both are "fool" here and 1097B prints them together, so if a peer has held them apart with two English words, mine is the half that should move; and (2) the clipped-lemma expansion split in §1d, which lists both classes on purpose, so that a later reader does not mistake the deliberate split for drift. Nothing in the peers' chunks was read or altered; I read cruces-0000.md only, and only for its vocabulary table.

Glossa ordinaria, Liber Proverbiorum (PL 113), CAPUT XXIV–XXXI — the tail of the book. Translated 2026-08-25, one translator, 4 chunks, 3,491 Latin words (frontmatter counts 1,068 + 807 + 757 + 859; body counts 1,074 + 811 + 764 + 869). With these four the work is complete: verify-english.mjs 8966 now reports verify OK: 14 chunks, columns/notes/sections aligned, no duplicate paragraphs for the whole work, where before it reported the missing tail. EN/LA ratio 1.54 with [var:] content stripped (peers: 1.56 in Part II).

Apparatus fired: 10 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [cn:] · 15 [n:] (8 + 1 + 3 + 3, exact parity with the Latin twins, contents verbatim and untranslated). Zero guillemets in either language. The plate spot check for this work is a checked zero — Migne prints no numbered Forte apparatus here, there is no [cn:] in the Latin and none in the English.

Parity, measured and not assumed: column anchors 11 + 9 + 8 + 10 = 38, verbatim and in order, each checked against the WORD it stands beside in the Latin (test 2b, §5). Italic spans: asterisk counts 170 / 110 / 106 / 160, identical in both twins once [var:] contents are stripped. Paragraph structure 1:1 (50 / 35 / 34 / 38 non-blank lines each side). VERS. n.-- addresses pass through verbatim, untranslated. Punctuation compared band by band across all 42 bands — see §5, which is where this stint's one repair came from.

§V · VOCABULARY — INHERITED FROM cruces-0000.md AND cruces-0005.md, NOT RE-DERIVED

Both earlier stints were on disk before I wrote a word and I read them first. Where they decided a word, I used their word, including the three renderings that Part II deferred to Part I:

Extensions this stint had to make, declared rather than smuggled:

• fortis → strong, everywhere, including mulier fortis. The [1114D] lemma is rendered A strong woman, not the Douay's "a valiant woman", and Fortitudo (1116B) → "Strength", fortia (1115C) → "strong things", fortissimum (1114A) → "most strong", fortis of the wise man (1107C) → "strong". ⚑ This is a deliberate refusal of the most familiar English phrase in the book. One Latin word, one English word, and the chapter's own argument — that the same fortitudo belongs to the Church, to the soul, and to the man of Prov. 24:5 — survives only if the word does not change costume.
• correptio → rebuke · correctio → correction, held apart, which is what makes the divergence at 1110B visible at all (§1a). This inherits chunk 0000's ruling at 1082A on the same pair.
• detractor / detrahere → backbiter / to backbite (1108A, 1109A ×3, 1116B).
• sacrum eloquium / sacra eloquia → the sacred word / the sacred words (1108C, 1116C).
• Proper names and technical words, form for form as Migne prints them: Lamuel → Lamuel · Ezechiae regis Juda → "Ezechias king of Juda" · Simon Barjona → Simon Barjona (his spacing) · Jeremias → Jeremias · Moyses → Moses · Nitria → Nitria · Antichristus → Antichrist · doma, institor, saphan, choerogryllus/choerogyllum, philargyria, ἐκκλησιαστής, Ecclesia (in the etymology at 1112B) left in the Latin/Greek as printed, because in every one of those places the gloss IS the word-study. None normalized toward a familiar English Bible form.

Head forms: CAPUT XXIV. → CHAPTER XXIV., and so through XXV, XXVI, XXVII, XXVIII, XXIX, XXX, XXXI — Roman numeral kept, per Ruth, 8954, and both earlier stints of this book. (8965 spells its numerals out and therefore diverges from the pilot; chunk 0000 flagged that and did not follow it, chunk 0005 did not follow it, and neither did I.)

§1 · THE LEMMA COLLATION — every italic span in cols 1107C–1116D, walked

All the italic spans of my four chunks were walked, in order, against sources/vulgate/clementine-flat.txt, cited book, chapter and verse explicitly (the house form ruled in 8964 §M). The agreements are listed as well as the divergences (§1c), because a list of findings alone cannot be told apart from a report by a reader who never looked. Of the spans, 93 are the usque ad abridgment formula (§2), which is not a scripture lemma at all.

1a · The ten divergences that fired a [var:]

1b · Divergences found, LOGGED, and declined a marker — each with its reason

None of these fires a [var:], because Pattern 14's bar is a divergence that CHANGES THE SENSE and because a [var:] is a public claim. They are listed because a blind reader will find them.

