Expositio in XX primos Psalmos

PL 114, cols 0751C–0794D · 18 chunks · 17,261 Latin words. Attributed by Pez and by Migne to Walafrid Strabo; in fact Anselmus Laudunensis et schola, and through Ps XX largely Augustine's Enarrationes abridged. Translated 2026-08-28 by four Opus stints, none of which could see the others.

This file is the four-way MERGE. The range files (cruces-0000.md, cruces-0005.md, cruces-0010.md, cruces-0014.md) are preserved below, unedited, each under its own heading. This head is the work-wide layer: what was reconciled ACROSS the four ranges, and what a later sweep must not churn.

⛔ Read §0b before anything else in this file. 9003 is the first work in the corpus collated against Migne's plate page by page — all 22 pages — and that changed almost every part of the apparatus. A reader who does not know it will misread the whole thing: markers that a range file argues at length are gone; markers the range files do not mention are here; and several sentences of English mean the opposite of what the same range file quotes.

§0a · Apparatus, counted from the English and not summed from four reports

48 [var:] · 12 [cj:] · 4 [sic:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [n:] — counted from src/english/9003/0*.md after the plate collation, the blind polarity read, and this merge. Do not sum the four range files. Every one of their headline counts is stale: cruces-0000 says 1 [cj:] where its own register then fires a second; cruces-0005 says ten [var:] before one was withdrawn and two more were recovered from the plate; cruces-0014 says five [sic:], three of which are withdrawn.

The work carries noteCount: 0 in every chunk and none was invented.

verify-english.mjs 9003 → OK, 18 chunks, columns/notes/sections aligned, no duplicate paragraphs. scan-raw-markers.mjs → no marker rendered as literal bracket text.

⚑ One marker is in the English and in no range file: 0767B [cj: cedere; read credere, "to believe"] (chunk 0007), on omnes, qui vetant cedere incognita. It is right — Augustine writes vetant credere incognita, and the sentence is about men who forbid believing what is not yet known — and it rests on collated plate (p. 389 was read; there is no patch at 0767). It is recorded here because cruces-0005's register does not list it, and a later reader auditing that register against the text would otherwise find an orphan.

§0b · ⛔ THE PLATE COLLATION — all 22 pages, and it is the spine of this work

2026-08-28. Every page of the work was read at Migne's plate (archive.org patrologiaecurs06saingoog, PDF page = (column + 11) / 2, corner numbers verified before each read; 300 dpi, one site at 1200). Five agents and the session lead read pp. 381–402 between them. 63 TEI patches were applied (data/tei-patches/9003.json) — ⚠ the four range files each say "61 patches"; they were written before the last two landed. The patches run in two directions on purpose:

• 45 normalized — Corpus Corporum had silently tidied Migne's own defective type, or moved his lemma toward the Clementine. Reverted to what Migne printed, so the defect is visible and can take an honest marker or an honest entry in the register.
• 18 corrupted — Corpus Corporum broke a word Migne set correctly. Repaired.

Rate: ~2 sites per page, against a standing corpus estimate of THREE INSTANCES IN TOTAL for this class. The rate, not the class, is the finding.

What the plate did to the apparatus

Five public claims against Migne were WITHDRAWN, all of them artefacts of the digitization.

⛔ 0770A is the one to carry away, and it is a class no procedure we had could catch. CC had written judicavit for Migne's judicabit. The lemma collation, run against that twin, duly "found" a divergence from the Clementine that does not exist on Migne's page, and the English went into the perfect tense to match. No collation against a Vulgate file can catch this, because the twin is already off the plate before the check begins: the check compares our transcription with the Vulgate and both of its inputs are innocent of the error. Only the plate can. The marker is gone and the English reads "shall judge the world in equity".

⚑ Two more were avoided only because a translator declined to fire and asked for the plate instead — 0763D potet (a real form with no grammatical slot) and 0769D misus (one dropped letter inside a scripture lemma), both logged in cruces-0005 §4 as plate candidates rather than marked. Both were CC's. Migne prints potest and missus. Had those two fired, this work would have shipped seven false accusations against Migne.

⭐ One marker was declined AGAINST the collation's own recommendation, and the decline was right. At 0788A the restored plate reads sive signa, quae impressi ad imitandum sequentibus, and the collation flagged it as the one place in the work where Migne does not construe, asking for a [sic:] or a [cj: impressa]. It construes exactly as printed: impressi is the first-person perfect active of imprimo and quae is its object, in a sentence that is first-person throughout (itinera mea, gressus meos, charitatem meam). The independent witness is on disk — the same clause in PL 21 at sources/pl/tei/7109.xml reads signa, quae impressi ad imitandum te. It would have been the sixth false claim against Migne. If a later reader wants the marker, the thing to overturn is the 7109 parallel, not this note.

⭐ Eight real Vulgate divergences RECOVERED, which CC had erased by normalizing toward the Vulgate

This is the collation's positive yield, and it is the mirror image of the withdrawals: where the digitization had quietly conformed Migne's lemma to the Clementine, the divergence Migne actually prints came back.

Four now carry a [var:]: 0758C Domine est salus (Ps 3:9, for Domini) · 0764B si fecit istud (Ps 7:4, for si feci — a change of person inside a first-person psalm) · 0771C Comprehendantur (Ps 9:23, subjunctive for indicative — a wish, not a fact) · 0755C …non erunt, ut in scriptura (Ez 13:9, for et).

⚠ 0755C is a contradiction inside cruces-0000 and this is its resolution: the marker STANDS. The plate-collation section of that file DECLINED it, on the ground that ut + future indicative construes as nothing and is therefore Pattern 12's business; the blind polarity read, working later and independently, FIRED it, on the ground that the English still read "and" and was therefore silently carrying CC's conformation. The blind reader is right about the English: a [var:] is the only marker that shows the reader both that Migne departs from the Vulgate here and that his departure does not construe. Both sections are preserved below; where they conflict, this paragraph governs.

Four were weighed and DECLINED, on a rule worth keeping: fire where the printed reading construes in place and yields a different sense; decline where it yields no sense, because that is broken type and not a divergent citation. — 0760C oribus for auribus (Ps 5:2) · 0764A rapiat at leo for ut leo (Ps 7:3) · 0778D dimitimus for dimittimus (Jo 11:48) · 0789D ut gigas ac currendam viam for ad currendam (Ps 18:6).

And one interpolation was removed rather than translated. At 0760D CC had supplied corde from Mt 5:8; Migne prints qui mundo conspicitur alone, and our English had rendered the supplied word ("with a clean heart"). It now reads "who is beheld by the pure". Never translate what the plate does not print.

Wilson's marker ruling of this session, and the 31 sorts it tables

Pattern 12 read literally would put a [sic:] on every typographic defect the collation restored and bury the two or three sites that matter under a crowd of notes about broken type. The ruling: mark where the plate's reading does not construe, or where a reader could not recover the intended word with confidence, or where the defect could be mistaken for a real variant; do NOT mark a plainly battered, turned, dropped or wrong sort whose intended word is not in doubt — record it in the register and disclose the count in the work-level page note.

31 sorts are tabled that way, in four PLATE COLLATION · REGISTER OF RESTORED TYPE tables in the range files below — 6 in 0751C–0762A, 10 in 0762B–0774A, 4 in 0774B–0784A, 11 in 0784B–0794D. Each row gives Migne's printed form, the intended word, and why the reading is not in doubt. The four tables are the single register for the work; nothing is duplicated between them and nothing in the English points at them.

Six sites cleared the bar and DID take a marker, and they share one shape — the defective sort leaves behind a real Latin word in a real position, so nothing on the page tells the reader to stop, and our English had been rendering CC's repair: 0753B [cj: quod si] (for quod sit, an indirect question) · 0761A [cj: operantur] (for operantem) · 0766A [cj: Providentia] (for the dative Providentiae) · 0771A [cj: Filium homines] (for Filium hominis) · 0774B [cj: apud ipso] (for apud ipsos) · 0794A [cj: gloria] (for the accusative gloriam). ⭐ 0794A states the general reason best: English has no case endings, so "glory" renders the printed gloria and the conjectured gloriam identically — the reader cannot see the defect in the English at all, and the marker is the only thing that tells him it is there.

⭐ Two generalizable lessons the plate produced, both worth carrying to the next work:

• A Pattern 10 run-together should always be checked against the line ending (0792B/C): a transcription that loses line breaks manufactures exactly that defect out of nothing.
• VERS printed without its period is not a class in this work — it was CC's, three times. See §0h.

§0c · VOCABULARY RECONCILED ACROSS THE FOUR RANGES — the merge's real work

The four ranges were diffed for this work's load-bearing vocabulary before the merge, by grepping the English and the Latin twins directly and not by reading the range reports. A lemma-and-gloss psalter commentary recurs on a small set of words across all four bands, and two of them had genuinely drifted.

⛔ 1. impius — the drift, and it is the work's key word. RECONCILED.

impius occurs ~47 times across thirteen chunks and it was being rendered three ways: "ungodly" in 0000–0004 and part of 0014–0017, "impious" in 0005–0009, "wicked" in 0010–0013. Worse, the three collided inside single chunks. The demonstrative case is 0769C–0769D, where Migne glosses his own lemma: the lemma periit impius was Englished "the wicked one perished", and his gloss four words later — Non enim appellantur impii, qui Domino vero credunt — was Englished "they are not called impious". The gloss no longer explained the lemma it was written for, which is exactly the 11551 stylus/epistyle failure.

Settled on "ungodly", and 35 English sites were changed. The reading is fixed by the work's own opening: at Ps 1:1 Migne argues the distinction between impii and peccatores (Si aliqua hic differentia accipienda est inter impios et peccatores), and chunk 0001 built that argument in "ungodly" against "sinners". Every later band now says the same word. Specifically:

• impius (noun and adjective) → "ungodly" — changed at 0005 ×3, 0006 ×4, 0008 ×7, 0009 ×3, 0010 ×4, 0011 ×2, 0013 ×3, 0014 ×1, 0016 ×1.
• impietas → "ungodliness" — changed at 0006, 0008 ×2, 0011 ×2, 0014, 0017. It had been uniformly "impiety", which is fine on its own but severs the tie Migne draws at 0765D (persecutores impios … ab illa impietate) and mismatches the Douay at Rom 11:26, avertat impietatem a Jacob, "shall turn away ungodliness from Jacob".
• impie (adverb, 0786C) keeps "wickedly" — English has no usable adverb from "ungodly", and nec impie gessi is a lemma. Recorded so the exception is not read as a survivor of the drift.
• "wicked" is now reserved for malus (0771A, cum mali florent), sceleratus (0757C, 0778B), nefandus (0772C) and iniquus (0766A). Those were checked one by one and left.

⚠ The preserved range files quote the pre-merge English at several of these sites — e.g. cruces-0005 quotes "Let them be caught, however, the impious" for 0771C. The text is right; the quotation is a snapshot.

⛔ 2. psallere vs canere — the distinction Migne argues. RECONCILED.

psallere occurs three times in the work and canere/cantare three, and at 0777B the whole gloss turns on the difference: Si autem aliqua hic differentia consideranda est, cantabo corde, psallam operibus. The 0010–0013 stint caught its own half of this and fixed it (psallam → "I will play", against cantabo → "I will sing"). But the verb's only other occurrence is in another band — 0770B, Psallite Domino — where it was "Sing ye to the Lord", the very English the gloss reserves for cantabo. Changed to "Play ye to the Lord." Canam at 0769B ("I will sing", Ps 9:3, the Hebrew edition's reading) is canere and stays.

3. speculatio — a split, and BOTH READINGS STAND. Do not level it.

Three occurrences, in three different bands, and all three are about the Church:

• 0756C (band 1) Sion speculationem quidam interpretantur → "Some interpret Sion as watching"
• 0770B (band 2) habitat in Sion, quae interpretatur speculatio → "which is interpreted watching"
• 0792B (band 4) quae te tanquam de speculatione exspectat → "which awaits thee as from a watchtower"

The first two are Migne naming what the Hebrew name means, and the gloss on a name has to be the abstract noun. The third is the place one watches from, and "as from her watching" is not English. A sweep that levels these will either turn the etymology into a claim about the Hebrew that Migne does not make in that form, or make 0792B unreadable. Recorded, not changed.

Checked across the bands and found ALREADY consistent — the negatives, which count

• occultum — substantival occulta / occultis / occultum → "hidden (things / thing / man)" in every band; adverbial occulte and in occulto / in occultis → "secretly / in secret" in every band. The adverb/substantive split is uniform across bands 2 and 4 and needs no work. ⚠ 0791A is settled and NOT to be re-opened: ipsum occultum is deliberately "hidden man" at VERS. 13 (two dative relatives requiring a person, picking up the man of VERS. 11) and "hidden thing" at VERS. 14 (one of the two origins of sin). The inconsistency is Migne's, and levelling it would make one of them false. The blind reader raised it; it was adjudicated and declined; see cruces-0014.
• tabernaculum → "tabernacle" in both bands that carry it (0779C ×4, 0785A, 0791A); the tabernaculum / mons and habitatio / habitaculum distinctions of Ps XIV are preserved on both sides of the band boundary.
• differentia → "difference" in both bands, and in both it introduces the same formula (Si aliqua hic differentia accipienda / consideranda est).
• in conspectu → "in … sight" everywhere; "presence" renders only praesentia. No drift.
• gentes — the four bands independently converged on a rule, which is worth recording as a rule rather than flattening: "nations" inside a lemma or a scripture quotation, "the Gentiles" in the gloss wherever the Jew/Gentile antithesis is in play. 0789A (VERS. 50) shows both in one breath (confitebor … in gentibus → "among the nations", glossed "the Gentiles will confess to thee for me"), which is Migne's own move.
• praesumptio does not occur in this work. Checked mechanically over all 18 Latin chunks, because the brief named it.
• 0774B mentem is deliberately UNMARKED (cruces-0010 §1.1) and is not to be re-opened: Migne prints Unum enim mentem teneo … in quo, a real word with the wrong gender, which Pattern 12 cannot wrap and which a [cj: montem] would contradict, since the English already says "mind".

§0d · STANDING CHECKS — the negatives, recorded so they count

• deservit / deserit: CHECKED ZERO across all 18 chunks, mechanically over the whole Latin work. Neither spelling, nor any deservi- / deseru- form, occurs in 0751C–0794D. The pair has fired in both directions elsewhere in the series, so the zero is worth having.
• Migne's own ( sic ) does not occur in this work. All four [sic:] are ours, all four on non-words, all four now standing on collated plate: 0751D partiorem (Pez's monitum), 0775B proptera, 0792B salavifica, 0792C magnifibabit.
• Every [cj:] and [sic:] in this work now rests on a page that was read. After the collation there is no marker in 9003 whose ground is the Corpus Corporum transcription alone.
• noteCount: 0 in all 18 chunks; Migne prints no footnotes in this work and none was invented.
• VERS.12.-- and VERS.13.-- at 0756C–D (chunk 0002) are set without the space in the Latin twin and in the English alike, on a page that was collated (p. 383). Migne's spacing, carried; not the same phenomenon as §0h's missing periods.

§0e · WHAT A BLIND READER SHOULD GO TO FIRST

Merged from the four range files, because each named its own. This section is NOT to be given to that reader — the runbook forbids showing the cruces before findings are formed; it is for whoever adjudicates afterwards.

1. 0788A impressi — the one deliberate dissent in the work. A first-person perfect that the collation itself read as a defect. If a reader raises it, the answer is sources/pl/tei/7109.xml.
2. 0774B mentem … in quo — a real word, wrong gender, no marker, by decision.
3. 0791A ipsum occultum — "hidden man" at VERS. 13 and "hidden thing" at VERS. 14. Settled; see §0c.
4. 0755C ut + future indicative — carries a [var:], does not construe, and is meant not to.
5. 0779D [cj: in corde; read sed non in corde] — a lost negation on the plate, in a simile that exists to distinguish lips from heart. Our English had also smoothed the connective around it; both are repaired.
6. 0764B reddidit mala pro nobis — almost certainly Migne's slip for pro bonis. Rendered as printed, deliberately unmarked, because the intended word is not in doubt.
7. 0794A [cj: gloria] — the defect is invisible in English; the marker is the only signal.
8. 0789A, VERS. 50 — the lemma reads confitebor and Migne's own gloss reads confitebuntur gentes. Carried as printed on both sides.
9. The 31 tabled sorts — a reader collating the English against the twin will meet restored defects (fremueruut, populii, poehas, snperbia, jnstitia) that carry no marker at all. That is the ruling, not an oversight.

§0f · PLATE-READ CANDIDATES STILL OPEN

None. That is the finding, and it is the first time in the corpus. All 22 pages are collated; data/tei-patches/9003.json carries an empty unread list and an empty refuted list. Every plate candidate any of the four stints logged — 0763D potet, 0769D misus, 0762B justorum, 0774B mentem, 0788A impressi, the VERS periods, the cortuum run-together — has been read and decided, and the decisions are in §0b and in the range files.

What a challenger would have to do, rather than what is left to buy: re-read a specific column at higher dpi. Only one site in the work was read above 300 dpi (0787A populum, at 1200), and only one carries a reading that a re-read could in principle overturn — 0788A impressi, where the question is not the sort but the parse, and where the PL 21 parallel would have to fall first.

§0g · THE BLIND POLARITY READ — and its headline is a negative

Four fenced readers took the work in four bands (0000–0004, 0005–0009, 0010–0013, 0014–0017), each reading Latin against English in full, with no cruces*.md file and no collation report open before or after, and with a mechanical negation-token count per chunk as backstop. Adjudicated 2026-08-28, after the plate collation, under Wilson's marker ruling.

Whole-work result across all 18 chunks: 13 confirmed sites. Ten carried a defect in our English and were re-rendered; four sat on Migne's plate and took a marker with the English left literal. The two figures overlap by one — 0779D is a lost sed non on the plate and a connective our English had smoothed, so it was both re-rendered and marked. Four new [cj:], one new [var:]. Eight further candidates were adjudicated and declined, with reasons, under their own ranges. Per range, ours / Migne's: 0000–0004: 5 / 0 · 0005–0009: 1 / 0 · 0010–0013: 3 / 3 · 0014–0017: 1 / 1.

⭐ The headline is what the readers did NOT find, and in this genre that is the result. 7a″ — a lemma quietly conformed to the familiar English Bible — is the predicted dominant failure mode of a lemma-and-gloss scripture commentary, and it is the reason a psalter commentary gets a blind read at all. Two of the four readers reported the profile ABSENT from their ranges, each having checked it positively rather than merely failing to notice it:

• 0005–0009 read every quoted verse in five chunks word by word against the plate and found no lemma conformed: wherever Migne diverges, the English diverges with him and the divergence is marked — Miserere mei, Deus, si fecit istud, terram manducabis, Consummetur nequitia, Templum Domini, Comprehendantur, humiliavit, ut tradat eos in manus suas.
• 0010–0013 checked all fifteen [var:] in its four chunks against the printed Latin and found each recording a real divergence and none concealing a conformation.

The other two ranges returned the same shape from the other side: 0000–0004 found the work's sharpest 7a″ trap not fallen into (0760C/0761A, where Migne prints quae veritatem consequitur in the lemma and quae haereditatem consequitur in the gloss, and the English carries both as printed); 0014–0017 returned zero unaided defects on our side and confirmed three counter-intuitive negations carried where a smoothing repair was most tempting, including non foras mittit consummata dilectio timorem hunc against the familiar 1 Jo 4:18.

⚠ The one conformation that did exist was not the readers' to find. It was CC's, not ours, and it lived in the Latin: 0755C, where the digitization had moved Migne's ut to the Vulgate's et and our English had followed the twin. The plate found it; the blind reader then found that the English had never been brought across. Both halves were needed.

§0h · CONVENTIONS SETTLED AT THE MERGE

1. impius → "ungodly"; impietas → "ungodliness"; impie stays "wickedly"; "wicked" is reserved for malus, sceleratus, nefandus, iniquus. 35 English sites changed. §0c.1.
2. psallere → "play"; canere / cantare → "sing". One English site changed (0770B). §0c.2.
3. speculatio is not levelled, and ipsum occultum at 0791A is not levelled. §0c.3 and §0c. Both are closed.
4. [var:] citation form is the compact one — Vulg. <Book> <ch>:<v>, with the corpus's abbreviations (Ps, Mt, Mc, Lc, Jo, Act, Rom, 1Cor, Gn, Ex, Ezech, Apc, Sap). The 0014–0017 stint had used Migne's own roman-numeral style (Vulg. Joan. XI, 48, Vulg. Ps. XIX, 8) and three of its markers carried no reference at all. Nine markers were normalized, and the three bare ones were given the references they were missing: Ps 17:21 (secundum puritatem manuum mearum), Ps 19:7 (Exaudiet illum), Ps 20:8 (non commovebitur).
5. ⛔ VERS without its period is NOT a class in this work — the claim is WITHDRAWN. cruces-0005 §0a records it at 0772B as Migne's and rides it as printed; cruces-0010 §4.4 says it happens "TWICE in this work, so it is a class". The plate says otherwise at all three sites: Migne prints VERS. 28.-- (0772B), VERS. 2.-- (0774B) and VERS. 7.-- (0781A), and Corpus Corporum dropped the period each time. The Latin twins were patched at the collation but two English chunks were left behind — 0010.md still read VERS. 2-- and 0012.md still read VERS 7.--. Both were corrected at this merge. Where those two range-file sections conflict with this one, this one governs.
6. The patch count is 63 (45 normalized, 18 corrupted), not the 61 the four range files report. They were written before the last two landed.
7. The range files below are preserved unedited and therefore quote pre-merge English at the sites changed under 1, 2, 4 and 5. Where a range file quotes an English sentence that the text no longer contains, the text is right.

Preserved unedited from cruces-0000.md; headings demoted one level. Where it conflicts with the head above, the head governs.

9003 · Expositio in XX primos Psalmos — cruces, chunks 0000–0004 (cols 0751C–0762A)

Translator's range: chunks 0000–0004, PL 114 cols 0751C–0762A, 4,154 Latin words. Apparatus fired: 1 [sic:] · 12 [var:] · 1 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [n:] — ⚠ revised twice on 2026-08-28: the blind polarity read added one [var:] @0755C (the Ezekiel ut) — see THE BLIND POLARITY READ at the foot of this file — and before that, was 10 [var:]: the plate collation added one @0758C (Ps 3:9 Domine for Domini), which the digitization had erased. See PLATE COLLATION — markers withdrawn at the foot of this file. (the work carries noteCount: 0 throughout, as the launch brief states, and none was invented). verify-english.mjs clean for these five chunks; column anchors, ?, «», : and ; all at band-by-band parity with the Latin twins (see §5); italic spans 1:1 in every chunk.

§0 · ⛔ THE HEADLINE FINDING: OUR LATIN TWIN DIVERGES FROM MIGNE'S PLATE AT ~ONE SITE PER COLUMN

This is the most important thing in this file and it is not a translation question.

I read the plate for my whole column range (archive.org patrologiaecurs06saingoog, PDF page = (column + 11) / 2 — re-verified against the printed corner numbers on p. 384, which reads 757 in the left corner, and on p. 386, which reads 761; renders at 300 dpi). I read pp. 381, 382, 383, 384, 385, 386. Thirteen places where our chunked Latin does not say what Migne printed. They run in both directions — some are corruptions introduced by the digitization, more are silent normalizations of Migne's own defective type — which is brief §4's class, at a density nobody has yet reported for a Glossa book.

