Expositio in Cantica canticorum

Merged 2026-07-30 from four agent batches (0000–0006 · 0007–0012 · 0013–0017 · 0018–0022). 23 chunks, 27,638 Latin words, 229 notes. verify-english OK. First English translation — workStatus: none, verified at triage.

The Song of Songs rule, and how it governed this work

Wilson's ruling, 2026-07-30: each commentary stands on its own; a lemma is never harmonized with another commentary. Sixteen PL commentaries expound this book, and imposing one English across them would be editorializing Migne. So every choice below was derived from THIS plate. Where the result differs from another Song commentary on the site, that is correct and must not be swept.

The rule's second case — "where a gloss depends on the word, the Latin stands" — did most of the work here, and the agents applied it four times to words nobody pre-set:

• murenulae (pre-set) — Murenulae autem dicuntur a similitudine piscis marini, qui murena vocatur. Kept in italic Latin.
• vermiculatas argento → "vermiculated with silver", because the next clause glosses it in modum vermium terrenorum; an English word without the verm- root breaks the gloss exactly as translating murenulae would.
• elatae palmarum — glossed through the Greek ἐλάτη at 1085C. Italic Latin.
• malum granatum / malum punicum — glossed etymologically at 1101A (granorum … Punica enim ipsa est Africa), so both stay Latin in that passage only; "pomegranate" everywhere else, matching the 12 instances in the peer chunks.

⚠ Site furniture is NOT a lemma. Vers. N. was unified to "Verse N." across the work (Migne prints the abbreviation uniformly; the English was split 76/40). That is apparatus, like the project's existing CAPUT → CHAPTER, and site-wide consistency there is right. The ruling governs the words of Scripture, not the furniture around them — do not read this sweep as licence to harmonize a lemma.

Work-wide conventions (reconciled at merge)

• ferculum (Song 3:9, 1070D) → "litter", reached independently by two agents. This work does NOT gloss it etymologically, only allegorizes it (designatur per ferculum, sancta videlicet Ecclesia), so the Latin need not stand. "Dish" is impossible — nothing is served on it; "palanquin" loses the couch. "Litter" carries both the conveyance Solomon made and ascended and the bed the saints recline on ad epulas, and keeps media hujus ferculi constrata sunt construable. It also stays clear of lectus/"bed" and lectulus/"little bed", which this work uses heavily and which 3:9 explicitly equates with the ferculum. Bruno of Segni (10804) glosses the same word as a banqueting table and renders it accordingly. Both are right on their own plates. Do not reconcile them.
• Pre-set and honoured throughout: dilectus/dilecta/dilecte → "beloved" · amica/amicus/amice → "friend" (both families occur here and must stay distinct) · sponsa/sponsus → "bride"/"bridegroom" · pulchra "beautiful" · formosa "comely" · speciosa "fair" · nigra "black" · fusca "swarthy" · columba "dove".
• Swept at merge: doctores → "teachers" (the 0018–0022 agent drafted "doctors" ×8, diffed against the peers 7-to-1, and swept itself) · maligni spiritus → "malignant" (×3) · castrorum acies ordinata → "a battle-line of the camps set in order" (×2 back-references, conformed to the peers' expounded lemmata).
• Independently converged, no reconciliation needed: decora → "seemly" (kept clear of pulchra/formosa) · electus → "chosen" · species/speciosus → "fairness/fair" · tornatilis → "turned" · pacificus → "peaceable" (the Salomon etymology depends on it). Back-referenced lemmata verified word-identical across batches.

Rendered as printed, NOT repaired — the ones that matter

• 1069B — quae circumcisione mundata, nec adhuc baptismo sanctificata jam currit ad Deum?, said of the Church of the Gentiles, which by this work's own argument was neither circumcised nor baptized. The sentence wants a first nec that is not printed, so as it stands it asserts she WAS cleansed by circumcision. Nothing supplied.
• 1097A — singular statura with plural assimilantur. English HAS a number exponent, so smoothing it would have been the textbook Pattern 7a silent repair. The plural is carried: "The stature therefore of the Church are likened to the palm."
• 1082D — in occupato et libero corde, which contradicts libero and the speech it sits in (the soul has just said it stripped off worldly cares). Conjecture non occupato; contradiction left standing.
• 1093C — Migne prints a doubled stop .:. Both marks carried per Pattern 8. This will look like a build defect on the page and is not one.
• Also: Matt 24:12 at 1082B was first drafted as "iniquity" (the familiar English) and reverted to "sin" for Migne's peccato — the scripture-substitution trap caught mid-draft by the agent itself.

## Two [var:] — Migne's citation diverges from the received text 2 Tim 1:8 quoted as crucem Christi against the Vulgate's testimonium Domini nostri, at 1074A and 1090D, marked identically by two agents independently. Load-bearing both times: the gloss makes the pomegranate rind the mystery of the passion, so "cross" is the hinge of the argument.

## For the indexer's operator — the inline-citation class again Two citations are set by Migne as running text rather than as [n:] notes, so they reach no index: (Isai. LXVI, 2) at 1058D (with a column anchor falling inside it) and Haggai 2:24 at 1102C. Both reproduced verbatim, both deliberately NOT tagged [f:] (Pattern 4's trigger is a citing PRACTICE, not a stray instance). This class is now roughly eight corpus-wide and still has no harvester.

Per-batch logs

The four batch logs follow verbatim, in chunk order.

Auctor incertus (Cassiodorus Vivariensis?), Expositio in Cantica canticorum, PL 70, cols 1055C–1070C. Batch: chunks 0000–0006 (Monitum, Prologus, Capp. I–III.7).