1c · Checked and found to AGREE with the Clementine EXACTLY

Listed verse by verse. Fragments count as agreeing when every word Migne prints is the verse's own word in the verse's own order.

• Prov. XXIV (1107C–1108B): 24:1 Ne aemuleris · 24:3 Sapientia aedificabitur domus · 24:5 Vir sapiens · 24:7 Excelsa stulto · 24:8 Qui cogitat · 24:10 Si desperaveris · 24:11 Erue eos · 24:13 Comede, fili mi · 24:15 Ne insidieris · 24:16 Septies enim cadet · 24:17 Cum ceciderit inimicus tuus · 24:21 Time Dominum, fili mi and Cum detractoribus · 24:23 Haec quoque sapientibus · 24:26 Labia deosculabitur · 24:27 Praepara · 24:28 Nec lactes quemquam labiis tuis (quoted at 1108D, verbatim) · 24:30 Per agrum hominis pigri transivi.
• Prov. XXV (1108B–1109A): 25:1 Hae quoque parabolae · 25:2 Gloria Dei and Investigare · 25:5 Aufer impietatem · 25:6 Ne gloriosus appareas · 25:11 Mala aurea · 25:13 Sicut frigus · 25:14 Vir gloriosus · 25:16 Mel invenisti, comede quod sufficit tibi · 25:20 Acetum in nitro qui cantat · 25:22 Prunas enim congregabis · 25:23 Ventus aquilo · 25:24 Melius est sedere in angulo domatis, quam · 25:25 Aqua frigida · 25:26 Fons turbatus · 25:27 Sicut qui mel · 25:28 Sicut urbs patens.
• Prov. XXVI (1109A–1110A): 26:1 Quomodo nix in aestate · 26:5 Responde stulto juxta stultitiam · 26:6 Claudus pedibus and Et iniquitatem bibens · 26:9 Quomodo si spina nascatur · 26:11 Sicut canis · 26:12 Vidisti hominem · 26:13 Dicit piger · 26:16 Sapientior · 26:17 Sicut qui apprehendit.
• Prov. XXVII (1110A–1111A): 27:1 Ne glorieris · 27:3 Grave est saxum and Ira stulti utroque gravior · 27:7 Anima saturata calcabit favum · 27:8 Sicut avis transmigrans · 27:14 Qui benedicit proximo · 27:17 Ferrum ferro · 27:20 Infernus · 27:23 Diligenter · 27:25 Aperta sunt prata · 27:27 Sufficiat tibi lac caprarum · and 28:14 Beatus homo qui semper est pavidus, quoted whole inside the VERS. 1 gloss at 1110B.
• Prov. XXVIII–XXIX (1111A–1112B): 28:1 Fugit impius and Justus autem quasi leo · 28:12 In exsultatione · 28:14 Beatus homo · 28:22 Vir qui festinat ditari · 28:28 Cum surrexerint impii · 29:5 Homo qui blandis · 29:6 Peccantem virum · 29:9 Vir sapiens · 29:11 Totum spiritum suum · 29:13 Pauper et creditor obviaverunt · 29:18 Cum prophetia · 29:20 Vidisti hominem · 29:22 Vir iracundus · 29:24 Qui cum fure participat, odit animam suam.
• Prov. XXX (1112B–1114B): 30:1 Visio quam · 30:2 Stultissimus and stultissimus sum virorum and Non didici · 30:4 Quis ascendit in coelum, Quis continuit, Quis suscitavit omnes, Quod nomen est ejus, et quod nomen filii ejus, Si nosti · 30:5 Omnis sermo and Clypeus est · 30:7 Duo rogavi te · 30:11 Generatio quae patri · 30:14 Generatio quae pro dentibus gladios habet · 30:15 Sanguisugae duae sunt filiae, Duae filiae, Tria sunt insaturabilia · 30:17 Oculum qui subsannat and Filii aquilae · 30:20 Talis est et via mulieris adulterae · 30:24 Quatuor · 30:25 Formicae populus infirmus · 30:26 Lepusculus (×2) · 30:27 Regem locusta · 30:28 Stellio manibus · 30:31 Gallus and Et aries · 30:33 Qui autem and Et qui provocat iras, producit.
• ⭐ Prov. XXXI — the mulier fortis acrostic, and EVERY LEMMA AGREES. 31:1 Verba Lamuelis · 31:3 Et divitias tuas · 31:4 Noli regibus · 31:6 Date siceram moerentibus · 31:8 Aperi os · 31:9 Aperi os tuum, decerne · 31:10 Mulierem fortem and Mulierem fortem quis inveniet · 31:11 Confidit in ea cor viri and Et spoliis non indigebit · 31:12 Reddet ei bonum et non malum · 31:13 Quaesivit · 31:14 Facta est quasi navis institoris de longe · 31:15 De nocte, Surrexit, Deditque praedam domesticis suis · 31:16 Consideravit and De fructu manuum suarum · 31:17 Accinxit · 31:18 Gustavit et vidit quia bona est · 31:19 Manum suam misit, Et digiti ejus, apprehenderunt fusum · 31:20 Manum suam aperuit inopi, et palmas suas · 31:21 Non timebit and Omnes enim domestici ejus vestiti sunt duplicibus · 31:22 Stragulatam vestem · 31:23 Nobilis in portis vir · 31:24 Sindonem fecit et vendidit · 31:25 Fortitudo · 31:26 Os suum and Lex clementiae · 31:27 Consideravit and Panem otiosa non comedit · 31:28 Surrexerunt filii ejus · 31:29 Multae filiae congregaverunt divitias: tu supergressa es universas · 31:30 Fallax gratia et vana est pulchritudo · 31:31 Date ei de fructu manuum suarum and Date ei de fructu. ⚑ This is the checked negative that matters most in my range. The brief warned that CAPUT XXXI is where familiarity would conform a lemma, and the answer is that Migne's own text needed no conforming: the only two departures in the whole chapter are a period for a question mark (31:10) and a colon for a semicolon (31:29), both punctuation, both carried as printed. The divergences in this stint are all in chapters XXV–XXX, not in the one I was told to fear.
• Non-Proverbs quotations checked and agreeing: Jo. 5:23 Qui non honorificat Filium, non honorificat Patrem, qui misit illum (1108A) · Mt. 16:17 Beatus es Simon Bar Jona, quia caro (1108B) · Mt. 16:19 Quodcunque ligaveris super terram erit ligatum et in coelis (1109C) · 2 Cor. 12:11 Factus sum insipiens, vos me coegistis (1109C) · Jo. 9:39 ut qui non vident videant, et qui vident caeci fiant (1109D) · Job 33:27 Respiciet homines, et dicet, Peccavi (1111C) · Jo. 15:5 sine me nihil potestis facere (1111C) · Ps. 49:18 Si videbas furem, currebas cum eo (1112B) · 1 Cor. 4:10 Nos stulti propter Christum (1112C) · 1 Cor. 2:6 Sapientiam autem loquimur inter perfectos (1112C) · Mt. 6:2 Amen dico vobis, receperunt mercedem suam (1115A) · Mt. 11:12 Regnum coelorum vim patitur, et violenti rapiunt illud (1115C) · Mt. 16:18 portae inferi non praevalebunt adversus eam (1116A) · Isa. 11:4 sed judicabit in justitia pauperes (1114D, the first half; the second half is §1a's fusion).