⚑ What follows from it, and it binds any later reader of this work:

1. No [sic:] or [cj:] may be fired anywhere in 9003 on the strength of the Latin twin alone. The twin has already been cleaned of most of the plate's defects; a marker keyed to it would either miss the real defect or, worse, blame Migne for a reading he does not carry. My single [sic:] (§1) is safe only because I read its plate line.
2. The fix is a data/tei-patches/9003.json + re-chunk, on the 8968 significaRuth precedent. That is outside a translating agent's remit and it must not be done mid-batch (re-chunking rewrites the Latin from TEI and can move chunk boundaries). I am escalating it, not doing it.
3. Until it is done, my English follows the TWIN, not the plate, everywhere except the one site where the twin is nonsense (0757A, below). I made that call deliberately and I record the reasoning so it can be overruled: our English sits in a parallel column beside the twin, every marker's safety check is a substring test against the twin, and an English column that contradicts its facing Latin misleads the reader worse than one that matches it. Pattern 7's "render the plate" is satisfied instead by this table, which names every place the two part.
4. A second witness is available and was not needed here — the on-disk PDF is legible at 300 dpi at every one of the thirteen sites. Nothing in my range is recorded as illegible.
5. ⚠ The rate should not be assumed to stop at col 0762A. I read six pages; thirteen sites is ~2.2 per page. Chunks 0005–0017 have not been collated and must not be assumed clean.

The one site where the English departs from the twin: 0756D→0757A, vergeret geret

The twin reads …ne idipsum vergeret [0757A] geret in effusionem temeritatis…. The plate settles it: col. 756 ends …ne idipsum ver- and col. 757 opens geret in effusionem temeritatis (p. 384, left column, first line — read at 300 dpi and again magnified). So Migne prints ONE word, vergeret, split across the column break, and our source has transcribed the fragment twice and joined it once.

• It takes no [d:]: that marker's tooltip says "Repeated in Migne's plate," and Migne repeats nothing. Firing it would be one of the seven false-[sic:] class the brief's Addendum B records.
• It takes no [sic:]: geret is not in Migne's text at all.
• English: "lest that same thing should incline [0757A] into an outpouring of rashness" — one verb, and the anchor placed exactly where Migne's column break falls, i.e. before in effusionem. This is Pattern 13a clause 1 (source damage, plate certain → restore silently, log the crux), applied to an added word rather than a lost one.

§1 · THE ONE [sic:] — 0751D partiorem, and it is Migne's own

Migne prints, in Pez's preface: qui me aliunde in propinandis theologicis non nihil partiorem esse cupiunt. Plate-verified (p. 381, left column): the plate reads partiorem, and our twin agrees — this is one of the few places in my range where twin and plate do not part.

partiorem is no Latin form: there is no positive partis to give a comparative partior, and the deponent verb partior has no such accusative. It occupies a predicative comparative slot (me … partiorem esse) that only an adjective can fill, so Pattern 12's first test applies — it has no grammatical slot — and it is carried type, not a real word merely misused.

• Marker: not inconsiderably more sparing [sic: partiorem] in the serving up of theological matters.
• 7a⁗ satisfied: the sense is in plain words OUTSIDE the marker. Strip the bracket and the sentence still reads and still asserts what Pez asserts. This was checked mechanically (§5).
• The conjecture is parciorem, "more sparing," and it lives here, not in the text. The argument fixes it: the very next clause is gravior essem, decrevi non nisi primorum viginti psalmorum expositionem … inserere — he cuts the edition down because these men want less theology from him, not more. The rival reading, taking partiorem from partior ("more given to sharing"), would make them ask for more, which the sentence's own purpose clause refutes. That rival is named and rejected on that ground; the marker takes neither reading into the text.

§2 · 7a″ — THE VULGATE COLLATION. Every lemma in cols 0751C–0762A, checked.

Collated against sources/vulgate/clementine-flat.txt, cited from the Clementine by book chapter:verse. 226 italic spans were listed for the work; 221 of them fall in my columns and every one was walked. Ten took a [var:]. The rest are recorded below as checked — including the ones that came back clean, because a findings-only list cannot be told from a list nobody made.

2a · The ten [var:] markers, with the Clementine reading each rests on

⚠ Why the 0760B site took a [var:] and not a [cj:]. Pattern 18's decline list is explicit: a divergence inside a QUOTED SCRIPTURE is Pattern 14, not this (the 7561 @0664 modio/medio ruling). A psalm title is quoted scripture. The [cj:] shape was weighed — the internal contradiction at 0761A is exactly the kind of evidence 18 wants — and rejected on where the divergence sits, not on how strong the argument is.

2b · Lemmata checked and found to AGREE with the Clementine (no marker, and that is the finding)

Ps 1: Beatus vir qui non abiit… (1:1–2, verbatim, including the sed in lege Domini tail) · Et erit tanquam (1:3) · in tempore suo · et folium ejus non defluet · Non sic impii · Sed tanquam pulvis · Ideo non resurgent · in judicio · in concilio justorum · Quoniam novit Dominus · iter peribit. Ps 2: Quare fremuerunt gentes… · Astiterunt reges · Dirumpamus vincula · dirumpamus vincula eorum · Qui habitat in coelis (both) · Tunc loquetur · ira Dei · Ego autem constitutus · Dominus dixit ad me · Postula a me · Reges eos in virga · In virga ferrea · Et nunc, reges · Servite Domino · cum tremore · Apprehendite disciplinam · de via justa · Cum exarserit in brevi · in brevi. Ps 3: Domine, quid multiplicati sunt (both) · Multi dicunt · Non est salus ipsi in Deo ejus · Tu autem, Domine, susceptor · Voce mea ad Dominum · De monte · Ego dormivi · Ego dormivi et soporatus sum · Non timebo millia · Quoniam tu percussisti omnes · dentes peccatorum. Ps 4: Cum invocarem exaudivit me · Deus justitiae meae · Miserere mei et exaudi · Filii hominum · gravi corde · Et scitote quoniam · Irascimini et nolite · quae dicitis in cordibus vestris · Sacrificate sacrificium · sacrificium justitiae (both) · Signatum est super nos · A fructu frumenti · frumentum et vinum et oleum · In pace in idipsum · Quoniam tu, Domine, singulariter · Quis ostendit nobis bona? Ps 5: Verba mea auribus percipe (the twin's reading; the plate's oribus is §0) · Intende voci · rex meus · Deus meus · Quoniam ad te orabo, Domine · Mane astabo · exaudies vocem meam · Neque habitabit · Odisti omnes qui operantur · Virum sanguinum et dolosum · Virum sanguinum · quae haereditatem consequitur · Introibo in domum · Introibo in domum tuam · in templo · ad templum · in timore tuo · Domine, deduc me · In conspectu tuo · viam meam · Quoniam non est in ore · Judica illos, Deus · Et laetentur · Et gloriabuntur · Domine, ut scuto bonae voluntatis. Inline scripture, agreeing verbatim: Jo 12:26 · Lc 23:31 · Apc 2:7 · Jo 7:38 · Jo 7:39 (truncated) · Mt 1:20 · Lc 24:44 · Lc 24:45 · Is 12:3 · Pr 18:4 · Lc 24:19 · Gal 4:4–5 (opening At dropped) · Pr 1:10–11 · Pr 9:13–14 (stopped at sellam, which is the word glossed) · Pr 9:17 · Job 21:14 · Jac 1:21 (Propter quod dropped) · Ez 13:9 (but see §0 for ut/et) · Pr 31:31 · Mt 7:22 · Mt 25:12 · Mt 7:23 (nunquam/numquam orthography only) · Mt 25:41 · Is 65:25 · Ps 48:14 · Ps 35:7 · Ps 50:19 · Ps 15:5 · Jo 12:36 · 1Cor 3:17 · Jo 2:19 · Mt 6:9 · Rom 8:33 · Jo 10:17–18 (see 2c).

2c · Divergences found, WEIGHED, and deliberately left unmarked — with the reason for each

A [var:] is a public claim, so the ones I declined are itemized rather than dropped:

• 0752D · 1 Jo 3:2 scimus quia for Vulg scimus quoniam.* A conjunction; no reader takes anything different away. (Also Charissimi/Carissimi*, orthography.)
• 0752D · tanquam lignum plantatum secus decursus aquarum against Ps 1:3 tamquam lignum quod plantatum est secus decursus aquarum. A relative clause compressed to a participle inside the gloss's own sentence — grammar, not a different text.
• 0753A · Jos 1:8, recast into indirect discourse by its own frame (Dominus praecepit ut … non deserat, sed meditetur), so ut dirigat viam suam, et intelligat cuncta quae ait against tunc diriges viam tuam, et intelliges eam. Person and mood follow the frame. ⚠ cuncta quae ait for eam is a real difference and I record it here rather than marking it, because the clause is visibly the gloss's own paraphrase, not a citation offered as verbatim.
• 0753B · Sap 2:6 drops ergo; Sap 2:10 silently omits et non parcamus viduae mid-verse. Ellipsis inside a quotation, of the kind Migne marks elsewhere with etc. Recorded, not marked.
• 0753D · Jac 4:7–8: appropiate Domino for Vulg Appropinquate Deo, with the Old-Latin appropiate/appropiabit forms and Subditi ergo estote Deo dropped. Domino for Deo changes no assertion in a sentence that has just named the devil as the other term. Declined, recorded.
• 0754A · Rom 6:4: resurrexit Christus for Christus surrexit, and Migne drops vitae (in novitate ambulemus, "walk in newness," not "in newness of life"). Weighed as the closest decline in my range. Left unmarked because surgere/resurgere is a synonym pair and the missing genitive removes a word without reversing anything; but it is the site a later reader should re-open first if the bar is ever restated.
• 0755A · Mt 24:35 non transibunt for Vulg non praeteribunt*, plate-verified (p. 383). A synonym: both are "shall pass away," and the gloss turns on the firmness, which survives either.
• 0755A · Is 65:22 Secundum dies enim… — particle position only.
• 0755B · Eph 2:2 in filiis diffidentiae for Vulg in filios. Ablative for accusative; no English exponent worth a public claim, and Pattern 9's spirit covers it.
• 0755B · Aquilo ventus durus, cited sicut scriptum est, IS NOT IN THE CLEMENTINE AT ALL. Searched the whole flat file for aquilo/ventus durus: no such verse (Sir 43:21's frigidus ventus aquilo is the nearest, and it is not this). Plate-verified as Migne's own (p. 383), so it is not our corruption. A [var:] needs a received reading to name and there is none, so the finding is recorded here instead: the gloss adduces as Scripture a sentence the Clementine does not contain. Recording a checked negative, per brief §4's instruction to keep them.
• 0755B · Jo 3:18 Qui non credit in Filium* — the Vulgate has a bare qui autem non credit* and supplies the object only in the following clause. Migne pulls it forward; nothing is asserted that the verse does not assert.
• 0757A/0757B · Ps 2:13 confidunt in Domino for Vulg confidunt in eo*. Same referent, named instead of pronominalized.
• 0757B · Ps 3:1 a facie filii sui Absalom for Vulg a facie Absalom filii sui*. Word order inside a title; plate-verified (p. 384) as Migne's.
• 0758B · Jo 10:17–18 is a conflation: nemo tollet eam a meipso welds 10:18's Nemo tollit eam a me to its own ego pono eam a meipso, and quoniam/et iterum sumam replace quia/ut iterum sumam. A conflation has no single "received reading" for a marker to name. Plate-verified (p. 384) as Migne's, and recorded here instead.
• 0758B · Is 2:2 word order (Erit mons domus Domini praeparatus) — reordered, nothing added.
• 0759C · Mt 15:8 Labiis me honorant for Populus hic labiis me honorat*: the subject is dropped and the verb follows it into the plural. Grammar following an ellipsis.
• 0759D · Quis novit si vera sunt, aut quis venit ab inferis, ut ista nuntiaret? is introduced qui saepe dicunt — the gloss putting words in the mouth of the foolish, echoing Sap 2:1's non est qui agnitus sit reversus ab inferis without quoting it. Not a citation, so Pattern 14 does not reach it; recorded so a later reader does not mistake the echo for a divergence.
• 0760C · Jo 14:6 Per me itur ad Patrem for Nemo venit ad Patrem, nisi per me. An impersonal compression introduced by secundum illud quod dictum est; it asserts what the verse asserts, from the other side of the negative. Recorded, declined.
• 0761A · Ps 5:7 Perdes eos, qui loquuntur mendacium for Vulg perdes omnes qui…. ⚠ This is the 2a trigger shape (a quantifier swapped out), so it was checked against the sense rather than the sound: the gloss's next clause is quia recedentes ab eo quod est … declinant, which is about the class, not about how many. Same referent. Declined, recorded.
• 0762A · Rom 5:10 as Migne prints it — Si enim cum adhuc inimici essemus, Christus pro nobis mortuus est, multo magis reconciliati salvi erimus ab ira per ipsum — is three verses woven into one: adhuc and Christus pro nobis mortuus est from 5:8–9, ab ira per ipsum from 5:9, the frame from 5:10 (whose own tail is in vita ipsius). A [var:] must name one received reading and there is no one verse to name, so it is recorded here. It is not our corruption: the weave is coherent and reads as a quotation from memory.

§3 · MIGNE'S OWN PRINTING, NOT TIDIED

• 0755C · The gloss contradicts the verse it glosses, and that stands. The lemma is Ps 1:5 Ideo non resurgent (impii in judicio), and the gloss then says Qui resurgent quidem in judicio, sed non in concilio justorum. Rendered as printed — "Who indeed shall rise in judgment, but not in the council of the just" — with both negatives where the Latin puts them. The exposition is reading the psalm's non as governing only the second member; that is the author's argument, not a defect, and smoothing it either way would destroy it.
• 0762A · The chunk ends on a verbless sentence. Hoc invictissimum scutum, quo pellitur inimicus, desperationem salutis suggerens multitudine tribulationum et tentationum. There is no finite verb in either language. Rendered as printed and stopped where Migne stops (Pattern 8: never supply).
• 0755B · Two construals, and I chose neither silently. projiciuntur a soliditate et stabilitate terrae in ventum praesentis, scilicet Ecclesiae et futurae (plate-verified p. 383, punctuation and all). Either (i) praesentis … et futurae depend on ventum — "into the wind of the present, that is, of the Church, and of that which is to come" — or (ii) they depend on the distant terrae, giving "cast out from the solidity of the earth, present (that is, the Church) and to come, into the wind," which yields the better theology. I rendered (i), because it keeps Migne's word order and therefore keeps both readings visible to a reader with the Latin beside him; (ii) is recorded here and is the reading I would take if forced to choose.
• 0752C/0756B · hominis Jesu Christi in Pez's quotation of the incipit, hominis Christi Jesu in the body at 0752C. Migne's own inconsistency between the preface and the text it prefaces; plate-verified in the preface (p. 381). Both rendered as printed.
• Orthography left alone throughout (Pattern 9): tanquam, coelum, Charissimi, nunquam, sepulcrum, quanquam, spiritales, aenigmata. None is a defect and none is marked.
• Migne prints no ( sic ) anywhere in my range, so nothing is stacked on one.
• Band-letter arithmetic: 0751D is followed by 0752C (the work's drop-title occupies 752A–B); 0755D by 0756A; 0758D by 0759A. All normal per runbook false positive 1. Not reported as gaps.
• Pattern 4 [f:] was NOT used. This is a lemma-and-gloss commentary, not a florilegium (runbook false positive 4). No inline citation tails exist here in any case.

§4 · CHUNK 0000 — PEZ'S LITOTES, AND WHAT EACH RENDERING DECIDED

The brief flags this preface as unusually dense in double negatives for 263 words. Negation fidelity (7a) requires every printed negative in the English; the construal decisions are these, and each one carries both printed negatives rather than collapsing to a bare positive:

Two further calls in this chunk:

• The guillemets are 1:1, one open and one close, exactly where Migne sets them (plate-verified, p. 381). The Mabillon quotation ends at tomus alter amissus videtur, and Huc usque Joannes Mabillonius is outside it, as on the plate.
• The Mabillon incipit is not a scripture citation and took no [var:]. Omnis Scriptura divinitus inspirata Veteris ac Novi Testamenti… opens on 2 Tim 3:16's words and then becomes Walafrid's own sentence; the Vulgate's continuation (utilis est ad docendum) is not what is being quoted. The whole span was rendered and stopped at Migne's etc., supplying no verb to complete it (the usque ad discipline, applied to a truncated incipit).
• Titles are NOT italicized in the English, because Migne sets Analectorum, Thesauro and Glossa ordinaria in roman. Pattern 2's title-casing governs attribution headnotes, not running prose, and italic-span parity is 1:1 in this chunk as a result.

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

• Test 1 (strip every marker and read). Run mechanically over all ten markers: each surrounding sentence reads aloud without its bracket, with no stranded article or preposition. The site that could have failed is §1's [sic: partiorem], where the sense word "sparing" is deliberately outside the marker. ⚑ And the fourth question — does the clause still have its object? — asked at that site: yes, me … esse is the object clause and it is intact.
• Test 2 / 2a (name the host of every negative). Walked every printed negative in all five chunks. Three sites fall in 2a's declared trigger population — a negative on a verb with a correlative or quantifier next to it — and each was proved from the sentence's structure, not from what read well:
• 0757A · non solum eos non attinget … sed etiam — two negatives, two hosts: the first on solum, the second on attinget. Rendered "shall not only not touch." The sed etiam clause proves the split.
• 0756A · Astiterunt autem non praesentiam significat, sed voluntatem — the sed pivot puts the negative on praesentiam, not on significat. Rendered "signifies not presence, but will," never "does not signify presence."
• 0760A · Non ergo foris quaerenda est laetitia, sed intus — the same shape, and the same answer: the negative belongs on foris. Rendered "it is not without that gladness is to be sought, but within." This is the 8990 @0614B solus class exactly, and it was the site I most expected to get wrong.
• Also checked and found NOT to be of that class: non perturbationem … sed vim (0756B, negative correctly on the predicate noun), non optat ut eveniat, sed cernit (0761D, correctly on the verb), non quidem ubique, sed ubi (0758C).
• Test 2b (anchor position, not mere presence). Every anchor in all five chunks was checked against the WORD it stands beside in the Latin, not merely that it was present and in order. Six had drifted and were moved: 0753C (belongs before peccatores, had migrated past it), 0755D (before dicit), 0760A (before foris), 0760B (between sperandum and est), 0760D (before tenebris), 0761D (between se and recipiunt). All six now stand beside their Latin word. 0757A is a deliberate mid-construction split, per §0.
• Test 3 (punctuation BY BAND, never by total). Column anchors partition both twins identically, so :, ;, ?, !, «, » were counted band by band across all 42 bands of the five chunks. The first pass returned 30 mismatched bands and they did not cancel — they were 21 colons softened to full stops or commas and 9 semicolons supplied. Every one was restored to Migne's mark (including the odd-looking …retribueret: Legimus enim: at 0758B, and the full stop Migne prints before Amen dico vobis at 0755D where an English ear wants a colon). Final: zero mismatched bands, on all six marks. ⚑ This test earned its place in my range: nothing else I ran would have found a single one of those thirty.
• Test 4 (§4 normalization / suspiciously tidy forms). This is what produced §0. The trigger was brief §4's own instruction — a form that reads too well in a rough passage — fired at 0758B's incovenienter (a non-word, so Addendum A said raise it) and answered by the plate in the opposite direction: the plate is clean and our file is corrupt. That one page read then yielded four more sites within four inches of it, exactly as Addendum C predicts.
• Test 5 (does it ASSERT what he asserts?). Read each finished paragraph back against the Latin with the counts already clean. Three sites were re-worked on this pass and none of them was findable by any count: the Ps 1:5 contradiction (§3) where the first draft had softened resurgent toward the Vulgate's non resurgent; Solent vero abominati dici exhaeredati at 0761A, where the case endings and not the sense decide which noun is subject; and the verbless close at 0762A, which a smooth English would have quietly completed.

§6 · WHAT I DID NOT DO, SO THE NEXT READER KNOWS

• I did not patch the Latin. §0's thirteen sites are escalated, not fixed. Re-chunking is the merge step's call and would rewrite chunk boundaries mid-fleet.
• I did not collate cols 0762A–0794D. Chunks 0005–0017 belong to other agents; §0's rate applies to my six pages only and must not be inherited as either a clean or a dirty verdict there.
• I read the plate for every page in my range (pp. 381–386) but not every column inch of pages 382, 383 and 386 — on 382 and 383 I read the full left column and the lemmata I was firing on; on 386 the full left column (col. 761) and not col. 762, of which my range holds only five lines. A complete collation of this work is a separate, cheap job (six page renders) and I recommend it before ship.
• No [cj:] anywhere. Two candidates were weighed and both declined: partiorem (already inside a [sic:], which Pattern 18 excludes explicitly) and 0760B's veritatem (quoted scripture, so Pattern 14's territory, not 18's — §2a).
• No [nt:], [n:], [ed:] or [d:]. The work has noteCount: 0 in all eighteen chunks and I wrote none; data/plate-notes/coverage.json records 9003 as spot, checked zero, and my own six pages saw no foot rule, no numbered Forte apparatus and no raised-asterisk layer — which is a sixth and seventh page of evidence for that zero, recorded here because a checked negative is worth having.

PLATE COLLATION — markers withdrawn

2026-08-28. All 22 pages of the work were read at Migne's plate and 61 TEI patches applied, so the Latin twins for these chunks now say what Migne printed rather than what Corpus Corporum transcribed. No marker was withdrawn in chunks 0000–0004 — the withdrawals fall in cruces-0005 (one [var:]) and cruces-0014 (three [sic:], one [cj:]). What changed here:

Fired — one new [var:] the digitization had erased.

• 0758C · Ps 3:9 · [var:] FIRED. Migne prints Domine est salus (vocative); the Clementine (sources/vulgate/clementine/Ps.lat) has Domini est salus, "salvation is the Lord's". Corpus Corporum had silently moved the lemma to the Clementine's genitive, and the English had duly rendered "Salvation is the Lord's". Both readings construe — Domine, est salus is good Latin ("O Lord, there is salvation"), and it answers v. 3 in the same chunk, Non est salus ipsi in Deo ejus. Migne is now rendered* — "O Lord, there is salvation, etc." — with [var: Vulg. Ps 3:9 Domini est salus, "salvation is the Lord's"] beside it. ⚑ Note that Migne's own gloss two words later goes the other way (quoniam Domini est salvos facere*), which is precisely the tension a [var:] exists to make visible rather than to smooth away.

Declined, and why — a [var:] is a public claim about the Clementine and must not be spent on broken type. The rule I applied, stated so it can be checked: fire where the printed reading construes in place and yields a DIFFERENT SENSE; decline where it yields NO sense, because that is Pattern 12's business (type the plate got wrong), not Pattern 14's.