Work-wide conventions (for merge reconciliation)

### The pre-set terms, as applied here

• murenulae — kept in italic Latin, untranslated, throughout (0002, Song 1:10 and its exposition). The work's own etymology (a similitudine piscis marini, qui murena vocatur) requires it. Note the knock-on choice: vermiculatas argento is rendered "vermiculated with silver" — not "inlaid" or "set" — because the very next clause glosses it in modum vermium terrenorum, quos lumbricos dicunt. An English word without the verm- root makes that gloss explain nothing, exactly as with murenulae. lumbricos is likewise kept in italic Latin, since it is the glossing term.
• dilectus/dilecta/dilecte → beloved; amica/amicus → friend; sponsa/sponsus → bride/bridegroom; pulchra → beautiful; formosa → comely; speciosa → fair; nigra → black; fusca → swarthy; columba → dove. Applied without exception in 0000–0006.
• Named etymologies translated, name kept as printed: Cedar … qui interpretatur tenebrae (0001), Salomon … qui pacificus interpretatur (0001), Engaddi … fons haedi interpretatur (0002), Bethel … domus consurgens, vel domus vigiliarum (0005), Israel, id est vir videns Deum (0006). Also Chrisma Graece, Latine unctio (0001) — both Greek and Latin terms kept in italic Latin with an English gloss ("anointing"), since unde Christus is the point.

ferculum (Song 3:9) — DECISION: "litter"

ferculum does not occur in this batch (it enters at 0007, col. 1070D), but the decision is recorded here for propagation, per the batch brief.

Rendering: "litter." Reasoning, from this work's own framing and nothing else:

1. The work does not gloss the word etymologically. It allegorizes it — Quod superius per lectulum Salomonis, sic nunc designatur per ferculum, sancta videlicet Ecclesia, in qua velut in lecto requiescunt sancti Dei, et velut in lecto discumbunt ad epulas. So Pattern-9 case 2 (gloss depends on the word) does not apply, and the Latin need not stand.
2. The two things the exposition needs the object to do are be reclined on as a bed (velut in lecto requiescunt) and be reclined on at a banquet (discumbunt ad epulas). "Litter" — a portable couch — carries both, and carries them at once, which "palanquin" (conveyance only) and "bed" (already spent on lectus/lectulus) do not.
3. lectulus is heavily used in this work (Song 1:15, 3:1, 3:7) and is rendered "little bed"; lectus is "bed." The exposition at 1070D turns on ferculum being the same thing under a different name as the lectulus Salomonis — so ferculum must be a word that is clearly distinct from those two and still recline-able. "Litter" is.
4. Downstream at 3:10 the work reads media hujus ferculi constrata sunt — "the middle parts of this litter were spread/paved" — which "litter" supports without strain.

Merge note: if a later agent has already shipped a different word, this is a WITHIN-work consistency question (the boundary in translation-style.md §Song of Songs), so it must be reconciled to one term, not left to stand.

### Other terminology fixed in this batch (not pre-set)

• decorus / decora → "seemly", decor → "grace." Forced: the obvious English ("comely") is reserved for formosa and "fair" for speciosa, and all three co-occur in this work (formosa mea at 1065A, speciosa mea at 1066A, pulcher … et decorus at 1061B, facies tua decora at 1066B). Distinctness on this work's own pages required a fourth word.
• turtur → "turtle-dove", kept distinct from columba → "dove." Both are expounded, differently (chastity/widowhood vs. simplicity), at 1059C and 1061B/1065C.
• paries → "wall"; maceria → "stone wall." The work glosses maceria as built ex siccis lapidibus for vineyard fortification (1066A–B), and both words stand within two columns of each other, so they must be tellable apart.
• lectulus → "little bed"; lectus → "bed"; accubitus → "reclining." The last is driven by the gloss Accumbere enim est reclinari vel discumbere … se … inclinavit (1060B).
• caprea → "roe"; cervus → "hart"; hinnulus → "fawn." Required by the exposition at 1064C, which turns on the roe being the smaller animal (minoribus videlicet animantibus).
• cellaria → "cellars" (1057C), cella vinaria → "wine cellar" (1062D) — same root kept visible, since the second passage recalls the first.
• senarius / denarius → "the senary" / "the denary" at 1070A, so that the arithmetic (sexies deni, vel decies seni) reads, while denarium … accepisse stays "a denarius" — the pun is the argument.
• grossi → "green figs" (1065D), because the gloss defines them as primitivae et immaturae ficus.
• per speculum et in aenigmate → "through a glass and in a riddle" throughout (1064D, 1067A); in figura et aenigmate (1067B) → "in figure and in a riddle."

Carried type — [sic: …] (Pattern 7 / 10 / 12)

### 1. disperi for dispersi — 0001, col. 1058C→D Printed: ad hoc enim divinitus disperi sunt, ut plures Ecclesiae fierent. disperi is not a Latin word. It stands three clauses after Migne prints dispersi sunt correctly twice in the same sentence (ubi nos legimus, dispersi sunt), which makes the conjecture dispersi effectively certain and the defect a compositor's dropped s. Carried untranslated in italics inside [sic: disperi]; the surrounding clause is rendered ("for to this end were they divinely [sic: disperi], that more Churches might be made"). No governing particle or negation is inside the marker (7a′ check: clear).

### 2. il est for id est — 0004, col. 1066A Printed: ibi nidificat, il est fetus bonorum operum congerit. il is not a Latin word; the run is a Pattern 10 division/substitution defect on the standard formula id est, which this work prints scores of times. The whole run is carried ([sic: il est]), not just il, per Pattern 10 — rendering "that is" for half of it would misrepresent the plate while looking considered. The same defect recurs at 0009; a later agent has marked it identically, so no reconciliation is needed. 7a′ check: the run is an appositive formula governing nothing; the clause it introduces (fetus bonorum operum congerit) is translated in full and loses no force.

### 3. unversi for universi — 0006, col. 1069D Printed: Et unversi (inquit) pulveris pigmentarii, id est congerie omnium virtutum. unversi is not a word; the lemma is quoted correctly as universi pulveris pigmentarii two columns earlier at 1069A, so this is a dropped i in the re-quotation, not a variant reading. Carried as [sic: unversi]; Et is a real word and stays outside the marker, translated. Note the marker sits inside an italic lemma — nested italics are not expressible, so the carried run takes the single italic pair of the marker content only. 7a′ check: the gloss id est congerie omnium virtutum renders omnium ("all"), so the universality the mangled word carries is not lost from the translated text.

Readings rendered literally, no marker (Pattern 7 / 12 boundary)

These are real words that are merely wrong or odd. Per Pattern 12 they are rendered as ordinary English — the word on our page is ours — and the conjecture lives here.

