Liber III Regum

Anselm of Laon and his school, on the third book of Kings. Translated 2026-08-25 by two translators working in parallel — chunks 0000–0005 (cols 0581A–0599D, 7,348 Latin words) and chunks 0006–0010 (cols 0600A–0610D, 4,270 words) — and merged here. 11 chunks, 11,618 Latin words. verify-english.mjs clean over the whole work.

Apparatus fired, work-wide: 13 [var:] · 1 [sic:] · 1 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [cn:]. Column anchors and [n:] notes at exact parity throughout; zero em-dashes in the English; question-mark parity exact in every one of the eleven chunks.

Ruth (8968) cruces.md §1 governs and was not reopened. Tobit (8970) §1(b) governs the abridgment formula: usque ad → as far as, with no verb ever supplied to complete a resumption tail — which this book needed repeatedly, since three of its sentences assert nothing at all once the abridgment has taken their verb, and they are rendered verbless.

§M · THE MERGE — one real drift, repaired; the rest checked and sound

Two translators cannot see each other's work and diverge on terminology without knowing it. So the merge is a check, not a concatenation, and it was made against the English files themselves, not against the two stints' vocabulary tables — a table records what a translator believed it did.

⛔ The drift that was real: a lowercased Ecclesia

At 0600C the second stint rendered Migne's benedixit omni Ecclesiae Israel as "he blessed all the church of Israel." Its Latin twin capitalizes all four of its Ecclesia forms and prints no lowercase one anywhere in the chunk. Repaired to "the Church of Israel".

⚑ Two things make this worth recording rather than fixing quietly. First, that stint's own table claimed the mirror-the-plate rule as an inherited convention — it stated the rule correctly and then broke it once, which is exactly the failure a merge exists to catch and exactly the failure a self-audit cannot. Second, the site was afterwards read at the plate (Gallica f300, right column band C, for an unrelated citation check) and Migne prints omni Ecclesiae Israel with the capital. The repair is confirmed from the photograph, not merely from our transcription.

Checked across the seam and sound, needing no intervention

• Migne's Solomon-etymology, "the peaceable one" — carried identically across the seam (0000 and 0005 in the first stint, 0006 and 0007 in the second), including in the four glosses that turn on it.
• doctores → teachers, never "doctors", in both stints, ten occurrences.
• The Temple vocabulary, which is where this book was most at risk. The first stint held cols 0586A–0599D — CAPUT VI and VII, Solomon's building account and the densest notes in the book — and fixed the members before translating them: oraculum = oracle, tabulatum = boarding, latera = sides, humeruli = little shoulder-pieces, and anaglypha left in Latin because the gloss is an etymology of the Greek word (Migne prints Ἀνάγλυφα and glosses it Latine dicuntur caelaturae). Anglicizing that one would have left the gloss explaining nothing — the 11551 failure exactly. The second stint inherited the table and its chapters carry the vocabulary over without divergence.
• The clipped-lemma expansion rule, applied ~20 times in the second stint, which listed both classes — expanded and unexpanded — so the split is not misread later as inconsistency. Two lemmata were deliberately NOT expanded (In sanctum, Praebente regi) because Migne's own lemma diverges there and expanding would be conformation.
• usque ad, censused per occurrence in both halves and never by rule: 94 occurrences in the first range of which 14 are not the formula, 27 in the second of which 2 are not. The traps were caught, including the lemma A labio usque ad labium, where the words are the verse's own.

Nothing else was rewritten. A merge that silently edits a translator's chunk destroys the only record of what that translator actually read.

§P · SIX MISPRINTED CITATIONS — ALL SIX READ AT THE PLATE, ALL SIX MIGNE'S

The first stint found six citations whose numerals do not match the words they are attached to, left every one exactly as printed, and referred them to data/citation-corrections.json — the right move twice over, since the text keeps what Migne prints and only the index is corrected.

Each was identified from the QUOTED TEXT and never from the numeral, which is the settled discipline, and each was then read on the second witness rather than asserted from our file. Five crops settled all six, at one HTTP request apiece:

⭐ The 0592D one is the striking case: it is not a wrong numeral inside a book but a wrong BOOK, and not even a wrong Testament — the words are the very verse the gloss is expounding, III Reg. 6:36, cited as I Corinthians. On the facing page Migne cites I Cor. correctly twice (I Cor. IV, Ibid. VI), so the fault is local to that note and not a compositor losing the thread.

⚑ And the 0589D one shows why the quoted-text discipline is not pedantry. Ps. 118 does have a verse 72. A reader checking the numeral alone would find a real verse there and stop; only the words (Bonum mihi lex oris tui — not these) show the citation is wrong.

What was done: all six printed forms stand verbatim in the text, refDisplay keeps them, and the index resolves to the true reference with the printed key retained and auditable. Unparsed citations for this work: 0.

§Q · WHAT THE PLATE ALSO SETTLED, UNASKED

The five crops were fetched for the citations, and read whole — which is the addendum's own advice, that the normalization class lives in the neighbouring words rather than the target one. Three of the first stint's declines came back confirmed:

• 0593C usque ad ... ulla in parte declinemus — a negative is plainly wanted (ne ... ulla in parte declinemus) and the translator did not supply it, judging that the ne fell inside what the abridgment elides. The plate prints exactly that, with nothing lost by us.
• 0599D Rivocorulam — a strange real word, [sic:] declined per addendum A. Migne prints it.
• 0600C capere non poterat standing without its object — his, not ours.

⚑ All three were declines, and all three were right. That is the addendum's finding restated from the other end: the argument for a marker loses, and the crux is what survives.

Translated 2026-08-25, one translator, 6 chunks (7,348 Latin words by the frontmatter counts; 7,417 counting the body as written), on Opus. verify-english.mjs reports no error on these six files (the run fails only on 0006–0010, which are a peer's batch and not yet written). EN/LA ratio 1.43–1.52.

Apparatus fired in this half: 6 [var:] · 0 [sic:] · 0 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 161 [n:] (body count; the frontmatter noteCount fields sum to 156 — runbook false positive 3, count the body). Exact [n:] parity with the Latin twins, contents verbatim. Zero guillemets in either language. Zero em-dashes in the English. Structure: CAPUT PRIMUM, CAPUT II–VII, with CAPUT VI. (cont.) and CAPUT VII. (cont.) at the chunk splits.

The plate spot check for this work is a checked zero — Migne prints no numbered Forte apparatus here, no [cn:] is in the Latin and none is in the English.

§0 · WORK-WIDE VOCABULARY — FIXED AT THE HEAD, FOR THE MERGE

⚠ My range is the Temple (CAPUT VI–VII, chunks 0002–0005). The 11551 failure — one half of a split work anglicizing a column's members while the other kept the Latin — is the risk this section exists to close. These are the choices I made and held; a peer taking 0006–0010 should either adopt them or say why not, and the merge should reconcile against this list, not against taste.

Chapter heads follow Ruth's shipped English (8968), as II Regum (8953) did: CAPUT PRIMUM. → CHAPTER ONE., CAPUT II. → CHAPTER II., and so on with the roman numeral kept. (cont.) rides through unchanged.

§1 · usque ad — 94 OCCURRENCES, AND 14 OF THEM ARE NOT THE FORMULA

Counted mechanically and then decided one by one from what stands on either side, per addendum D. 94 occurrences of usque in the six chunks: 81 fall inside an italic span, 13 stand in ordinary roman prose. Of the 81, 80 are the abridgment formula; the remaining one is a lemma. So the non-formula population is 14, and here is every one of them:

• 0581B in incommoditate frigoris usque ad mortem permansit — "even unto death."
• 0583D inclinavit usque ad humilitatem hyssopi, id est, usque ad stultitiam crucis — twice, inside the resumption tail of a formula that ends four words earlier. ⚑ This is the trap the addendum names: the formula's own words and the tail's words are the same words.
• 0586C non usque ad tectum templi, sed usque ad coenaculum inferius pertingebant — twice, and see §4.1: the printed negative belongs to the first phrase, and the sed proves it.
• 0588C–D five in one sentence — a pavimento domus usque ad summitatem parietum et usque ad laquearia … a primis fidei rudimentis usque [0588D] ad perfectionem … et usque ad ingressum patriae coelestis … a primis justis usque ad ultimos. ⚑ The third of these has a column anchor inside it, exactly as the formula sometimes does; it is not the formula.
• 0592D Usque ad hoc atrium universi conscendunt electi — capitalized, sentence-initial.
• 0594C the lemma A labio usque ad labium. (III Reg. 7:23) — the verse's own words, and the one usque inside an italic span that is not the formula. Rendered "From brim to brim." and not italicised as a formula.
• 0594C a primo baptizato … usque ad ultimum qui in fine saeculi crediturus est.
• 0595B a triginta cubitis circuitus … usque ad decem est coactum.

§2 · MIGNE'S OWN IRREGULARITIES, REPRODUCED NOT TIDIED

• 0583D · [0583D]VERS. 32.-- — the anchor set hard against the verse address with no space. Reproduced without a space. Same at 0584B, 0585A, 0588C, 0593A, 0597C, 0599B.
• 0584B · [n: (BEDA, ut supra. )] — the note's own internal space before the closing parenthesis is kept verbatim. Same class at 0584D (Latomorum. [n: (BEDA, ibid.)]. — a full stop after the closing italic as well as inside it, both reproduced), 0596A (Duodecim. [n: (BED. ubi supra)].), 0598A (Eratque. [n: (ID., ibid)].).
• 0591C · (BEDA, [0591C] ibid., cap. 15.) — this siglum is not an [n: …] note: Migne sets it inside an italic span as running text, and a column anchor falls inside it. Carried verbatim, unanglicized, anchor in place. Exactly the class 8953 logged at 0565D.
• 0593A · Psal. VI) — an unbalanced closing parenthesis inside the italic scripture span (Domine, ne in furore tuo arguas me, neque in ira tua corripias me Psal. VI)), with no [n: wrapper and no opening bracket. Carried verbatim and unanglicized, on the 0591C footing. Not marked: it is apparatus, not prose, and stacking a [sic:] on a citation would be a claim about the plate I cannot make.
• 0590A · possunt intelligi.-- Olivarum. — a second lemma introduced mid-paragraph by .-- rather than on a new line. Reproduced.
• 0599A · …in modum repandi lilii erat efformata — the sentence ends with no full stop, and the English ends without one (Pattern 8: never supply).
• 0596B · Grossitudo. printed twice — once as the VERS. 26 lemma and again four lines below as unde subjungitur, Grossitudo. This is not dittography and takes no [d:]: Migne is re-quoting his own lemma to hang a second gloss on it, which is this recension's ordinary habit (cf. 0594B Malogranatorum. re-quoted the same way).
• 0596C · VERS. 27.-- [n: (ID., ibid.)] Mille perfectionem significat — a verse address with no lemma at all, the note standing where the italic span would be. Reproduced as printed.
• 0590C · [n: (ID., ibid.)] Et opus. — the note stands inside the italic lemma span and before the lemma words. Reproduced in place.
• Wrong citation numerals, five of them, left exactly as printed (they are data/citation- corrections.json business at index time, never text repairs — the Ruth precedent): (Joan. III) at 0590B for 1 Joan. 3:2 · (I Cor. II) at 0592D for III Reg. 6:36, which is not a New Testament text at all · (Psal. CXVIII, 72) at 0589D for Ps. 72:28 · (II Cor. III) at 0593B for 1 Cor. 3:6 · (II Cor. XV) at 0599C for 1 Cor. 15:41 · (Joan. XV) at 0599D for Joan. 14:2.