1d · Clipped lemmata — BOTH classes listed, so the split is not misread later as drift

The 2026-08-19 addendum's rule: a fragment whose governed word is missing is rendered as the phrase it stands for; a complete one-word or one-phrase lemma has nothing elided and stands as it is. Both classes occur here and the split is deliberate.

Expanded, because the governed word is missing and English will not carry the stump: Ne aemuleris → "Emulate not evil men" (transitive verb, object elided) · Excelsa stulto → "Wisdom is too high for a fool" · Qui cogitat → "He that deviseth to do evil" · Si desperaveris → "If thou despair in the day of distress" · Erue eos → "Deliver them that are led to death" · Septies enim cadet → "For the just shall fall seven times" · Praepara → "Prepare thy work without" · Sicut frigus → "As the cold of snow in the day of harvest" · Melius est sedere in angulo domatis, quam → "…than with a contentious woman" · Sicut qui mel → "As he that eateth much honey" · Sapientior, etc. → "The slothful man seemeth to himself wiser," etc. · Sicut qui apprehendit → "As he that taketh a dog by the ears" · Sicut avis transmigrans → "As a bird that removeth from her nest" · Noli regibus → "Give not wine to kings" · Stellio manibus → "The stellio supporteth itself on hands" · Et qui provocat iras, producit → "…bringeth forth discords".

⭐ One expansion supplies a NEGATIVE and is therefore declared on its own line: 1114A, Regem locusta. → "The locust hath no king." Prov. 30:27 reads regem locusta non habet, and the two printed words alone ("A king the locust") are not English at all. Convention 3 requires the phrase the fragment stands for, and the gloss beneath is built on the absence of the king — tamen instanter in suis locis Christo militare ("nevertheless it is wont to serve Christ instantly in its own places"), which is only a nevertheless if there is no king. ⚠ Every other negative in these four chunks is one Migne prints. This one is not, and it is the only one, and it is here.