### 4. Itera for Iterat — 0006, col. 1068B Printed: Itera autem difficultatem inventionis, et dicit: quaesivi, et non inveni. Itera is a well-formed imperative ("repeat!"), but the sentence is third-person narration about the author's technique (et dicit), so Iterat ("he repeats") is near-certainly meant. Rendered as printed: "Repeat, however, the difficulty of the finding, and he says…". The English is as odd as the Latin, deliberately. Not marked — it is a real word, and marking it would blame the plate for a rendering that is ours.

### 5. Missing first nec — 0006, col. 1069B ⚠ NEGATION Printed: Quanta laude et admiratione digna, quae circumcisione mundata, nec adhuc baptismo sanctificata jam currit ad Deum? The subject is Ecclesia de gentibus — the Church of the Gentiles, which by this work's own argument was neither circumcised nor yet baptized. Sense requires nec circumcisione mundata, nec adhuc baptismo sanctificata; the printed text has only the second nec, so it asserts that she was cleansed by circumcision. Rendered exactly as printed — "who, cleansed by circumcision, and not yet sanctified by baptism, already runs to God?" The printed nec is carried; no nec is supplied. This is a Pattern 7a case of the purest kind: the smooth reading would require inventing a negation Migne does not print, and the contradiction with the surrounding argument is precisely the evidence the edition exists to expose.

### 6. labores aeternos — 0003, col. 1062B Printed: Videns sponsam suam ad requiem festinare, et labores aeternos fugere, ac per quietem praesentis temporis ad aeternam beatitudinem tendere. "Fleeing eternal labours" contradicts the sentence's own destination (ad aeternam beatitudinem tendere) and the moral that follows (per pressuras et labores hujus saeculi illuc pervenire). Conjecture: externos ("outward"), i.e. the labours of public action she is being told she cannot skip — which is exactly this chapter's theme (cf. 1065A, de secreto quietis egredere ad publicam actionem). Rendered as printed: "flee eternal labours." No emendation.

### 7. Inter omnes virtute primatum tenet charitas — 0003, col. 1063A Printed virtute (ablative singular) where virtutes ("Among all the virtues") is expected and gives the standard sentence. Rendered as printed: "Among all, charity holds the primacy in virtue." An ablative of respect is defensible Latin, so this may not be a defect at all; logged for the record, no marker.

### 8. sexcenta et sex millia — 0001, col. 1059C Printed as the number of Israelites who left Egypt, with [n: (Num. II, 32; Exod. XIV, 22)]. Num. 2:32 gives 603,550. Rendered as printed: "six hundred and six thousand." No [var: …] supplied — the divergence is in the exposition's own arithmetic, not in a quoted scripture text, so Pattern 14's trigger is not met.

### 9. qui sodales Christi sunt — 0001, col. 1059A Of the conventicula haereticorum: the plate says the heretics are companions of Christ, justified only by quia et ipsi greges suos pascunt. The looser sense ("who are called companions") is not what is printed. Rendered literally.

### 10. Aemulantur vos non bene — 0004, col. 1065D The [n: (Galat. VI, 13)] citation covers a conflation: Aemulantur vos non bene is Gal. 4:17, sed volunt vos circumcidi, ut in carne vestra glorientur is Gal. 6:13. Rendered as printed, conflation intact, with non carried ("They are zealous for you, not well"). The note stays verbatim per the marker rule; per the standing false-positive list an [n:] locator is never corrected.

### 11. (Joan., IV, 19) — 0003, col. 1063A Migne's note for Diligamus Deum, quoniam ipse prior dilexit nos, which is 1 John 4:19 (and whose Vulgate reads Nos ergo diligamus, quoniam Deus prior dilexit nos). Note reproduced verbatim; text rendered as printed. Logged only so a later reader does not read it as our slip.

### 12. (Isai. LXVI, [1059A] 2) — 0001, col. 1058D–1059A This citation is not in an [n: …] marker — Migne prints it inline, inside the italic run of the quotation, and a column anchor falls inside it. Reproduced exactly as printed, anchor included. It is therefore not counted in noteCount and will not be harvested as a note; flagged in case the indexer should pick it up by another route.

### 13. Punctuation kept as printed (Pattern 8)

• 0006, col. 1068C: plenitudo gentium, introierit, tunc omnis Israel salvus fiet — the comma after gentium breaks the clause and is preserved in the English.
• 0006, col. 1068D: quomodo requiescebat? sed dulcissimus… — question mark followed by lower-case sed; both preserved, no capital supplied.
• 0000: the whole MONITUM is set in italics by Migne (it is the Maurist editor's note, in his own voice, and it argues against the Cassiodoran attribution the byline carries). Italics preserved across the whole paragraph. It is editorial prose but it is body text, not a note, so Pattern 15 [nt: …] does not apply.

Not defects — checked and deliberately left alone

• Column bands: 1055C→1055D→1056C skips a whole band; 1057A closes 0000 and 1057B opens 0001; 1064C closes 0003 and reopens as colContext of 0004. All positional-guide behaviour per the standing false-positive list. No text breaks mid-word or mid-construction at any boundary in this batch — checked at all six.
• (cont.) on the repeated ## CAPUT PRIMUM. / ## CAPUT II. heads (0002, 0004, 0005): kept as-is, marker untranslated.
• [n: (Ibid. X, 22)] (0006) and [n: (Ibid., 12)] (0006) resolve to Act. VIII and Ephes. VI respectively; left exactly as printed, no correction proposed.
• Migne's Vers. numbering restarts per caput and occasionally lags the lemma it heads; not touched.
• The work quotes Song lemmata that diverge from the received Vulgate in small ways (e.g. Osculetur me osculo oris sui for osculo oris sui pl.; pelles Salomonis; Nolite me considerare quod fusca sim). None changes the sense on which an exposition turns, so no [var: …] was raised. Per Pattern 14's own caveat, absence of a [var:] here is not a claim that the lemmata agree with the Vulgate. # Cruces — 7914 (ps.-Cassiodorus, Expositio in Cantica canticorum), chunks 0007–0012

Agent batch: chunks 0007–0012 (cols 1070D–1083C). Frontmatter copied verbatim from each Latin twin. All column anchors, [n:] notes, ## heads, page marks (514 …521) verified 1:1 against the Latin by script.