§3 · THE SIX [var:], AND WHY EACH EARNED A PUBLIC CLAIM

Every one rests on sources/vulgate/clementine-flat.txt, cited by book chapter:verse.

⚠ Every [var:] content string was checked NOT to be a verbatim substring of its Latin twin (the Pattern 14 mirror test) — which is why 0598A's reads rotunditas dimidii cubiti whole rather than the bare dimidii cubiti, a fragment the twin does contain.

§4 · THE LEMMA COLLATION — EVERY ITALIC SPAN IN 0581A–0599D WALKED

data/briefs/8954-lemmata.txt was walked line by line for the whole band. The agreements are listed as well as the divergences, because a findings-only list cannot be told from a list by a reader who never looked. Verse references are to the Clementine on disk; a lemma clipped at either end is counted as agreeing.

4.1 Checked and AGREEING with the Clementine (clip at either end excepted)

CAPUT I–IV (0581A–0584A): Et rex David senuerat. (3Rg 1:1) · Cumque operiretur, (1:1) · Adonias autem. (1:5) · Tu quoque. (2:5) · Venit ergo Bethsabee ad regem. (2:19, clipped before Salomonem) · Et non egredieris, (2:36) · Si autem ambulaveris in viis meis, (3:14) · Ego et mulier. (3:17) · Peperi apud eam in cubiculo. (3:17) · Mortuus. (3:19) · Tulit filium. (3:20) · Quem diligentius intuens, (3:21) · Dixit ergo. (3:24) · Dividite. (3:25) · Dixit autem. (3:26) · Cibus Salomonis. (4:22) · Viginti boves. (4:23) · Excepta venatione. (4:23) · Quadraginta, (4:26) · Et duodecim. (4:26) · Hordeum, (4:28) · Et fuerunt carmina ejus. (4:32) · Hyssopum. (4:33) · Et disseruit. (4:33) · Ignem veni mittere in terram (Lc 12:49, verbatim) · Nonne cor nostrum ardens erat (Lc 24:32, clipped before in nobis, question mark as Migne prints it) · Ubi posuisti eum (Jo 20:15) · Quis me tetigit (Lc 8:45).

CAPUT V (0584A–0585D): Et quia non potuerit aedificare domum, (5:3, the negative is the verse's own) · Praecipe igitur ut praecidant mihi, (5:6) · Scis enim quomodo non est in populo meo vir qui noverit ligna, (5:6, clipped before caedere sicut Sidonii) · Itaque Hiram dabat Salomoni ligna cedrina et ligna abiegna, juxta omnem, (5:10, the Clementine's comma after cedrina not printed) · Elegitque rex Salomon. (5:13) · Operarios de omni Israel, et erat indictio triginta millia virorum. (5:13, verbatim) · Absque praepositis. (5:16) · Trium millium. (5:16) · Praecepitque rex ut tollerent. (5:17) · Lapides grandes. (5:17) · In fundamentum templi. (5:17) · Porro Giblii praeparaverunt ligna et lapides. (5:18, verbatim) · Aperi, Libane, portas tuas (Za 11:1, verbatim) · Superaedificati super fundamentum apostolorum et prophetarum (Eph 2:20, verbatim) · Mundemus nos ab omni inquinamento, (2Cor 7:1, clipped before carnis et spiritus) · Corripite inquietos, consolamini pusillanimes, (1Thes 5:14, verbatim).

CAPUT VI (0586A–0593A): Anno quarto. (6:1) · Mense Zio (ipse est mensis). (6:1, clipped before secundus — and the gloss supplies it, Quod mense secundo) · In longitudine, · Viginti cubitos. · In latitudine. · Triginta cubitos. · In altitudine. (all 6:2) · Porticus erat, (6:3) · Viginti cubitorum longitudinis, juxta mensuram latitudinis templi. (6:3, verbatim) · Fenestras obliquas. (6:4) · Et aedificavit, (6:5) · Latera. (6:5) · Trabes autem posuit in domo per circuitum forinsecus. (6:6, verbatim) · Trabes, · forinsecus. (6:6) · Domus autem, (6:7) · Et malleus et securis. (6:7) · Ostium lateris. (6:8) · Ascendebant in medium coenaculum et a medio in tertium. Et aedificavit. (6:8–9, a join of the two verses across the verse boundary, both verbatim) · Laquearibus. · Cedrinis. (6:9) · Tabulatum super omnem domum quinque cubitis altitudinis. (6:10, verbatim) · Et aedificavit parietes. · Parietes domus. · A pavimento. (6:15) · Et texit pavimentum. (6:15) · Porro. (6:17) · Et cedro omnis. · Habens tornaturas. · Caelaturas. · Omnia cedrinis, (6:18) · Oraculum. (6:19) · Porro oraculum. · Sed et altare. (6:20) · Domum quoque ante oraculum. · Clavis aureis. (6:21) · Nihilque erat in templo. (6:22) · Cherubim de lignis. · Olivarum. · Decem cubitorum. (6:23) · Quinque cubitorum. · Ala cherub altera. (6:24) · Decem quoque cubitorum. (6:25) · Et opus. (6:25) · In medio templi interioris. (6:27) · Extendebant autem alas, (6:27) · Texit quoque cherubim. (6:28) · Sculpsit variis coelaturis. (6:29, Migne's coe- against the Clementine's cae-; orthography only) · Et palmas. · Et picturas varias. (6:29) · Et in ingressu oraculi, (6:31) · Postesque angulorum. (6:31) · Anaglypha. (6:32) · Fecitque in introitu. · Postes de lignis. (6:33) · Altrinsecus. · Et utrumque ostium. · Et se invicem. (6:34) · Et sculpsit cherubim. (6:35) · Aedificavit atrium. · Aedificavit atrium interius tribus ordinibus · Et uno ordine. (6:36) · Ipse est mensis. · Aedificavitque eam. (6:38) · Christi bonus odor sumus Deo (2Cor 2:15, clipped) · De regionibus congregavit eos. A solis ortu et occasu, ab Aquilone et mari (Ps 106:2b–3, verbatim) · Ego diligam eum, et manifestabo ei meipsum. (Jo 14:21, verbatim) · Venit hora cum jam non in proverbiis loquar vobis, (Jo 16:25, verbatim, clipped before sed palam) · Mihi autem adhaerere Deo, (Ps 72:28, clipped before bonum est) · et quae sursum sunt sapite, non quae super terram (Col 3:2, verbatim, both negatives intact) · In hoc cognoscent omnes, quia discipuli mei estis, si dilectionem habueritis ad invicem (Jo 13:35, quia for the Clementine's asyndeton after omnes) · Imitatores mei estote sicut et ego Christi (1Cor 4:16, verbatim) · Nescitis quoniam angelos judicabimus, quanto magis saecularia (1Cor 6:3, verbatim; Migne prints neither of the Clementine's two question marks and none is supplied) · Domine, ne in furore tuo arguas me, neque in ira tua corripias me (Ps 6:2, verbatim, both negatives intact) · in spiritu et virtute Eliae (Lc 1:17, verbatim) · Beati immaculati in via, (Ps 118:1).

CAPUT VII (0593B–0599D): Fecit omne opus. (7:14) · Et finxit. · Decem, · Et linea. (7:15) · Duo, (7:16) · Septena, (7:17) · Et duos. · Ut tegerent. · Malogranatorum. (7:18) · Quasi, · In porticu. (7:19) · Et rursum · Malogranatorum. (7:20) · Columnam. (7:21) · Fecit quoque. · A labio usque ad labium. · Quinque cubitorum. · Et resticula triginta cubitorum cingebat illud. (7:23, clipped before per circuitum) · Et sculptura. (7:24) · Duodecim. · E quibus tres. (7:25) · Grossitudo. ×2 · Lilii. (7:26) · Et fecit. · Quatuor cubitorum longitudinis. (7:27) · Et ipsum opus basium. (7:28) · Et subter leones. (7:29) · Et quatuor rotae. · Humeruli. · Contra. (7:30) · Unum cubitum. · Media. · pariterque habebat unum cubitum et dimidium, (7:31, verbatim) · In hunc modum fecit decem bases, fusura una, et mensura, sculpturaque, (7:37, clipped before consimili) · Quadraginta batos capiebat luter unus, · Eratque. (7:38) · Et constituit decem bases, quinque ad dexteram partem templi, et quinque ad sinistram. (7:39, verbatim) · Mare autem posuit ad dexteram partem templi, (7:39, clipped before contra orientem) · Jordanis fudit ea rex in. (7:46, a fragment clipped at BOTH ends — see §5.6) · Argillosa terra. (7:46) · Fecitque Salomon omnia vasa in domo, (7:48, clipped before Domini) · Altare aureum. (7:48) · Mensam. (7:48) · Quasi lilii. · Lucernas desuper aureas. · Forcipes. (7:49) · Hydrias. · Fuscinulas. · Domus interioris. (7:50) · Perfecit omne opus. · Et intulit quae sanctificaverat. · Reposuitque in thesauris domus Domini. (7:51) · Ex parte enim cognoscimus, et ex parte prophetamus, (1Cor 13:9, verbatim) · Non alligabis os bovi trituranti, (1Cor 9:9, verbatim) · Funiculus triplex difficile rumpitur (Ecl 4:12, verbatim) · Poenitentiam agite, (Mt 3:2) · Capitella autem quae erant super capita columnarum, (7:19, verbatim) · In domo Patris mei mansiones multae sunt (Jo 14:2, verbatim) · Habemus firmiorem propheticum sermonem, cui bene facitis attendentes, (2Pt 1:19, Et dropped at the head and benefacitis printed as two words).