Left as they stand, because nothing is elided: Vir sapiens · Comede, fili mi · Ne insidieris · Cum detractoribus · Hae quoque parabolae · Gloria Dei · Coelum sursum et terra deorsum · Aufer impietatem · Ne gloriosus appareas · Mala aurea · Vir gloriosus · Acetum in nitro qui cantat · Ventus aquilo · Aqua frigida · Fons turbatus · Sicut urbs patens · Quomodo nix in aestate · Nam sicut avis · Claudus pedibus · Sicut canis · Vidisti hominem (×2) · Dicit piger · Ne glorieris · Grave est saxum · Anima saturata calcabit favum · Melior est vicinus · Ferrum ferro · Infernus · Diligenter · Aperta sunt prata · Fugit impius · In exsultatione · Beatus homo · Vir iracundus · Verba congregantis · Visio quam · Omnis sermo · Duo rogavi te · Quatuor · Lepusculus · Gallus · Et aries · Verba Lamuelis · Aperi os · Mulierem fortem · Quaesivit · De nocte · Surrexit · Consideravit (×2) · Accinxit · Fortitudo · Os suum · Lex clementiae · Non timebit · Stragulatam vestem.

⚑ One lemma was deliberately NOT expanded, and the reason is worth keeping: 1108B Investigare. Prov. 25:2 reads investigare sermonem, so the stump wants its object — but Migne's gloss opens with that very word: Investigare. Sermonem, etc. Expanding the lemma would print sermonem twice on our side of the page and once on his. Rendered "To search out.", with the gloss's own "The speech, etc." completing it exactly as he arranged.

⚑ And one control the range provides for free: Vidisti hominem stands as a complete lemma at 1109D (Prov. 26:12) and again at 1112A (Prov. 29:20), for two different verses. Neither was expanded, and both were checked against their own verse — the same discipline chunk 0005 applied to its two Ubi non stumps.

§2 · usque ad — CENSUSED PER OCCURRENCE, NEVER BY RULE

96 tokens of usque in the bodies of my four chunks. 93 are the abridgment formula (30 · 17 · 13 · 33), each inside its own italic span and each answered by exactly one italic as far as in the English — verified 1:1 per chunk, not in total. Three are not the formula:

• 1112B — Huc usque parabolae Salomonis — "Thus far the parables of Solomon", the glossator's own transition at the head of CAPUT XXX.
• 1114D — Hucusque verba Lamuelis regis — the same adverb written as one word, at the head of the mulier fortis section.
• 1115B — Hodie et usque in finem saeculi — "to-day and even to the end of the world", ordinary usque in.

Three formula spans carry something other than the bare two words, and all three are reproduced as Migne arranges them rather than normalized:

• 1110C — usque ad, in Jericho, et incidit, etc — the resumption tail (Lc. 10:30's own words) is fused into the same italic run, Migne's commas included, and the paragraph ends on etc with no period. Both are carried: the English ends "as far as, into Jericho, and fell, etc".
• 1114B — usque ad postquam elevatus est in sublime. — tail inside the italics.
• 1116C — usque [1116C] ad — ⭐ the column anchor falls INSIDE the formula, between the two words. The English splits its own formula at the same place: "as far [1116C] as". This is the deliberate mid-construction split that 8995 established at 0605B.

⚠ The trap the brief names did not fire in this range: no lemma in cols 1107C–1116D has usque ad among the verse's own words. Recorded as a checked negative, because Part II met that trap once (Apoc. 2:10, Esto fidelis usque ad mortem) and 8954 met it fourteen times.

§3 · CRUCES — including everything declined, with the reason

3a · Plate candidates — raised, never patched, and NO MARKER FIRED ON ANY OF THEM

Ship test B: an unread candidate is safe if and only if it fires no marker. No [sic:] and no [cj:] was fired in this range, so these four chunks assert nothing whatever about Migne's plate. Gallica ark:/12148/bpt6k5505319w is the second witness for PL 113, f = (column + 1) / 2 (Part II's map, not the archive.org one), one fetch each.

⭐ The clipped left margin — looked for, and NOT found. Our archive.org copy of PL 113 clips the left margin on some pages, and 8965 found split runs sitting at left-column heads because of it. In cols 1107C–1116D there is no split run and no truncated word; the two suspect tokens above (inaequitate, nostium) are a fusion and a letter substitution, neither of which is the margin's shape. Recorded as a checked negative alongside Part I's, for cols 1079C–1091D.

3b · Three misprinted citations, each identified from the QUOTED WORDS and never from the numeral

All three sit inside [n:] notes, which pass through verbatim; the text keeps exactly what Migne prints and only the index should be corrected, through data/citation-corrections.json. None was read at the plate, so each is a candidate and not a claim. I have written no entries — the printed forms stand and the index is where they resolve.