• 0755C · Ez 13:9 — DECLINED. Migne prints …non erunt, ut in scriptura domus Israel non scribentur; Clem. et in scriptura. But ut + future indicative construes as nothing at all: it is not a purpose clause and not a comparison. This is a broken sort (the same class as this work's fremueruut and populii at 0755D), not a divergent citation. English keeps "and", unmarked, per the standing ruling that the work's restored typographic defects are out of scope.
• 0760C · Ps 5:2 — DECLINED. Migne prints Verba mea oribus percipe; Clem. auribus. oribus is a form (os, oris, "mouths") but "receive my words with mouths" is not a sense, it is a dropped sort. English keeps "Give ear to my words." Same class as above.
• 0760D · Mt 5:8 — DECLINED as a [var:], but the ENGLISH WAS CORRECTED. Migne prints Deum videre, qui mundo conspicitur — unitalicized, in the glossator's own prose, an echo of Beati mundo corde and not a lemma; Pattern 14 governs citations, so no marker. ⚠ What did have to change is that Corpus Corporum had SUPPLIED corde, and our English had translated the supplied word ("who is beheld with a clean heart"). The plate carries no corde. The English now reads "who is beheld by the pure" — Pattern 8, never supply what the plate does not print.

PLATE COLLATION · REGISTER OF RESTORED TYPE, cols 0751C–0762A

Wilson's ruling, 2026-08-28. Pattern 12 read literally would put a [sic:] on every one of the ~31 typographic defects the plate collation restored to this work, and would bury the two or three sites that matter under a crowd of notes about broken type. The ruling is: mark only where the reading is in doubt; record the rest here and disclose the count in the work-level page note. The operative test, applied site by site: mark it where the plate's reading does not construe, or where a reader could not recover Migne's intended word with confidence, or where the defect could be mistaken for a real variant; do not mark a plainly battered, turned, dropped or wrong sort whose intended word is not in doubt.

MARKED in this range — two.

• 0753B · quod si for quod sit — [cj: quod si; read quod sit, "what the counsel is"]. Migne prints Si autem apertius quis requirit quod si consilium impiorum, multipliciter Scriptura loquitur. The indirect question needs sit; CC had supplied the t. This is the dangerous shape: the dropped sort leaves behind a perfectly idiomatic Latin phrase (quod si, "but if"), so nothing on the Latin page tells the reader anything is wrong, and our English had silently rendered the conjecture ("what the counsel of the ungodly is"). Marker placed after the rendering; the rendering stands.
• 0761A · operantur for operantem — [cj: operantur; read operantem, "him that worketh"]. Migne prints Virum sanguinum operantur iniquitatem, dolosum autem diligentem mendacium intellige — a third-person plural finite verb where the clause, governed by intellige, requires an accusative participle in parallel with diligentem. It has no grammatical slot, it is not a one-sort slip, and the English ("as him that worketh iniquity") was rendering CC's repair.

Already marked, and standing: 0758C [var: Vulg. Ps 3:9 Domini est salus, …], argued above.

CRUCES ONLY — recorded here, unmarked in the text. Migne's printed form → intended word.

Not a defect, recorded for completeness: 0760D, where Corpus Corporum had supplied corde from Mt 5:8. The plate prints qui mundo conspicitur alone; the interpolation is gone and the English reads "who is beheld by the pure" (see the note above). Nothing of Migne's is wrong here.

THE BLIND POLARITY READ

Four fenced readers took the work in four bands (0000–0004, 0005–0009, 0010–0013, 0014–0017), each reading Latin against English in full, with no cruces.md file and no collation report open before or after. Their reports were adjudicated on 2026-08-28, after the 22-page plate collation, under Wilson's marker ruling of the same session: mark where the plate's reading does not construe, or where a reader could not recover the intended word with confidence, or where the defect could be mistaken for a real variant; do NOT mark a plainly battered, turned or dropped sort whose intended word is not in doubt.*

Whole-work result across all 18 chunks: 13 confirmed sites. Ten carried a defect in our English and were re-rendered; four sat on Migne's plate and took a marker with the English left literal. (The two figures overlap by one: 0779D is a lost sed non on the plate and a connective our English had smoothed, so it was both re-rendered and marked.) Four new [cj:] markers, one new [var:]. Eight further candidates were adjudicated and declined; they are listed under their own ranges, with reasons.

Per range — ours / Migne's: 0000–0004: 5 / 0 · 0005–0009: 1 / 0 · 0010–0013: 3 / 3 · 0014–0017: 1 / 1.

Repaired in this range (chunks 0000–0004)

1. 0755B — the Church was identified with the devil's wind. Migne: projiciuntur a soliditate et stabilitate terrae in ventum praesentis, scilicet Ecclesiae et futurae. The genitives depend on terrae, not on the accusative ventum; Migne's hyperbaton drops in ventum between them. Our English had read "into the wind of the present, that is, of the Church, and of that which is to come" — and the very next sentence is Ventus autem ipse diabolus intelligitur. Now: "cast forth from the solidity and the stability of the earth of the present, that is, of the Church, and of that which is to come, into the wind". 7a mechanism 5 (reference mis-bound), and the page had been contradicting its own neighbour.

1. 0755C — a 7a″ conformation the plate collation exposed. Migne prints In concilio populi mei non erunt, ut in scriptura domus Israel non scribentur; the Clementine has et, and Corpus Corporum had normalized Migne toward it. The Latin was corrected to ut in the collation, but the English still read "and". Now rendered "that in the writing of the house of Israel they shall not be written", with a [var:] recording the Vulgate's et and the fact that ut + future indicative does not construe. ⚠ The failure to construe is Migne's; nothing has been smoothed to hide it.

1. 0761C, VERS. 10 — the lemma had lost the word its own gloss expounds. Migne prints Quoniam non est in ore, etc.; our English gave "For there is no," etc., with no exponent at all for in ore, in a stump whose gloss turns on precisely that phrase (Quomodo potest esse in ore eorum veritas…). Completed to "For there is no truth in their mouth," etc., on this work's settled fragmentary-lemma convention (cf. 0769D Inimici defecerunt → "The enemy's swords have failed").

1. 0760A, VERS. 9 — speratur a talibus was read as a source, not an agent. "hoped for from such men" → "hoped for by such men". a + ablative with a passive is the agent, and tales are the vir fidelis of the sentence that introduces the verse ("The faithful man therefore rejoices and says"); the source reading detached the clause from its own cue.

1. 0756B — who foreknows. per quas utique praescientes quid futurum sit, deridebit Deus et subsannabit: praescientes is accusative plural agreeing with quas, the holy souls, not with Deus. The paragraph's whole argument is that God's derision is the force of foreknowledge he gives to his saints (ea vis accipienda est, quam dat sanctis suis, ut futura cernentes… intelligant). Now "through whom, since they indeed foreknow what shall come to pass, God will deride and mock."

Adjudicated and DECLINED in this range

• 0762A — desperationem salutis suggerens attaching to the shield rather than to inimicus (reader's confidence: medium-high). Declined. The reader's stated ground — that suggerens "is nominative and agrees with inimicus" — is only half right: the present participle's nominative singular is identical in all three genders, so scutum and inimicus are equally available in the Latin, and Migne's sentence is anacoluthic with no finite verb to settle it. Our English reproduces the same ambiguity with the same nearest-noun default (inimicus / "the enemy"). Nothing is asserted in English that the Latin does not equally assert, so a re-binding would be our disambiguation of Migne's, not a repair.

Positive results worth recording

• 0000 (the Pezian monitum) is clean at every one of its six litotes — nullam non operam … dedi, non nihil … commodaretur, non nihil partiorem, non nisi primorum viginti psalmorum, Nec … unquam mutassem, nisi … deprehendissem, ne … omnino depereat. Both printed negatives survive in all six; the double-negative idiom is nowhere smoothed.
• 0760C / 0761A — the psalm title, this work's 7a″ trap, was not fallen into. Migne prints pro ea quae veritatem consequitur in the lemma and quae haereditatem consequitur in the gloss; the English carries both as printed, with a [var:] on the first.
• 0760D qui mundo conspicitur — taken as the adjective ("beheld by the pure"), which is right; the alternative ("by the world") would have contradicted the preceding clause outright. Noted because it is exactly the seat a silent reversal would take.

Preserved unedited from cruces-0005.md; headings demoted one level. Where it conflicts with the head above, the head governs.

Glossa ordinaria, Expositio in XX primos Psalmos (PL 114) — cruces, chunks 0005–0009

Cols 0762B–0774A (Ps VI, VII, VIII, and Ps IX to v. 39), 5 chunks, ~4,930 Latin words. One translator, 2026-08-28.

verify-english.mjs: my five chunks pass (the run's only failures are chunks 0011–0013 and 0017, not yet filed by other agents). Frontmatter copied verbatim from each Latin twin and diffed. Column anchors: 48 across the five chunks, verbatim, in order, and each placed against the same word it stands beside in the Latin (test 2b — checked one by one, not merely for presence). [n: …] in my English: 0, matching noteCount: 0 in all five twins, as the launch brief requires. Guillemets: zero in either language. Question marks: 2 · 1 · 1 · 7 · 2 — exact parity with the Latin twins, band by band. Em-dashes: 0 (three were written and all three removed; Migne prints none).

Punctuation was compared band by band, not by total (test 3), using the column anchors as the partition: colons, semicolons, question marks, exclamation marks, guillemets and italic asterisks, in all 53 bands of the five chunks. All bands now match. Six real mismatches were caught this way and repaired — five softened or supplied colons and two comma↔semicolon swaps — plus three merged italic spans (§6 below), which nothing else would have found.

Apparatus fired: 11 [var:] · 3 [cj:] · 0 [sic:] · 0 [d:] · 0 [ed:] · 0 [nt:]. (⚠ [cj:] count corrected 2026-08-28 from 1 to 3; the three stand at 0766A, 0767B and 0771B. The blind polarity read added none in this range — see its section at the foot of this file.) ⚠ Revised 2026-08-28, was 10 [var:]: the plate collation withdrew the one @0770A (it was an artefact of the transcription, §2f) and added two the digitization had erased, @0764B and @0771C. See PLATE COLLATION — markers withdrawn at the foot of this file.

§0 · THE NUMBERING IS MIGNE'S AND I HAVE NOT TOUCHED IT

Migne numbers by the Vulgate/Septuagint reckoning and counts the title as verse 1, so his VERS. 2.-- on Psalm VI addresses what a modern English Bible prints as 6:1. The Clementine uses the same reckoning and agrees with him throughout. Nothing here is renumbered and the offset is not reported as a divergence. Three places where his own count runs off the Clementine's within that reckoning are recorded as his ordering, not corrected:

• Ps VI — he prints VERS. 9.-- Discedite a me, and then takes Exaudivit Dominus (Clem. 6:10) as a second lemma inside the same address, so his VERS. 10.-- Erubescant, sits on Clem. 6:11. From v. 10 on, his count is one behind.
• Ps VIII — his VERS. 7.-- Omnia subjecisti is Clem. 8:8 (Clem. 8:7, et constituisti eum, is not glossed); his VERS. 8.-- Volucres coeli is Clem. 8:9; there is no VERS. 9, and VERS. 10.-- Domine, Dominus noster lands back on Clem. 8:10.
• Ps IX — Narrabo omnia mirabilia tua (Clem. 9:2) is glossed under his VERS. 3.--.

§0a · VERS 28.-- at 0772B — the period after VERS is missing, and it rides as printed

Chunk 0009 opens VERS 28.-- where every other one of the 72 verse marks in my range prints VERS. n.--. It is in the Latin twin's incipit field as well. Under the settled Glossa convention the address passes through verbatim, untranslated, so the English prints VERS 28.-- too. ⚑ This is the exact class of Ruth's VERS.9 at 0537D, which turned out to be a Corpus Corporum slip and was fixed by data/tei-patches/8968.json, not by a marker. Plate candidate, p. (0772+11)/2 ≈ 391 of the PL 114 scan: if Migne prints VERS. 28.--, this is a tei-patch for 9003 and the English should follow the patch. I have fired nothing.

§1 · THE LEMMA COLLATION — every italic span in cols 0762B–0774A

Collated against sources/vulgate/clementine-flat.txt; readings cited from sources/vulgate/clementine/Ps.lat, Mt.lat, Jo.lat, Gn.lat, Apc.lat, 1Cor.lat, Rom.lat, 2Cor.lat, Jac.lat, Gal.lat, 1Ptr.lat, Zach.lat, Phlp.lat, 2Thes.lat. A lemma named here with no comment agrees with the Clementine at its own verse, verbatim, or is a clean clip of it — that is why they are named. The psalter here is the Gallican, which is what the Clementine prints, so the clean run is expected and a genuine divergence is correspondingly more significant.

Psalm VI (0762A–0763D) — chunk 0005

In finem in carminibus pro octava psalmus David. (6:1 — Clem. punctuates and orders as In finem, in carminibus. Psalmus David. Pro octava.; Migne runs it as one clause and moves pro octava forward. Order and punctuation only, no word gained or lost) · Domine, ne in furore tuo, (6:2) · arguuntur, · Corripiuntur (both 6:2, the gloss's own third-person forms of arguas/corripias) · Miserere mei, Deus, (6:3 — Clem. Miserere mei, Domine. [var:] FIRED, §2a) · ossa (6:3) · Et anima mea, (6:4) · Convertere, Domine, (6:5) · Convertimini ad me, et convertar ad vos, dicit Dominus. (Zach 1:3 — Clem. Convertimini ad me, ait Dominus exercituum, et convertar ad vos, dicit Dominus exercituum; Migne drops both exercituum and the interposed ait Dominus exercituum. Omission only, nothing substituted, and the argument turns on the reciprocity, not on the title of hosts. No marker; see §5a) · Quoniam non est, (6:6) · mortem · infernum (6:6) · Laboravi, · Lectus · lacrymis lavat, (6:7 — the gloss's own third-person recasting of lacrimis meis stratum meum rigabo / lavabo … lectum meum; not a fresh citation) · mente servio legi Dei, · carne autem legi peccati, (Rom 7:25, verbatim) · mane adstabo tibi. (Ps 5:5 — Clem. Mane astabo tibi; adstabo/astabo is orthography, no marker) · stratum, · rigabo · lavabo: (6:7) · Turbatus est, (6:8) · oculum · turbatum. (6:8) · inveteravi inter omnes inimicos meos, (6:8, verbatim) · inveteravi. In omnibus · inimicis meis, (see §3a — the re-quotation does not match the lemma Migne has just printed) · prope est Dominus obtritis corde, (Ps 33:19 — Clem. Juxta est Dominus iis qui tribulato sunt corde. Candidate weighed and DECLINED, §5b) · Discedite a me, (6:9) · Discessuri enim sunt, (not scripture — the gloss's own prophetic paraphrase of Discedite) · Exaudivit Dominus, (6:10) · Erubescant, (6:11, printed under his VERS. 10) · valde velociter, (6:11, verbatim).

Psalm VII (0763D–0766C) — chunk 0006

Chusi · filius Jemini, (7:1) · Domine Deus meus, in te speravi, (7:2) · Ne quando rapiat, · ne quando rapiat ut leo animam meam. · ne quando rapiant; (7:3 — Clem. sets nequando as one word throughout; word-division only, and the third is the gloss's own counterfactual plural, expressly marked as such by Non enim ait) · Adversarius vester diabolus tanquam leo rugiens circuit, quaerens quem devoret. (1Ptr 5:8 — Clem. tamquam; orthography only, the rest verbatim) · salvum me fac ex omnibus persequentibus me, (7:2) · dum non est qui redimat, (7:3) · Domine Deus meus, · si feci istud, · si est iniquitas in manibus meis. (7:4) · Si reddidi, · si reddidi facientibus, (the gloss's own counterfactual, again marked Non ait) · retribuentibus mihi mala. · decidam merito ab inimicis meis inanis. · inanis (7:5) · Persequatur inimicus, · Inimicum · leonem · terram (7:6) · terram manducabis. (Gn 3:14 — Clem. terram comedes. [var:] FIRED, §2d) · in pulverem deducit, (7:6 — Clem. in pulverem deducat, subjunctive; here indicative because the gloss has taken the phrase into its own sentence. No marker) · Exsurge, Domine, · iram · fines, · exaltari · Exsurge, Domine Deus meus, (7:7) · Synagoga · populorum, · circumdedit, (7:8 — Clem. circumdabit te, future; Migne's perfect. Candidate weighed and DECLINED, §5c) · Caecitas ex parte contigit in Israel, donec plenitudo gentium intraret. (Rom 11:25, verbatim) · Et propter hanc, (7:8) · Dominus judicat populos, · Judica me, Domine, · secundum innocentiam · super me, (7:9) · si reddidi retribuentibus mihi mala. (7:5, verbatim re-quotation) · Consummetur nequitia, and Consummetur, (7:10 — Clem. Consumetur nequitia peccatorum. [var:] FIRED; the single most consequential divergence in my range, §2b) · sordescat (Apc 22:11, verbatim) · justus justior fiat, (Apc 22:11 — Clem. qui justus est, justificetur adhuc. [var:] FIRED, §2c) · diriges justum, scrutans corda et renes Deus. (7:10, verbatim but for a comma before Deus) · dirigitur · cordis · renium · Scrutans · cor · dirigit justum (7:10, the gloss's own indicative for diriges) · Justum adjutorium, (7:11) · Justum auxilium ×2 (§3b — Migne re-quotes his own lemma with a synonym; kept distinct in the English as help / aid) · quoniam, si cum adhuc peccatores essemus, Christus pro nobis mortuus est, quanto magis nunc justificati salvi erimus ab ira per ipsum. (Rom 5:8–9 — candidate weighed and DECLINED, §5d) · Deus judex justus, · Justus, · fortis, · patiens, · irascitur per singulos dies, (7:12 — Clem. numquid irascitur per singulos dies?; Migne turns the rhetorical question into a flat negation by putting a roman Non in front of the italic. That is his gloss doing the work, not a variant text: the italic span itself is verbatim. No marker) · Nisi conversi, · Et in eo paravit, (7:13–14) · Judex justus · Arcum · sagittae · ardentibus effecit, (7:14) · Vasa · mortis vel (§6a — the italic span runs one word past the verse) · Vel · aliis · odor vitae in vitam, aliis odor mortis in mortem. (2Cor 2:16 — Clem. prints the two clauses in the reverse order, aliis quidem odor mortis in mortem: aliis autem odor vitae in vitam, and has quidem/autem. Candidate weighed and DECLINED, §5e) · Ecce parturiit, · concepit dolorem, (7:15) · parturit injustitiam, · parit iniquitatem; (7:15 — the gloss's own present tenses for parturiit/peperit, expounding generically; no marker) · peccatum, cum consummatum fuerit, generat mortem. (Jac 1:15 — Clem. peccatum vero cum consummatum fuerit; the dropped vero is a clip) · Lacum aperuit, (7:16) · in operibus manuum suarum comprehensus est peccator. (Ps 9:17, verbatim) · Convertetur dolor, (7:17) · iniquitates meae supergressae sunt caput meum; sicut onus grave, gravatae sunt super me. (Ps 37:5 — Clem. caput meum, et sicut onus grave; the dropped et and Migne's semicolon, nothing more) · Confitebor Domino, (7:18).

Psalm VIII (0766C–0768C) — chunk 0007

Torcularia · torcularia (8:1, pro torcularibus) · Domine, Dominus noster, · Domine, · Dominus noster, (8:2) · Ex ore infantium, · infantes · lactentes (8:3) · Tanquam parvulis in Christo, lac vobis potum dedi, non escam. (1Cor 3:1–2 — Clem. tamquam; orthography only, the rest verbatim across the two verses) · Inimicos · ultor (8:3) · Quoniam videbo coelos, · Videbo, (8:4) · coelos opera digitorum tuorum: (8:4 — Clem. videbo caelos tuos, opera digitorum tuorum; Migne drops tuos and the comma. Clip, no marker) · Luna · stellae · stellarum (8:4) · Quid est homo, (8:5) · homines et filium hominis (the gloss's own accusative pairing, not a citation) · aut filius hominis, (8:5, verbatim) · Homo · memor · Filium · hominis, · visitat (8:5) · Minuisti eum, (8:6) · Minues eum paulo minus a Deo: (not the Vulgate and not offered as such — Migne says in Hebraico legitur, i.e. Jerome iuxta Hebraeos. Not collated as a Vulgate lemma; see §5f) · Omnia subjecisti, · omnia. (8:8) · Excepto eo qui ei subjecit omnia. (1Cor 15:27 — Clem. sine dubio praeter eum qui subjecit ei omnia. Candidate weighed and DECLINED, §5g) · Oves et boves, · oves et boves (8:8) · nonaginta novem oves in montibus, (Mt 18:12 — Clem. nonne relinquit nonaginta novem in montibus; Migne's oves is an addition, but the whole is inside the gloss's own indirect statement, Dominus dicit reliquisse se. No marker) · bubus · boves · Pecora · campi · Campus (8:8) · Volucres coeli, · volucres coeli (8:9) · posuerunt in coelum os suum: (Ps 72:9 — Clem. caelum, orthography) · Linguam nostram magnificabimus, labia nostra a nobis sunt, quis noster Dominus est? (Ps 11:5, verbatim; Clem.'s lower-case dominus only) · pisces maris · qui perambulant semitas maris, · Perambulant (8:9) · Domine, Dominus noster, (8:10).