4.2 DIVERGENT but taking no marker — every one named, with the reason

• 0581A · Hi sunt qui venerunt de calore domus Rechab (I Par. II). 1Par 2:55 reads Hi sunt Cinaei, qui venerunt de Calore patris domus Rechab. Two dropped words. No marker: the gloss adduces the verse only for the calor, which is intact.
• 0582B · Introducam te in domum matris meae, et in cubiculum ejus, quae concepit me (Ibid. III). Ct 3:4 reads donec introducam illum in domum matris meae, et in cubiculum genetricis meae. Two divergences: the object pronoun is switched from third person to second, and genetricis meae is paraphrased. No marker, and the reasoning is the point: the two words the gloss builds on — domum matris meae and cubiculum — agree exactly, and this quotation is adduced after* the one at 0582B that does take a [var:], as a second witness to the same word. Marking the pronoun would put a public claim on the half of the citation that is not doing any work. Logged, not marked.
• 0582C · Vere hic erat Filius Dei (Marc. XV). Mc 15:39 reads Vere hic homo Filius Dei erat. A dropped noun and a word order change; the centurion's confession asserts the same thing either way.
• 0582C · Pater, ignosce illis, Lc 23:34 reads Pater, dimitte illis. A verb substitution with the same sense ("pardon" for "forgive"), and nothing in the gloss turns on it — the paragraph is about the three and five thousand who believed afterward. Contrast the Matt 27:24 site above, where the substituted word IS the gloss's image; that is the whole difference between a [var:] and a line here.
• 0581A · Abiit itaque Salomon. 3Rg 3:4 reads Abiit itaque in Gabaon. Migne supplies the subject from later in the same verse (obtulit Salomon). Not a divergence in substance.
• 0584C · Adoniram erat super hujuscemodi indictionem. 3Rg 5:14 has indictione, ablative. One letter, and sources/vulgate/README.md says this copy is not to be trusted to that precision.
• 0584D · Fuerunt itaque Salomoni septuaginta millia… 3Rg 5:15 has Fueruntque. A connective.
• 0585C · Ad aedificandum domum. 3Rg 5:18 has ad aedificandam domum. A gerundive ending; the English is identical either way.
• 0586A · Aedificare coepit. ⚑ 3Rg 6:1 has aedificari coepit domus Domino — passive. Migne's lemma is active: "he began to build" against "the house began to be built." ⚠ This one is worth the next reader's attention and I want to be plain about why I declined it. Two things pull opposite ways: the divergence is a single letter (-e for -i), which the README says not to adjudicate from this file; but the difference is a change of voice, and the gloss's own re-quotation three lines above prints the passive (coepit aedificari domus). I rendered Migne's active and fired nothing. The rival — a [var:] on the voice — lost because the marker is a public claim about a one-letter difference in a copy whose own README disclaims one-letter precision, and because nothing in the gloss is falsified either way.** It is a good plate candidate if this work is ever read at the plate.
• 0586A · Quae praeparaverat David, — ⚑ located nowhere in the Clementine. There is no such clause at III Reg. 6, and the nearest text is 2Par 3:1 in loco quem paraverat David, which the gloss then expounds (paraverunt … locum Domino). Recorded as unlocated rather than as divergent: it may be Bede's text of the Chronicles parallel. No marker.
• 0586A · Levate oculos vestros … ut qui seminat simul gaudeat et qui metit (Joan. IV). A conflation of Jo 4:35 with 4:36, both halves printed accurately, joined at ad messem: et qui metit. Abridgment family (8970 precedent: an abridgment is not a variant). The Clementine's ut et qui seminat loses its et. No marker.
• 0586D · Porticum vero ante frontem quae tendebatur in longum juxta mensuram latitudinis cubitorum viginti (II Paral. III). 2Par 3:4 has juxta mensuram latitudinis domus, cubitorum viginti. One dropped noun. No marker: the gloss argues from the twenty cubits and the orientation, not from whose breadth it is.
• 0588A · Noe vir perfectus in generationibus suis, Gn 6:9 has Noe vir justus atque perfectus fuit in generationibus suis. Migne compresses. No marker: the argument turns on perfectus, which is the word that survives, and the sentence is adduced precisely to be qualified (nullus vere perfectus … incedere potest).
• 0588D · Super omnia autem charitatem habentes, and 0591A · super omnia autem charitatem, quae est vinculum perfectionis (both Col. III). Col 3:14 reads Super omnia autem haec, caritatem habete, quod est vinculum perfectionis. Migne drops haec, turns the imperative into a participle at the first site and into a bare accusative at the second, and writes quae for quod. No marker at either: both are woven into the glossator's own sentence (the second continues a list of virtues in the accusative), which is quotation-as- syntax, not a rival text.
• 0588D · Aedificavit viginti. 3Rg 6:16 has Aedificavitque. An enclitic.
• 0590B · Cum apparuerit, similes ei erimus et videbimus eum sicut est (cited Joan. III, in fact 1Jo 3:2). The Clementine has quoniam videbimus eum sicuti est. A causal connective flattened to a coordinating one, and one letter on sicuti. No marker:* the gloss adduces the verse for the imago* recovered, and neither reading disturbs that.
• 0591C · Duo ostiola de ligno olivarum. ⚑ 3Rg 6:31 reads fecit ostiola de lignis olivarum (no number), and 6:32 reads Et duo ostia de lignis olivarum. Migne's lemma is a conflation of the two verses, taking duo from 6:32 and ostiola from 6:31, and printing the singular ligno. This one is a near miss and I say so: the gloss's first sentence is Duo sunt ostiola, so the number is load-bearing — but it is load-bearing on a word the received text does print one verse later, which makes this a conflation and not a substitution. Conflations take no [var:] in this series (8953 @0571B, @0580B). Logged.
• 0593B · Ego plantavi, Apollo rigavit, Dominus autem incrementum dedit (cited II Cor. III, in fact 1Cor 3:6). The Clementine has sed Deus incrementum dedit. A connective and a divine name. No marker:* Dominus and Deus have the same referent here and the gloss (opus Dei operantur*) is unaffected.
• *0594A · ambulant de virtute in virtutem (Psal. LXXXIII). Ps 83:8 has ibunt* de virtute in virtutem. A verb substitution of the same sense, and the phrase is woven into the glossator's own clause (quod sancti ambulant). No marker. The same idea stands unquoted at 0585D (proficiendum de virtute in virtutem).
• 0594B · Inter vestibulum et altare plorabant; Joel 2:17 has plorabunt sacerdotes. A tense and a dropped subject. No marker: the quotation is adduced only to prove that vestibulum and porticus name the same place.
• 0596B · Duo millia batos capiebat et tria millia, ⚑ 3Rg 7:26 has duo millia batos capiebat and nothing more; tria millia comes from 2Par 4:5. A conflation — and Bede's gloss on it is precisely about reconciling the two accounts, so the conflated lemma is doing exactly the work the gloss needs. No marker.
• 0596D · Inter et coronulas et plectas… 3Rg 7:29 reads Et inter coronulas. The et has moved one word. Rendered by taking the printed et adverbially ("Between the little crowns also and the plaited work"), which is a real sense of et and needs no emendation. Logged; no marker.
• 0597B · Initium Evangelii Filii Dei sicut scriptum est in Isaia propheta; Mc 1:1 reads Initium Evangelii Jesu Christi, Filii Dei, and Sicut scriptum est in Isaia propheta is 1:2. A dropped name and a join across the verse boundary. Conflation family; no marker.
• 0597B · ut adimpleretur Scriptura prophetarum (Matth. XXVI). Mt 26:56 has ut adimplerentur Scripturae prophetarum. Singular for plural, twice over, asserting the same thing. No marker.
• 0599C · quamvis in ea stella a stella differat in claritate (cited II Cor. XV, in fact 1Cor 15:41). The Clementine has Stella enim a stella differt in claritate. The subjunctive follows from the glossator's own quamvis, which is inside the italic span; this is quotation-as-syntax again. No marker.
• 0598A · ita fabrefacta ut luter superimponi posset. 3Rg 7:35 has ut luter desuper posset imponi. A compound verb for a verb-plus-adverb, same sense. Recorded beside the [var:] that the same lemma earns for its number, and deliberately kept out of the marker.

4.3 The name- and etymology-italics — rendered as the Latin says

These are not scripture and 7a″ does not reach them; they are rendered from the Latin printed and never from a Hebrew lexicon. In my range there are only two, plus one that is not italicised: dominator dominus → "ruler lord" (0581B, of Adonias) · cherubim … id est, multiplicata scientia, vel scientiae multitudo → "multiplied knowledge, or a multitude of knowledge" (0590A) · and the roman etymology of Adoniram at 0584C, Dominus meus excelsus → "My lord is exalted."

§5 · CRUCES — the plate, and what was declined

5.1 Non-word candidates — SEVEN, all raised for the plate, NONE marked

Per the 2026-08-23 addendum, a form that is no Latin word at all has been our digitization's corruption ten times out of ten, never Migne's, and 8953 then made it twelve. Each is rendered for its evident sense, logged here, and carries no [sic:], because that marker is a public claim that Migne's plate is defective. ⛔ I have no plate access and did not patch anything.

⚑ Qui dam is on the addendum's own confirmed list (found in the 2026-08-23 batch and proved to be ours). It recurs here, in a different volume, which is worth recording.

5.2 0593B · Domum autem mulieris vi. — the chunk's opening lemma, and a truncation

Migne's lemma for CAPUT VII opens the chapter and reads, as we hold it, Domum autem mulieris vi. The gloss is Ecclesiae, scilicet praesentis, pro qua Christus vir suus morte gustata surrexit — the Church whose husband died, i.e. the widow. Bede's own lemma at that place is III Reg. 7:14, filium mulieris viduae de tribu Nephthali, and the [n:] on this very paragraph cites lib. de templo Salom., cap. 17, which is the chapter on 7:13–14.

So vi. is with near-certainty a truncation of viduae, and the sense is not in doubt.

What I did, and what I refused. I carried the stub into the English inside the lemma's own italic span — "But the house of the woman vi." — and fired no marker.

• Expanding it to "the widow woman" was the rival, on Glossa convention 3 (a fragmentary lemma is rendered as the phrase it stands for). It lost because convention 3 is about lemmata Migne clipped deliberately, and this is a broken word, not a clipped phrase: expanding it would put a word on the page that no witness we hold prints, exactly the silent repair Pattern 7a forbids.
• A [sic:] was the other rival. It lost on the ship test: it is a public claim that Migne's plate is broken, and the seven-for-seven finding above says this class is ours.
• 7a⁗'s fourth question was asked: the quarantine takes no verb, no negation, no conjunction and no object — vi. is an attributive on mulieris and the phrase stands without it.

⚠ This is the single best plate candidate in my half. PL 113 col. 593 is p. 302 of the archive.org copy by the recorded map, and CLAUDE.md records that that copy clips the left margin on some pages — so if it comes back clipped, the Gallica witness (ark:/12148/bpt6k5505319w, f = (column + 1) / 2) settles it in one request.

5.3 0589A · usque et ad — the abridgment formula split by an intruded et

Migne prints, in the gloss on Omnia cedrinis: Lapides parietis vel pavimentum, et tabulae, et aurum, etc., usque et ad qui gratiam Evangelii perfecte susceperunt, aeterna vita pariter perfruuntur. The two words of the formula stand in two separate italic spans with a roman et between them — the only occurrence of that shape in 94 uses of usque in this half.

Rendered as printed, with the italic structure 1:1: "as far as and to those who have perfectly received the grace of the Gospel alike enjoy eternal life." The English is awkward because the printed line is.