⛔ One note considered and DECLINED as a misprint, with the reason, because a blind reader will raise it: 1109D [n: (Eccl. X)]. It closes ut qui non vident videant, et qui vident, caeci fiant, which is John 9:39 — and John 9:39 is already correctly cited by the note four words earlier, [n: (Joan. IX)]. So the second note is not a rival attribution of the same words; it is almost certainly a cross-reference for the gloss as a whole, and Eccles. 10:3 (cum ipse insipiens sit, omnes stultos aestimat — "since he himself is a fool, he counts all men fools") is precisely the parallel to Prov. 26:12, the verse under gloss. A note that points somewhere else on purpose is not a misprint, and citation-corrections.json must not be fed a guess.

3c · Migne's own irregularities, reproduced and not tidied

• ⭐ 1110A/B — VERS. 1. with NO --. Every other verse address in my four chunks (and in the whole work) reads VERS. n.-- . This one, opening CAPUT XXVII, prints a bare VERS. 1. . The English reproduces it exactly, because the address passes through verbatim and this is his setting, not a class of ours. ⚠ It is not the emdash; defect Part II found at 1107B — that one is an undecoded entity in our TEI with a visible tail; this is simply a missing rule.
• Two verse numbers out of register, both reproduced, never renumbered: 1107C VERS. 4.-- Vir sapiens. is Prov. 24:5, and 1111C VERS. 1.-- Homo qui blandis, etc. is Prov. 29:5. ⭐ Neither offset persists — the very next address after each is exact (VERS. 7. = 24:7; VERS. 6. = 29:6) — so this is the same shape Part I found at 1089C and it must be checked one address at a time rather than assumed to run.
• 1109A VERS. 4.-- gathers TWO verses under one address, Non respondeas . . . Responde stulto juxta stultitiam (Prov. 26:4 and 26:5), because the gloss exists to say they are not contrary. His spaced ellipsis . . . is carried with his spacing.
• 1112B–1112C — the Greek. ἐκκλησιαστής stands in the Latin unitalicized and is carried into the English unitalicized and untranslated, with congregans and Ecclesia beside it: Migne's sentence IS the etymology, and Englishing its terms would leave it explaining nothing (the 11551 stylus precedent).
• 1109B — doma. The gloss says the Greek of Acts 10 reads doma where the Latin has superiora. Migne prints doma roman, not italic, all three times; italic-span parity is sacred, so the English prints it roman too, untranslated. Logged because a reader will expect the register's italic-Latin treatment and it is not available here.

3d · Passages that will not construe cleanly, rendered as closely as the printed words allow

• 1107D — usque ad si algore perituros, victu, veste recrea. The resumption tail opens on a bare si whose protasis is inside the text the abridgment removed, and perituros is an accusative with no expressed governing verb but recrea. Rendered literally: "if those about to perish of cold, refresh thou with food, with raiment." ⚑ Parsed as strictly as ordinary prose, per the brief's ferias warning; the roughness is the abridgment's cut, not a licence to smooth.
• 1110C — Bona consolatio consilium sapientum: qui dum se invicem consolando instruunt, ferrum ferro acuitur. A masculine plural relative (qui, the wise) opens a clause whose main verb (acuitur) has ferrum for its subject — an anacoluthon. Rendered as printed: "who, while by consoling they instruct one another, iron is sharpened by iron." No conjecture offered.
• 1115C — usque ad sic sed et nocte accensa lucerna manibus operatur. sic sed et is not a sequence English can smooth without inventing a connective. Rendered "so but also by night, the lamp being lighted, she works with her hands."
• 1113C — quot cogitationibus juvenis agitur, qui viam veritatis deservit. ⚠ deservit is a real, well-formed verb ("serves diligently") and is rendered as such: "who serves the way of truth". The rival is deserit ("deserts"), which the moral shape of Prov. 30:19's via viri in adolescentia might seem to want. ⛔ The [cj:] is declined and the reason is stated rather than left implied: both readings construe, so the printed one does not make the sentence assert something the author is arguing against, and 18a declines exactly that class; and the addendum's rule for a real word is decisive. The rival is named here and loses on those two grounds — the apparatus records only what the marker says, and a reader never sees a rival weighed in silence.
• 1108C — Sicut superius praelatum, etc. Rendered "As was set forth above", taking praelatum as the neuter participle of praeferre. ⚠ The rival is the noun praelatus, "the prelate", which the surrounding matter (kings, rigid rulers, plebis odium) makes attractive. It loses because the sentence has no verb for a nominative subject to govern and because Sicut superius + a perfect participle is Migne's ordinary back-reference formula in this book. Named because it is a genuine fork.