Psalm IX to v. 39 (0768C–0774A) — chunks 0008 and 0009

De occultis filii (§6b — the italic span swallows the gloss's own De) · Si vos Filius liberaverit, vere liberi eritis. (Jo 8:36 — Clem. Si ergo vos filius liberaverit; the dropped ergo is a clip) · occulta filii, ×2 (the gloss's own nominative/accusative) · Victori super morte filii. · victori pro morte filii, (not the Vulgate and not offered as such — ex Hebraeo translata editio, §5f) · in finem pro occultis filii, (9:1, verbatim) · Confidite, ego vici mundum, (Jo 16:33, verbatim) · finis, · Non veni solvere legem, sed adimplere. (Mt 5:17 — Clem. Nolite putare quoniam veni solvere legem aut prophetas: non veni solvere, sed adimplere; Migne conflates the two halves, carrying legem down into the second. A conflation of the verse's own words, adding nothing. No marker) · Confitebor tibi, Domine, (9:2) · Non in toto corde confitetur · Narrat · omnia mirabilia (9:2, the gloss's own negated and third-person forms) · Laetabor et exsultabo, · in te, (9:3) · Narrabo · omnia mirabilia tua, (9:2, verbatim) · Omnia quaecunque audivi a Patre meo nota feci vobis: (Jo 15:15 — Clem. quaecumque, orthography; verbatim otherwise) · In convertendo, · Inimicum meum, (9:4) · Redi retro, Satanas. (Mt 16:23 — Clem. Vade post me Satana. Candidate weighed and DECLINED, §5h) · quae retro sunt obliviscens, et se ad ea quae ante sunt extendens. (Phlp 3:13 — Clem. quae quidem retro sunt obliviscens, ad ea vero quae sunt priora, extendens meipsum. Candidate weighed and DECLINED, §5i) · Canam nomini tuo, Altissime, · Cum ceciderint inimici mei retrorsum, et corruerint, et perierint a facie tua. (not the Vulgate and not offered as such — sicut Hebraica docet editio, §5f) · Quoniam fecisti judicium, · meum · causam · meam (9:5) · Increpasti gentes, (9:6) · increpavit gentes, et periit impius, · periit impius, · Deletumque · nomen (9:6, the gloss's own third-person and passive recastings) · Inimici defecerunt, · frameae defecerunt, (9:7 — Clem. Inimici defecerunt frameae in finem; Migne's second span reorders the verse's own two words. No marker) · In finem (9:7) · civitates ejus destruxit, (9:7 — Clem. civitates eorum destruxisti; the gloss recasts to third singular with ejus because it has made Christ the subject. Recasting, not a variant text. No marker) · Princeps hujus saeculi misus est foras. (Jo 12:31 — Clem. nunc princeps hujus mundi ejicietur foras. [var:] FIRED, §2e; and misus is a plate candidate, §4a) · Periit memoria, (9:7) · Paravit in judicio thronum, (9:8) · judicio sedem suam. (9:8 — Clem. thronum suum; Migne's re-quotation substitutes sedem for the thronum he has just printed. Candidate weighed and DECLINED, §5j) · Et ipse … judicavit orbem terrae in aequitate. (9:9 — ⛔ [var:] WITHDRAWN 2026-08-28; Migne prints judicabit with the Clementine, the twin was corrupt, §2f) · Et factus est Dominus refugium pauperi, · adjutor in opportunitatibus, · in tribulatione. (9:10) · Et sperent in te, (9:11) · cognitio nominis ×2 (the gloss's own noun for qui noverunt nomen tuum) · Psallite Domino, · habitat in Sion, · speculatio, (9:12; the last an etymology, not a citation) · Templum Domini sanctum est, quod estis vos. (1Cor 3:17 — Clem. Templum enim Dei sanctum est. [var:] FIRED, §2g) · Quoniam requirens, (9:13) · Domine, quis credidit auditui nostro? (Rom 10:16, verbatim — Is 53:1 has no Domine, so Migne is quoting Paul, not Isaiah) · Propter te mortificamur tota die. (Ps 43:22, verbatim) · quoniam requirens sanguinem eorum recordatus est. (9:13, verbatim) · Miserere mei, Domine, (9:14) · Non est oblitus clamorem pauperum. (9:13, verbatim) · Qui exaltas me, · Exaltatur · portae mortis, (9:15) · portae filiae Sion ×2 (9:15, Clem. in portis filiae Sion; the gloss's nominative) · Exsultabo in salutari tuo, ×2 · salutari tuo, (9:16 — Clem. exultabo, orthography) · In laqueo isto, · Laqueus · pes · comprehensus est pes eorum, (9:16) · Cognoscetur Dominus, · judicia (9:17) · Convertantur peccatores, (9:18) · Quoniam non in finem oblivio, · Pauperes · patientia · non peribit, (9:19) · Exsurge, Domine, non confortetur, · in conspectu tuo, (9:20) · Constitue, Domine, · quoniam homines sunt. (9:21) · Cum revelabitur homo peccati, (2Thes 2:3 — Clem. et revelatus fuerit homo peccati. Candidate weighed and DECLINED, §5k) · Utquid, Domine, recessisti, (9:22 — Clem. sets Ut quid as two words; word-division only) · Dum superbit impius, · Comprehenduntur · in consiliis, (9:23) · Quoniam laudatur peccator, (9:24) · Exacerbavit Dominum, (9:25) · Non est Deus in conspectu ejus, · in omni tempore, · Auferuntur judicia tua, · Omnium inimicorum suorum dominabitur, (9:26, all verbatim) · in templo Dei sedebit; extollens se super omne quod colitur et quod dicitur Deus. (2Thes 2:4 — candidate weighed and DECLINED, §5k) · Dixit enim in corde suo, (9:27) · Loquitur in corde suo: Non movebor: in generatione et generatione ero sine malo. (not the Vulgate and not offered as such — Hebraica translatio, §5f) · Cujus maledictione os plenum est, · maledictione os · plenum est. (9:28) · Cum amaritudine et dolo, (9:28 — Clem. et amaritudine, et dolo. Candidate weighed and DECLINED, §5l) · Sub lingua ejus labor et dolor, · sub lingua ejus, (9:28, verbatim) · Non movebor, (9:27) · Sedet in insidiis, · in insidiis · sedere · In occultis · Innocentem · interficere (9:29) · Oculi ejus in pauperem respiciunt, (9:30) · Beati pauperes spiritu. (Mt 5:3, verbatim — note the span runs on into Leonem, §6c) · Leonem · in spelunca · Leo: · in spelunca sua. · Insidiatur ut rapiat, · insidiae · rapere (9:30) · In laqueo suo humiliavit, (9:31 — Clem. In laqueo suo humiliabit eum. [var:] FIRED, §2h) · Laqueus · inclinabit et cadet, (9:31 — Clem. inclinabit se, et cadet; the dropped se is a clip) · Dixit enim in corde suo, (9:32) · Exsurge, Domine Deus, exaltetur, (9:33) · Propter quid irritavit impius Deum, (9:34) · Vides quoniam tu laborem, (9:35) · non requiret Deus, (9:34 — Clem. Non requiret. stops there; Migne supplies the subject Deus inside the italic. Candidate weighed and DECLINED, §5m) · ut tradat eos in manus suas, (9:35 — Clem. ut tradas eos in manus tuas. [var:] FIRED, §2i) · Tibi derelictus est, · Orphanum (9:35) · Mihi mundus crucifixus est, et ego mundo. (Gal 6:14, verbatim clip) · Ne dicatis vobis patrem in terris. (Mt 23:9 — Clem. Et patrem nolite vocare vobis super terram. [var:] FIRED, §2j) · Contere brachium peccatoris, (9:36) · omnium inimicorum suorum dominabitur. (9:26, verbatim re-quotation) · Brachium · Quaeretur · peccatum ejus, (9:36 — Clem. quaeretur peccatum illius; ejus/illius is a pronoun synonym, no marker) · nec invenietur, (9:36 — Clem. et non invenietur; nec for et non is the same negation in one word. Explicitly checked as a negation-fidelity site: the English carries the negative, §3c) · Dominus regnabit in aeternum, · Gentes (9:37) · Desiderium pauperum, (9:38) · Praeparatio · cordis (9:38, the gloss's nominative for praeparationem) · Paratum cor meum, Deus, paratum cor meum: (Ps 56:8, verbatim) · si autem, quod non videmus, speramus, per patientiam exspectamus. (Rom 8:25, verbatim) · aurem Dei (9:38, the gloss's own for auris tua) · Judicare pupillo et humili, (9:39, verbatim) · judicare pupillum, · judicare pupillo. · Judicat pupillum · judicat · pupillo, (the gloss's own constructed contrast) · Homines · homo (9:39) · Constitue, Domine, legislatorem super eos, ut sciant gentes, quoniam homines sunt; (9:21, verbatim) · ut non apponat ultra magnificare se homo super terram, (9:39, verbatim).

§2 · THE TEN [var:] MARKERS — each one a claim resting on the file, not on memory

Every one below was read off clementine-flat.txt and confirmed in the per-book .lat. In every case Migne's reading is what the English prints; the marker stands beside the faithful rendering and never in place of it (Pattern 14).

2a · 0762B/C · Ps 6:3 — Miserere mei, Deus, where the Clementine has Domine

Ps 6:3 Miserere mei, Domine, quoniam infirmus sum ; sana me, Domine, quoniam conturbata sunt ossa mea. Migne prints VERS. 3.-- Miserere mei, Deus, etc. A different divine name in the psalm's own opening address. English: "Have mercy on me, O God," [var: Vulg. Ps 6:3 Miserere mei, Domine, "Have mercy on me, O Lord"].

2b · 0765A/B · Ps 7:10 — Consummetur for Consumetur, and the whole gloss turns on it

Ps 7:10 Consumetur nequitia peccatorum, et diriges justum, scrutans corda et renes, Deus. Consumetur is from consumo: the wickedness of sinners shall be consumed. Migne prints Consummetur (from consummo, to complete) and then expounds precisely that: Consummetur, inquit, perficiatur, id est perveniat ad summam nequitiam, ut possit istud jam venire judicium — "let it be completed … let it come through to the height of wickedness, that this judgment may now be able to come." ⚑ This is the highest-value site in my range, because the exposition is unintelligible on the Clementine's word and exact on Migne's: a translator who reached for the familiar "shall be consumed" would leave the next sentence arguing from a word no longer on the page (7a″). Migne also drops peccatorum. English: "Let wickedness be consummated," [var: Vulg. Ps 7:10 Consumetur nequitia peccatorum, "the wickedness of sinners shall be consumed"], and Consummetur is rendered "consummated / brought to completion" wherever it recurs.

2c · 0765A/B · Apc 22:11 — justus justior fiat for qui justus est, justificetur adhuc

Apc 22:11 Qui nocet, noceat adhuc : et qui in sordibus est, sordescat adhuc : et qui justus est, justificetur adhuc : et sanctus, sanctificetur adhuc. ⚑ The two halves of Migne's own sentence behave differently, which is what makes this decisive rather than a stylistic impression: sordescat adhuc is the Clementine verbatim, and the very next clause, introduced by his sed etiam, is not — a different verb (fiat for justificetur) and a comparative (justior) the Vulgate does not have. English: "let the just become more just," [var: Vulg. Apc 22:11 qui justus est, justificetur adhuc, "he that is just, let him be justified still"].

2d · 0764B/C · Gn 3:14 — terram manducabis for terram comedes

Gn 3:14 … super pectus tuum gradieris, et terram comedes cunctis diebus vitae tuae. A different verb, not a form of the same one. English: "thou shalt eat earth," [var: Vulg. Gn 3:14 terram comedes, "thou shalt eat earth"] — the note gives the received reading; the sense of the two is the same, which is why the marker states the words and argues nothing (Pattern 14's "never speculate about why they differ").

2e · 0769D · Jo 12:31 — Princeps hujus saeculi misus est foras

Jo 12:31 Nunc judicium est mundi : nunc princeps hujus mundi ejicietur foras. Two divergences at once: saeculi for mundi, and a perfect for the Clementine's future ejicietur. The gloss needs the perfect — Christ exclusit inde principem, already. English: "The prince of this world was sent forth." [var: Vulg. Jo 12:31 princeps hujus mundi ejicietur foras, "the prince of this world shall be cast out"]. On misus itself see §4a.

2f · 0770A · Ps 9:9 — ⛔ WITHDRAWN 2026-08-28. THE DIVERGENCE NEVER EXISTED.

Not corrected — withdrawn. The plate collation of 2026-08-28 read col. 0770A and Migne prints judicabit, exactly the Clementine's own reading at Ps 9:9. Corpus Corporum had changed it to judicavit, and this collation, run against a twin that had already been moved off the plate, "found" a divergence that is not on the page. The [var: Vulg. Ps 9:9 judicabit, "shall judge"] has been removed from 0008.md, and the English tense corrected with it: "hath judged the world in equity" → "shall judge the world in equity".

⚠ The reasoning printed above — the perfect against v. 8's in occulto — was constructed to explain a reading Migne never printed. No lemma collation could have caught it; only the plate could. Recorded, not deleted, because the failure mode is the finding: a [var:] fired off a transcription is a public claim about Migne resting on Corpus Corporum.

2g · 0770B · 1Cor 3:17 — Templum Domini for Templum enim Dei

1Cor 3:17 Si quis autem templum Dei violaverit, disperdet illum Deus. Templum enim Dei sanctum est, quod estis vos. Domini for Dei, in a proof-text adduced for the Church as God's dwelling. English: "The temple of the Lord is holy, which you are," [var: Vulg. 1Cor 3:17 Templum enim Dei sanctum est, "for the temple of God is holy"].

2h · 0772D/0773A · Ps 9:31 — humiliavit for humiliabit eum

Ps 9:31 In laqueo suo humiliabit eum ; inclinabit se, et cadet cum dominatus fuerit pauperum. Perfect for future, and the object eum dropped, in the verse address itself. English: "In his net he hath humbled, etc." [var: Vulg. Ps 9:31 In laqueo suo humiliabit eum, "in his net he shall humble him"].

2i · 0773B · Ps 9:35 — ut tradat eos in manus suas for ut tradas eos in manus tuas

Ps 9:35 Vides, quoniam tu laborem et dolorem consideras, ut tradas eos in manus tuas. The Clementine is second person addressed to God; Migne prints third person with a reflexive, and the gloss reads it that way throughout (quasi laborare et dolere timeat, et propterea illis parcat). Two words changed consistently, which is what makes it a text and not a slip. English: "that he may deliver them into his own hands," [var: Vulg. Ps 9:35 ut tradas eos in manus tuas, "that thou mayest deliver them into thy hands"].

2j · 0773C · Mt 23:9 — Ne dicatis vobis patrem in terris

Mt 23:9 Et patrem nolite vocare vobis super terram : unus est enim pater vester qui in caelis est. A different verb (dicatis for vocare), a different negation construction (ne + subjunctive for nolite), and in terris for super terram. English: "Call not to yourselves a father upon earth," [var: Vulg. Mt 23:9 Et patrem nolite vocare vobis super terram, "and call none your father upon earth"].

§3 · SITES INSIDE THE GLOSS PROSE — where the lemma list does not reach

The brief is explicit that a named-lemma list does not cover the sentences between the lemmata (test 2's own worked case, 8987 @0624B). These are the ones I stopped on.

3a · 0763B · Migne re-quotes his own lemma in a form he did not print

The verse address prints inveteravi inter omnes inimicos meos, — the Clementine verbatim. Four lines later the gloss picks the phrase up again as In omnibus autem inimicis meis, — a different preposition and a different case, in omnibus for inter omnes. Nothing in the plate is broken and both are good Latin. ⚑ The danger here is 7a″ in its purest form: the natural move is to render the second occurrence with the same English as the first ("among all my enemies") and let the divergence vanish. The English keeps them apart — "I have grown old among all my enemies" for the lemma, "by in all my enemies he means…" for the re-quotation — so the reader can see that Migne's gloss is expounding a wording his own lemma does not carry. No marker: this is Migne against Migne, not Migne against the Vulgate, and Pattern 14 is for the latter.

3b · 0765C · Justum adjutorium glossed by Justum auxilium

The verse address prints VERS. 11.-- Justum adjutorium, (Ps 7:11, verbatim). The gloss then expounds it twice as Justum auxilium / justum auxilium — a synonym, italicised as though it were the lemma. Kept distinct in the English (help / aid) for the same reason as §3a. No marker.

3c · Negation fidelity — the sites I named a host for (test 2 and 2a)

Every printed negative was traced to the single Latin word it attaches to, and the English checked to carry it on that word — not merely to contain a negative. The sites where a quantifier, correlative or predicate sat next to a plausible verb, which is 2a's trigger population, were:

• 0762B ut non solum mortem non timeat, sed nec urentis … manus sustineat — the first non is on solum, the second on timeat, the third (nec) on sustineat. Three negatives, three hosts. English: "may not only not fear death, but may not even have to bear the hand…".
• 0763C usque adeo non erubescunt impii, ut nobis insultare non desinant — on erubescunt and on desinant. English keeps both ("so far from being ashamed … do not cease").
• 0764A tanquam cui non resistat jam perfecto nisi invidus diabolus — on resistat, restricted by nisi. English: "none resists but the envious devil".
• 0765A quoniam non solum dictum est sordibus … sed etiam — on solum, and the sed etiam is the structural proof of it (§2c).
• 0765C in renibus nullas inferiores delectationes — on delectationes, a quantifier, not on the participle prospiciens. English: "in the reins no lower delights".
• 0766B ut non tua, sed sua malitia puniantur — ⚑ the sharpest 2a site in my range. The negative stands before nothing but the possessive tua, and the sed sua correlative proves the host from structure: the claim is not that sinners go unpunished by God, but that the malice that punishes them is theirs and not his. English: "punished not by thy malice, but by their own", with the negative on the possessive.
• 0767D quamvis non omnis homo Filius hominis possit intelligi — on omnis, not on possit. English: "although not every man can be understood as the Son of man". Putting it on the verb would make the sentence deny what the next line affirms of Christ.
• 0768B ipsi homines non ob aliud boves dicti sunt, nisi quod… — on ob aliud. English: "for no other reason than that".
• 0768C non hominum tantum, sed etiam angelorum — on tantum.
• 0770B Non enim convertitur anima … nisi cum / Nec opportunus … avertitur, nisi — on the two verbs, each restricted by nisi.
• 0770C ⚑ non sine magno fructu aeternitatis morituros … Christianos — a litotes, and the host is the prepositional phrase sine magno fructu, not the participle morituros. English: "Christians will die not without** a great fruit of eternity", which keeps the negative on the phrase; "will not die … with a great fruit" would invert it.
• 0771C in quibus non solum non metuitur reprehensor, sed etiam laudator auditur — two negatives, on solum and on metuitur, with sed etiam fixing the scope.
• 0774A Homines autem sunt non solum de quibus … sed etiam ille — on solum.

3d · The unexpressed subject — every supplied pronoun, and what fixed it

The brief names this the dominant defect class, and a psalm gloss changes speaker constantly. The ones that were genuinely open:

• 0762D–0763A — deprecatur … molitur … contexit … conatur … cogitur. The subject is anima, printed at tutam se praeberi anima deprecatur, so the English says "she" from there to the end of the verse, and switches to "he" at quem lacrymis lavat, qui se ab illo conatur extrahere where Migne prints a masculine relative. The switch is Migne's, not mine.
• 0763D — quia gementem et tam saepe diuque flentem Deus exaudivit, intelligatur liberata … orare. Both participles and liberata are feminine, so the subject is the anima pia of 0763B and not the impii who are the nearest plural. English: "she is understood, being set free from her sins, to pray even for her enemies".
• 0763D/0764A — qui licet vitio suo ceciderint … tamen in natura sua boni a Deo creati sunt. The relative before it is singular (Qui etiam filius Jemini … scribitur) and the verbs here are plural. Migne's own number shift, carried: "Who is also written son of Jemini … who, although they fell by their own fault". Rendered as printed and not smoothed in either direction.
• 0764D — quibus poena est, cum aufertur ab eis ista possessio, in qua dominari desiderant. The datives and desiderant are the devil and his angels, not the sinners who are the possession: it is the possessors who want to lord it. English names them by keeping "to whom" attached to diabolum et angelos ejus.
• 0765D — qui ipsos, qui persecuti sunt, non statim post resurrectionem ad supplicium rapuit. Subject = God the judex patiens; object = the persecutors. English holds both.
• 0771D — ⚑ Multum irascitur, dum non exquirit, dum quasi obliviscitur…. The subject is God, and the sentence is Augustine's paradox: God's anger consists in not punishing. Read with the Antichrist as subject it becomes nonsense, and it construes either way at a glance. The closing clause Sed quanta ista ira Dei sit docent sequentia names the subject explicitly and settles it.
• 0773A — dum animus humanus … parci sibi putat, cum excaecatur. Subject = animus humanus, printed. English: "the human mind … thinks itself spared, when it is blinded".

§4 · PLATE CANDIDATES — logged, and NOT marked

⛔ Zero [sic:] in these five chunks, and that is a decision. Brief addendum A measured the split 10 for 10: a suspicion aimed at a non-word came back as our digitization's corruption, not Migne's, every time — which is the class that made seven false [sic:] withdrawable in the 2026-08-23 batch, each publicly blaming Migne for Corpus Corporum's error. The two sibling Glossa OT books settled the practice for this series (8951 §, 8952 §2): render the evident word, log the crux, fire nothing, exactly as Ruth's significaRuth was handled by a tei-patch and never by a marker. I have followed that. All three below are worth the read; none is asserted on the page.

4a · 0769D · misus for missus — one dropped letter inside a scripture lemma

Princeps hujus saeculi misus est foras. misus is no Latin word. The reading missus est foras is the form Augustine quotes throughout, and it is what the gloss needs (exclusit inde principem). ⚑ Doubly worth the read because it sits inside a lemma that is independently divergent (§2e): if the plate reads missus, the [var:] stands unchanged and the defect is ours; if the plate reads misus, this is a [sic:] site and the [var:] should say so. Rendered "was sent forth", which keeps missus visibly distinct from the Clementine's ejicietur that the marker quotes.

4b · 0763D · potet for potest — a real form with no grammatical slot

quod sine confusione et conturbatione fieri non potet. potet is a real word — third singular present subjunctive of poto, "let him drink" — which is exactly why it is dangerous: nothing looks broken. But poto cannot govern the infinitive fieri, so the form has no slot in its own clause (Pattern 12's first test), and the clause's only possible verb is potest. Rendered "cannot be done", crux logged, nothing fired. ⭐ Recorded here rather than passed over precisely because addendum A's finding runs the other way for real words: a real word that reads oddly is Migne's until a plate says otherwise. This one does not read oddly — it does not read — which is what separates it from the refuted cases.

4c · 0762B · justorum in septem dierum justorum repetitione — suspiciously placed

Et quoniam omnia tempora septem dierum justorum repetitione volvuntur. The sentence says all times revolve by the repetition of seven days; justorum has no work to do there and no antecedent in the clause (the just have been named in the previous sentence, animae justorum, which is how an eye-skip of exactly this shape happens). Rendered as printed — "the repetition of the seven days of the just" — under Pattern 7. Real word, so no marker and no conjecture in the text, per addendum A. Worth one line of a plate read while the column is open.

4d · Three roughnesses rendered literally, no candidate raised

• 0764B qui acceptis beneficiis reddidit mala pro nobis — the antithesis wants pro bonis ("evil for good"), and pro nobis is odd in a clause about a private injury. Real words, and the sentence construes, so it is rendered as printed ("rendered evils for us") and nothing is fired. Raised here so a blind reader does not think it unseen.
• 0766B Verus justus es, Domine where Vere justus es is the expected idiom. Verus is a real word and the phrase construes as "Thou art the true just one", which is what the English prints. No marker.
• 0768C quo tritus idem sermo referendus est. Rendered "the same discourse, now threshed out" — which is not merely tolerable but apt, since the whole chunk has been expounding threshing-floor and winepress. No candidate.
• 0770B Nec opportunus ab hoc saeculo avertitur — a nominative adjective where an adverb (opportune, which Migne uses four lines earlier) would be expected. Rendered "Nor is one seasonably turned away". Gender/case oddity with no clean English exponent; no marker, and Pattern 9's reasoning applies.
• 0769A Siquidem et victoriam feret de mundo, finis est futurorum bonorum. The two clauses do not join cleanly. Rendered literally, unsmoothed.

§5 · CANDIDATES WEIGHED AND REJECTED — with the reason each lost

⛔ Brief §5's rule: naming a rival is not adjudicating between them. Each of these was a real divergence from the Clementine that I considered marking and did not.