The reading I believe is behind it, and did not put in the text: usque ad et qui gratiam Evangelii perfecte susceperunt… — the formula whole, followed by a resumption tail opening on et qui, which construes perfectly and is this work's ordinary shape. The rival — treating et as a stray to be dropped — lost because dropping a printed word is an emendation, and because the italic markup may be ours rather than Migne's (11632's informen t is the precedent for suspecting our own asterisks). Plate candidate, and cheap: p. 300 of the archive copy.

5.4 Real words that read oddly — rendered as printed, markers DECLINED, rivals named

Per addendum A: a real, well-formed word is Migne's until a plate says otherwise (refuted five times out of five), so none of these fires anything.

• 0591A · ne eo minus ab arcae justitiae cadant — arce is what ab governs, and the work's own control is one column away in my other chunk: 0584D prints ad arcem virtutum quae est in unitate Ecclesiae, the same image in the same voice. Rendered "from the citadel of justice". Declined: a case slip has no English exponent (Pattern 9 family), and 18a leaves case and mood anomalies that merely fail to construe declined.
• 0596A · Apostolus evangelistas, imo omnes verbi ministros; — a nominative heading a series of accusatives (evangelistas … ministros), where Apostolos is required. Rendered as a flat list, "The Apostle, the evangelists, nay rather all the ministers of the word", with no copula supplied, because supplying one would decide the case question in the English. Declined: same class as above.
• 0599C · Quia deos suos abscondit in abdito vultus sui a conturbatione hominum. ⚑ This is Ps 30:21, Abscondes eos in abscondito faciei tuae a conturbatione hominum — third-person and freely paraphrased throughout (abdito for abscondito, vultus for faciei), and deos for eos is one letter more than the rest of the paraphrase. The [cj:] was seriously considered and DECLINED, and here is why the rival lost. deos is a perfectly good Latin word, and in this sentence it reads: "he hides his gods in the hidden place of his countenance" — the saints, who are the treasures of the house of the Lord, called gods by adoption after Ps 81:6 Ego dixi: Dii estis. That is a live patristic reading in this author's own idiom, and the paragraph is about the saints hidden among many treasures in one house. Pattern 18's bar is that the English asserts something false, and under that reading it does not. Rendered as printed.
• 0593C · …a recta via qua ad promissam patriam gradimur, ulla in parte declinemus. ⚑ ullus wants a negative, and there is none on the page. No negative was supplied (7a is unconditional), and the English reads "that in any part we may turn aside from the straight way." The reason it is not a defect of Migne's: the missing ne falls in the text the usque ad elides — this is the tail of an abridgment, and the formula removes words. The brief's own warning applies in reverse here: the formula removes words, it does not suspend grammar, and the grammar of what survives is exactly what was printed. Logged, nothing marked.
• 0587D · Tantae longitudinis erat, ut capita earum forinsecus prominerent. A singular verb against a plural genitive. earum fixes the referent as trabes two words above in the lemma. Rendered "It was of such length that their heads projected on the outside." Declined: number agreement has no English exponent here.
• 0592A · cum sobrie, et juste, et pie cogitant de eis per ipsos corporis sensus agere decernunt. Two finite verbs with no conjunction between them. Rendered as a cum clause answered by a main clause ("when they think soberly, and justly, and piously, they determine to act upon those things through the senses of the body themselves"), which is the only reading that gives both verbs a slot. Declined: asyndeton is not a false claim.
• 0587C · luriculas, twice — for loriculas, little parapets. Listed apart from the non-words above because it may not be one: u for o is ordinary medieval orthography (Pattern 9), and the form is printed consistently twice in the same gloss, which is what a spelling looks like and not what a slip looks like. Rendered "little parapets", no marker, no crux beyond this line.
• 0590B · veniunt denarium — an accusative of the goal without ad. Rendered "they come to the denary". Declined, Pattern 9 family.
• 0594B · Notandum autem in hac sententia Paralipomenis… — the book name without a preposition. Rendered "in this sentence of Paralipomenon". Declined, same class.

⭐ Recording the negatives, as the addendum asks. Four real words were suspected and refused a marker on the rule that a real word is Migne's until a plate says otherwise: deos (0599C), arcae (0591A), Apostolus (0596A), aedificare (0586A). Two of the four had good arguments behind them — the first two — and the argument is exactly what the addendum measured as losing.

5.5 A form that struck me as suspiciously TIDY (brief §4, the normalization class)

0588A–B · Qui dam hunc locum male intelligentes — and its neighbours in the same look, per addendum C. Reading the whole line rather than the target word: et bene unum ostium, propter consonam in omnibus sanctis fidem et dilectionem veritatis runs clean, and Notandum quod triginta cubiti altitudinis de quo supra legitur has the well-known quod … de quo mismatch that reads like Migne rather than like a tidy-up. Nothing in my range struck me as normalized, which I record as a checked negative rather than as silence.

5.6 A lemma clipped at BOTH ends

0598C · Jordanis fudit ea rex in. — III Reg. 7:46 reads In campestri regione Jordanis fudit ea rex in argillosa terra, inter Sochoth et Sarthan. Migne's lemma begins mid-phrase (the genitive Jordanis with its governing In campestri regione cut away) and stops on the preposition in, because the words it stops before are the next lemma, Argillosa terra. Rendered as the printed words stand — "Of Jordan the king cast them in." — and not completed, per the brief's rule on clipped lemmata: nothing is elided inside it, so convention 3's expansion does not apply.

§6 · UNEXPRESSED SUBJECTS — every supplied pronoun named, and what named it

The dominant defect class. Every "he/she/it/they" in my English that the Latin does not print:

• 0581A–B · "he grows not warm", "he wasted away", "he remained" — David, from the lemma Et rex David senuerat and Frigidus senex.
• 0586A · "she was made possessed of that inward peace" — feminine, and not decidable from the sentence: it is the tail of an abridgment (compos effecta est) and the noun it agrees with was elided by the usque ad. Bede's chapter is on Jerusalem/the Church, both feminine, and either reading construes. Said here rather than chosen silently.
• 0586C, 0586D · "she patiently bears", "she may rejoice", "she despises" — the Church, named in the first clause of the same gloss (longanimitatem Ecclesiae significat).
• 0587B · "they could not hear his doctrine" — the antiqui justi of the same sentence.
• 0590B · "they have twin wings", "they preach", "they show" — the cherubim, i.e. the two Testaments, named in the note-paragraph immediately above (Possunt per duo cherubim duo Testamenta figurari) and confirmed by suis auditoribus.
• 0590C · "they embrace one another" — the two cherubim.
• 0590D · "their writers … take care for our salvation" — printed (scriptores eorum).
• 0591A · "he makes palms", "he fixes" — the Lord/Solomon of the lemma Sculpsit; the same unexpressed subject runs through Facit picturas varias and Texit ergo Salomon, where Migne names him.
• 0591D · "they do somewhat", "they think", "they determine" — the qui … Domino serviunt of the same sentence.
• 0592D · "he speaks briefly of the inner court" — Bede, the cited author, whose voice the whole usque ad block is.
• 0594C · "he foretold" — the Lord, from the parable of the vineyard the clause alludes to.
• 0596B · "he makes them glow" — the subject of the abridged tail; not printed, and the tail is Bede's. Named here.
• 0598B · "I consider that they are alien from the flame of divine charity" — ⚑ the verb is printed first person (considero, Bede's own voice, which this recension keeps), and the subject of aliena esse is not printed at all; the neuter plural aliena is left without an antecedent by the abridgment. Rendered with "they" and flagged here.
• 0599B–C · "he does indeed the work", "he grants", "he finishes" — the Lord, named two clauses later (cum noster pacificus … glorificat).
• 0599C · "he hides his gods" — God, from Ps 30:21 (see 5.4).

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

1. [sic:] swallow test — vacuously clean: no [sic:] was fired anywhere in my half. The test was still run in its stripped form (every marker deleted, the remainder read aloud) on all six chunks, because the [var:] markers occupy the same position class: no stranded article or preposition, no doubled mark, every sentence readable without any bracket.
2. Negation attachment (2 + 2a) — every printed negative was traced to the single Latin word it attaches to. The trigger population (a negative on a verb with a quantifier, correlative or predicate nearby) fired five times, and each host was proved from structure, not from what reads well: - 0582D non ut possideret, sed ut occideret — hosts are the two purpose clauses. - 0582D Non enim fecit hoc filii amore, sed odio aemulae — the negative is on the ablative phrase, not on fecit; the sed odio correlative proves it. (Rendering it on the verb — "she did not do this" — would have read better and been wrong.) - 0584C Vivunt enim non agrum colendo, sed negotiando — on the gerund phrase. - 0586C triginta cubiti altitudinis non usque ad tectum templi, sed usque ad coenaculum — on the first prepositional phrase, not on pertingebant. - 0591C non solum animas electorum aula coelestis recipit, sed et corporibus … fores aperit — on animas electorum. - *0597B non quod quatuor sint tantum libri prophetici, sed* omnia quae locuti sunt prophetae … testimonium praebuere — ⚠ the sharpest of the five, because tantum sits right beside the verb. The sed clause affirms that ALL prophecy witnesses to the four Gospels, so what is denied is the whole quod clause with its tantum inside it, not the number four. Rendered "not that the books of the prophets are only four." Total negatives carried: every non, nec, neque, nisi, ne, nullus, nemo, nihil, nondum, nunquam the Latin prints, and none supplied — including at 0593C, where ulla plainly wants one (§5.4).
3. Column anchors, position not merely presence (2b) — all 76 anchors reproduced verbatim and in order, and each was then checked against the WORD it stands beside in the Latin. Six were found displaced on the first pass and moved: 0584C (belongs before colendo, "not by [0584C] tilling the field"), 0584D (before et visitatore), 0589D (before veniens), 0595C (before rerum historias), 0595D (before Domino — it had migrated a phrase downstream), 0596C (before surgat). Anchors split mid-construction to match Migne where he splits them: usque [0594B] ad, usque [0588D] ad perfectionem, and the anchor inside the 0591C siglum.
4. Punctuation, band by band, never by total (3) — the six chunks partition into 76 bands, and : ; ? ! « » — were counted in each band on both sides with the markers stripped. The first run returned 23 mismatched bands, and every mismatch was a mark of mine, not of his: 20 semicolons written where Migne prints a comma, and 5 colons — three of his restored where I had written a comma, two of mine removed where he prints none. All were repaired to the plate. Second run: zero mismatched bands in all six chunks. ⚠ Migne prints semicolons himself in this book — 23 of them across my half — so the Tobit exception does not apply and no comma→semicolon sweep was attempted; the repairs went the other way, back to his commas. Question marks: 4 in chunk 0000, 1 in chunk 0003, 0 elsewhere, matched exactly. ⚑ 8b did not fire. 0593A prints Quaeritur quomodo dicatur domus Domini in mense octavo perfecta… with a comma, not a question mark, so the indirect question was rendered indirectly and no colon was introduced. Supplying one would have been the mirror of the defect 8b exists to prevent.
5. The fourth test — does it ASSERT what he asserts? Every finished paragraph read back against the Latin with that question only. It returned three repairs that no count would have caught: qui de suo situ vel abstracti sunt (0585D) had lost its vel — "even drawn away" — restored; quia non nisi per Evangelium (0594B) had been softened to "not otherwise than", restored to "not except through the Gospel", which carries the exceptive as an exceptive; and the resumption tail at 0589A had been given "they who" where the clause needs "those who" as the subject of perfruuntur (§5.3). It also confirmed the filii amore and non solum … sed et attachments above, which had been the two I was least sure of.
6. Em-dashes: zero in all six English chunks.