Work-wide conventions set by this batch (for merge reconciliation)

Pre-set terms observed as instructed (murenulae untranslated — does not occur in 0007–0012; dilectus/dilecta → beloved; amica/amicus → friend; sponsa → bride, sponsus → bridegroom; pulchra → beautiful; formosa → comely; speciosa → fair; nigra → black; fusca → swarthy; columba → dove). Additional choices made here:

### ferculum (Song 3:9, col. 1070D) — RULED: "litter" The prompt flagged this as the known crux. This work does not gloss ferculum etymologically; it allegorizes it: sic nunc designatur per ferculum, sancta videlicet Ecclesia, in qua velut in lecto requiescunt sancti Dei, et velut in lecto discumbunt ad epulas. The framing is therefore (a) a conveyance Solomon made of Lebanon wood, with silver pillars, a golden reclining-place and a purple ascent — one climbs up into it — and (b) a thing one rests on as on a bed and reclines on to feast. "Dish" (the commonest lexical sense of ferculum) is impossible: nothing is served on it. "Palanquin" and "portable throne" carry the conveyance but not the couch. "Litter" carries both — an English litter is a couch that is carried and ascended — and it keeps the later media hujus ferculi constrata sunt ("the midst of this litter is paved") construable. Used consistently: 0007 ×4 (the lemma, the sic nunc designatur, the media…constrata, the hujus ferculi). Since the gloss is allegorical and not etymological, keeping the Latin was not required (Song-of-Songs rule, case 2 does not fire).

### Other terminology fixed in this batch

• reclinatorium → "reclining-place" (0007, 1071A–B). Glossed as requies aeterna; "back-rest"/"headboard" would fight the rest-gloss.
• ascensus purpureus → "purple ascent" (0007). purpureus → "purple", coccineus → "scarlet", coccus → "scarlet" (0008).
• charitas → "charity" throughout; dilectio → "love"; diligere → "to love". Both occur side by side (0008 gemina charitate et dilectione; 0010 charitatem Dei et proximi), so they must stay distinct.
• lectus → "bed", lectulus → "couch" (0007 has both in one sentence).
• decora → "seemly" (0008 lemma), reserving "comely" for formosa per the preset.
• aromata → "aromatical spices" (Song lemma vocabulary, 0010–0012); unguenta → "ointments"; thus → "frankincense"; fistula → "sweet cane"; cyprus → "cypress"; nardus → "spikenard"; crocus → "saffron"; stacte kept as "stacte".
• malum punicum → "pomegranate"; malum granatum → "the grained apple" (0008, where the work is explaining the second name, so it must be rendered, not equated).
• doctor → "teacher" — reconciled at draft time, note for the merge. This batch was first drafted with "doctors" (to keep ordo doctorum legible against ordo martyrum), then swept to "teachers" because the peer chunks already on disk render doctores that way by a wide margin (0001, 0004, 0006, 0013, 0014, 0015, 0016 = "teachers"; only 0018 = "doctors"). Within-work consistency is the required merge step, so the minority reading was the one to move. 0018 still reads "doctors" (6 instances) and should be swept to match. magistri stays "masters" (0008 @1073A prints sancti doctores et magistri Ecclesiae, so the two must stay distinct).
• Galaad acervus testimonii interpretatur → "the heap of testimony"; Libanus … candidatus, vel candidatio … vel dealbatio → "white-robed / whitening / made white"; turris David … manu fortis, vel visu desiderabilis → "strong of hand / desirable in sight". Names kept as printed (Galaad, Libanus→Lebanon, Sanir, Hermon, Amana, Sion).
• anagoge → "anagogy, that is, the higher sense" (0009, the work glosses it itself).

Carried type — [sic: …] (4 instances) + supporting notes

1. 0008 @1072B — esset for es et. Lemma prints Quam pulchra es, amica mea; quam pulchra esset decora (Song 4:1). The commentary in the very next sentence says Vox Christi bis repetit, quam pulchra es — twice, which requires a second es. This is the Pattern 10 run-together class (exactly sedet for sed et), so the whole broken token is carried untranslated and marked; it sits inside an italic lemma, so it takes no second italic marker. Conjecture: quam pulchra es et decora.
2. 0008 @1073B — spollatione for spoliatione. Non-word; carried in italics inside plain prose. Conjecture certain from context (rerum spollatione gaudentes, paired with voluntaria paupertate).
3. 0009 @1077C — ac for lac. Prints Sub lingua ergo Ecclesiae mel et ac est where the lemma two lines above is Mel et lac sub lingua tua and the whole gloss turns on lac ("milk is fitting for little ones"). Rendered literally the token would produce "honey and and", which renders nothing; carried and marked instead. Conjecture: lac.
4. 0010 @1078D — metalia for metalla. Non-word; inter omnia metalia pretiosius est aurum. Carried in italics.
5. 0012 @1081A — aromalibus for aromatibus. Non-word, inside the italic lemma Messui myrrham meam cum aromalibus meis, so no second italic marker. The commentary's own next sentences use aromata and cum aromatibus suis correctly, which fixes the conjecture.

[var: …] (1 instance)

• 0008 @1074A — 2 Tim. 1:8. Migne prints Noli erubescere crucem Christi, neque me vinctum ejus; the Vulgate has testimonium Domini nostri. The divergence is load- bearing here: the whole paragraph argues that the doctors nec erubescunt crucem ejus, so the proof-text as printed says exactly what the argument needs and the received text does not. Migne rendered as printed; [var: Vulg. testimonium Domini nostri, "the testimony of our Lord"] set beside it. No speculation offered as to the cause.