• 5a · Zach 1:3 @0762C. Migne drops exercituum twice and the whole interposed ait Dominus exercituum. Rejected: omission, not substitution. Nothing is said that the Clementine does not say, and the gloss adduces the verse for the reciprocity of turning, which both texts carry identically. A [var:] here would report an ellipsis as a variant.
• 5b · Ps 33:19 @0763B, prope est Dominus obtritis corde. The Clementine reads Juxta est Dominus iis qui tribulato sunt corde — wholly different wording, and this was the strongest rejected candidate. It lost because it is not the psalm under exposition: it is adduced in passing to license the claim that the prayer was heard, and the sense of the two forms is the same ("the Lord is near to the broken-hearted"). ⚑ The reading Migne prints is the Roman/Old-Latin form of the verse, and Pattern 14 forbids speculating about why two texts differ, so a marker here would either say nothing useful or say more than I can support. Recorded in full instead, which is where the finding belongs.
• 5c · Ps 7:8 @0764D, circumdedit for circumdabit te. Perfect for future. Rejected: the word has been absorbed into the gloss's own syntax — ita eum circumdedit, ut verissime diceretur — where a future indicative could not stand. Contrast §2f and §2h, where the tense change sits in a lemma quoted as the verse, not in a clause of the gloss's own making. That is the line I drew between the fired and the rejected tense cases, and it is the same line as §3a's: Migne recasting his own citation is not Migne's text diverging.
• 5d · Rom 5:8–9 @0765C. Migne adds si, drops secundum tempus and in sanguine ipsius, and reads quanto magis for multo igitur magis. Rejected: the divergences are a clip and a connective, and the a fortiori the gloss is arguing from survives all of them untouched. The dropped in sanguine ipsius is the one I hesitated over; it is a real loss, but nothing in the gloss touches the blood, and Pattern 14 asks for a divergence that changes the sense of this argument.
• 5e · 2Cor 2:16 @0766A. Migne prints aliis sunt odor vitae in vitam, aliis odor mortis in mortem; the Clementine has death first. Rejected: order only, both members present. ⚠ Order reversal is a real class (7914 @1079B's cinis et pulvis), and it was weighed on that ground; it lost because that case turned on a gloss that expounded ashes, and here the gloss expounds neither member — it uses the pair whole, to explain how one bow shoots both apostles and heretics. The English keeps Migne's order.
• 5f · The ex Hebraeo citations @0767D, 0768C–D, 0769B, 0772A. Minues eum paulo minus a Deo · Victori super morte filii · victori pro morte filii · Canam nomini tuo, Altissime · Cum ceciderint inimici mei retrorsum… · Loquitur in corde suo: Non movebor… — not collated as Vulgate lemmata at all, and no [var:] is possible on any of them. Migne labels every one (in Hebraico legitur, ex Hebraeo translata editio, sicut Hebraica docet editio, Hebraica translatio): they are Jerome iuxta Hebraeos, adduced precisely because they differ from the Gallican he is expounding. ⚑ Firing a Vulgate variant on a reading the author has just told you is not the Vulgate would be the marker asserting as a divergence what the page already prints as one.
• 5g · 1Cor 15:27 @0768A, Excepto eo qui ei subjecit omnia for praeter eum qui subjecit ei omnia. Rejected: a synonymous prepositional phrase and a word-order swap. The exception clause is the whole point and it is identically made.
• 5h · Mt 16:23 @0769B, Redi retro, Satanas for Vade post me Satana. A different verb and a different adverbial. Rejected because the gloss's argument is about the retro, and the retro is Migne's own word here — retrorsum conversus est, retro factus est, retro esse facit, three times in four lines. A [var:] would be true, but the reading it reports is not the reading the argument uses, and the Clementine's Vade post me is itself a "behind me". Logged rather than marked; ⚠ I record it as the closest call of the four rejected NT citations, and a later reader may reasonably reopen it.
• 5i · Phlp 3:13 @0769B. Migne has et se ad ea quae ante sunt extendens for the Clementine's ad ea vero quae sunt priora, extendens meipsum. Rejected: this is a quotation woven into the gloss's own third-person sentence (qui … portaverit imaginem coelestis, ea scilicet, quae retro sunt obliviscens), so se for meipsum is required by the syntax the gloss has built, and ante for priora follows it.
• 5j · Ps 9:8 @0770A, judicio sedem suam where the address four words earlier prints thronum. Same class as §3a and §3b — Migne re-quoting himself with a synonym. Kept visible in the English ("throne" / "seat"), not marked.
• 5k · 2Thes 2:3 and 2:4 @0770D, 0772A. Cum revelabitur homo peccati for et revelatus fuerit homo peccati; and in templo Dei sedebit; extollens se super omne quod colitur et quod dicitur Deus for qui adversatur, et extollitur supra omne, quod dicitur Deus, aut quod colitur, ita ut in templo Dei sedeat ostendens se tamquam sit Deus. The second is a substantial reworking. Both rejected: Migne flags them as paraphrase himself — de quo Apostolus dicit, etiam secundum Apostolum — and the second reorders and compresses rather than reading a different text. Recorded in full so a reader can see how far it travels.
• 5l · Ps 9:28 @0772B, Cum amaritudine et dolo for the Clementine's et amaritudine, et dolo. Rejected: this is the gloss picking the phrase up as a phrase, and cum + the ablative is how it must be picked up once detached from plenum est. The verse address three lines above prints Cujus maledictione os plenum est, which is verbatim.
• 5m · Ps 9:34 @0773B, non requiret Deus for the Clementine's bare Non requiret. Rejected: Migne supplies the subject inside the italic, which is an expansion for clarity in a sentence that has been about God for a page, not a variant reading.
• 5n · Ps 9:36 @0773C, nec invenietur for et non invenietur, and peccatum ejus for peccatum illius. Rejected: one negation written as one word instead of two, and a pronoun synonym. Both flagged only because the brief asks for the negation sites to be named; the English carries the negative either way.

⭐ Two negatives worth recording as negatives (addendum §4's point that a checked negative is worth having): I looked for the normalization class — a form suspiciously tidy in a rough passage — and found no candidate in these five chunks; and I ran addendum D on usque ad, which occurs five times in cols 0762B–0774A and is the abridgment formula in none of them: two are usque adeo ("so far", 0763C usque adeo non erubescunt and 0771D usque adeo beatus videbitur), and three are ordinary prepositional uses with a real object — 0763A fletum usque ad cordis intima, 0767A ut sonus usque ad aures valeat, 0767C Scripturae praedicatae usque ad infantium tarditatem. Each was decided from what stands on either side of it, per occurrence, and none was allowed to swallow a lemma's own words. Both checks were run and both came back negative.

§6 · ITALIC-SPAN BOUNDARIES — three spans in the digitization do not fall where the verse does

All three are reproduced in the same place in the English, per the brief's rule that every italic span in the Latin is an italic span in the English. None is patched, none is marked. They are logged because the resulting English reads oddly and a reader is entitled to know the oddity is in the file and not in the translation.

• 6a · 0766A · Vasa vero mortis vel haeretici intelliguntur. The span runs one word past the verse: vel is not part of Ps 7:14 but the first limb of the gloss's own vel … vel ("either heretics … Vel the same apostles"). English: "But by vessels of death, either heretics are understood … Or the same apostles". ⚑ Reading vel as part of the lemma instead of as the correlative would break the sentence's structure, and the second Vel, italicised alone four lines later, is what proves the pairing.
• 6b · 0768C · De occultis filii quaeri potest. The span opens with the gloss's own preposition De; the psalm title (9:1) reads in finem pro occultis filii, and Migne prints exactly that, unitalicised pro and all, ten lines later. English: "Concerning the hidden things of the son it can be asked."
• 6c · 0772C · Beati pauperes spiritu. Leonem autem in spelunca. One span runs across a full stop, joining the end of a Matthew quotation to the first word of the next sentence. Reproduced as one span. The same shape recurs at 0773C, omnium inimicorum suorum dominabitur. Brachium ergo ejus dixit potentiam ejus, and at 0774A, judicare pupillo. Judicat [0774A] pupillum. Three instances of one class, which is what makes it a markup artifact rather than three accidents.
• 6d · Two spans were split by English word order and one merged, all recorded because band-level asterisk parity is how they were found: secundum innocentiam suam (0765A) is rendered "according to the innocence that is his own" so that the roman suam stays outside the span; Filium autem hominis (0767D) keeps its two spans as "the Son of man"; and maledictione os / plenum est (0772B) is rendered "cursing … mouth is full", which moves os from the first span into the second because English cannot put "mouth" anywhere else. That last one is the only span whose contents I redistributed, and it is stated here rather than left to be discovered.

§7 · PUNCTUATION CARRIED AS PRINTED

• 0765C · prospiciens, non nos acquiescere carni et sanguini, sed delectari in Domino — ⚠ CORRECTED 2026-08-28 at the plate: Migne prints a COMMA here, not a full stop. The full stop was Corpus Corporum's. The note printed here in the original crux — "a full stop in the middle of a non … sed correlative" — was therefore an observation about the transcription, not about the page, and the English had duly broken the sentence in two ("…flesh and blood. but are delighted…"). The English now runs on: "…that we do not acquiesce in flesh and blood, [0765C] but are delighted in the Lord". The column anchor still falls at the same point, now inside a comma-joined clause. Pattern 8 cuts both ways: never supply a mark Migne does not print.
• 0763C · quia etiamsi pariter atque iisdem conventiculis continentur; tamen — a semicolon where a comma would be expected; carried.
• 0766B · sed justitiae Dei, ita loquimur: — a comma splice; carried as a comma.
• 0765B · quod anima justa est: Et inquam — a capital after a colon; carried.
• 0770C · cum infidelitas gentium saevitura sit? convenienter sequitur — a question mark closing an indirect question, with a lower-case word after it. ⚑ Pattern 8b was considered and is NOT needed here: the question the mark falls on (quomodo annuntiabitur…?) is already direct, standing behind the colon of quaereret dicens:, so English can take the mark exactly where Migne puts it. The 8b recasting is for a mark that English grammar refuses, and this one it does not refuse. No colon was inserted and nothing was moved.
• Question-mark counts per chunk (2 · 1 · 1 · 7 · 2) match the Latin twins band for band. The seven in chunk 0008 include the two one-word interrogatives Quare? (0769C) and Sed quare? (0771C), both carried as printed.

§8 · TEST 5 — does the English assert what he asserts

Run last, with the counts already clean, paragraph by paragraph against the Latin. Three readings were changed as a result, all of them cases where my first draft read more smoothly than the Latin does:

1. 0765C — Scrutans ergo cor nostrum, et prospiciens, non nos acquiescere carni… My draft read "and looking upon it", supplying an object for prospiciens that the Latin does not have; prospiciens governs the accusative-and-infinitive that follows. Changed to "and looking forward, that we do not acquiesce…". Nothing was missing in the draft — that is the point of the test.
2. 0762D — In qua morte, hoc est, esse immemorem Dei … ut infernum dixerit is an anacoluthon: the ut clause has no main verb to depend on. My draft had quietly supplied one by turning the parenthesis into a dash-bounded aside. Rewritten with commas, and the broken join left broken.
3. 0771D — Multum irascitur, dum non exquirit… was drafted with the Antichrist as subject, which construes perfectly and inverts the paragraph. Repaired from the sentence's own closing clause (quanta ista ira Dei sit). See §3d.

The remaining paragraphs were read back and assert what Migne asserts. The two places where I would send a reader first if I am wrong are §4a (misus, inside a lemma that already carries a [var:]) and §5b (Ps 33:19, the largest divergence in my range on which I fired nothing).

PLATE COLLATION — markers withdrawn

2026-08-28. All 22 pages read at Migne's plate; 61 TEI patches applied; the Latin twins for these chunks now say what Migne printed. Chunk boundaries did not move.

Withdrawn — one, and it is the worst class of error this collation found.

• 0770A · Ps 9:9 · [var: Vulg. Ps 9:9 judicabit, "shall judge"] — WITHDRAWN. ⛔ The divergence never existed. The plate prints judicabit, exactly the Clementine's reading; Corpus Corporum had written judicavit, and the collation of §2f, run against that twin, reported a departure from the Vulgate that is not on Migne's page. Marker removed from 0008.md and the English tense corrected with it: "hath judged the world in equity" → "shall judge the world in equity". §2f above is rewritten in place, not deleted, because the failure mode is the lesson: a [var:] fired off a transcription is a public claim about Migne resting on Corpus Corporum, and no amount of lemma collation can catch it — only the plate can.

Punctuation corrected — §7's first bullet.

• 0765C. Migne prints a comma after carni et sanguini, not a full stop; the sentence runs on into sed delectari in Domino. The full stop was the digitization's. The English had faithfully carried a mark Migne does not print, breaking one non … sed correlative into two sentences ("…flesh and blood. but are delighted…"). It now runs on: "…that we do not acquiesce in flesh and blood, [0765C] but are delighted in the Lord." The anchor sits at the same point. ⚑ Pattern 8 has two halves and only one of them was being enforced here.

Sense changed by the corrected Latin.

• 0764D. Migne prints ut desinat possidere quod possidet (CC: quos). The English said "that he may cease to possess those whom he possesses"; it now reads "what he possesses" — which is also what the surrounding gloss requires, since the thing possessed is the possessio / fines of the next clause, not the sinners as persons.

Recovered Vulgate divergences weighed in this range. Rule applied: fire where the printed reading construes in place and yields a different sense; decline where it yields no sense at all, because that is Pattern 12's business, not Pattern 14's.

• 0764B · Ps 7:4 · [var:] FIRED. Migne prints si fecit istud against the Clementine's si feci istud — a whole change of person, in a psalm spoken throughout in the first person, and Migne's own next clause is si est iniquitas in manibus meis. CC had normalized it to feci. Migne is rendered: "What he says: if he hath done this," with [var: Vulg. Ps 7:4 si feci, "if I have done"]. The strain between the third person of the lemma and the first person of its explanation is Migne's, and is now visible.
• 0771C · Ps 9:23 · [var:] FIRED. Migne sets Comprehendantur in italic — subjunctive — where Clem. Ps 9:23 has the indicative comprehenduntur. CC had supplied the indicative. The word is italicized as the psalm's, so it is a citation and Pattern 14 reaches it; the mood is construable and changes what is asserted (a wish, not a fact). English: "Let them be caught, however, the impious, in the counsels, [var: Vulg. Ps 9:23 comprehenduntur, "they are caught"] because their thoughts become bonds to them." ⚠ Italic parity with the twin is preserved: two spans, autem impii outside them, as Migne sets it.
• 0764A · Ps 7:3 — DECLINED. Migne prints ne quando rapiat at leo animam meam; Clem. ut leo. at is a real word but produces no clause here, and Migne prints the same lemma correctly (Ne quando rapiat, etc.) three lines above. Broken type, not a variant reading; English keeps "as a lion", unmarked, per the standing out-of-scope ruling on this work's restored typographic defects.

PLATE COLLATION · REGISTER OF RESTORED TYPE, cols 0762B–0774A

Wilson's ruling of 2026-08-28 governs (stated in full at the head of the matching section in cruces-0000.md): mark only where the reading is in doubt; record the rest here and disclose the count in the work-level page note.

MARKED in this range — two.

• 0766A · Providentia for Providentiae — [cj: Providentia; read Providentiae, "to the Providence"]. Migne prints et ob hoc divinae Providentia etiam ista disposito tribuenda est. tribuenda est takes a dative, and divinae is already dative and left stranded by a nominative. The dropped e leaves a clean, ordinary Latin nominative behind, so the Latin page gives the reader no signal; the English ("must be ascribed to the divine Providence") was rendering CC's repair. Marker placed after the sentence, before Or the same apostles.
• 0771A · Filium homines for Filium hominis — [cj: Filium homines; read Filium hominis, "the Son of man"]. Migne prints qui nolunt liberari a Filio Dei, et pertinere ad Filium homines, et esse filii hominum. A nominative/accusative plural where the genitive singular is required; it does not construe, and homines is a real word standing in a real position (the same sentence ends quoniam homines sunt), so nothing marks it as broken.

Already marked, and standing: 0764B [var: Vulg. Ps 7:4 si feci, …] and 0771C [var: Vulg. Ps 9:23 comprehenduntur, …], both argued above.

CRUCES ONLY — recorded here, unmarked in the text.

Orthography, not defective type — no marker and no conjecture. 0771A Inploratur is Migne's unassimilated spelling of imploratur: correct Latin for its period, Pattern 9 exempt, and recorded here only so that a reader collating the twin is not left to wonder.

Punctuation in this range is dealt with above (0765C, the comma Migne prints where the digitization had a full stop).

THE BLIND POLARITY READ

Fenced reader on chunks 0005–0009 (Pss VI–IX, cols 0762B–0774A), reading Latin against English in full with no crux file and no collation report open. Adjudicated 2026-08-28 under Wilson's marker ruling of the same session (stated in full in cruces-0000.md). Whole-work figures are in that file.

Result for this range: 1 site, ours; 0 markers placed.

Repaired

• 0774A, VERS. 39 — extulit patrem is a burial, not an exaltation. Migne: qui exuit veterem hominem, atque hoc modo tanquam extulit patrem. Our English read "as it were exalted the father". Efferre here is the funeral idiom (funus efferre, patrem efferre) — carried his father out to burial — and the control is two paragraphs back in the same run at 0773C, where the expositor defines the pupillus: Orphanum autem dixit, cui moritur pater hic mundus, per quem carnaliter genitus est, and Talibus enim orphanis pater fit Deus. The clause is supplying the reason this man qualifies as the orphan the Son of man judges for. Tanquam is the tell: there is a metaphor to flag in burying a father one never literally had, none at all in exalting one — and the old reading made exuit veterem hominem a stripping-off and an elevation at once. Now: "and in this way as it were carried his father out to burial." 7a mechanism 1; no lexical trace, and verify-english cannot see it.

Adjudicated and DECLINED

• 0769C Princeps hujus saeculi missus est foras → "was sent forth". Momentarily ambiguous in English (it can read as commissioning), but the governing clause immediately before — cum exclusit inde principem, "when he shut out thence the prince" — fixes the direction for the reader before he reaches it. A clarity question, not a polarity one.
• 0770A paravit in occulto judicio sedem suam. Our distribution ("prepared in secret his seat in judgment") and the reader's ("his seat in the hidden judgment") assert the same thing, which is the sentence's own contrast with palam manifesteque venturus est.
• 0764B reddidit mala pro nobis — almost certainly Migne's slip for pro bonis, the reading the maxim needs. Our English renders what is printed and does not repair it, which is 7a-correct. No marker:* the intended word is not in doubt (mala pro bonis is the maxim's own shape and acceptis beneficiis* stands three words earlier), so it falls on the cruces-only side of Wilson's ruling.

Positive results worth recording

• ⭐ The 7a″ profile is ABSENT from this range. Every quoted verse in five chunks was read word by word against the plate, and no lemma was found conformed to the familiar English. Where Migne diverges, the English diverges too and the divergence is already marked: Miserere mei, Deus (0762C), si fecit istud (0764B), terram manducabis (0764D), Consummetur nequitia and justus justior fiat (0765B), Templum Domini (0770B), Comprehendantur (0771C), humiliavit (0773A), ut tradat eos in manus suas (0773B), Ne dicatis vobis patrem in terris (0773C). For a scripture commentary — the genre whose predicted dominant failure mode 7a″ is — that is the notable finding of the read, not an absence of one.
• Every printed negation in the five chunks is carried, including where carrying it leaves the sentence odd: usque adeo non erubescunt … ut … non desinant keeps the idiom's double negation rather than smoothing it. No negation appears in the English that the Latin does not print.
• 0769C Convenientius hoc Domino Jesu Christo dici, quam ipsum dicere — passive to the Lord, active by him — is rendered in the right direction ("said to the Lord … than said by him"). This is the exact 7a mechanism-5 shape, verified rather than assumed.

Preserved unedited from cruces-0010.md; headings demoted one level. Where it conflicts with the head above, the head governs.

9003 · Expositio in XX primos Psalmos — cruces for chunks 0010–0013 (cols 0774B–0784A)

Range: IN PSALMUM X (from 0774B) through IN PSALMUM XVI (to 0784A) — Psalms 10, 11, 12, 13, 14, 15 and 16 by the Vulgate/Septuagint reckoning, which is what Migne prints and what sources/vulgate/clementine-flat.txt uses. Nothing below renumbers toward the Hebrew, and the title-as-verse-1 offset is not treated as a divergence anywhere.

Apparatus placed in this range: 16 [var:] · 1 [sic:] · 4 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [n:] — ⚠ [cj:] count revised 2026-08-28, was 0 at filing: the plate collation placed one @0774B (apud ipso) and the blind polarity read three more (@0779D, @0781A, @0775A); see THE BLIND POLARITY READ at the foot of this file. (the work has noteCount: 0 throughout, as expected). Zero guillemets in either language. Band-by-band punctuation and italic-span counts are 1:1 with the Latin twins in all four chunks with one deliberate exception, recorded at §4.1.

§0 · WHAT WAS READ AT THE PLATE, AND WHAT IT FOUND

The on-disk scan (raw/scans/pl114/patrologiaecurs06saingoog.pdf) was read at 300–600 dpi for four pages covering the whole of cols 773–784 that carry my candidates: p. 392 (cols 773/774), p. 393 (cols 775/776), p. 395 (cols 779/780), p. 396 (cols 781/782). Page map used and confirmed against the printed corner numbers on p. 392 (773 / 774): PDF page = (column + 11) / 2, the figure coverage.json already records for this volume.

⛔ The single most important finding in this batch is not a translation call. Our Latin for this work diverges from Migne's plate at a high rate — nine sites in four pages — and it diverges in BOTH directions (brief §4). Every one of these was found by reading the whole line for a different word, exactly as addendum C predicts. None is detectable from the Latin alone. Recommend a data/tei-patches/9003.json and a re-chunk after the whole work's agents have filed; I have not patched, because re-chunking mid-fleet would rewrite the chunks the other agents are working from.

0a. Migne's own defective type, silently NORMALIZED by Corpus Corporum (5 sites)

0b. Corpus Corporum's OWN corruptions, where Migne is clean (4 sites)

⛔ THREE OF THESE WOULD HAVE TAKEN A FALSE [sic:]. Dominius, Juant and infirmi ates are textbook Pattern 10/12 shapes — a non-word, a J/l substitution, a split run — and a [sic: …] on any of them would have blamed Migne publicly for the digitization's error, which is the exact class of seven withdrawn markers in the 2026-08-23 batch (brief addendum B). The addendum's rule held 5 for 5 here: every non-word suspicion turned out to be a transcription question, not a plate question. The rule for a non-word in this work is: render it for its evident sense, log it here, fire nothing, and ask for the plate — and where the plate is on disk, as it is for PL 114, just read it. Cost: four page renders.

⭐ Recorded negative: proptera at 0775B was suspected on exactly the same grounds as the four above and came back agreeing with the plate — Migne set it. That negative is what makes the four positives credible, and it is the one candidate in the batch that earned a marker.

§1 · THE TWO SITES WHERE THE PLATE CHANGED WHAT THE ENGLISH SAYS

1.1 · 0774B — ✅ RESOLVED 2026-08-28. The patch landed; the English now reads Migne.

Read this box first; the section beneath it is the record of the compromise, not the current state. The tei-patch the section asks for (montem → mentem at 0774B) was applied in the full plate collation of 2026-08-28. The Latin twin now prints Unum enim mentem teneo, and the English has been changed from "one mountain do I hold" to "one mind do I hold".

⚠ Two things this section proposed that were NOT done, deliberately:

1. No [sic: mentem] was placed. §3.1 of cruces-0014 settles the principle for this work: Pattern 12 excludes "a real word that is merely wrong", and mentem is a real word. A [sic:] here would be the same false public claim the collation has just withdrawn four times over.
2. No [cj:] either. Pattern 18 renders Migne and proposes the emendation; but here the instruction was to render Migne's mentem ("mind"), so a marker reading "read montem, 'mountain'" would sit beside an English word that already says something else. The incoherence goes in the crux, which is this paragraph.

What is genuinely wrong on Migne's page, and is now visible in both languages: Unum is masculine/neuter and mentem feminine, and the relative in quo two words later demands a masculine antecedent — the mons of the sentence before (etiam mons appellatus est), which is what the heretics' Ps 10:2 Transmigra in montem sicut passer is about. The argument of the gloss wants montem; the plate prints mentem; we print what the plate prints and say so here. The English keeps "in which", which is genderless and therefore neutral between the two.

1.1 (superseded) · 0774B — Migne prints mentem; our Latin twin prints montem. RENDERED FROM THE TWIN, AND THAT IS A COMPROMISE.