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

• No [sic:], no [cj:], no [d:], no [ed:], no [nt:]. Seven non-word candidates and four real-word suspicions are logged above and none of them fires anything; the ship test governs and the crux is the safe place. Nothing here asserts a defect in Migne's plate.
• No plate read was performed — I have no plate access from this seat. The three candidates worth one page each, in priority order: 0593B vi. (col. 593), 0589A usque et ad (col. 589), 0586A aedificare / aedificari (col. 586). All three fall inside the range where our archive.org copy of PL 113 is known to clip the left margin, so the Gallica witness is the one to use.
• Vocabulary drift is the merge's real risk and §0 is the answer to it. The peer's chunks 0006–0010 are CAPUT VIII–XXII — the dedication, the queen of Saba, the schism — and only the first of those still handles Temple furniture. If the peer's half renders oraculum, tabulata, latera or anaglypha differently, §0 states the reason each choice was made so the merge can adjudicate rather than split the difference.
• Recurring names checked against I and II Regum (8952, 8953) and kept: Solomon, David, Bethsabee, Adonias, Hiram, Nathan, Sadoc, Joab, Semei, Libanus, Gabaon, Jerusalem, Israel, Sion. Two conventions carried without change: Migne's Hierusalem/Jerusalem variance is kept per site (my half prints the name once only, at 0581C, and prints it Jerusalem), and noster pacificus / verus Salomon render as "our peaceable one" / "the true Solomon", as 8953 has them.

Translated 2026-08-25, one translator, 5 chunks (4,270 Latin words), cols 0600A–0610D (III Reg. 8:6 – 22:19). verify-english.mjs clean across all 11 chunks of the work. Apparatus fired in this range: 1 [sic:] · 1 [cj:] · 7 [var:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 83 [n:] (exact parity with the Latin twins). Zero guillemets, zero em-dashes, zero [cn:] (this volume's plate apparatus is a checked zero — see the Ruth pilot, §5).

⚠ Chunks 0000–0005 are a peer's stint and are not covered here. Where a rendering of theirs governs mine it is named below (§0).

§0 · CONVENTIONS INHERITED, AND THE ONE I HAD TO STATE

0.1 Followed from the peer's 0000–0005, not re-derived

• usque ad → as far as, italics kept as one span. 25 occurrences in my range.
• Pacificus noster → "our peaceable one", lowercase — the peer's §0 traces it to 8953, and I carry it at 0601A, 0601B (twice) and 0602B. Not "our Peacemaker" — 8958 uses "peacemaker" for pacificus in the Beatitude sense and that is a different word doing a different job.
• Chapter heads: CAPUT VIII. → CHAPTER VIII., roman numeral kept, per the peer's §0 and Ruth's shipped English. My range runs CHAPTER VIII to CHAPTER XXII with no (cont.).
• Migne's own name-spellings ride as printed. The peer records that their half prints Jerusalem; my 0602B lemma of Isa. 62:6 prints Hierusalem and is rendered "Hierusalem", while the gloss prose at 0603B–C prints Jerosolymis / Jerusalem and is rendered "Jerusalem". Three spellings in one work is what he set, and the English does not level them.
• In Paralipomenis → "in Paralipomenon" (peer, five times). Verba/Verbis Dierum is a separate thing and is rendered literally, "the Words of the Days" (0600D), after 8951's ruling at 0700A — Migne prints two names for the book and the English keeps two.
• Names take the Latinate/Douay forms already standing in 8952 and 8953: Elias, Eliseus, Josue, Jeroboam, Naboth, Nathan, Esdras, Juda, Abdias, Nicodemus, Caiphas. Extended here, with no divergence, to Jezabel, Achab, Abia, Roboam, Baasa, Sesac, Micheas, Semer, Thersa, Emath, Ophir, Amorrhites. The one place I departed from Migne's own spelling is Michaeas → Micheas (0610B, twice), the received English form of the name; recorded here so it is not read as a slip.

0.2 The expansion rule I used for clipped lemmata, stated because I applied it ~20 times

Convention 3 (a fragment is rendered as the phrase it stands for) does most of the work in this range, which is the abridgment at its densest. The rule I held to, and it has a boundary:

A clipped lemma is completed to the smallest phrase from the glossed verse that stands as English on its own. A lemma that is grammatically complete as printed is rendered as it stands, however short. And words are supplied from the verse ONLY where Migne's own lemma does not itself diverge from the Clementine there — where it diverges, the printed words alone are rendered, because expanding would be conformation, which is the thing forbidden.

• Expanded (clipping leaves a dangling verb, conjunction, subject or governed word): Cumque. → "And when the staves stood out." · Subter alas. → "Under the wings of the cherubim." · In die. → "In that day." · Ab introitu. → "From the entrance of Emath." · Universum. → "The whole people." · De filiis. → "Of the children of Israel." · Ligna. → "Wood." · Ducenta. → "Two hundred shields." · Et duodecim. → "And twelve little lions." · Rex autem. → "And king Solomon." (both at 10:13 and at 11:1) · Erat autem. → "Now Jeroboam was a valiant man." · Tolle tibi. → "Take thee ten pieces." · Venit autem. Durissimum. → "And Roboam came. A most hard yoke." · Fecit duos. → "He made two calves of gold." · Ecce vir. → "Behold a man of God." · Sepelite, etc. → "Bury me," etc. · Aedificavit Hiel. → "Hiel built." · Apprehendite. → "Take the prophets of Baal." · Multae pluviae. → "Of a great rain." · Et ecce angelus Domini tetigit. → "And behold an angel of the Lord touched him." · Fecitque. → "And he made a trench." · Erit anima. → "Thy life shall be for his life." · Et furibundus. → "And in a rage he came into Samaria." · Haec faciant mihi. → "These things may the gods do to me." · Tulitque Elias. → "And Elias took the child."
• Left as printed (complete English as it stands): In arca. "In the ark." · Stetit. "He stood." · Et benedixit. "And he blessed." · Altare. "The altar." · Festivitatem. "A festival." · Laetantes. "Rejoicing." · Aedificavit. "He built." · Et Palmiram. "And Palmira." · Quinquaginta. "Fifty." · Classem. "A fleet." · In Ophir. "In Ophir." · Cantoribus. "For the singers." · Saltus Libani. "The forest of Libanus." · Thronum de ebore. "A throne of ivory." · Voraginem. "The gulf." · Dies autem. "Now the days." · Adolescentes. "The young men." · Scorpionibus. "With scorpions." · Ad Roboam. "To Roboam." · Prophetes. "A prophet." · Haec dicit Dominus. "Thus saith the Lord." · Ecce. "Behold." · Qui mortuus fuerit. "He that shall die." · In Abiram. "In Abiram." · Et dixit. "And he said." · Abscondere. "Hide thyself." · In torrente Carith. "By the torrent of Carith." · Mulier. "A woman." · Abdias autem. "And Abdias." · Curavit altare. "He repaired the altar." · Et spiritus. "And a wind." · Non in spiritu. "Not in the wind." · Quicunque. "Whosoever." · Profectus. "He departed." · Secutus. "He followed." · Fugit quoque. "And he fled." · Abiit ergo. "He went therefore." · Post verba, etc. "After the words," etc. · Naboth. "Naboth."
• NOT expanded, because Migne's lemma diverges there: In sanctum (0600A) and Praebente regi (0601A) — see §2.1 and §2.2.

⚑ Per the standing ruling, the split between these two lists is class, not inconsistency: the first list is clipped mid-phrase, the second is complete. Do not contrast them.

0.3 usque ad — 27 occurrences, and two of them are not the formula

Decided per occurrence, from what stands either side, as the brief requires.

• 0600A, inside the scripture quotation Fuit itaque arca ibi usque ad praesentem diem: — the verse's own words (2 Par. 5:9). Rendered "until this present day", not the formula.
• 0600A, in the gloss prose significante usque ad tempora aetatis suae eam ibi permansisse — ordinary prose. Rendered "down to the times of his own age".
• The other 25 are the abridgment formula and take as far as. One of them, at 0604A, has the tail INSIDE the italic span (usque ad et cujus principes vescuntur in tempore suo.) and is carried as one span, as far as and whose princes eat in due season. — the tail is Eccl. 10:17 verbatim.

§1 · THE DIVERGENCE LIST — every lemma and inline scripture span in 0600A–0610D

⛔ Checked against sources/vulgate/clementine-flat.txt, cited by book chapter:verse. Nothing below rests on recollection. The agreements are listed because a findings-only list cannot be told apart from a report by an agent that never looked.

1a. [var: …] fired — 7 sites, material divergence

1b. Divergences found and DELIBERATELY not marked — crux only

Each of these differs from the Clementine and none changes what the sentence asserts, so Pattern 14's "in a way that changes the sense" is not met. Recorded so a later reader knows they were seen and weighed.

• 0600A Fuit itaque arca ibi usque ad praesentem diem ‖ 2 Par. 5:9 usque in praesentem diem. Preposition only. (3 Reg. 8:8 also reads usque in.)
• 0601A Praebente regi ‖ 3 Reg. 9:11 praebente Salomoni. Same referent, and this is why the lemma was NOT expanded — see §2.2.
• 0602B Quaecunque petieritis Patrem in nomine meo, dabit vobis ‖ Jo. 16:23 si quid petieritis Patrem in nomine meo, dabit vobis, conflated with Jo. 15:16 ut quodcumque petieritis Patrem in nomine meo, det vobis. A loose conflation; the assertion is identical. Migne prints no citation here, so there is nothing to bind it to one verse.
• 0606D Expedit vobis ut unus moriatur pro populo ‖ Jo. 11:50 ut unus moriatur homo pro populo. One word dropped; sense untouched.
• 0608D Qui in Christo baptizati estis ‖ Gal. 3:27 Quicumque enim in Christo baptizati estis. A clipped opening.
• 0608D nisi quis renuntiat omnibus, non potest meus esse discipulus ‖ Luc. 14:33 Sic ergo omnis ex vobis, qui non renuntiat omnibus quae possidet, non potest meus esse discipulus. Recast from a relative to a conditional; the proposition is the same, and the negation count is unchanged (nisi … non).
• 0610C Vidi universum Israelem dispersum ‖ 3 Reg. 22:17 Vidi cunctum Israel dispersum in montibus. Synonym plus a declined form of the indeclinable.