§4 · SUPPLIED PRONOUNS AND SECOND PERSON

4a · Every "he / she / it / they" the Latin does not print, named, with what names it

12 of 25 sites in 11613's blind read were this class, and in a gloss the antecedent is very often in the lemma above or inside the text the abridgment removed.

4b · Pattern 17 — thou is the singular, you is the plural, decided by the morphology

The book addresses one hearer and takes thou throughout my range: Ne desideres · videris · recrea · Comede, fili mi · ne desistas · quod faciebas · Melior est tibi · si tuo tempore bene paveris · vestieris · comparas · acquiris · Sufficiat tibi · Si videbas furem, currebas cum eo · tuam sapientiam jactans … vocas · noli corrumpere · non misereberis · esto fortis · Vade et amplius noli peccare · tu supergressa es universas.

Four passages are plural and take "you", and every one is plural in Migne's own morphology:

• 1108C — ut exaltet vos in tempore visitationis and Humiliamini (1 Pt. 5:6): "Be you humbled … that he may exalt you."
• 1109C — vos me coegistis (2 Cor. 12:11): "you** have compelled me."
• 1111C — nihil potestis facere (Jo. 15:5): "you can do nothing."
• 1115C — ut bonis operibus fulgeatis: "that you may shine with good works."

⭐ The 1116D lemma is the instructive one: tu supergressa es universas is singular inside a sentence about multae filiae, and the English keeps "thou hast surpassed them all" so that the turn from the many daughters to the one woman is visible with no apparatus.

§5 · THE TESTS, RUN, AND WHAT THEY RETURNED

Test 1 — a [sic:] must not swallow the word it marks. ⭐ Vacuously clean, and for the right reason: this range fires ZERO [sic:]. Nothing is quarantined, so nothing can be quarantined with its verb, its negation or its object inside. I also ran the mechanical form — stripped every marker from the English and read what was left — and found no stranded article, no stranded preposition and no doubled mark; every sentence reads aloud without its brackets. All ten [var:] stand beside a complete rendering, never in place of one. ⚑ The fourth question (does the clause still have the thing it acts on?) has nothing to bite on here for the same reason.

Test 2 and 2a — name the single Latin word each printed negative attaches to. 83 negative tokens stand in the four chunks (non 50 · ne 15 · nunquam 4 · noli 4 · nihil 3 · nec 2 · and one each of nisi, neque, nullus, nemo, noluit, nequeat). Proverbs is aphorism, so this was run on every proverb rather than every paragraph. The 2a trigger population — a negative resting near a quantifier, a correlative, a predicate or a bare case-phrase — is TWENTY sites, and each host was proved from the sentence's own structure and never from what reads well:

The remaining ~61 negatives rest on their own finite verb with no correlative, quantifier or bare case-phrase in reach, and were checked one at a time.

Contradictions carried, not repaired: a checked zero. No printed negative in cols 1107C–1116D makes its own sentence contradict its neighbour, and I looked specifically for the 8964 shape (a negative Migne supplies where the received text has none) at every scripture quotation in §1c. The one place a supplied negative could hide — Et spoliis non indigebit at 1114D and portae inferi non praevalebunt at 1116A — carries a non that Prov. 31:11 and Mt. 16:18 both print themselves.

Test 2b — each anchor against the WORD it stands beside. All 38 were checked at the site, not for presence. Migne breaks mid-construction constantly here and the English follows him there, keeping his word order where English would prefer its own precisely so the anchor lands where he puts it:

• Non omnis fortis sapiens, [1107C] sed → "wise, [1107C] but the contrary"
• sapientiae suae gloriam [1109D] perdit → "the glory of his own wisdom does he [1109D] lose" — ⚑ the object is fronted in English for no other reason than the anchor.
• audientium [1113B] animas → "the hearers' [1113B] souls it slays"
• minores videntur [1113D] justi → "the just seem [1113D] lesser" — the anchor sits immediately after videntur / "seem" in both.
• sperare salutem [1114A] didicit → "in the help of God, it hath [1114A] learned"
• proximorum [1115B] curam → "girds himself solicitously to take [1115B] the care of his neighbours"
• quid [1115C] potest → "what [1115C] she can"
• cum late [1116A] longeque → "when far [1116A] and wide"
• Colus [1115D] in sinistra → "The distaff [1115D] in the left hand"
• et [1114D] arguet → "and [1114D] shall reprove"
• usque [1116C] ad → "as far [1116C] as" — the formula split at his break, deliberately.
• ad superiora [1109B] ascendit → "to the upper places [1109B] ascendeth"

⚠ One anchor is approximate and is declared: 1112B. Migne breaks at a consensu [1112B] abstinere docet sapientia; my English fronts the whole non solum … sed etiam pair so that the negative stays on a peccato (test 2a above), which puts the anchor after "consent" and immediately before "does wisdom teach us to abstain". It is inside the same clause and beside the same verb phrase; the two requirements pulled against each other and 2a won, which is the right order.