Printed on the plate: Unum enim mentem teneo, in quo confido, quomodo dicitis… Our TEI: Unum enim montem teneo…

The gloss's whole argument is that Christ is called mons (etiam mons appellatus est), and the heretics quote Ps 10:2's Transmigra in montem sicut passer. Migne's mentem does not construe — Unum is masculine/neuter, mentem feminine, and the relative in quo demands a masculine antecedent. So the digitization did not corrupt this word; it repaired it.

I have rendered "one mountain do I hold", following the Latin twin, and I want the compromise on the record rather than hidden:

• Pattern 7 says render what Migne prints. Migne prints mentem.
• Pattern 12 says a [sic: …]'s content must appear verbatim in the Latin twin. mentem does not, so the marker is not merely unwise here, it is impossible — the verifier would reject it, and rightly, since that check is what stops us gluing our own conjectures onto Migne's page.
• Rendering "one mind do I hold" against a Latin column reading montem would put an English word on the page with no witness anywhere in our files.

⚑ The fix is a tei-patch, not a translation choice. Patch montem → mentem at 0774B with this crux as provenance, re-chunk, then the English becomes "one mind do I hold [sic: mentem]" with the montem conjecture in the crux. Until that happens the English on this line is faithful to our Latin and not to Migne's, and no reader can tell. That is the whole reason the finding is at the top of this file.

1.2 · 0775B — proptera, and the marker does NOT swallow the force

Plate and TEI agree: Calicem autem proptera puto appellatum, ne quid praeter modum atque mensuram… fieri arbitremur. proptera is no Latin word (haplography for propterea), so it is carried untranslated under Pattern 10 and marked under Pattern 12:

A cup, however, I think is for this reason [sic: proptera] so named, that we may not judge anything to be done beyond mode and measure…

7a⁗ four-question check, run with the marker's contents deleted:

1. Verb? yes — puto appellatum, "I think … so named".
2. Negation? yes — the ne clause, "that we may not judge".
3. Interrogative/conjunction? ⚠ this is the one that fires. propterea is the forward-pointing half of a propterea … ne correlative; quarantine it and the sentence loses the pointer to its own reason clause. So "for this reason" is rendered in plain words beside the marker, which is 7a⁗ obeyed rather than admired.
4. Object? (the fourth question 7a⁗ does not ask) — yes, Calicem, "a cup", untouched.

Strip-the-marker test: "A cup, however, I think is for this reason so named, that we may not judge…" reads aloud with nothing missing. ✓

§2 · THE 7a″ COLLATION — EVERY LEMMA AND EVERY QUOTED SCRIPTURE IN COLS 0774B–0784A

265 italic spans were listed for this range. What follows names the ones that were checked against sources/vulgate/clementine-flat.txt and the ones that came back clean, because a findings-only list cannot be distinguished from no list at all.

2a · READ CLEAN — Migne agrees with the Clementine word for word

These were each collated against the cited verse, not merely against "somewhere in the Clementine" (which is all the pre-scan's ✓ means). All agree exactly, punctuation apart:

Psalm 10: In Domino confido, quomodo dicitis (10:2) · montem (10:2) · Quoniam ecce peccatores (10:3) · Quoniam quae perfecisti destruxerunt / destruxerunt / quid fecit (10:4) · Dominus in templo sancto suo / templo sancto suo / Oculi ejus in pauperem / Filios / hominum / palpebrae / interrogant / coelum (10:5) · Dominus interrogat justum et impium (10:6) · Pluet super peccatores / Pluet / laquei / igne / pars / calicis eorum / Calicem (10:7) · Quoniam justus Dominus / justitiae / aequitatem vidit vultus ejus / aequitas (10:8). Psalm 11: Octava / octavam (title, 11:1) · Salvum me fac, Domine, quoniam / Defecit sanctus / veritates (11:2) · Vana locuti sunt unusquisque / Proximum / in corde et corde (11:3) · Disperdat Dominus universa labia / Universa / Linguam / magniloquam (11:4) · Qui dixerunt linguam nostram (11:5) · Propter miseriam inopum / inopes et pauperes / Ponam in salutari / In salutari / Fiducialiter (11:6) · Eloquia Domini, eloquia casta / Casta / probatum terrae (11:7) · Tu, Domine, servabis nos (11:8) · In circuitu impii ambulant / In circuitu ambulant / secundum altitudinem (11:9). Psalm 12: Usquequo, Domine, oblivisceris me in finem? / Avertere / faciem (12:1) · dolorem in corde meo per diem (12:2) · Usquequo exaltabitur inimicus / Inimicus (12:3) · Respice / Illumina oculos meos, ne unquam / oculos / illuminari (12:4) · Qui tribulant me exsultabunt / motus / motum (12:5) · Exsultabit cor meum in salutari / In salutari tuo / exsultabit cor meum / cantabo / psallam (12:6). Psalm 13: Dixit insipiens in corde suo / in corde / Corrupti sunt et abominabiles / Non est Deus / abominabiles / usque ad unum (13:1) · Dominus de coelo prospexit super filios hominum / filios hominum / Prospicit / Dominus / de coelo / ut videat (13:2) · Omnes declinaverunt / Sepulcrum patens est / Linguis suis dolose agebant / Venenum / aspidum / Quorum os maledictione / Pedes / veloces / Ad effundendum sanguinem / Contritio et infelicitas in viis eorum (13:3) · Nonne cognoscent omnes / Omnibus / cognoscere / Qui devorant / plebem meam sicut escam panis (13:4) · Dominum non invocaverunt / invocat / ubi non erat timor (13:5) · Quoniam Dominus in generatione / Consilium inopis confudistis (13:6) · Quis dabit ex Sion / Quis dabit? / Quis dabit ex Sion salutare Israel? / ex Jacob / Israel / Exsultabit Jacob / laetabitur Israel (13:7) · and Rom 11:26 et sic omnis Israel salvus fieret — exact. Psalm 14: Domine, quis habitabit in / tabernaculum / Mons / In tabernaculo / quis habitabit? / in monte, quis requiescet? (14:1) · Qui ingreditur sine macula et operatur justitiam. Qui loquitur (14:2) · veritatem / corde / locutus est in corde / in corde / In lingua / dolus / Nec fecit proximo suo malum / Proximum (14:3) · Ad nihilum deductus est / conspectu ejus / Timentes autem Dominum glorificat / Qui jurat proximo suo et non decipit / loqui veritatem in corde suo / dolum in lingua sua / et non facere proximo suo malum (14:4) · Qui facit haec, non movebitur in aeternum (14:5) · and Ps 138:17 Mihi autem nimis honorificati sunt amici tui, Deus — exact. Psalm 15: Conserva me, Domine, quoniam speravi in te. Dixi Domino: Deus meus es tu, quoniam bonorum meorum non eges / bonis meis (15:1–2) · Sanctis qui sunt in terra ejus / voluntates meas (15:3) · Multiplicatae sunt infirmitates eorum / infirmitates / acceleraverunt / Non congregabo conventicula eorum de sanguinibus / de sanguinibus (15:4) · Dominus pars haereditatis meae et calicis mei / haereditatem / calicis mei / mei / congregabo conventicula eorum / restitues (15:5) · Funes ceciderunt mihi in praeclaris / ceciderunt (15:6) · Benedicam Dominum qui tribuit mihi intellectum / intellectu (15:7) · Providebam Dominum in conspectu meo semper / providens / a dextris est (15:8) · Propter hoc laetatum est cor meum / Insuper et caro (15:9) · Quoniam non derelinques animam meam in inferno / Notas mihi fecisti vias vitae / Notas fecisti / vias / vitam / adimplebis / laetitia / me adimplebis (15:10) · and Jo 19:19 Jesus Nazarenus rex Judaeorum, Mt 11:29 Discite a me, quia mitis sum, et humilis corde — both exact. Psalm 16: Exaudi, Domine, justitiam meam / justitiam / Non in labiis dolosis / in labiis dolosis (16:1) · De vultu tuo judicium meum prodeat / de vultu tuo (16:2) · Probasti cor meum et visitasti nocte / nox / ignis (16:3) · Ut non loquatur os meum opera hominum / opera hominum (16:4) · Perfice gressus meos in semitis tuis / Perficiatur / vestigia (16:5) · Ego clamavi, quoniam exaudisti me (16:6) · Mirifica misericordias tuas, qui salvos facis sperantes in te (16:7) · A resistentibus dexterae tuae / A resistentibus / custodi me ut pupillam oculi / Sub umbra alarum tuarum protege me (16:8) · Inimici mei animam meam circumdederunt / Inimici / animam / circumdantes (16:9) · os eorum locutum est superbiam (16:10) · Projicientes me nunc circumdederunt me / Projicientes / circumdederunt me / statuerunt declinare (16:11) · Susceperunt me sicut leo paratus ad praedam / Susceperunt (16:12) · Exsurge, Domine, praeveni eum et supplanta eum / Exsurge / supplanta eum / animam meam / framea (16:13) · Domine, a paucis de terra divide eos / Pauci / divisi / a paucis / terra / in vita / absconduntur / adimpleta / Saturati sunt filiis, et dimiserunt reliquias suas parvulis suis (16:14) · in justitia tua / satiabor, cum apparuerit gloria tua (16:15) · and Jo 19:15 Non habemus regem nisi Caesarem, Mt 27:40 qui destruis templum, Jo 8:44 Vos ex patre diabolo estis — all exact.

⚑ The clean list is long on purpose, and it is what makes the divergences below mean something. The Gallican psalter is what the Clementine prints, so these lemmata ought to collate cleanly; the sixteen places where they do not are correspondingly heavier, not lighter.

2b · MARKED [var: …] — 16 divergences, each resting on the file, none on recollection

The firing rule I applied, stated so it can be checked: a [var:] where the divergence is (a) certain from clementine-flat.txt at the verse actually being glossed, (b) lexical or grammatical, not orthographic and not a clipping, and (c) sitting in a lemma or in a scripture the gloss expounds. Everything failing (b) or (c) is in §2c with its reason.

2c · DIVERGENCES FOUND AND DELIBERATELY NOT MARKED, with the reason for each

⚑ These are the rejected candidates. Each is a real difference from the Clementine; none earned a public claim.

1. 0774B Paraverunt sagittas in pharetra — Clem Ps 10:3 has sagittas suas in pharetra. Plate-confirmed as Migne's. Not marked: an omitted possessive is not sense-bearing, and the lemma is clipped (it ends in a comma and the gloss picks it up mid-verse), so the omission is as likely a clipping as a text.
2. 0776B Erat enim docens eos tanquam potestatem habens — Mt 7:29 has sicut. Not marked: tanquam and sicut are synonyms here; no reader takes anything different away. (Mc 1:22's quasi shows the same slot is fluid.)
3. 0777A/B Finis legis Christus — Rom 10:4 is Finis enim legis, Christus, ad justitiam omni credenti. Not marked: it is an incipit used as a chapter motto, not a proof-text, and the omissions are the clipping.
4. 0777B tribuit mihi bona — Ps 12:6 has qui bona tribuit mihi. Not marked: word order only, and the words are woven into the gloss's own syntax.
5. 0777B egit in eo — Ps 12:6 has fiduciali­ter agam in eo (1st sg. future). Not marked: the gloss has recast the verse into the third person about Christ throughout this sentence; the person and tense are the gloss's grammar, not a variant text.
6. 0778A Nunc cognovi, quod timeas Deum — Gn 22:12 has times. Not marked: subjunctive in indirect statement is standard medieval usage and English has no exponent for it here (Pattern 9's reasoning). Rendered "that thou fearest God".
7. 0778C non cognoverunt viam pacis — Ps 13:3 has et viam pacis non cognoverunt. Not marked: order only, inside the gloss's Unde … clause.
8. 0780B opprobrium non accipere adversus proximum suum — Ps 14:3 has opprobrium non accepit adversus proximos suos (plural). Not marked: Migne has recast the whole catalogue into infinitives here, and the singular agrees with his own gloss two paragraphs above, Proximum omnem hominem accipi oportere.
9. 0781D Delectatio — Ps 15:10 has delectationes. Not marked: the gloss needs a singular subject for est; this is its syntax.
10. 0783A Qui adipem suum concluserunt — Ps 16:10 begins adipem suum concluserunt. Not marked: the Qui carries over the relative of 16:9, which Migne quotes as the preceding lemma. Nothing is added to the text.
11. 0783B et sicut catulus ejus — Ps 16:12 has et sicut catulus leonis habitans in abditis. Not marked, and this was the closest call in the batch. ejus refers to the leo named four words earlier in Migne's own lemma, so the two readings denote the same animal and the gloss (populus scilicet) turns on the whelp, not on whose it is. A [var:] here would report a difference no reader could act on. Logged instead.
12. 0784A Scimus quia — 1 Jo 3:2 has Scimus quoniam. Not marked: interchangeable conjunctions.
13. Orthography throughout, Pattern 9, no markers and no cruces beyond this line: Quandiu/quamdiu · unquam/umquam · coelo/caelo · coecitas/caecitas · Sepulcrum/Sepulchrum.
14. The two Hebraica editio readings are NOT Vulgate divergences and take nothing. 0776B separatum a terra (against probatum terrae) and 0781A Et tu haereditas speciosa mea es — Migne himself labels both as the Hebrew, and 0783D quotes a third (Ego in justitia videbo faciem tuam; implebor, cum evigilavero, similitudine tua). A reading the page itself identifies as another version is not a divergence in the Gallican lemma.

§3 · THE UNEXPRESSED SUBJECT — every supplied pronoun in this range, and what fixed it

The launch brief named this the dominant defect class for a psalm gloss, where the speaker moves between psalmist, Christ, the Church and the persecutor inside one sentence. Below is every place I supplied a subject English cannot leave out, with what decided it. Where two readings construe, both are named.

1. 0775A/B Pluet, ad verba praedicantium retulit… — supplied "he referred". The subject is the psalmist, not the commentator: the sentence explains what the verse's own word does, and the surrounding formulas (Quod autem ait, Ait enim, quod dixit veloces) all take the psalmist as subject. ⚠ The rival is Augustine, whose Enarratio this gloss abridges throughout; on that reading it is "he [Augustine] referred it". I took the psalmist because Migne's abridgment nowhere names its source in this range, so a reader has no third person to attach it to.
2. 0774D Quippe cui derelictus est pauper — "to him the poor man is left". The antecedent is God, carried from Oculi ejus in the lemma directly above and confirmed by Ps 9:35/10:14's tibi derelictus est pauper. Not the poor man, and not the psalmist.
3. 0776D ut perfecte operetur misericordiam — "that it may perfectly work mercy". ⚠ Two readings construe and I say so rather than choosing silently. The subject is either the soul (in anima mea stands three words earlier, and the whole psalm's frame is the soul's deferral) or God (the sentence's other actor, who is deferring). I took the soul, because the quandiu ponam consilium clause it depends on has the soul as its location and the speaker as its agent; a reader preferring God loses nothing in the argument.
4. 0780A certissime noverit malignum non esse — "let him know most certainly". The subject is the perfect man of Ista perfectio est, ut nihil in homine valeat malignus, i.e. the Qui ingreditur sine macula of the lemma. Third-person jussive, not an address to the reader.
5. 0776A in quo intelligitur posuisse — "in whom he is understood to have set". God the Father*, from Hoc de persona Dei Patris accipiendum, qui … Filium suum misit two sentences above. Not Christ, who is the in quo*.
6. 0781A manifestius in Hebraico … sic exponit; — "he sets forth more plainly in the Hebrew". ⚠ Two readings. Either the psalmist/text (my rendering) or the Hebrew edition as subject, which is exactly the construction Migne uses at 0776B four columns earlier: manifestius Hebraica editio posuit. That parallel is the strongest argument for the rival, and it is why I have named it here rather than settled it in the text; the English is unaffected either way except for the pronoun.
7. 0783B ut per ipsam regna debellet iniquitatis — "that through it it may war down". The subject is thy hand / thy eternal power (quam assumpsit manus tua, id est aeterna virtus tua), not the soul, which is the ipsam the power acts through.
8. 0782A quae res tacitas intelligentiae suae virtute cognoscit — rendered "which knows silent things by the power of its own understanding". ⚠ quae is feminine and can only attach to justitiam or vocem; the sense plainly wants the Lord as knower. I have followed the grammar rather than the sense and logged it, per Pattern 7. A reader who takes quae as loosely attracted to the Lord reads "by the power of his understanding".
9. 0778B cogitans de insidiis (0783B in chunk 0013) — "he devising of the snares". Subject is the adversary/whelp, i.e. the people, from catulus ejus, populus scilicet.
10. 0784A non solum … gratulatur … confidit; — "he not only rejoices … but trusts". The subject is the believer of gloria resurrectionis fidelium, the topic of the whole paragraph, not Christ and not John.

§4 · CONSTRUCTION-LEVEL CALLS

4.1 · Pattern 8b was invoked once, and it is the batch's one band mismatch

0776D: In eo etiam, quod subjungit: dolorem in corde meo per diem, subauditur, quandiu ponam? Migne closes an indirect question with ? (verified on the plate, p. 393, where the mark is unambiguous). English cannot end "it is understood how long I should set it?", so 8b's recast applies: "it is understood: how long shall I set it?" — the colon carries the subauditur frame and the direct question carries the mark. Nothing supplied, nothing deleted. ⚑ This is the single band where my colon count exceeds the Latin's (chunk 0011, band 1, 1 → 2), and the excess colon is precisely the one 8b requires. Recorded so a later checker does not read it as drift.

4.2 · Question-mark parity, verified

? counts are 1:1 with the Latin twin in every chunk (0010: 3 · 0011: 6 · 0012: 2 · 0013: 0). Two places tempt a mark that Migne does not print, and neither got one:

• 0779A ut sit sensus: Nonne cognoscent omnes… non in eis qui diligunt saeculum. — a Nonne clause closed with a full stop. Rendered as printed, ending in a period.
• 0784A Quid enim est aliud… nisi quod beatus Joannes Apostolus ait: … — a Quid enim question closed with a full stop. Likewise rendered with a period.

4.3 · Sentences Migne leaves without a terminal mark, reproduced as such

0780B Tamen etiam ista minora ita concludit — no full stop before the VERS. 5 line that completes it. Carried through unpunctuated in English ("Yet even these lesser things he thus concludes"), per Pattern 8's "never supply".

4.4 · VERS printed without its period — TWICE in this work, so it is a class

0774B prints VERS. 2-- (period after VERS, none after the numeral) and 0781B prints VERS 7.-- (no period after VERS). ⚑ Chunk 0009 carries a third, VERS 28.--, so this is a recurring feature of the work's transmission and not a one-off. The plate at p. 392 sets VERS. 2. — correctly, so at least the first is our transcription. Convention 1 says the verse address passes through VERBATIM, so all of them are reproduced exactly as our Latin prints them, unnormalized. Flagged for the assembler: if the work is ever patched, these are patch candidates alongside §0.

4.5 · usque ad — five occurrences, NONE of them the abridgment formula

Brief addendum D warns that the formula has look-alikes. Decided per occurrence, from what stands on either side:

• 0777D ×3 — usque ad unum is Ps 13:1's own words, an italic lemma, and the gloss expounds it at length across four sentences (vel cum ipso uno … vel praeter unum). This is the exact trap addendum D describes and it is the verse, not an abridgment.
• 0781B usque ad mortem me erudivit inferior pars — the ordinary preposition, "even unto death", parallel to Ps 15:7's usque ad noctem.
• 0781D perducens usque ad finem gloriae conspectus tui — ordinary preposition. ⚠ Its parse is genuinely ambiguous: either "even unto the end, [namely] the beholding of thy glory" (my rendering, taking conspectus as nominative in apposition) or "unto the end of the glory of thy countenance" (conspectus genitive). Neither is an abridgment; logged because the English commits to one.

4.6 · Test 2a — negations resting on a quantifier or correlative, proved from structure

The trigger population (a negative with solus/omnis/tantum/non solum … sed/non … sed nearby) fires eight times in this range. In every one the host is proved from the correlative, not from what reads well:

1. 0775C quia non se praebet cognoscendum malis, sed bonis — the sed bonis answers malis, so the negation is on malis: "he offers himself to be known not to the evil, but to the good", not "he does not offer himself".
2. 0777B quod non illi sed nobis est utile — host is illi: "useful not to him, but to us".
3. 0777C quia non tantum labiis dixerunt, sed in corde — host is tantum: "not only with the lips, but in the heart".
4. 0779A in Deo solo spem ponerent, non in rebus transeuntibus — host is the phrase in rebus transeuntibus, answering in Deo solo.
5. 0780C Non enim Deo, sed nobis praestamus — host is Deo: "it is not to God, but to ourselves, that we render service".
6. 0781A/B Non enim restitues mihi quod non amisi, sed restitues his… — the pivot is mihi ↔ his, so the negation governs mihi: rendered "thou wilt restore not to me what I have not lost, but thou wilt restore to those who have lost it". Rendering it as "thou wilt not restore** to me" also construes and reads more smoothly — which is exactly why 2a says to take the structure, not the smoothness.
7. 0782A/B Non autem nox tantum, quia perturbare assolet, sed etiam ignis — host is tantum on nox: "is to be called not night only*… but also fire". Putting the negative on vocanda est* would deny that tribulation is night at all, which the previous sentence has just asserted.
8. 0783C Non solum autem ista poena eos consecuta est visibilis, sed etiam… — host is solum: "not only** did this visible punishment follow them, but also…".
9. 0784A non solum de tam excellenti dono gratulatur, sed insuper… — host is solum.

⚑ Also checked and found to have no quantifier in scope, so the verb is the host: 0780C bono servi non eget — solus stands in the previous clause (solus ille veraciter Dominus sit) and does not govern the negation; "he alone is truly Lord who has no need" is right, and "he is not alone in needing" would be the 8990 @0614B error.

4.7 · Italic spans — three places where English had to draw a roman word inside a span

Every italic span in the Latin is an italic span in the English, in the same order, and the counts are 1:1 band by band. Three spans could not be reproduced without absorbing a possessive that Migne sets in roman, because English cannot separate the possessive from its noun:

• 0779D locutus est in corde suo → hath spoken in his heart (Migne's suo is roman).
• 0783C in vita ipsorum → in their life (Migne's ipsorum is roman).
• 0780D miras fecit Pater voluntates meas — kept as two spans with "the Father" roman between them, matching the plate exactly; noted only because the temptation was to merge them. Conversely, four Latin single spans were nearly split into two by English word order and were kept whole: quid fecit (0774C), Multiplicantur + filii hominum (0776C, kept as two as Migne prints), Perficiatur (0782B), statuerunt declinare (0783A), me adimplebis. Delectatio (0781D — one span crossing a full stop, reproduced crossing the full stop).