1c. Checked and found to AGREE with the Clementine, exactly, at their own verse

In arca (8:9) · Nebula implevit domum Domini, et non poterant (8:10 + 8:11, a lemma spanning two verses, not a divergence) · Stetit / Et benedixit (8:55) · In die (8:64) · Festivitatem / Ab introitu (8:65) · Laetantes (8:66) · Subter alas (8:6) · Cumque / Qui et fuerunt (8:8) · Aedificavit (9:17) · Et Palmiram (9:18) · Universum (9:20) · De filiis (9:22) · Quinquaginta (9:23, the second word of the Clementine's quingenti quinquaginta) · Offerebat quoque (9:25) · Classem (9:26) · In Ophir (9:28) · Videns autem (10:4) · Verus est sermo (10:6) · Beati viri (10:8, clipped of tui) · Ligna (10:11) · Cantoribus (10:12) · Rex autem (10:13) · Ducenta (10:16) · Saltus Libani (10:17) · Thronum de ebore (10:18) · Et duodecim / Non est factum (10:20) · Sed et omnia vasa / Nec alicujus (10:21) · Deferens inde aurum (10:22) · Congregavitque Salomon (10:26) · Invenimus eam in campis sylvae (Ps. 131:6, sylvae for silvae is orthography only) · Multae filiae congregaverunt divitias, tu supergressa es universas (Prov. 31:29) · fons aquae salientis in vitam aeternam (Jo. 4:14) · Rex autem (11:1) · Voraginem (11:27) · Erat autem (11:28) · Tolle tibi (11:31) · Dies autem (11:42) · Venit autem (12:1) / Durissimum (12:4) · Adolescentes (12:8) · Scorpionibus (12:11) · Ad Roboam (12:23) · et cujus principes vescuntur in tempore suo (Eccl. 10:17) · Fecit duos (12:28) · Ecce vir (13:1) · Prophetes / Omnia opera (13:11) · Haec dicit Dominus (13:21) · Et leo stabat (13:24) · Sepelite (13:31) · Ascendit Sesac (14:25) · Factus est autem sermo (16:1) · Prophetas tuos occiderunt (19:10 and 19:14, clipped of gladio) · Ecce (16:3) · Qui mortuus fuerit (16:4) · Aedificavit Hiel / In Abiram (16:34) · Et dixit / In cujus conspectu (17:1) · Converte, Domine, captivitatem nostram (Ps. 125:4, verbatim) · Abscondere / In torrente Carith (17:3) · Mulier (17:10) · Quid mihi et tibi (17:18) · Tulitque Elias (17:23 — see §2.6 on the verse number) · Nunc in isto cognovi quoniam vir Dei es tu (17:24) · Abdias autem (18:3) · Curavit altare (18:30) · Fecitque (18:32) · Dominus ipse est Deus (18:39) · Apprehendite (18:40) · Multae pluviae (18:41, clipped of sonus) · Timuit ergo Elias et surgens abiit (19:3) · Et ecce angelus Domini tetigit (19:5) · Et ecce ad caput suum (19:6) · Et spiritus / Non in spiritu (19:11) · Ego solus (19:14) · Quicunque (19:17) · Profectus (19:19) · Secutus (19:21) · Haec faciant mihi (20:10) · Ne glorietur accinctus and the full Ne glorietur accinctus aeque ut discinctus (20:11 — the Clementine prints a comma after glorietur, and only that) · Fugit quoque (20:20) · Abiit ergo (20:38) · Erit anima (20:39) · Et furibundus (20:43) · Post verba (21:1) · Vidi Dominum (22:19) · Contra spiritualia nequitiae in coelestibus (Eph. 6:12, coelestibus for caelestibus is orthography only).

The Hebrew-name etymologies — vivens Deo, domus Dei, Deus meus Dominus, Deus meus, Deus fortis, captivans, convertens, cognitio, concisio, divisio — are Rabanus's onomastica, not scripture citations, and no Clementine comparand exists for them. Rendered, italics kept 1:1, no marker.

§2 · THIS RANGE'S CRUCES

2.1 · 0600A · In sanctum — the Clementine reads in Sanctum sanctorum. NOT expanded.

Migne's lemma at 8:6 prints In sanctum; the Clementine prints in Sanctum sanctorum. Rendered "Into the holy place." — the printed words, which stand as English on their own. Expanding to "Into the holy of holies" was considered and rejected: it would supply from the Vulgate at exactly the point where Migne's plate diverges from it, which is conformation, and conformation is the failure this range's brief exists to prevent. The gloss's own sancta sanctorum, two words later, is rendered "the holy of holies" as printed — so the reader sees both forms and can see that Migne set them differently.

2.2 · 0601A · Praebente regi — regi where the Clementine has Salomoni. NOT expanded.

Rendered "Furnishing to the king." For the same reason: the lemma diverges, so no words come in from 3 Reg. 9:11. The ablative absolute's subject (Hiram) is not printed and is not supplied. Not marked — "the king" and "Solomon" are the same man and the sentence asserts the same thing (§1b).

2.3 · 0603D · [sic: vides] — the one [sic:] in this range, and here is the argument

qui reddebant regi per vides singulis mensibus annonam. per governs the accusative and vides is a second-person finite verb. There is no reading of the printed form that occupies the slot: not "thou seest", not any noun. This is Pattern 12's first test — a form in the wrong person or mood cannot be carried by any English (the 11632 Pecte / ama / sternis ruling, 2026-08-09) — and it is the one case where the marker is the only honest treatment.

⚑ The rival, and why it lost. The obvious conjecture is per vices, "by turns", which fits 3 Reg. 4:7's per singulos enim menses in anno, singuli … ministrabant exactly. It was not entered in the text: a conjecture in the running text wears Migne's voice, and c → d is precisely the shape of a Corpus Corporum misread, which would make this our defect and not his. So the marker records the type and this crux records the conjecture, and neither pretends to be the other.

⚠ Plate read wanted. If the plate prints vices, the fix is data/tei-patches/8954.json and the [sic:] is withdrawn in the same commit. I have no plate access and have not asserted one. 7a⁗ check: with the marker's contents deleted the clause still has its verb (reddebant, "rendered") and its object (annonam, "the provision") — the quarantine took only the prepositional object.

2.4 · 0609B · [cj: morte; read more, "after the manner"] — the one [cj:]

Samaria morte civitatum habebat terram interius ipsis muris pene aequalem. morte is a perfectly good ablative and renders in place ("by the death of cities"), so [sic:] cannot reach it — and the English it yields asserts something the passage does not: the sentence is describing how Samaria's ground was banked up after the manner of cities (more) as a support against the ram. This is Pattern 18's founding shape exactly. The gloss is additive: Migne's word keeps its English and the conjecture stands beside it.

2.5 · 0605D · In Thersa regna — a real word that will not construe. RENDERED AS PRINTED.

Two things are wrong on this line and neither was patched.

1. The verse number. Migne prints VERS. 6.--, but 3 Reg. 16:6 says only sepultusque est in Thersa; the gloss (Sex annos fecit in Thersa) expounds 16:23, in Thersa regnavit sex annis, which is Amri, not Baasa. Migne's ordering and numbering are his and are reproduced as printed — never renumbered.
2. The word. regna is a real, well-formed Latin word (imperative of regno, or the plural of regnum), and it occupies no slot the sentence has. Under the 2026-08-23 addendum — a real word that reads oddly is Migne's until a plate says otherwise, refuted 5 of 5 — no marker was fired. Rendered "In Thersa reign.", which is the imperative the letters give. The expected form is regnavit, and if a plate read shows it, this is a tei-patch and not an emendation.

⛔ Note this is deliberately treated differently from §2.3: vides has no slot after a preposition; regna stands in a bare lemma with no governing structure at all, so "carried untranslated" would be a stronger claim than the evidence supports.

2.6 · Verse numbers that do not match the verse glossed — all reproduced as printed

• 0605D VERS. 6.-- on a gloss of 16:23 (§2.5).
• 0606D VERS. 19.-- on Tulitque Elias, which is 17:23 (Tulitque Elias puerum, et deposuit eum … et tradidit matri suae); 17:19 has Tulitque eum de sinu ejus. The gloss ("Elias took the boy and restored him to his mother") follows 17:23. Not renumbered; the lemma is expanded from 17:23, which is the verse it actually opens.
• 0610B VERS. 43.-- Et furibundus is 20:43 and is correct; but the gloss attached to it ("Achab commanded the prophet to be kept shut up … confounded by the words of Micheas") is the matter of 22:26–27, and Michaeas does not appear in III Reg. before ch. 22. Likewise 0610A VERS. 38.-- Abiit ergo is 20:38, where the prophet is unnamed, and the gloss names Micheas. This is the abridgment folding two chapters together. Reproduced; nothing moved.

2.7 · Real words that read oddly, rendered as printed, NO marker — the refuted-class discipline

Each was raised, weighed and left. None is a non-word; per addendum A the corpus argument loses.

• 0600D Rivocorulam — "another stream which flows into the sea beside Rivocorula". The place is Rhinocorura, the town at the Brook of Egypt, and every other spelling in the tradition is closer to that. Migne's form is well-formed as a Latin proper name and is transliterated as printed.
• 0600C Quid est quod Salomon aedificavit medium atrii — 3 Reg. 8:64 has sanctificavit rex medium atrii, and Migne's own next paragraph reads medium atrii sub divo sanctificavit. This is gloss prose, not a quoted lemma, so Pattern 14 does not reach it. Rendered "built".
• 0604B Gaddo for the seer's name (2 Par. 12:15 Addo). Rendered as printed.
• 0607B Abdias princeps exercitus Israel — 3 Reg. 18:3 makes him dispensatorem domus suae, the steward. Gloss prose; rendered as printed.
• 0608D Elias enim interpretatur Dominus Deus — the same work interprets Elias as Deus meus Dominus at 0606B. The work contradicts itself two columns apart and both stand; that is a fact about this recension, and harmonizing them inside the English would conceal it.
• 0609B ducenta millia et triginta duo — 3 Reg. 20:15 counts ducentorum triginta duorum, 232. Rendered "two hundred thousand and thirty-two", the printed words, not 232.
• 0610A Si prosperitate concessa Deo gratiae agentur — future passive where agantur is wanted. Real form, real slot; rendered "thanks shall be rendered to God".
• 0610C Si Dominus est, inquit eis, in Deum — will not construe (vobis in Deum, "God to you", is the obvious want). Rendered literally, "If the Lord is, saith he to them, for God". No conjecture entered.
• 0610D voluntas electorum spirituum divinae deseruit potestati — the dative potestati settles it: desero takes the accusative, so the verb is deservio, "is subject to". Rendered "has been devoted to the divine power", keeping the perfect against the parallel present obtemperat. This is a reading, not a repair, and it is decided by the case ending.