Test 3 — punctuation compared BY COLUMN BAND, never by total. Colons, semicolons, question marks and exclamation marks were counted in all 42 bands (12 + 10 + 9 + 11, with [n:] and [var:] contents stripped from both sides). First pass: one mismatched band, and it was a mark I had supplied.

It is the documented shape exactly — a maxim in two halves invites a semicolon and the translator supplies one without noticing. ⚑ And the totals would not have caught it: chunk 0010's Latin carries 5 semicolons of its own, so a chunk-level total of 6-against-5 would have looked like a rounding error rather than a located fault.

⛔ No comma→semicolon sweep was made in either direction, and the reason is measured on this book's own plate: Migne prints 19 semicolons of his own in these four chunks (5 · 4 · 7 · 3), including four in one band at 1113B. Tobit's exception rests on a book where he prints zero; this is not that book. After the repair, 42 of 42 bands match exactly, and there are zero question marks and zero exclamation marks on either side — which is itself a parity worth stating, since Prov. 31:10 and Prov. 30:4 both end in ? in the Clementine and Migne closes them with periods (§1b).

The fourth test — does the paragraph ASSERT what he asserts? Run last, with the counts already clean, reading each finished paragraph against its Latin and asking that one question. Four paragraphs were changed by it, and not one of them was missing anything:

1. 1109A, Prunas enim congregabis — I had written "do good things to an enemy for the sake of his evil", which quietly made the sentence about the enemy's wickedness. mali causa is "for the sake of evil", i.e. in order to harm him — the reading the paragraph exists to reject, since the coals are ardorem charitatis. The supplied "his" was doing an argument's work.
2. 1110C, Ferrum ferro — I had rendered the relative qui as "which", which tidied the anacoluthon into a sentence about the counsel. qui is masculine plural and means the wise men; restored, and the sentence left rough (§3d).
3. 1110C, Qui benedicit proximo — "he may finish by human favour" read as a purpose; ut … finiat here is consecutive and the flatterer's effect is a result, not his aim. Changed to "finishes".
4. 1112B, Verba congregantis — perpenditur rendered "is weighed", which is the dictionary and not the assertion: Migne is saying the point is discerned from the name. Changed.

⚑ Two habits, applied: a sentence that reads smoothly where the Latin is rough was treated as a suspect (that is what caught 2), and the abridgment resumption tails were parsed as strictly as ordinary prose rather than excused as abridged — which is what produced §3d's list instead of a silent smoothing of four awkward tails.

§6 · WHAT I DECLINED, AND WHAT THE MERGE SHOULD KNOW

• Four markers a strict reading of the patterns would allow, all declined, each named above with its reason: the [sic:] at 1114D (inaequitate), the [sic:] at 1116A (nostium), the [cj:] at 1113C (deservit), and the [sic:]/[cj:] at 1108D (Deus for Dens, where the [var:] fired instead). ⚑ The common reason is not timidity. In the first two the fault shape (a fusion, an h/n substitution) points at our file rather than at his plate, and a marker would publicly blame Migne for a fault that is probably ours — the reason seven false [sic:] were withdrawn from an earlier batch. The third and fourth are real-word cases that the addendum rules are Migne's until a plate says otherwise.
• No data/citation-corrections.json entries written for the three misprints of §3b: the printed forms stand verbatim in the text, and the index is where they resolve. No Ibid. in this range depends on any of them (runbook false positive 2 checked, not assumed — there is no Ibid. in these four chunks at all).
• No tei-patch written. ⚠ Part II's open item stands and is not mine to close: the VERS. 35.emdash; at 1107B has been repaired in the Latin since — my chunk 0010 begins at 1107C, just past it, and I confirmed the corpus-wide grep now returns no emdash; anywhere in src/latin/. Recorded so the merge knows that item is done and need not be re-raised.
• [cn:]: a checked zero for this range, consistent with the work's own spot check.
• For the merge, the four things most likely to need reconciling with the peers' stints: 1. fortis → "strong", including mulier fortis → "a strong woman" (§V). It is the most visible vocabulary choice in the book's most famous chapter, and if Wilson or the merge prefers the received "valiant woman", mine is the half that should move — but it should move everywhere, including the eight other fortis/fortitudo sites in my range, or the argument the chapter makes about one strength stops being visible. 2. insipiens → "unwise" carried into 2 Cor. 12:11 at 1109C, against the familiar "foolish". This is Part II's convention applied where it costs something; flagged so it is not read as a slip. 3. The correptio/correctio pair, which now has a [var:] in Part I (1082A) and another in Part III (1110B). Two sites, one drift; the merge may want a single note on the work page. 4. The clipped-lemma split of §1d, which lists both classes on purpose, exactly as Parts I and II did, so that a later reader does not mistake the deliberate split for drift — and the one expansion that supplies a negative (Regem locusta → "The locust hath no king") is declared on its own line rather than buried in the list.
• Nothing in the peers' chunks was read or altered. I read cruces-0000.md and cruces-0005.md for their vocabulary and their rulings, and 8968 §1, 8964 §M and 8954 §M for the conventions, and wrote only 0010–0013 and this file.