4.8 · Real words that read oddly — rendered as printed, no [cj:], per addendum A

1. 0775A crebra apertio atque operatio; — the figure is God's eyelids, which need opening and shutting; opertio ("covering") would complete it and operatio ("working") does not. But operatio is a perfectly good Latin word that renders in place, so Pattern 12 has nothing to wrap and Pattern 7 governs: rendered "this frequent opening and working in the holy books". [cj:] declined — the English is odd, not false; it does not assert anything the author is arguing against, which is 18a's bar, and 18a's census is closed. The conjecture is recorded here and nowhere else, by design.
2. 0779A et servire creaturae quam Creatori — a quam with no comparative anywhere to complete it (cf. Rom 1:25, servierunt creaturae potius quam Creatori). ⚠ This is the 11535 @1231B shape — a correlative whose answering member is printed and whose own member is not — and that shape is admitted to Pattern 18. Declined anyway, and the reason is the bar rather than the class: nothing false is asserted. My English writes "to serve the creature rather than the Creator", which supplies in English the comparative force potius would carry, and every reader takes exactly the sense Migne intends. 11535's quam earned its marker because the sentence would not otherwise resolve; this one resolves on sight.
3. 0777A Tribulantes sunt diaboli et angeli ejus — diaboli is nominative plural ("devils") while ejus is singular and wants a singular diabolus. Rendered "the devils and his angels", following the printed form. No marker: a number slip with a singular possessive is 18a's declined class (person/number anomalies that merely fail to construe), and Job's tempters are plural in the gloss's own argument either way.
4. 0782D sic per Christi humanitatem divinitas judicii, discernens inter justos et peccatores. — a sentence with no finite verb at all. Migne's ellipsis; rendered elliptically ("so through the humanity of Christ the divinity of judgment, discerning between the just and sinners"), nothing supplied.
5. 0782A Ex illuminatione notitiae tuae verum judicem. — likewise verbless unless judicem is read as the 1st-sg. present subjunctive of judico. ⚠ Two readings. I took the subjunctive — "may I judge what is true" — because the paragraph is a first-person prayer throughout and a bare accusative "a true judge" has nothing to depend on. The rival ("[let there come forth] a true judge", supplying the lemma's prodeat) is defensible and is why this is logged rather than settled in the text.
6. 0782A adjuncta Ecclesia, quae corpus, ejus est. — Migne's comma falls between corpus and ejus, inside the predicate. English cannot carry a comma there; rendered "which is his body". Recorded because it is a deleted mark, and Pattern 8 forbids deleting marks silently even where English has no slot for one.

4.9 · Rendering decisions fixed for the range (drift control)

One voice within the work, per the merge rule. Terms fixed at first occurrence and held: conventicula = assemblies · claritas = brightness (never "glory", which is reserved for gloria) · framea = sword · insipiens = the fool · malignus = the malignant one · saeculum = the world (and saecula = the ages) · conversatio = conversation · inopes = the needy and pauperes = the poor, never interchanged, because the gloss at 0776A distinguishes them (inopia et paupertate) · tabernaculum = tabernacle and mons = mountain, held apart because Psalm 14's whole gloss turns on the contrast. ⚠ Chunks 0000–0009 and 0014–0017 are with other agents; these are offered to the merge step, not imposed.

Pattern 17: every second person in this range was checked for number. The plural vos of Mt 11:28 (0778C), Jo 8:44 (0783B) and the confudistis / contempsistis / vidistis of 0779A take you; the singular tu … es … eges of Ps 15:2 (0780C), the ieris / eris of the robber parable (0779D) and every address to God take thou. No archaism is used as reverence.

§5 · WHAT THIS BATCH DID NOT SETTLE

• The tei-patch for §0 is not written and the work is not re-chunked. Nine sites across four pages; the remaining fourteen chunks of this work have not been collated against the plate at all, and on this rate they will carry more. A full-work plate collation is cheap here — the PDF is on disk, the page map is confirmed, and cols 0751–0794 are 22 pages.
• ~~§1.1 (mentem/montem) ships reading from our Latin twin, not from Migne, and will keep doing so until that patch lands. It is the only place in these four chunks where the English is knowingly faithful to the transcription rather than to the plate.~~ ✅ CLOSED 2026-08-28: the patch landed, the twin prints mentem, and the English now reads "one mind do I hold". Nothing in these four chunks is now knowingly faithful to the transcription over the plate.
• §3.1, §3.3, §3.6, §4.8.5 and §4.5's third item each name a rival reading that also construes. None is marked; all are argued here so a blind reader arrives at them already knowing what was weighed.

PLATE COLLATION — English brought to the corrected plate

2026-08-28. All 22 pages read at Migne's plate; 61 TEI patches applied. No marker was withdrawn in chunks 0010–0013 — but the one open compromise in this file is now closed.

• 0774B · mentem — §1.1 RESOLVED. The patch this file asked for landed. The Latin twin now prints Unum enim mentem teneo, and the English reads "For one mind do I hold, in which I put my trust" where it read "one mountain". See the resolution box at the head of §1.1 for why no marker was placed — neither the [sic: mentem] §1.1 proposed (Pattern 12 excludes a real word that is merely wrong, which is exactly the ground on which four other markers in this work were just withdrawn) nor a [cj:] (it would propose montem, "mountain", beside an English word that now says "mind"). Migne's gender mismatch — Unum … mentem, and in quo two words later — is recorded here and rendered as printed.
• 0778D · Jo 11:48 — [var:] DECLINED, and the existing one stands. The collation restored Migne's dimitimus (single t) where CC printed dimittimus. That is not a divergent reading — it is a dropped sort in a word that is not otherwise a Latin form — so it takes no [var:] and, per the standing out-of-scope ruling, no [sic:] either. ⚑ The real divergence in this same lemma (et tollent nobis et gentem et locum for Clem. et tollent nostrum locum, et gentem) was already caught and already carries its [var:] in 0011.md; the plate confirms it.

PLATE COLLATION · REGISTER OF RESTORED TYPE, cols 0774B–0784A

Wilson's ruling of 2026-08-28 governs (stated in full at the head of the matching section in cruces-0000.md): mark only where the reading is in doubt; record the rest here and disclose the count in the work-level page note.

MARKED in this range — one.

• 0774B · apud ipso for apud ipsos — [cj: apud ipso; read apud ipsos, "with them"]. Migne prints haereticos, qui nos ab uberibus matris Ecclesiae avertere moliuntur, affirmantes quod apud ipso sit Christus. apud governs the accusative, so the ablative does not construe; and the defect is not confined to case — ipso is singular where the whole sentence is about the heretics in the plural, which our English ("that Christ is with them") had silently supplied. A dropped s that leaves a real pronoun in a real position: the Latin page gives the reader nothing to stop on. Marker placed after the rendering.

Already settled, and not re-opened: 0774B mentem carries no marker — the reasoning is at §1.1 and in the section above.

CRUCES ONLY — recorded here, unmarked in the text.

Orthography, not defective type: 0778D dimitimus (single t) for dimittimus, Jo 11:48 — argued in the section above, where the [var:] was declined.

THE BLIND POLARITY READ

Fenced reader on chunks 0010–0013 (Pss X–XVI, cols 0774B–0784A), Latin against English in full, with a mechanical negation-token count per chunk as backstop (0010 15/16, 0011 45/44, 0012 39/38, 0013 21/22; the 0013 shortfall is nisi collocations, checked by hand). No crux file and no collation report was open. Adjudicated 2026-08-28 under Wilson's marker ruling of the same session (stated in full in cruces-0000.md). Whole-work figures are in that file.

Result for this range: 5 confirmed sites — three carried a defect in our English, three sit on Migne's plate; 0779D is both, which is why the two figures overlap. Three new [cj:] markers. The reader found no dropped negation anywhere in the range and no word translated as its opposite.

Marked — Migne's own defects, our English literal and unchanged

1. 0779D · [cj: in corde; read sed non in corde, "but not in the heart"]. Migne: Verum ille locutus est in corde suo, aliud enim putabat, et nesciens verum dixit. The whole simile exists to show a man who speaks truth with the lips and not in the heart — that is the lemma under exposition (qui loquitur veritatem in corde suo), it is stated three lines up (in labiis habent veritatem, in corde non habent), and the next sentence draws the moral (parum est verum loqui, nisi etiam in corde ita sit). The line has lost its sed non. ⚠ Our English changed here too, and it is not a repair. It had taken Verum as the adversative connective ("But that man hath spoken in his heart"), which leaves the clause with no object and reads as smooth prose — 7a mechanism 3, connective smoothing: the contradiction survives but loses its visible edge. Verum is now taken as the object it most likely is, matching nesciens verum dixit four words later: "Truth that man hath spoken in his heart". The clash with aliud enim putabat is now on the surface, where the marker can name what fell out. A dropped word is glossable under Pattern 18 (7561 @0648): the gloss opens on words Migne did print, so the verbatim-in-the-Latin-twin guard holds.
2. 0781A · [cj: amici; read inimici, "enemies"]. nec a me jam et peccatores, et amici, et homines, sed justi et fratres mei, et filii Dei vocabuntur. A three-term before/after list — peccatores→justi, X→fratres mei, homines→filii Dei — whose middle term sits on the wrong side of the ledger: as printed they will no longer be called friends, but will be called brethren. A single lost in-, in a gloss about their being made better. Marked because the printed word is real, in a real position, and a reader would take it for a variant rather than a defect.
3. 0775A · [cj: operatio; read opertio, "covering"]. ista in sanctis libris crebra apertio atque operatio tanquam palpebrae sunt Dei, in a sentence built out of clausis oculis and apertis oculis. Eyelids are named precisely because they do two opposite things; "opening and working" is not one of them. ⛔ This REVERSES the decline at §4.8.1 above, and deliberately. That decline was argued on 18a's bar — the English is odd, not false — and 18a's census was closed. Wilson's ruling of 2026-08-28 states a different test: mark where a reader could not recover the intended word with confidence, or where the defect could pass for a real variant. opertio is a rare word; nothing on the page lets a reader reach it, and operatio reads as a genuine reading. Under the new test it is marked. §4.8.1 stands as the record of the earlier reasoning; the text now follows this paragraph.

Adjudicated and DECLINED

• 0774B mentem. Re-raised by the reader as a high-confidence site. Not open — it is settled at §1.1 above and deliberately carries no marker: mentem is a real word (so Pattern 12 has nothing to wrap), and a [cj: read montem, "mountain"] would sit beside an English word that already says "mind". The incoherence lives in §1.1, which is its home.
• 0777A Tribulantes sunt diaboli et angeli ejus. Already adjudicated at §4.8.3 and unchanged: a number slip beside a singular possessive, 18a's declined class, and the intended diabolus is not in doubt.
• 0780A — three unanchored "he/his" in Et ut nihil sit in conspectu ejus, certissime noverit malignum non esse… (reader's confidence: low). Declined. in conspectu ejus is a verbatim re-quotation of the lemma (Ps 14:4, ad nihilum deductus est in conspectu ejus malignus), and the pronoun is unglossed in the psalm as it is in Migne; supplying an antecedent inside a lemma re-quotation would put our gloss on the page in the psalm's own italics. Our English disambiguates it correctly ten lines later ("in whose sight the malignant one is brought to nothing"), which is where the disambiguation belongs.

Repaired on our side

1. 0777B — the cantabo / psallam distinction was half-collapsed. The gloss announces the distinction itself (Si autem aliqua hic differentia consideranda est, cantabo corde, psallam operibus), and our lemma gave psallam the very English word the gloss reserves for cantabo, so the reader met "I will sing … that is, I use my body in the most ordered way". Lemma and its [var:] now read "I will play to the name of the Most High", the same English the gloss already gives psallam. Not the fragmentary-lemma convention: this is a whole clause, not a stump.
2. 0777B — the relative bound to the wrong antecedent. In salutari tuo, hoc est in Christo Jesu, exsultabit cor meum, qui tribuit mihi bona spiritalia: qui is masculine and can only be Christ, but the English put it against "my heart", where an English reader binds it. Re-ordered to "In thy salvation, that is, in Christ Jesus, who hath bestowed good things upon me that are spiritual … shall my heart rejoice." Three italic spans before, three after. 7a mechanism 5.

Positive results worth recording

• ⭐ The 7a″ profile is ABSENT from this range too. All fifteen [var:] notes in the four chunks (Ps 10:3, 12:2, 12:6, Rom 1:28, Act 10:13, Mt 11:29-30, Jo 11:48, Ps 13:6, Rom 11:25, Rom 11:26, Ps 15:3, Ps 16:4, 16:5, 16:13, 16:15) were checked against the printed Latin: each records a real divergence and none conceals a conformation. No lemma in the range was found quietly conformed to the familiar English Bible — which, in the genre where 7a″ is the predicted dominant failure, is the second independent negative result of the read.
• Ps XIV's tabernaculum/mons and quis habitabit/quis requiescet contrasts are fully preserved, as is Ps XV's humanitas meae divinitatis / divinitas humanitatis chiasmus, and Ps XV:5's congregabo against v.4's Non congregabo — an opposition the Latin builds deliberately, not a dropped negative.

Preserved unedited from cruces-0014.md; headings demoted one level. Where it conflicts with the head above, the head governs.

Glossa ordinaria, Expositio in XX primos Psalmos (PL 114) — cruces, chunks 0014–0017

Cols 0784B–0794D, Psalms XVII–XX (Vulgate numbering), 4,249 Latin words → 6,315 English (ratio 1.48). This range is the end of the work. verify-english.mjs clean for these four chunks. Frontmatter copied verbatim from each Latin twin.

Apparatus placed in this range: 2 [sic:] · 9 [var:] · 3 [cj:] (the third @0790D, from the blind polarity read — see its section at the foot of this file) · 0 [d:] · 0 [ed:] · 0 [nt:] · 0 [n:] — ⚠ revised 2026-08-28, was 5 [sic:] · 9 [var:] · 2 [cj:]: the plate collation withdrew three [sic:] and one [cj:] and placed one new [cj:] @0794A. See PLATE COLLATION — markers withdrawn at the foot of this file. (the work carries noteCount: 0 throughout — no note was written, and none was expected). Zero guillemets in either language; zero ( sic ) of Migne's own in this range (checked). Marker/anchor/italic/paragraph parity verified 1:1 against each Latin twin, and ? ; : compared band by band across the 43 column anchors, zero mismatched bands.

§0 · HOW THE WORK ENDS — not a defect, do not mark it

Chunk 0017 ends at 0794D with the last words of the gloss on Ps 20:14: Nos autem corde et opere celebrabimus, et nota faciemus mirabilia tua. — "But we with heart and with work will celebrate, and will make known thy wondrous works."

Two things a merge reader should know and neither is a hole:

1. The work simply stops at the twentieth psalm. That is Pez's deliberate choice to print twenty psalms as a specimen. No closing formula, no subscription, no FINIS. Nothing has been supplied and nothing marked.
2. The last gloss also stops short of its own verse. Migne's lemma is Exaltare, Domine, in virtute tua, etc.; the Clementine's Ps 20:14 continues cantabimus et psallemus virtutes tuas, which Migne's etc. covers and the gloss then paraphrases in its own words (celebrabimus … nota faciemus). This is the ordinary lemma-plus-etc. shape of the whole book, not a truncation.

§1 · THE VULGATE COLLATION — the deliverable

Method, stated so a later reader can tell a checked lemma from an unchecked one: every one of the 246 italic spans on the named-lemma list for these columns was searched in sources/vulgate/clementine-flat.txt, and the four psalms were pulled out whole (grep -P "^Ps (17|18|19|20):") so that each lemma could be read against its own verse rather than against a match somewhere else in the canon. Non-psalm quotations were searched individually. Citations below are from sources/vulgate/clementine/.

1.1 The firing rule I used, stated before the list

The 48 spans the pre-scan marked "⚠ NOT in Clementine verbatim" are overwhelmingly innocent: a fragment cut mid-phrase, a gloss re-inflecting its own lemma (commoti sunt, humilias, permanet, emundabimur), a gloss supplying a connective (sed, cujus), or a re-ordering inside the gloss's own sentence. Those are logged in §1.3 and take no marker.

Where a real divergence stands, I fired [var: …] on this rule and no other:

• A whole-word or multi-word substitution that changes what the sentence asserts → marker.
• A one-letter tense/mood shift in a re-quotation → crux only, no marker. sources/vulgate/README.md warns that this copy is not to be trusted to single-letter precision, and brief §4 records that the digitization alters spelling in both directions; a public claim about the Clementine should not rest on one letter of our own file. The one exception is @0792D Exaudiat, and §1.2 gives the reason.
• A synonym substitution that leaves the assertion intact → crux only. (aculeus for stimulus, omnes for universos, viventium for vivorum, super vos for in vos.)

1.2 The nine [var:] markers — what was claimed and why

⚠ Every marker's content was checked not to appear verbatim in the Latin twin (the [ed:]/[var:] mirror test), programmatically, after the files were written.

1.3 Read clean, or diverging but NOT marked — the negatives, which are the point

Psalm XVII lemmata that collate exactly (Clementine Ps 17, verse by verse): VERS. 3 Deus meus adjutor meus… and Protector meus, et cornu salutis meae, et susceptor meus. · VERS. 4 Laudans invocabo Dominum · VERS. 5 Circumdederunt me dolores mortis · VERS. 6 Dolores inferni circumdederunt me · VERS. 7 In tribulatione mea invocavi Dominum, De templo, introivit in aures ejus · VERS. 8 Commota est et contremuit terra, fundamenta montium conturbata sunt · VERS. 9 Ascendit fumus in ira ejus · VERS. 10 Inclinavit coelos et descendit · VERS. 11 Et ascendit super cherubim et volavit · VERS. 12 Et posuit tenebras latibulum suum · VERS. 13 Prae fulgore in conspectu ejus nubes · VERS. 14 Et intonuit de coelo Dominus · VERS. 15 Et misit sagittas suas, et dissipavit eos, fulgura · VERS. 16 Et apparuerunt fontes aquarum, Ab increpatione tua, Domine, ab inspiratione spiritus irae tuae · VERS. 17 Misit de summo et accepit me · VERS. 18 Eripuit me de inimicis meis fortissimis · VERS. 19 Praevenerunt me in die afflictionis meae, factus est Dominus protector meus · VERS. 20 Et eduxit me in latitudinem, salvum me fecit, quoniam voluit me · VERS. 21 Et retribuet mihi Dominus secundum justitiam meam · VERS. 22 Quia custodivi vias Domini, nec impie gessi a Deo meo · VERS. 23 both spans · VERS. 24, 25 Et ero immaculatus cum eo… · VERS. 26, 27 Cum sancto sanctus eris … et cum perverso perverteris · VERS. 29, 30, 31, 33, 34, 35, 36, 37, 38, 39, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51 lemma-heads. Psalm XVIII: VERS. 2, 3, 4, 5–6 (the In sole half), 8, 9, 10, 11, 12, 13, 14 Si mei non fuerint dominati, 15 all collate. Psalm XIX: VERS. 2, 3, 4, 5, 6, 7 (Impleat Dominus omnes petitiones tuas), 8, 9, 10 all collate. Psalm XX: VERS. 2, 3, 4, 5, 6, 7, 8 (head), 9, 10, 11, 12 (head), 13, 14 all collate.

Divergent but deliberately unmarked, with the reason:

• 0784A–B Implebor, cum evigilavero, similitudine tua. (Ps 16:15; Clementine satiabor cum apparuerit gloria tua). Not a divergence at all: chunk 0013 introduces this very wording as juxta editionem Hebraicam and prints it there twice. Chunk 0014 is quoting back the Hebrew edition the work itself named. Rendered "I shall be filled, when I shall have awaked, with thy likeness."
• 0784B ubi est, mors, aculeus tuus? — 1 Cor 15:55 reads stimulus tuus. Synonym ("sting" either way); the argument does not turn on it. Crux only.
• 0784B nec laetatur in perditione viventium — Sap 1:13 vivorum. Synonym. The companion quotation Impii autem manibus et verbis accersierunt illam (Sap 1:16) is verbatim.
• 0785A accensi — the verse (Ps 17:9) has carbones succensi sunt ab eo. The italic span is a single word inside the gloss's own clause (rursus accensi et illuminati revixerunt), with no carbones anywhere; it is the gloss re-forming the verse's participle, not a citation. Rendered "kindled". Crux only.
• 0786A Appropinquavit super vos regnum Dei — Luc 10:9 in vos. Preposition only, sense unchanged. Crux only.
• 0786D retribuit mihi Dominus against the lemma's own retribuet four lines above. This one matters for the English: I held the divergence and rendered the lemma "the Lord will reward me" and the re-quotation "the Lord rewards me". Conforming the second to the first is precisely the silent repair 7a″ item 3 exists to prevent. One letter, so no marker.
• 0786D sed in conspectu oculorum ejus — Ps 17:25 has the phrase without sed; the connective is the gloss's, answering its own non quod homines vident. Not a scripture divergence.
• 0787A humilias — Ps 17:28 humiliabis. The gloss re-inflects to the present in its own sentence. Crux only.
• 0787A Non est recta via Domini — checked and exact: Ez 33:20. (It was worth checking: Ez 18:25/29 print non est aequa via Domini, and picking the wrong parallel would have manufactured a divergence.)
• 0787A illuminabit tenebras meas (roman, not italic) — Ps 17:29 has the imperative illumina. Not set as a citation. Crux only.
• 0788A Et praecinxisti me virtute ad bellum, supplantasti… — Ps 17:40 has et supplantasti. A dropped connective, sense intact.
• 0788A Clamaverunt, nec erat qui salvos faceret ad Dominum… — the Clementine's semicolon (qui salvos faceret ; ad Dominum) is absent, which changes the parse of the lemma. Migne's own gloss then reads it the Clementine's way (Nec ad quemlibet, sed ad Dominum clamaverunt). Punctuation is Pattern 8 territory, not Pattern 14; carried as printed.
• 0789A facit misericordiam Christo suo — Ps 17:51 faciens misericordiam christo suo David. The gloss needs a finite verb for its own Deus magnificat … et facit; David is outside the span. Gloss grammar, not a variant.
• 0789D Exsultavit ut gigas ad currendam viam suam* — Ps 18:6 ends ad currendam viam. An added possessive; the gloss's next sentence drops it again (non ad habitandum, sed ad currendam viam*). Sense unchanged.
• 0789D A summo coeli egressio ejus — Ps 18:7 a summo caelo. A change of case (genitive for ablative), not of orthography. Rendered on the printed genitive, "from the height of heaven". Declined because the gloss turns on a summo = a Patre, not on the case, and because a case ending is exactly the precision this copy is not warranted to. Worth a plate read if one is going that way anyway.
• 0790B testimonium fidele — Ps 18:8 testimonium Domini fidele; the gloss replaces Domini with its own Cujus. Gloss syntax.
• 0791A multa retributio — Ps 18:12 retributio multa. Word order inside the gloss's sentence.
• 0791B A superbis quoque libera servum tuum; si mei non fuerint dominati, tunc immaculatus ero: — diverges wholesale from the Gallican, and Migne says so in the sentence that introduces it (ex Hebraeo translata editio). A [var:] here would be reporting as a divergence the very thing the text announces. Declined on that ground.
• 0791C si mei non fuerint dominata, tunc immaculatus ero — likewise introduced by Migne as ut in quibusdam codicibus habetur. He is citing a variant, not printing one. See §3.4 for how the gender difference was carried into English, which has no exponent for it.
• 0791C ipse est rex super omnes filios superbiae — Job 41:25 super universos filios superbiae. Synonym.
• 0792B Pater, salavifica me ex hac hora — Jo 12:27 Pater, salvifica me ex hora hac. Two separate things: the word order (ex hac hora) is a trivial inversion, logged only; salavifica is broken type and carries [sic:] (§2).
• 0792C Et confirmet omne consilium tuum — Ps 19:5 et omne consilium tuum confirmet. Word order only.
• 0793B exsultavit vehementer — Ps 20:2 exsultabit. One letter (past for future), in a re-quotation. Rendered as printed, "he hath exulted exceedingly", which is what the gloss's resurrection reading wants. Crux only, per the rule in §1.1.
• 0793C benedictionem dulcedinis — accusative in the gloss's own clause (hauserat); the lemma above has the Clementine's in benedictionibus dulcedinis. Gloss syntax.
• 0793C Pater, clarifica Filium tuum — Jo 17:1 Pater, venit hora: clarifica Filium tuum. An ellipsis of the opening, not a variant.
• 0794C Venite, occidamus eum, et habemus haereditatem ejus — Mt 21:38 habebimus. Tense in a quoted saying; two letters. Crux only. (Note Mc 12:7 gives the same saying in a different form again, et nostra erit haereditas; Migne's is nearer Matthew.)
• 0786D criniculis peccatorum suorum unusquisque constringitur — the allusion is Prov 5:22, which the Clementine gives as funibus peccatorum suorum constringitur. Corpus frequency here is a negative, and I am recording it as one: criniculis occurs once in the whole src/latin/ corpus, funiculis ten times. Per addendum A that argument is worthless against a real word, and criniculus is a regularly formed diminutive of crinis used for a plaited cord; it is the Old-Latin reading Augustine's text carries here, not broken type. Rendered "cords". No marker (the phrase is woven into gloss prose, unitalicized and unattributed — an allusion, not a citation).