2.8 · Grammatical mismatches with no English exponent — Pattern 9 family, no marker

• 0601B palma, quae Graece vocantur Palmira — singular noun, plural verb. English carries it visibly ("the palm, which in Greek are called Palmira") and nothing is concealed.
• 0601A/0600D Qui non fuerat de filiis Israel, fecit Pacificus noster tributarios — singular relative, plural predicate. Carried: "He who had not been … made tributaries."
• 0603D quare vitulus fecerit — nominative singular where the object of fecerit is wanted, and against the lemma's own Fecit duos. Marker considered and declined. English nouns have no case exponent, so this sits with manum meum (8715) and the gender class rather than with 11632's number disagreement; and the ship test forbids an unread [sic:]. Rendered "Why Jeroboam made a calf".
• 0608C nec discipulis ejus Elisaeus unxit — dative/ablative plural where the nominative singular discipulus is wanted. Same treatment: "nor Eliseus his disciple, anointed".
• 0604B sicque regi persuadere possent, sicut fecit — plural then singular in one clause. Rendered "they might be able to persuade the king, as he did".

2.9 · Sentences that do not construe, rendered as closely as the printed words allow

• 0600A (opening, VERS. 6) Illata in sancta sanctorum arca assumptam Christi humanitatem intra velum regiae coelestis inductam. — no finite verb at all; the abridgment has removed the designat/significat that designant supplies in the sentence before. Rendered as the verbless apposition it is, and no verb was supplied: "The ark brought into the holy of holies, the assumed humanity of Christ led within the veil of the heavenly court."
• 0602A (opening of CAPUT X) Haec regina venturam Ecclesiam de gentibus, etc. — the same shape, an accusative with the verb abridged away. Rendered as apposition, verb not supplied. ⚑ This is the ferias trap the brief names: the tail of an abridgment is parsed as strictly as ordinary prose, and a missing verb is reported, not invented.
• 0600C quia illud angustum erat, et capere non poterat — the object (3 Reg. 8:64's holocaustum, et sacrificium) is not printed. Rendered "and was not able to contain", with the gap left open rather than filled with a supplied "them".
• 0605B Sed nulla virtus illic est, ubi (licet cum multis millibus) licet unus pugnat injuste. — licet twice, the second resuming after the parenthesis. Not marked as dittography: the two are not an eye-skip of one run, they are a resumption across an interrupting parenthesis, and Pattern 11 wraps only an intrusive copy. Both rendered.
• 0605B quae scilicet apud nos est, servantes a principio leges et verum Deum venerantes — nominative plural participles with no subject to agree with. Carried as loose participles.
• 0605B Nam quanta nullo bello refertur caedes fuisse commissa, neque Graecorum, neque Barbarorum, mirabilem perceperunt, Deo conferente, victoriam. — the quanta clause has no main-clause hook. Rendered in the order printed, with both negations on their printed hosts (nullo bello, neque … neque), and left rough.
• 0610A …alio statim vincere conatur — the sentence stops with no terminal mark. Pattern 8: no full stop supplied. The English stops too.

2.10 · VERS. 28--. at 0601D — the dash and period transposed

Every other verse address in this work prints VERS. n.--. This one prints VERS. 28--. The verse address passes through verbatim under convention 1, so the transposition rides as printed. Not a candidate for normalization: Pattern 9's section-label licence covers a translated structural head, and this address is not translated at all.

2.11 · 0602C · ferculum — why "carriage" and not "litter"

Thronus vel solium … quae in canticis ferculum appellatur, eo quod residentes ferat, vel de loco ad locum circumferat. The gloss is the etymology: ferculum from ferre. An English word sharing no root with "bear/carry" would make the gloss explain nothing (the murenulae / stylus precedent). "Litter" was rejected on exactly that ground; "carriage" keeps the derivation visible in English — carriage … carries … carries about — with no italic Latin added, so span parity is untouched.

2.12 · 0603D · divina Scriptura rendered "divine Scripture"

Pattern 3 lists divina Scriptura among the forms promoted to "Holy Scripture". I used "divine Scripture" instead, which is attested in the shipped corpus (11064) and renders the word Migne prints; sanctae Scripturae at 0602A is rendered "Holy Scripture", so the promotion is applied where the Latin says sancta. Flagged because it is a deliberate narrowing of a pattern, not an oversight.

2.13 · Unexpressed subjects supplied — named, with what fixes each

The dominant defect class. Every "he/she/it/they" I supplied that Migne does not print:

• 0600A nunc dicitur posuisse → Solomon, from Salomon addidit two words earlier.
• 0600C medium atrii sub divo sanctificavit … quod fecerat → Solomon, from the lemma In die (8:64, sanctificavit rex) and from 2 Par. 4:1's twenty-cubit altar.
• 0600D designat → the sacred writer, the subject Rabanus is expounding throughout this block; "he designates".
• 0600B quibus … Dominus de se etiam aperta narravit → the priests, from sacerdotes earlier in the same sentence.
• 0603B–C constituit ducibus → the king; egreditur, suadebat → Jeroboam. Both fixed by the named subjects standing in the same sentence.
• 0604B excogitavit → the false prophet; transgrederetur → the man of God; illius iram → God's. Fixed by Iste pseudopropheta and virum Dei in the clause.
• 0605D Sex annos fecit and nominavit eam → Amri, the subject of 16:23, which is the verse the gloss expounds (§2.5); rendered "he", not named, because Migne does not name him.
• 0606A restauravit → Hiel, from the lemma Aedificavit Hiel; destructa … anathematizata agree with Jericho, feminine, not with Hiel.
• 0605D ab eo Samaria appellatur → from him, i.e. Semer, named in the previous clause — not "from it" (the city), which would make the etymology circular.
• 0610A Si viderint coelestia desiderare → the accusative subject is unexpressed; supplied as them, i.e. the animas of the preceding sentence.

⚑ Where two readings both construe I have said so rather than choosing silently: at 0601A quas non potuerunt filii Israel delere the antecedent of quas may be civitates or reliquiis — both are feminine and both construe. Rendered "which the children of Israel could not destroy", which keeps the ambiguity that the Latin keeps. 3 Reg. 9:21 has masculine quos (the sons), so neither Latin reading is the Vulgate's.

§3 · READ CLEAN — sections examined and found sound

• Negation fidelity. Every printed negative in the five chunks was located, its host word named, and the negative checked to be on that host in the English: 17 in 0006, 15 in 0007, 22 in 0008, 16 in 0009, 7 in 0010. Seven sites fell in test 2a's trigger population (a negative by a verb with a quantifier or correlative nearby) and each was decided from the sentence's structure, not from what read well:
• 0601B qui non ancillae filii sunt, sed liberae → "sons not of the bondwoman but of the free", not "who are not sons of the bondwoman". The sed liberae pivot proves the host is ancillae.
• 0602B Non solum quod petimus, imo etiam … → "Not only what we ask, but even …"; the non is on solum, and largitur is affirmative.
• 0603A Non in persuasibilibus … verbis consistit, sed in ostensione → the negative is on the phrase, shown by sed in.
• 0605B quae vestrae sunt impietatis et non religionis indicia → "of your impiety and not of religion"; the host is religionis, against the impietatis contrast.
• 0604A Juvenes Scriptura non semper juxta aetatem appellat, sed juxta instabilitatem → "does not always … but according to".
• 0606C quam Dominus non frumenti, sed verbi pane pascit → "feeds not with the bread of wheat but of the word"; host frumenti.
• 0608A/0608C quibus non sufficit simplex pastus, sed duplex and dum non solum se remansisse cognosceret → both resolved onto the quantifier, not the verb.
• Punctuation, band by band. Colons, semicolons, question marks, exclamation marks and guillemets compared within each of the 44 column bands in this range, never by total. Zero mismatched bands at file time. Three supplied semicolons were caught this way and reverted to Migne's commas (0600A, 0600C, 0608C); two more were caught while drafting (0605C, 0609A). All nine question marks Migne prints are preserved (1 · 2 · 4 · 1 · 1 across the five chunks) and all nine are direct questions, so 8b did not fire anywhere in this range — there is no indirect Quaeritur here.
• Column anchors. 44 anchors (8 · 9 · 12 · 8 · 7), verbatim, in order, and each checked against the word it stands beside in the Latin, not merely for presence. Two are deliberate mid-construction splits matching Migne: 0601B (regio eo | quod nusquam) and 0602A (provocata | est). 0601D stands before aureos in both twins ("because it has [0601D] golden mountains").
• [sic:] swallowing test. Both in-text markers stripped and the sentences read aloud: no stranded article or preposition on punctuation, no doubled mark, verb and object intact in both.
• Marker parity. 83 [n:], same count and order, contents untranslated, Migne's own internal spacing and his lowercase (Id.) at 0602B kept as he sets them.
• The fourth test. Every paragraph re-read against its Latin asking only does it assert what he asserts — which is what produced §2.9 (three sentences that assert nothing, because the abridgment removed the verb) and §2.11.

§4 · Candidates rejected, with the reason

• Pattern 4 [f:] — not applicable. This is a lemma-and-gloss commentary, not a florilegium (runbook false positive 4). No tag was placed.
• Column-band arithmetic — not reported (false positive 1). No text breaks mid-word or mid-construction across any anchor in this range.
• Em-dashes — none used, none reported.
• [nt:] — every note in this range is a citation siglum or a locator; none is editorial prose. Zero conversions.
• [ed:] — no demonstrated source loss. The brief warns that our archive.org copy of PL 113 clips the left margin on some pages; no word in this range is truncated at a column head, so nothing was attributed to that cause and nothing was patched.
• A [var:] on the etymologies (§1c) — declined: onomastica are not scripture citations.
• A [sic:] on vitulus (0603D) and discipulis (0608C) — declined, §2.8.
• A [sic:] on regna (0605D), Rivocorulam (0600D), agentur (0610A), morte (0609B as a [sic:] rather than a [cj:]) — declined; all are real, well-formed words, and addendum A's 10-for-10 split says the corpus argument loses against a plate we do not have.

§Z · THE BLIND POLARITY READ (step 4a)

Run 2026-08-25 by a reader fenced from this file. All 11 chunks read against their Latin twins. Every printed negation in cols 0581A–0610D is present in the English and on the right word. Zero polarity defects of the classic kind — the same profile as every other book in this batch.

Repaired, four sites

1. 0587C — agency inverted. Domus quae tabulata portant became "The houses which the boardings bear." quae is nominative (an accusative referring to domus would be quas), so the houses bear the boardings. ⚑ The column's own continuation is the control and our English renders it correctly nine lines later: non muris templi infixa, sed juxta muros trabibus … superposita. Repaired.
2. 0592A — a referent inverted for the reader. Iste ingressum nostrum ad fidem, ille designat ad spem was "The one … the other", which in English binds to the nearest antecedent — the entrance of the oracle — so the reader took faith to the oracle and hope to the temple, exactly backwards. Iste is the nearer, the entry into the present Church, just named. The next sentence proves it: postes hujus introitus quadrangulati sunt propter quatuor Evangelii libros, quorum doctrina in fide veritatis erudimur. Repaired to "This … that".
3. ⛔ 0610D — a silent repair that reversed the printed word. voluntas electorum spirituum divinae deseruit potestati was rendered "has been devoted to the divine power" — which is the emendation deservit, printed with no marker at all. As Migne prints it the sentence says the elect spirits deserted the divine power, the opposite of the paragraph's claim that omnis exercitus assistit Deo. ⚑ The dative potestati is the argument: desero takes the accusative, deservit the dative. This is the 7a specimen exactly — a repair hidden inside the translation, passing every automated check. Now rendered "serves" with [cj: deseruit; read deservit], so the reading is shown rather than smuggled.
4. 0610C — a supplied pronoun with the wrong referent. "if he had had a good and pious shepherd, he would have overcome" — but the lemma is Vidi universum Israelem dispersum, and Achab is the shepherd, so as it stood Achab needed a shepherd and Achab won the battle. The one lacking a shepherd is the people. Repaired to "it".