## Scripture divergences logged but NOT marked Marked only where the sense turns on the difference (Pattern 14 is deliberately sparing). These are recorded so a later sweep can reconsider:

• 0009 @1075C, Job XIV, 4: Migne prints Nemo est super terram sine peccato, nec infans unius diei; the Vulgate verse is Quis potest facere mundum de immundo conceptum semine? The printed form is the familiar patristic paraphrase and the argument does not turn on the wording.
• 0010 @1077D, note (Luc. VIII, 1) for Oportet semper orare — Luke 18:1. Note contents are verbatim; no correction proposed (and the standing rule forbids correcting a printed reference).
• 0012 @1082B, Matt. XXIV, 12: Migne prints Abundante peccato refrigescit charitas where the Vulgate has abundavit iniquitas. Rendered "When sin aboundeth" — a first draft had smoothed this to "iniquity" and it was reverted; recording the near-miss because it is precisely the Pattern 7a substitution class.
• 0012 @1083B, note (II Cor. IX, 27) for Castigo corpus meum — I Cor. 9:27. Verbatim, no correction.

## Punctuation carried per Pattern 8 (no marks supplied, none deleted)

• 0010 @1079A — Fistula. brevis est arbuscula. A full stop with a lowercase continuation, where every parallel lemma-gloss in the same paragraph runs on (Crocus aurosi coloris habet florem, Cinnamomum et ipsa brevis est arbuscula). The stop and the lowercase are both reproduced: "Sweet cane. it is a low shrub…". Reads oddly by design; the mark is Migne's.
• 0012 @1083A — no terminal stop after …haec Conditor pro nobis agere dignatus est before Nam quod venter cor significet. None supplied.
• 0008 @1072B — In opere, quia, a capitalized continuation after a comma (in opere videlicet et praedicatione, In opere, quia…). Both comma and capital kept.
• 0007 — Vers 9. (no stop after Vers) where every other verse label prints Vers. This is the Pattern 9 section-label class, normalized to "Verse 9." in English; the misprint stays visible on the facing Latin column.
• Verse-label form — MERGE ITEM. This batch uses "Verse N." throughout. The peer chunks already on disk are split: "Verse N." in 0013–0016, 0018 (and this batch); "Ver. N." in 0001–0006. The Latin prints Vers. everywhere. One form must win at merge; "Verse N." is currently the majority and is the fuller normalization, but the choice is the merge's, not this batch's.

## Grammar rendered as printed, no emendation (Pattern 7)

• 0008 @1074D — sicut duo hinnuli gemelli capreas esse dicuntur. capreas is accusative where the construction wants capreae. Rendered "just as two twin fawns are said to be roes", carrying the word without asserting a case the plate does not print. The same sentence's per capreas designatus est (of Christ) is printed and rendered as a plural, which is itself odd against the singular caprea of the earlier chunks — flagged for the merge in case a peer agent has fixed a singular convention.
• 0009 @1076B — propter talium conversationem in a clause otherwise entirely about conversion (quando principes saeculi ad fidem Christi convertuntur … quando superbos … convertit, et pro illorum quoque conversione …). Rendered "manner of life" — the word Migne prints. Conjecture: conversionem. Not silently swapped; this is the 8625/11547 substitution class.
• 0010 @1079A — Cinnamomum et ipsa brevis est arbuscula … ideoque designant.* Plural verb, singular subject. Rendered "they designate" (the plural has an English exponent, so it is carried, unlike the Pattern 9 gender class). Probably attracted to the following fistula et cinnamomum* pair.
• 0011 @1080C — flatibus persecutionem modum imponit quemdam. Two accusatives; the construction (modum imponere alicui rei) wants a genitive with flatibus. Rendered "he imposes upon the blasts a certain measure of persecution" — the word is rendered, the case is not asserted. Conjecture: persecutionis / persecutionum.
• 0012 @1082D — quae nunc in occupato et libero corde divino fruor contuitu. occupato contradicts libero and contradicts the whole speech (the soul has just said it stripped off the cares of the world). Rendered literally, "in an occupied and free heart" — the self-contradiction is the plate's and is left standing. Conjecture: non occupato or expedito. Flagged for the polarity sweep: this is a dropped- negative candidate of exactly the 11536 shape.

## Attested/period spellings — translated normally, no marker (Pattern 9) Judeae (0009 @1076A) for Judaeae, beside Judaea in 0010; dulcidinem (0012 @1081A) for dulcedinem; spiritualia/spiritualis, charitas, coelestis, moerores, praelio, thymiamatis, sollicitudo, quotidie. None italicized, none marked.

## Renderings worth a merge check

• 0012 @1080D — the lemma Veni in hortum meum. The commentary's whole point is that veni here is not imperative but perfect indicative (Nam veni, hic, non est imperantis, sed praeteriti temporis indicativum). English cannot print the ambiguity, so the lemma is rendered on the author's own reading ("I am come into my garden"), and the Latin word veni is kept in italic Latin inside the gloss so that the observation still has something to observe. If a peer agent renders a parallel imperative/perfect crux elsewhere, match this treatment.
• 0007 @1071C — quae quanto majoris fragilitatis sibi sunt consciae rendered "the more conscious they are of their own greater frailty", keeping the comparative (Pattern 7a mechanism 4).
• 0009 @1075B — omnes revelata facie gloriam Domini contemplabuntur: the [n:] falls inside the quotation between Domini and contemplabuntur, so the English splits the same way ("all with unveiled face shall contemplate the glory of the Lord [n: …]"). Marker position preserved, italics rebalanced around it.
• 0010 — the frankincense natural history (Thus arbor est Arabiae … mannam vocamus, cols 1077D–1078A) is Isidore/Pliny matter and is translated as running prose, not treated as apparatus. masculum / mannam are rendered as the plain English trade names "male" and "manna" without italics, matching the Latin, which does not italicize them.

## Negation audit (Pattern 7a / 7a′) Every printed non, nec, neque, nisi, nullus, nullatenus, nunquam, nihil, nemo, ne in 0007–0012 was checked into the English one by one after drafting. Two carry notes:

• 0010 @1077D — Nec semper ante Dominum hoc aliter impleri potest, nisi…. Both nec and nisi carried ("Nor can this be fulfilled always before the Lord in any other way, unless…"); the semper is kept where Migne puts it even though it strains the sentence.
• 0012 @1082C — sine quibus officium praedicationis aut vix aut nullatenus agitur. nullatenus is printed correctly here (the prompt's mullatenus trap does not fire in this batch) and is rendered "or not at all". No [sic: …] marker in this batch encloses a negation, particle or conjunction governing translated text — the 7a′ condition was checked explicitly for all five markers. # Cruces — 7914, chunks 0013–0017 (cols 1083D–1095D)

Auctor incertus (Cassiodorus Vivariensis?), Expositio in Cantica canticorum, PL 70. Batch: chunks 0013, 0014, 0015, 0016, 0017.