All 14 chunks read against the Latin twin, fenced from this file until findings were formed. Every printed negation in cols 1079C–1116D is carried: non, nec/neque, ne, nihil/nil, nemo, nullus, nunquam, nisi, nondum, and the non solum … sed etiam frames — roughly 150 sites, not one dropped, not one supplied. ⭐ Zero polarity defects. Chunks read and found clean: 0000, 0001, 0002, 0003, 0005, 0008, 0009, 0010, 0011, 0013. Two chunks gave sites (below).

Sites — both are OURS, both are referent, neither is a negation

• ⭐ 0004 · 1091C · Hucusque locutus est parabolice ad Filium — was "spoken in parables to the Son", which makes Solomon address Christ; nothing in the book addresses anyone but fili mi. The control is inside the same sentence: ostendit et commendat eam secundum utramque naturam, scilicet divinam et humanam — the passage is about what is predicated of the Son, not about who was spoken to. ad + acc. carries "with reference to" in this register. Now "spoken in parables concerning the Son." ⚠ Migne's capital F is kept; only the preposition moved.
• ⭐ 0007 · 1099C · David homicidium et adulterium septem dierum fame purgavit — was "by a famine of seven days". fames here is David's own seven-day fast (2 Reg. 12:16–18), the episode Jerome cites de morte Fabiolae; a famine is a public calamity nobody chooses, and cannot purge the penitent who suffers it. ⚑ The work had the word right elsewhere — §897 renders famae at 1103A "the hunger of lowest covetousness" — so this is a slip, not a convention. Now "by a hunger of seven days."

⚖ ONE STANDING RULING RE-OPENED — for Wilson, nothing changed pending it

1113C · qui viam veritatis deservit (§1529) declined the [cj:] on the stated ground that "both readings construe". ⛔ They do not. deservio ("serve diligently") governs the dative; viam veritatis is accusative, which only desero takes. So the printed word does not construe as "serves" at all, and the rival deserit/deseruit is not a rival but the only parse. ⭐ This is the exact converse of 8954 @0610D, found earlier in this same batch (deseruit rendered "has been devoted to"), which makes deservit/deserit a live confusion pair in this corpus. The English is untouched; the ruling turned on a fact about Latin syntax, so the correction belongs to whoever owns 18a.

Candidates raised and REJECTED, with reasons

• 1094D ille … iste — "the former … the latter" maps correctly; the prize fixes it.
• 1096A hic … ille — "the one … the other" binds first/second in English and lands right.
• 1096D Destruet → "She shall destroy" — read as a live trap (Prov. 14:1's subject is insipiens), then found already decided at §1008. Not a new site. ⚠ It remains reader-facing: after "A wise woman, etc." an English "she" binds to the wise woman.
• 1084C palma, 1108D Deus for Dens, 1116A nostium — all three met blind, all three already logged and marked.
• 1083B usque ad juventam … comparavit — an elided tail; both parses underdetermined.
• 1086C Non enim doceret sapientia, 1092B Noli arguere — checked, faithful.

§1529 declined the conjecture on the stated ground that "both readings construe". ⛔ They do not, and the decline rested on that one wrong fact. deservio ("serve diligently") is intransitive and governs the dative; Migne's object viam veritatis is accusative, which only desero takes. So "who serves the way of truth" is not a weaker reading of the printed words — it is not a possible parse of them at all, and deserit is not a rival but the only one.

18a's bar is therefore met: the English asserts something false, and a verb substitution is on 18a's own ✅ list. The text is unchanged, as the convention requires — nothing is removed and nothing replaced — and the marker now stands beside it:

who serves [cj: deservit; read deserit, "forsakes"] the way of truth

⭐ This is the converse of 8954 @0610D in the same batch (deseruit silently rendered "has been devoted to"), which makes deservit / deserit a live confusion pair in this corpus and worth a standing check in every commentary from here.

⚠ This is NOT a re-opening of the 18a census, which is closed and stays closed: that census ran 2026-08-18 over candidates declined on the bar. This decline was made 2026-08-25, after it, and on a claim about Latin syntax rather than on the bar. ⭐ A stint can apply the right rule to a wrong fact. Check the fact the rule is standing on.