§2 · CARRIED TYPE — five [sic:], and what each one would have swallowed

All five are non-words, which is the class addendum A says is worth raising and has been right every time. Each was rendered in plain English as well as marked, per test 1; the strip-the-markers read was run mechanically over all four chunks afterwards and no clause was left without its verb, object, negation or conjunction.

Rejected as [sic:] candidates, and why (a blind reader will look at all three):

• 0790B resisti — see §3.1. ⛔ The whole question is moot as of 2026-08-28: Migne prints resistit, with a fully formed final t. The [cj:] is withdrawn and the English carries the plain reading, "since God resisteth the proud, but giveth grace to the humble." The reasoning below (a real Latin form, so outside Pattern 12) was correct as reasoning and was answering a question the page does not pose.
• 0787B manducaverint — a real word. Took [cj:] instead; see §3.2.
• 0786D criniculis — a real word and, I believe, the author's own; see §1.3.

§3 · THE HARD PLACES

3.1 0790B · quoniam Deus superbis resisti, humilibus autem dat gratiam — [cj:]

Jas 4:6 / 1 Pet 5:5, woven unitalicized into the gloss. resisti is a real form (present passive infinitive of resisto) — so Pattern 7 renders it and Pattern 12 has nothing to wrap. But with Deus nominative and superbis dative there is no construction at all: the clause has no finite verb for its subject, and the only reading that construes is resistit. Note that the same work prints resistis correctly at 0792A (qui superbis resistis, second person, addressing God), so the book knows the tag.

Two candidates were weighed and one lost, explicitly:

1. Pattern 9, "attested medieval spelling" — loss of final -t is a real medieval orthographic habit, which would make resisti a spelling of resistit and licence "God resisteth the proud" with no marker at all. ⛔ Rejected: resisti is also a perfectly good classical form in its own right, so it cannot be dismissed as a spelling of something else; treating it as one is the silent repair Pattern 7 exists to forbid, and it leaves no trace.
2. Pattern 18 [cj:] — taken. The marker is additive: Migne's word keeps its English and the conjecture stands beside it, and the form's appearance in the Latin twin is what makes the claim checkable. Shipped as since God to the proud is to be resisted [cj: resisti; read resistit, "God resisteth the proud"].

⚠ Declared as a borderline firing: 18a's bar is "the English asserts something false", and what the literal rendering does here is closer to asserting nothing. It is filed under 18 rather than left in the cruces because the un-glossed English ("God … is to be resisted") is actively misleading about who resists whom, which is the founding munde case's shape. Also a plate-read candidate: a lost final -t is exactly the brief §4 class, in both its directions.

3.2 0787B · nisi manducaverint viam fidei — [cj:]

Deus meus non venit in homines, nisi manducaverint viam fidei, qua veniat ad eos, quia impolluta est via ejus. Rendered "unless they have eaten the way of faith", which is what the printed word says and is not what the sentence is about. The conjecture is mundaverint, "have cleansed": the clause's own quia impolluta est via ejus is a purity pivot, and the parallel at 0787C (sicut impolluta est fides, qua venit ad me) makes cleanness the whole figure on both sides of the coming.

The rival, named and adjudicated: manducaverint could be a eucharistic figure — "unless they have eaten the way of faith" is not unthinkable in this register. It loses because nothing in the paragraph or its neighbours mentions eating, food, bread or the table, while impollutus is printed twice within four lines. Shipped additively: unless they have eaten [cj: manducaverint; read mundaverint, "have cleansed"] the way of faith.

3.3 0786D / 0787A / 0793A · Migne's retribuit, humilias, exsultabimus — the pressure to conform

Recorded here because it is the class 7a″ was written for and it is invisible in the finished English if you do it right. In three places the gloss re-quotes its own lemma with the verb changed (retribuet → retribuit @0786D, humiliabis → humilias @0787A, invocabimus → exsultabimus @0793A). In each the English follows the re-quotation as printed, not the lemma above it. Only the third took a marker; the first two are one letter.

3.4 0791B–C · dominati / dominata — a gender difference English cannot carry

The gloss's whole VERS. 14 argument is that the verse reads two ways depending on the participle's gender: with masculine dominati the subject is the proud alien spirits; with neuter dominata it is occulta mea et aliena, the hidden and alien things. English participles carry no gender (Pattern 9), so a literal rendering makes the two readings identical and destroys the argument on the page.

Resolved by putting the difference on the subject pronoun, which English does have: dominati → "If they shall not have had dominion over me"; dominata → "if those things shall not have had dominion over me". Nothing is added that the Latin does not say — the gloss itself supplies the neuter subject in the next clause (si mei non fuerint dominata occulta mea et aliena). Flagged so a blind reader does not read the two Englishes as an inconsistency.

Also note mei is genitive in both readings (dominari + gen. = "have dominion over"), not the nominative "my own" it looks like; the gloss's own paraphrase confirms it.

3.5 Unexpressed subjects — the ones I supplied, and what decided each

The launch brief is right that this book shifts speaker constantly. The sites where I supplied a pronoun the Latin does not print, with the evidence:

• 0784A eripiendum se novit — "it [the body of Christ] knew that it was to be rescued". Subject is corpus Christi, printed in the same clause. Then gratulabitur (0784A) takes "she", because the subject has become Ecclesia Christi, printed two words earlier. The shift from "it" to "she" inside one paragraph is Migne's, not mine.
• 0784B Loquitur haec Christus et Ecclesia — printed subject; no supply needed. The rest of Psalm XVII's first person is totus Christus, caput et corpus, which the work says outright, and says again in its closing paragraph at 0789B. That closing sentence is the work's own instruction for reading every "I" in the psalm and I let it govern the whole of chunks 0014–0015.
• 0784B Qui mihi prius adjutorium … praestitisti — second person singular, God. "Thou who first didst furnish me" (Pattern 17: tu → thou).
• 0784B qui me firmasti, qui refugi ad te — two relatives, two different subjects, one second-person (firmasti) and one first-person (refugi). Rendered "who hast made me firm, I who have fled for refuge to thee". Easy to flatten into a single "who"; the verb endings forbid it.
• 0785A commoti sunt et contremuerunt — the subject peccatores stands outside the italic span in Migne, so the English keeps it outside too ("sinners were moved and trembled").
• 0788C Populus … servivit mihi: neque oculis me vidit — the subject is Populus gentium, singular, so "it hath served me: neither did it see me". Not "they". Named because the plural is the natural English reflex and would quietly turn a people into persons.
• 0790A et ipsam penetravit verbi calor — ipsam is feminine accusative and its antecedent is umbra mortis four words back, not mortalitatem. Rendered "penetrated that shadow itself" and the noun named, because a bare "it" would have been ambiguous between the two feminine nouns in the sentence and a reader could not recover which.
• 0791D ut consentiendo suum faceret — subject is homo, printed; object suum is the alienum of the previous clause. "so that by consenting he made it his own".
• 0794B inveniat — the subject is claritas, printed, not the king; the verb is jussive in parallel with Inveniatur. Rendered "may the brightness … find".

Where two readings both construe, per instruction rather than choosing silently:

• 0789A nec terrenam ipsam salutem sed in excelso de illo sperem (VERS. 47). The clause has no expressed object for sperem, and nec terrenam ipsam salutem has no verb of its own — it may hang on the preceding sentio ("nor do I judge that salvation itself to be earthly") or be an object of sperem ("nor do I hope for that salvation itself as earthly, but on high"). Rendered on the first, which keeps the non … nec pair on the same verb: "Not after an earthly custom do I judge of the God of my salvation; nor that salvation itself as earthly, but on high concerning him may I hope." Neither reading is marked. The English is deliberately as loose-jointed as the Latin.
• 0794C in te occidendum retorserunt — "they turned back upon thee, to be slain". occidendum agrees with te, so it is "upon thee, who wast to be slain", but it can also be read as a gerundive of purpose governing the turning. Rendered on the first; declined a note in the text.

3.6 Column anchors — three that had to be placed by hand

Anchors were checked one by one against the word each stands beside in the Latin, not merely for presence and order (test 2b). Three fall inside constructions English reverses, and I record the choice made:

• 0784C Migne breaks at Protector [0784C] meus — the anchor sits between noun and possessive, and English puts the possessive first. Placed after the whole phrase: "My protector, [0784C] because I presumed not…". One word downstream, same clause.
• 0785B qui terrena [0785B] sapiunt — same reversal. Placed after the relative clause: "who savour of things earthly, [0785B] knew not…".
• 0789B bonis operibus [0789B] meis — placed as "by my good [0789B] works", which keeps the anchor one word from the end of the sentence exactly as Migne has it.

Placed inside the construction, deliberately, where English allows it (the 8995 precedent): in [0787B] te → "in [0787B] thee" · non eos [0785C] deserens → "not [0785C] forsaking them" · Si mei non fuerint [0791C] dominati → "If they shall not have had [0791C] dominion over me" · quo cecidit [0791D] diabolus → "by which [0791D] the devil fell".

⚠ No column-band arithmetic is reported (runbook false positive 1). The bands in this range run B–D, A–D and skip freely; nothing is broken mid-word or mid-construction across an anchor.

§4 · REGISTER DECISIONS FIXED FOR THIS RANGE

These are one-work equivalences, chosen once and held. A merge across agents should reconcile them with chunks 0000–0013.

• latitudo = "breadth" everywhere (VERS. 20–22 of Ps XVII), including the lemma "into breadth". "A broad place" reads better in the lemma and would break the gloss, which picks the word up three times to build latitudo fidei → latitudo bonorum operum → longitudo perseverantiae → altitudo spei, a dimensional figure that only works if one English word carries it.
• alienus = "alien" throughout Ps XVIII (ab alienis, alienum, alieni maligni spiritus). The English preposition varies with the syntax ("by alien things" / "from alien" desires) because English requires it; the lexeme does not vary.
• occultus = "hidden", so Ab occultis meis munda me is "From my hidden things cleanse me", not "my secret ones" — the gloss depends on occultum suum being the same word as the lemma's.
• desiderabilis = "desirable" across Ps 18:11, so that desiderabilia, desiderentur, desideret and desiderabiliora stay visibly one family.
• dorsum = "a back" in both places it is glossed (Ps 17:41 dorsum mihi eos fecisti esse; Ps 20:13 pones eos dorsum), following the gloss's own literal reading rather than the idiomatic "make them turn their back".
• in saeculum saeculi = "for ever and ever", uniform across Ps XVIII and Ps XX.
• VERS. n.-- passes through verbatim; heads are Englished (IN PSALMUM XVII. → ON PSALM XVII., Roman numeral kept). (cont.) on chunk 0015's repeated head is carried untranslated (runbook false positive 7).
• Pattern 17: God, Christ and the psalmist are addressed in the singular throughout and take thou/thee/thy; the only plural addressee in the range is vobis at 0793C (relinquo vobis), which takes "you".
• Punctuation as printed. Migne's period inside the lemma before etc. at 0791D (eloquia oris mei. etc.) is reproduced, though every other lemma in the range ends in a comma. Not a defect and not normalized.

§5 · WHAT I DID NOT DO

• No [n:] anywhere. The work's noteCount is 0 in all eighteen chunks and these four carry no attribution siglum at all — unusual for the Glossa, and worth a merge note: this exposition is a continuous abridgment of Augustine's Enarrationes with no inline sigla, unlike Ruth.
• No [ed:]. Nothing in this range looks like lost text: the paragraph structure matches the Latin twin 1:1 in all four chunks (37/17/16/27 paragraphs) and no sentence stops mid-construction.
• No [d:]. The repeated multum at 0790C is Migne's deliberate three-way construal (sive multum … sive multum … sive multum), not dittography.
• No plate was read. Three sites would repay one if a run is going to PL 114 anyway, in this order: 0790B resisti (§3.1, a lost final -t is the brief §4 class), 0787B manducaverint (§3.2), and 0789D coeli for caelo (§1.3, a case ending that could equally be a digitization tidy-up). The five [sic:] non-words are not on that list: addendum A records that suspicion aimed at a non-word has been confirmed ten times out of ten.

PLATE COLLATION — markers withdrawn

2026-08-28. All 22 pages read at Migne's plate; 61 TEI patches applied; the Latin twins now say what Migne printed. Chunk boundaries did not move. Four of this range's markers were false public claims against Migne and are gone. The rows and sections above are struck and annotated in place, never deleted.

Withdrawn.

• 0787A [sic: poputum] — WITHDRAWN. At 1200 dpi Migne prints populum, the l at full height and matched against the l of humilem on the same line. poputum was Corpus Corporum's. English: For thou wilt save the humble people,.
• 0789D/0790A [sic: cut] — WITHDRAWN. Migne prints Exultavit ut fortissimus. English: He rejoiced as one most mighty.
• 0792B/0792C [sic: cortuum] — WITHDRAWN, and there was never a run-together. Migne sets cor at the end of a line and tuum, at the head of the next, with the column break between them; the twin now reads secundum cor [0792C] tuum,, the anchor inside the italic span. The English anchor has been moved to the matching position: but according to thy own [0792C] heart,. ⚑ Generalizable: a Pattern 10 run-together should always be checked against the line ending, because a transcription that loses line breaks manufactures exactly this defect.
• 0790B [cj: resisti; read resistit] — WITHDRAWN; no conjecture is needed. Migne prints resistit with a fully formed final t. English carries the plain reading: "since God resisteth the proud, but giveth grace to the humble." §3.1 above stands as method and is void as to this site.

⚠ All four had the same shape: the marker accused Migne of a defect that the digitization had introduced. Counting 0770A's withdrawn [var:] in cruces-0005, five of this work's public claims were artefacts of Corpus Corporum.

Punctuation corrected at the plate — Pattern 8 in both directions.

• 0789A (VERS. 48). Migne prints …liberator meus de inimicis meis. Judaeis, clamantibus. Crucifige, crucifige! — a stop inside the italic span, a stop (not a colon) after clamantibus, and an exclamation at the end. The English had smoothed all three into one running sentence with a colon. It now reads: my deliverer from my enemies. The Jews, crying out. Crucify him, crucify him! The fragment is Migne's and is carried as a fragment.

Sites where the corrected plate changed the sense.

• 0789D · lemma ut gigas ac currendam viam suam (CC: ad). The English keeps "to run his way", which is the evident sense and the reading the gloss itself repeats three lines later in Migne's own words (non ad habitandum, sed ad currendam viam). No marker: this is one more of the work's restored one-sort defects, which are out of scope for marking by the standing ruling. Recorded here so the reader of the twin is not left to wonder.
• 0785D · Ut n profundo rerum humanarum audirentur coelestia — Migne's in has lost its i. The English already rendered the evident sense ("That in the depth of human affairs heavenly things might be heard") and is unchanged. Same class, no marker.

0788A · sive signa, quae impressi ad imitandum sequentibus — ⛔ NO MARKER, and this is a dissent

The plate collation restored impressi (CC had tidied it to impressa) and flagged the site as the one place in this work where the plate genuinely does not construe, recommending a [sic: impressi] or a [cj:] proposing impressa. I decline to fire either, because the premise is a misparse and the marker would be the sixth false claim against Migne in this work.

The reading construes exactly as printed. impressi is not a participle that has failed to agree with signa; it is the first-person perfect active of imprimo, and quae is its direct object:

et non sunt infirmata sive itinera mea, sive signa, quae impressi ad imitandum sequentibus — "and they are not weakened, whether my journeys, or the marks which I impressed for those who follow to imitate."

The whole sentence is first-person and addressed to God (Dilatasti gressus meos, latam fecisti charitatem meam), so a first-person verb is not merely possible here, it is the expected voice.

⭐ And there is an external witness on disk. The same passage in the parallel psalm commentary at sources/pl/tei/7109.xml (PL 21, Commentarius in LXXV psalmos, the paragraph marked (Ex Augustino)) reads:

Et non sunt infirmata vestigia mea. Id est, signa, quae impressi ad imitandum te.

Same clause, same finite verb, an independent book and an independent transcription. Migne's page is not defective here; only the parse was. The English has been changed from the passive "the marks which are stamped for those who follow to imitate" — which was rendering CC's impressa — to the active "the marks which I impressed for those who follow to imitate."

⚠ If a later reader wants the marker anyway, the thing to overturn is the 7109 parallel, not this note.

0794A · sed adhuc gloria et magnum decorem addes ei — [cj:] FIRED

The collation restored gloria (CC had gloriam). Here the plate really does not construe, and unlike 0788A there is no reading that rescues it: addes ei is transitive and takes magnum decorem in the accusative, so a nominative gloria is stranded in a coordinated pair with an accusative. It is a real Latin word wrongly inflected, which is Pattern 12's explicit exclusion and Pattern 18's proper territory.

• Marker: [cj: gloria; read gloriam, "glory"], placed immediately after the italic span and before the [0794A] anchor, mirroring the twin's own order.
• The compromise, stated plainly: English has no case endings, so "glory" renders the printed gloria and the conjectured gloriam identically — the reader cannot see the defect in the English at all, and the marker is the only thing that tells him it is there. That is exactly why it takes a marker where the one-sort defects above do not: those are invisible in Latin too (n for in, ac for ad), whereas this one is a real word that a Latin reader will stop on and an English reader would never suspect.
• ⚠ [cj:] inverts the [var:] content test: gloria must appear verbatim in the Latin twin, and it does. verify-english.mjs passes.

PLATE COLLATION · REGISTER OF RESTORED TYPE, cols 0784B–0794D

Wilson's ruling of 2026-08-28 governs (stated in full at the head of the matching section in cruces-0000.md): mark only where the reading is in doubt; record the rest here and disclose the count in the work-level page note.

MARKED in this range — one, and it was already argued above: 0794A [cj: gloria; read gloriam, "glory"]. Nothing further is marked here. In particular 0788A impressi stays unmarked — it construes, and the dissent is set out above.

CRUCES ONLY — recorded here, unmarked in the text.

Orthography and case, not defective type — no marker and no conjecture.

• 0789D · Exultavit for Exsultavit in the gloss (the lemma two words earlier keeps the s). An accepted spelling, not broken type. ⚑ This is the site whose neighbour produced the withdrawn [sic: cut], above.
• 0792B · Mittat tibi auxilium de sancto — Migne sets the lemma lower-case where the digitization had capitalized it. A compositor's case decision with no English exponent beyond capitalization, which the facing Latin column already shows.

Punctuation in this range is dealt with above (0789A, three marks).

THE BLIND POLARITY READ

Fenced reader on chunks 0014–0017 (Pss XVII–XX, cols 0784B–0794D — the end of the work), clause by clause against the Latin twin, with a mechanical negation-token count as backstop. No crux file and no collation report was open. Adjudicated 2026-08-28 under Wilson's marker ruling of the same session (stated in full in cruces-0000.md). Whole-work figures are in that file.

Result for this range: 2 sites — 1 ours, 1 Migne's. One new [cj:]. The reader returned zero confirmed defects on our side unaided; the one repair below came out of a low-confidence candidate that survived adjudication.

Marked — Migne's own defect, our English literal and unchanged

• 0790D, VERS. 11 · [cj: relinquetur; read eliquetur, "shall be pressed out"]. …cui opus sit aliqua pressura, non opprimentis, sed exprimentis manus Dei, qua de temporali vita in aeternum relinquetur. The whole sentence is the wax-and-honeycomb image, built on a hand that does not crush but presses out; relinquetur ("shall be left") nearly reverses the movement its own neighbour describes, and de … in … hangs on it awkwardly. Our English is correctly literal and stays ("by which he shall be left … from a temporal life unto eternity"); the marker carries the conjecture. It sits beside the [cj:] this chunk already had at VERS. 6 (gloria → gloriam, 0794A).

Repaired on our side

• 0787C, VERS. 34 — the gloss overshot its own lemma's preposition. Lemma et super excelsa statuens me, rendered "setting me upon high places"; the gloss re-quotes the same preposition — et super coelestem habitationem figit intentionem meam, ut implear in omnem plenitudinem Dei — and our English gave "fixes my purpose above the heavenly habitation", which makes the aim overshoot heaven and is not what the purpose clause supports. Now "upon the heavenly habitation". The translator's default "above" is right nine columns earlier at VERS. 11 (Exaltatus est super plenitudinem scientiae) and appears to have been carried across.

Adjudicated and DECLINED

• 0791A, VERS. 13 — Quae ipsum occultum claudunt, cui suavitas est veritas, rendered "that very hidden man", against the next gloss's occultum suum / alienum rendered "hidden thing" twice (reader's confidence: low; the reader flagged the inconsistency across two adjacent glosses, not either rendering). Declined — both renderings are right, and the inconsistency is Migne's (Augustine's), not ours. At VERS. 13 the two dative relatives (cui suavitas est veritas, cui desiderabiliora et dulciora sunt judicia Dei) require a person, and they pick up the man of VERS. 11 (dulciora illi sunt judicia Dei, quam sibi ipsi est); "the hidden man" is the standing idiom (absconditus cordis homo, I Pet. III, 4). At VERS. 14 the same word is doing an entirely different job — Non enim est tertia origo peccati praeter occultum suum … et alienum — where it names one of the two origins of sin and can only be a thing. Levelling the two would make one of them false.

Positive results worth recording

• Every counter-intuitive negation in the range is carried, and these are precisely the seats where a smoothing repair would have been tempting: 0790C, VERS. 10, non foras mittit consummata dilectio timorem hunc — perfect love does not cast out this fear — carried against the familiar I Joan. IV, 18; 0789D, VERS. 4, Per quas non auditae sunt voces evangelistarum, the double negative of Ps XVIII, 4, not "fixed" into an affirmative; 0794B, Ps XX VERS. 12, three negatives in one clause, all three present.
• 0788C, VERS. 31 nisi manducaverint viam fidei — the printed manducaverint is kept and the conjecture carried in a [cj:], not substituted. That is the shape 7a exists to prevent, handled right.
• 0789B, VERS. 44 — Migne's conflation of Joan. XI, 48 and XII, 19 is rendered literally with the divergence in a [var:]; 0789C, VERS. 50 — the lemma confitebor against the gloss's confitebuntur gentes is a real lemma/gloss mismatch and it is Migne's, carried as printed on both sides. Two more 7a″-shaped seats, neither conformed.