Work-wide conventions set in this batch (for merge reconciliation)

Pre-set by the session brief and followed throughout: murenulae untranslated; dilectus/dilecta = beloved; amica/amicus = friend; sponsa = bride, sponsus = bridegroom; pulchra = beautiful; formosa = comely; speciosa = fair; nigra = black; fusca = swarthy; columba = dove. Named etymologies translated with the proper name kept as printed.

Additional choices made here, none of them pre-set. These need reconciling across the other three agents' batches:

No guillemets occur anywhere in chunks 0013–0017; the Latin twins have none either, so nothing was supplied (Pattern 6).

Carried type — [sic: …] (Patterns 7, 10, 12)

### 1085C — multi udines for multitudines Migne prints Comae sponsi sunt multi udines fidelium. Split type (Pattern 10); the whole broken run is carried, spacing included, and marked [sic: multi udines]. The word is certainly multitudines — the same phrase runs unbroken three lines later (quia fidelium multitudines et virore fidei gaudent), which is the internal evidence.

### 1086B — perserutantur for perscrutantur sancti doctores assidue … mysteria perserutantur. Not a word (r for c); carried untranslated per the non-word class of Pattern 7 and marked. Conjecture perscrutantur, "search out" — near-certain, and the sense of the sentence is not in doubt.

### 1086D–1087A — crant for erant Aureae enim crant, quia operatio illius divinitatis peragebatur potentia. Not a word; carried and marked. Conjecture erant ("For they were golden"), which the imperfect peragebatur in the following clause requires.

### 1087C — Crara for Crura (inside an italic lemma) The lemma of Song 5:15 is printed Crara illius columnae marmoreae; the very next sentence of the exposition prints Crura sponsi itinera sunt Salvatoris, so the plate itself supplies the correct reading. The broken run is carried inside the lemma. Per Pattern 10 it takes no second italic marker — nested italics are not expressible — so on the page it reads His [sic: Crara] are pillars of marble….

### 1092C — quaies for quales …quaies erant illi quibus Apostolus dicebat. Not a word; carried and marked. Conjecture quales, "such as".

### 1094B — me cedis for mercedis pro sola dilectione et supernae me cedis exspectatione hoc facere debet. Split type (Pattern 10); the whole run including the real word me is carried, per the Ad dam rule. Conjecture mercedis, "of the [heavenly] reward" — required by exspectatione and paralleled at 1091A (intuitum supernae remunerationis).

### 1095D — aeras for aerias / aereas Significat etiam aeras potestates, quae animarum nostrarum cruorem sitiunt. Borderline case, and the judgment is recorded here because it sits on the Pattern 12 boundary. aeras is a real form (accusative plural of aera, "counters, eras"), which would normally send it to the "real word merely wrong" side and out of the marker. But in this construction it stands where an adjective agreeing with potestates must stand, and rendering it as the noun would produce English that misrepresents the sentence rather than reproducing it. Carried and marked; conjecture aerias/aereas, "the powers of the air" — the standard phrase, and the same phrase the work prints uncorrupted eight columns earlier at 1092B (Terribilis aeris potestatibus).

Rendered as printed — no marker (Pattern 7, real words merely wrong)

### 1094D — clari for clavi, Eccl. 12:11 Migne prints Verba sapientium quasi stimuli, et quasi clari in altum defixi. The Vulgate has clavi, "nails", which is what in altum defixi ("fixed deep on high") governs. clari is a real word ("bright ones"), so under the Pattern 12 boundary it is rendered as ordinary English — "as bright ones fixed deep on high" — and not marked. The exposition does not build on the word (it goes straight to quae data sunt a pastore uno, "namely Christ"), so nothing in the argument is lost by the plate's reading. Conjecture clavi recorded here only.

### 1088C — plurumque for plerumque Ad hoc enim plurumque differt vota et orationes sanctorum. Attested Old-Latin / medieval orthography, not a defect — Pattern 9. Translated normally ("very often"), no italics, no marker. Flagged only so a later sweep does not churn it.

### 1091C–D — colombae for columbae qui in specie colombae apparuit. Attested late/medieval spelling (Pattern 9); translated normally as "dove". Same non-action as above.

### 1088A — Judeae for Judaeae Standard medieval orthography, Pattern 9. Rendered "Judea".

Punctuation carried per Pattern 8

### 1093C — the doubled stop .: Migne prints …ideoque a gressibus coepit eam laudare.: Gressus enim Ecclesiae sunt velox praedicatio apostolorum. Both marks are printed and both are carried; neither was deleted and none supplied. The English reads "…began to praise her from her steps.: For the steps of the Church are…". This looks like a build defect on the reading page and is not one; it is the plate.

### 1087B — comma absent after frigidissimae …et naturae frigidissimae unde et a dracone, qui calidae naturae est, vehementer appetitur. No mark supplied; the English runs on as the Latin does.

Scripture divergence — [var: …] (Pattern 14)

### 1090D — 2 Tim. 1:8 Migne prints Noli erubescere crucem Christi, neque me vinctum ejus; the Vulgate has Noli itaque erubescere testimonium Domini nostri, neque me vinctum eius. Migne is rendered ("Be not ashamed of the cross of Christ, nor of me his prisoner") and the received reading is noted in a [var:]. The divergence is load-bearing here: the whole point of the sentence is that the cortex mali punici signifies mysterium passionis Christi, and the proof-text as printed says "cross" where the Vulgate says "testimony". Whether the substitution is Migne's compositor, his copy-text, an Old Latin reading or the author quoting from memory is not adjudicated (Pattern 14 forbids speculating).

### Divergences noted but NOT marked Marked sparingly per Pattern 14; these were judged not to change the sense of the argument, and each is recorded here instead.