⬜ NOT repaired — 0593C, and the reason is that a deliberate decision already stands there

The reader argues that usque ad … ulla in parte declinemus cannot be affirmative: ulla is negative-polarity, the governing ne fell inside what the abridgment elides, and our page therefore tells a reader that the apostolic linea exists so that we may stray.

The argument is good and it is not being dismissed. But three things stand against acting on it unilaterally: the translating stint considered this exact site and declined it deliberately, recording that the ne lies inside the elision; the plate confirms the printed text (Gallica f297, read this session for an unrelated citation, prints usque ad a recta via qua ad promissam patriam gradimur, ulla in parte declinemus with nothing lost by us); and 7a forbids supplying a negative Migne does not print, which is the rule that has protected this edition more often than any other.

⚑ So the question is genuinely open, and it is a question about a rule, not a fact: does 7a's prohibition reach a negative that the abridgment formula has swallowed, when the surviving word (ulla) is itself negative-polarity? That is Wilson's to rule. Recorded here and in the resume note; the text is unchanged meanwhile.

0599C · deos suos for Ps. 30:21's eos — the crux, settled 2026-08-26

Migne prints Quia deos suos abscondit in abdito vultus sui a conturbatione hominum. The English reads "Because he hides his gods in the hidden place of his countenance from the disturbance of men."

The psalm reads otherwise. Ps. 30:21 is Abscondes eos in abscondito faciei tuae a conturbatione hominum — "Thou shalt hide them in the secret of thy face from the disturbance of men." The glossator is paraphrasing throughout, not quoting: he turns the second person into the third (abscondes → abscondit), abscondito into abdito, faciei into vultus. Against that, deos for eos is one letter further along the same road, and it is impossible to say from the page whether the letter is the glossator's, a copyist's, or the printer's.

What was done, and why. ⭐ Rendered as printed, with no [cj:] and no [sic:]. deos is a perfectly good Latin word, and in this sentence it carries a live patristic sense: the saints, who are the treasures of the house of the Lord, are called gods by adoption after Ps. 81:6 Ego dixi: Dii estis — and the paragraph the sentence opens is precisely about the saints, many treasures differing in merit within the one house. Pattern 18 fires only when the English asserts something false; under that reading it does not, so the bar is not met and the printed word stands.

⚠ What a reader should know. The sentence is not about idols. Read it with Ps. 81:6 in hand and the "gods" are the saints of the next sentence; read it against Ps. 30:21 and the word is one letter from eos, "them", which is the same people under a plainer name. Both roads end at the saints, which is why the choice between them changes no doctrine and did not warrant a marker. Not read at the plate — Gallica f300 was read this session for the (II Cor. XV) citation in the same column and the crop was not extended to this line, so a second witness could still settle whose letter it is.

⚠ The 24 lemma expansions — RAISED, AND NOT ACTED ON. See §Z2.

§Z2 · THE 24 LEMMA EXPANSIONS — the finding, and why nothing was changed

The reader found 24 English lemmata carrying words the Latin twin does not print, every added word verbatim the Clementine continuation, and — this is the striking part — all 24 in chunks 0006–0010 and none in 0000–0005. That is a stint boundary, not scattered slips: the first stint keeps bare lemmata scrupulously (Aedificavit viginti. → "He built twenty."; Porro. → "Moreover."; Ecce. → "Behold."), and the practice changes at col. 0600A and does not change back.

⛔⛔ The reader classed all 24 as conformation. On the settled conventions they are not, and acting on that reading would have damaged the work. The same finding was made on the sister book II Paralipomenon in this session, acted on, and withdrawn:

• Convention 3 (the Ruth pilot) licenses expanding a FRAGMENT — a lemma whose governed word is missing — and renders it as the phrase it stands for rather than a word-for-word stump. Most of these 24 are exactly that: Subter alas. (wings of what), Cumque., Ducenta. (two hundred what), Sepelite., Fecitque.
• The standing ruling on a work that treats the same bare lemma two ways is to EXPAND THE STUMP, NOT CONTRACT ITS TWIN. So the half that should move is the first stint's, not the second's.
• ⛔ The defect is never expansion or its absence — it is CONFORMATION, following the familiar English Bible where Migne's plate diverges from it. Expanding a fragment per convention 3 is not conformation.

⚑ What the sister book cost, written down because it is the reason this one was left alone. There, two numeral lemmata were contracted on this reading. Both were wrong and both were reverted: Centum. and Ducenti. are clips of two-word numerals (2 Par 1:17 equum centum quinquaginta, 8:10 ducenti quinquaginta), so both are fragments. Worse, the contraction was justified on the ground that Migne's gloss supplies the figure itself — Qui sunt ter quinquageni — and that reading was backwards: ter quinquageni is thrice fifty, which is 150. The gloss was not withholding the number, it was confirming it, and contracting the English left the lemma saying one hundred above a gloss saying one hundred and fifty.

⭐ The general lesson, and it applies to every blind read this edition runs: a reader's finding can be right about the FACT and wrong about the RULE. These 24 English lemmata really do print words the Latin twin does not — that observation is correct, valuable, and was invisible to every other check. What a fenced reader cannot supply is whether the convention licenses them.

⬜ What is genuinely owed here is a ruling on scope, and it is Wilson's: whether to expand the first stint's stumps so the work is internally consistent, and how far convention 3 reaches when the supplied word is a proper name rather than a generic slot-filler (the sister book I Paralipomenon has Her pater. → "Her the father of Lecha", which is new lexical content). Nothing has been changed in either direction. The count is the useful thing to carry forward: 24 sites here, 3 in 8964 against 4 identical shapes left bare, 5 in 8965.

Wilson ruled: expand the bare stumps; and convention 3 STOPS AT GENERIC FILLERS — a proper name is new lexical content, "a name a reader can get only from the Vulgate, not from anything on Migne's page." Both halves were then applied to 8954, 8964 and 8965, and the second half turned out to carry the first.

⭐⭐ THE FINDING THAT NARROWED IT: MOST "BARE STUMPS" ARE NOT STUMPS

Nineteen lemmata across the three books end in a cardinal numeral — five already expanded, fourteen bare. The bare ones were read against their own glosses before anything was touched, and the glosses settle it: they are NUMEROLOGICAL. They expound the bare number as a number.

⛔ Expanding these would have been the 8965 error in mirror image. There, Centum. was contracted and left a lemma saying one hundred above a gloss saying one hundred and fifty. Here, expanding Quadraginta, to "Forty thousand stalls" would have put a counted noun above a gloss that never uses one — a gloss whose entire subject is the number forty as a figure of the ancient law. The counted noun is not what Migne is glossing. They stay bare, and that is not inconsistency with their expanded twins: the twins (Ducenta → "Two hundred shields") sit over glosses about the things counted.

✅ WHAT WAS EXPANDED — two sites, and the test is that the English was FALSE as printed

⭐ A clipped SPLIT numeral does not merely read tersely; it states the wrong figure. That is 18a's own bar, and it is the whole of what was expanded:

• 3Rg 7:15 · Decem, → was "Ten," for a column eighteen cubits high. ⭐ Migne's own gloss supplies the completion — Ter enim seni decem et octo faciunt — so nothing is imported. Now "Ten and eight,".
• 1Par 18:12 · Decem. → was "Ten." for eighteen thousand. Again the gloss carries the figure, arguing against the codices that read viginti et octo millia. Now "Ten and eight thousand."

Both are the Centum. → "A hundred and fifty." precedent exactly (2Par 1:17, already in the book), which is why they are consistent rather than novel. ⬜ 3Rg 9:23 Quinquaginta. is the same class and was NOT expanded: it is the second half of quingenti quinquaginta, so the missing digits precede it and cannot be recovered forward — and its gloss expounds fifty, not five hundred and fifty. Named so the asymmetry is not read as an oversight.

✅ PROPER NAMES — five contracted, three left, by Wilson's own criterion

The criterion is whether the name is on Migne's page, not whether it is a name:

• Contracted — 1Par 4:21 Her pater. ("of Lecha") · 3Rg 9:20 De filiis. ("of Israel") · 3Rg 10:4 Videns autem. ("the queen of Saba") · 3Rg 10:13 Rex autem. ("king Solomon") · 3Rg 20:43 Et furibundus. ("he came into Samaria") · 2Par 25:18 Carduus qui, ("is in Libanus").
• Left standing, because the name is in Migne's own gloss on the same line — 3Rg 8:65 Ab introitu. → "From the entrance of Emath" (per introitum Emath designat) · 3Rg 11:1 Rex autem. → "And king Solomon" (the gloss reproves Salomon by name) · 2Par 27:1, where Joatham stands in the printed lemma itself and was never an expansion at all.

⚑ The general lesson, and it is this session's own, turned on itself twice. 8965 taught that a reader can be right about the fact and wrong about the rule. 8966 @1113C taught that a stint can apply the right rule to a wrong fact. This one is the third turn: a correct ruling can be applied to the wrong population. The ruling said expand the stumps; the page said most of these were never stumps. Read the gloss before expanding the lemma it glosses.

Wilson ruled: leave the text, add the [cj:]. The question was whether 7a reaches a negative the abridgment swallowed — usque ad … ulla in parte declinemus, where ulla is negative-polarity and the governing ne fell inside what the formula elides.

⭐ The ruling threads it exactly, and the reasoning is worth keeping: 7a forbids supplying a negative IN THE TEXT, and the text is untouched. The English still reads "that in any part we may turn aside"; the marker tells the reader where the negative went. Nothing is smuggled in, and the page no longer says only the affirmative:

that in any part [cj: ulla; read ne ulla, "that we may not in any part turn aside"] we may turn aside

Precedent: 7561 @0648, where Migne's setting dropped the non of non solum … sed etiam and was glossed rather than supplied — a site that sits on 18a's own approved list. ⚑ A dropped word is glossable; that is the whole point of the form, and it is why the printed word (ulla) leads the marker and verify-english checks it against the Latin twin.

⬜ What is NOT claimed: that Migne erred. He did not — the plate confirms the printed text (Gallica f297), and the usque ad formula removing words is an abridgment, not a defect. The marker names the elision, not a fault.