• 1087B, Isai. 53:9. Printed peccatum non fecit, nec dolus inventus est in eo ullus, which is 1 Pet. 2:22 rather than Isaiah, and the note attributing it to Isaiah is Migne's. The same verse is quoted at 1085A with the correct 1 Pet. note and with in ore ejus rather than in eo ullus. Both quotations rendered as printed; the note contents are verbatim in both places. No reference correction proposed.
• 1090C, Matth. 5:15. Non potest civitas abscondi supra montem posita is Matt. 5:14; the note prints XV. Migne's note, left verbatim.
• 1094A, Prov. 3:26. Ne dixeris amico tuo: Vade et revertere… is Prov. 3:28; the note prints 26. Migne's note, left verbatim.
• 1090B, Ps. 63:7 vs 1083C, Ps. 63:8. The same verse (Accedet homo ad cor altum) is quoted twice in this batch with two different verse numbers in the two notes. Both are Migne's; both left verbatim. Not a defect on our side.

Construction notes

### 1093D — In feminibus, vel in femoribus The plate offers two Latin thigh-words in apposition (femen/femina and femur/femora), and English has one exponent for both. Rendered "In the thighs, or in the loins" so that the doubling is visible without reading as a dittography; the lemma juncturae femorum and the gloss Femina ergo Ecclesiae both keep "thighs", which is where the argument sits. This is a Pattern 9 situation (a Latin distinction with no English exponent), not a defect: nothing is emended, and the Latin column carries both words.

### 1090C — number shift in the Galaad gloss per Galaad mens sanctorum designatur, qui sunt acervus testimonii quoniam … solum Deum desiderare evidentissimis attestatur indiciis. The relative is plural (qui, agreeing in sense with sanctorum) while the verb is singular (attestatur, agreeing with mens). Rendered as printed — "the mind of the saints, who are a heap of witness, since it attests…" — with the shift preserved rather than smoothed to one number.

### 1086B — the Holy Spirit as subject of his own gloss quia Spiritus sanctus lactis nomine, id est Dei gratia intelligitur, quae per Spiritum sanctum Ecclesiae tribuitur. As printed, the Holy Spirit is what is "understood under the name of milk", and then the grace is what is given "through the Holy Spirit" — the subject is doing duty on both sides of the id est. Rendered as printed; no clause re-bound (Pattern 7a, mechanism 5).

### 1086B — quae lacte sunt lotae answered twice The lemma's relative clause is expounded once of the Holy Spirit (1086A–B) and again of the teachers of the Church (1086B). Both expositions rendered in full; the ut dictum est back-references are the author's, not ours.

### 1088C — Vers. 17 of Song 5 The plate numbers the Quo abiit dilectus tuus lemma as verse 17 of chapter V, and chapter VI opens at Dilectus meus descendit in hortum suum. The chapter/verse division here is the plate's; not adjusted. # Cruces — 7914, chunks 0018–0022 (cols 1096A–1106C)

Auctor incertus (Cassiodorus Vivariensis?), Expositio in Cantica canticorum, PL 70. Batch: chunks 0018, 0019, 0020, 0021, 0022 — the final chunks of the work.

Work-wide conventions set or confirmed in this batch (for merge reconciliation)

Pre-set by the session brief and followed throughout: murenulae untranslated; dilectus/dilecta = beloved; amica/amicus = friend; sponsa = bride, sponsus = bridegroom; pulchra = beautiful; formosa = comely; speciosa = fair; nigra = black; fusca = swarthy; columba = dove. Named etymologies translated with the proper name kept as printed.

Reconciled against the peer batches after drafting (chunks 0000–0017 were on disk by the time this batch was checked, so the following were aligned rather than invented):

New choices made in this batch that were not already fixed elsewhere:

No guillemets occur anywhere in chunks 0018–0022; the Latin twins have none either, so none were supplied (Pattern 6). The inner-edition page numbers 529–535 are reproduced verbatim in place.

Carried type — [sic: …] (Patterns 7, 10, 12)

### 1096D — pulmae for palmae (inside an italic lemma) The lemma of Song 7:7 is printed Statura tua assimilata est pulmae. Not a word; carried untranslated per the non-word class of Pattern 7 and marked. Per Pattern 10 it takes no second italic marker — nested italics are not expressible — so the page reads Thy stature is likened to [sic: pulmae]. The conjecture palmae is certain: the very next sentence but one prints Statura ergo Ecclesiae palmae assimilantur, and the whole exposition that follows is about the palm tree (Palma enim victricem ornat manum).

### 1106B — hinauli for hinnuli Sicut (inquit) capreae et hinauli cervorum, indomita animalia, humanum consortium refugiunt. Not a word; carried and marked. The plate itself supplies the reading twice in the same paragraph — the lemma of Song 8:14 at 1105C prints assimilare capreae hinnuloque cervorum, and the re-quotation a few lines above prints Similis esto capreae hinnuloque cervorum. Conjecture hinnuli, "young harts".

Rendered as printed — no marker

### 1097A — Statura … assimilantur, singular subject with plural verb Migne prints Statura ergo Ecclesiae palmae assimilantur. The subject statura is singular; the verb is plural. Unlike a gender mismatch (Pattern 9), English does have an exponent for number, so smoothing it to "is likened" would be a silent repair of exactly the class Pattern 7a forbids. The English carries the plural — "The stature therefore of the Church are likened to the palm" — and the reader is told here that the disagreement is the plate's. Conjecture: assimilatur, or staturae; the sentence construes either way and neither is preferable on internal evidence, so nothing is proposed in the text. This will look like our typo on the reading page and is not one.

### 1103A — nullius mundi substantiae nam nullius mundi substantiae, nullae opes comparatione dilectionis aliquid sunt. The genitive singular nullius mundi against nominative plural substantiae is odd; rendered as printed ("for the substances of no world, no riches, are anything in comparison of love") rather than normalized to "no substance of the world". Real words, so no marker (Pattern 12 boundary). The argument does not turn on it.

### 1105C–1106A — the two pacifici constructions Song 8:12 Mille tui pacifici is grammatically ambiguous (predicate adjective vs. vocative "O peaceable one"), and the work reads it as a predicate: mille argentei pacifici tui sunt, id est, manet te certa illa remuneratio. The lemma is therefore rendered "A thousand are thy peaceable ones", following the exposition rather than the commoner vocative reading. Recorded because a reader who knows the verse will expect the other construction, and because the choice is the work's, not ours.

Punctuation carried per Pattern 8

### 1102C–D — Rom. 8:35 punctuated as one clause Migne prints Quis nos separabit a charitate Christi, tribulatio, an angustia, an persecutio, an fames, an nuditas, an periculum, an gladius? — a single comma-run with one terminal ?, where the Vulgate and English convention would break the question after Christi. No mark supplied and none deleted: the English runs "…from the charity of Christ, tribulation, or distress, or persecution, or famine, or nakedness, or danger, or the sword?" with the one question mark the plate prints, outside the italic run exactly as printed.

### 1102C — the inline citation (Aggaei II, 24) Haggai 2:24 is cited inside the italic scripture run as running text, not as a [n: …] note, unlike every other citation in the batch. Reproduced verbatim in place, inside the italics, untranslated. Not tagged [f: …]: this is a lemma-and-gloss commentary, not a florilegium, and Pattern 4's trigger is a citing practice, not a stray instance. Flagged so the indexer's operator knows the reference exists and will not be harvested from a [n:].

Scripture divergences noted — none marked [var: …]

Marked sparingly per Pattern 14. Each of the following was judged not to change the sense the exposition builds on, so each is recorded here instead of on the page.

• 1096A, Jer. 4:2. Printed Circumcidimini Domino in cordibus vestris; the Vulgate (Jer. 4:4) has Circumcidimini Domino et auferte praeputia cordium vestrorum. The argument needs only "circumcision of the heart", which both readings give. The note prints II for IV; Migne's note, left verbatim, no correction proposed.
• 1096D, Exod. 6:8. Printed Ego sum Dominus Deus vester, qui eripui vos de ergastulo Aegyptiorum, ut incederetis recte — this is Lev. 26:13 (eduxi vos de terra Aegyptiorum … ut incederetis erecti), not Exod. 6:8, and ergastulum ("workhouse") and recte are both the plate's. Rendered as printed; the note is Migne's and stands. The exposition turns on recte / incurvari, which the printed wording supplies.
• 1097D, Rom. 9:5. Printed Quorum patres, et ex quibus Christus, an abbreviation of quorum patres et ex quibus est Christus secundum carnem. The continuation Qui est Deus super omnia benedictus in saecula is quoted a line later without a note of its own, exactly as the argument requires. Rendered as printed.
• 1097B, I Cor. 3:1. The quotation runs lac potum vobis dedi, non escam (Vulg. 3:2). The negation is carried.
• 1099D, Luc. 24:44. Printed Necesse est impleri omnia quae scripta sunt in Lege, et Prophetis, et Psalmis de me; the Vulgate has quoniam necesse est impleri omnia… Rendered as printed.
• 1103B, Philipp. 3:8. Printed Haec omnia contemno et arbitror ut stercora, ut Christum lucrifaciam; the Vulgate has omnia detrimentum esse duxi … et arbitror ut stercora. Rendered as printed; the gloss (id est, ut amorem illius acquiram) fits the printed wording.
• 1106C, II Cor. 2:1. The quotation Christi bonus odor sumus Deo in omni loco is 2 Cor. 2:15 (conflated with 2:14 odorem notitiae suae manifestat per nos in omni loco); the note prints II Cor. II, 1. Migne's note, left verbatim.
• 1106B, II Cor. 5:16. Printed Etsi noveramus Christum secundum carnem, sed nunc jam eum non novimus. Both the concessive etsi and the adversative sed stand in the Latin, which English will not normally carry together; rendered "And if we have known Christ according to the flesh, yet now we know him no more", keeping the negation intact.

Construction notes

### 1100A — Tu qui modo es in sinu Patris The relative qui is printed inside the italic run while Tu is outside it (Tu qui modo es in sinu Patris), so the italics fall across a clause boundary rather than round a quotation. Reproduced as printed: the English sets who italic and art now in the bosom of the Father italic, with Thou roman. This looks like a typesetting slip and is carried, not tidied.

### 1100D — Marc. IX with no verse The note attached to saepe etiam volentes interrogare apostolos Dominus praeveniebat gives a chapter only. Left verbatim; no verse supplied.

### 1101B — S. Gregorius, homil. 30 in Evang. and 1101C — S. Greg. in cap. II Cant. cant. Two notes citing Gregory in two different abbreviation styles within four columns. Both left verbatim per the [n: …] rule; the variance is evidence of Migne's citing practice and is not normalized (CLAUDE.md rule 9).

### 1100B–C — the Synagogue called the Church Nec mirum videri debet, si nunc Ecclesiam Synagogam dicimus; una est enim Ecclesia in praecedentibus et in sequentibus partibus. The work uses Ecclesia and Synagoga of the same subject throughout chapter VIII, and says so. Both words kept distinct in the English wherever the Latin distinguishes them; no clause re-bound to make the referent easier.

### 1104C — In ea quae habet populos The lemma is printed without the quae clause's antecedent (Song 8:11 in ea quae habet populos), and the gloss supplies it (Ecclesia vero habet populos). Rendered as printed — "In that which hath peoples" — rather than expanded to "in the vineyard which…".

### 1105A — the doubled hundred Sicut per millenarium plenitudo perfectionis, sic et per centenarium summa perfectio designatur: quia centenarius de laeva in dexteram transfertur. The remark about the hundred being "carried over from the left hand to the right" is the ancient finger-reckoning convention (a hundred is the first number counted on the right hand). Rendered literally; no explanatory word added to the text.

### 1096B — quas Graece βάτια vocant The relative quas is feminine while its antecedent canales is masculine. Pattern 9: Latin gender has no English exponent, nothing to emend, nothing rendered differently. Logged only so a later sweep does not churn it.

Four Opus units, blind, 23/23 chunks, no sampling. 14 sites: 7 source-side, 7 ours. Units in data/sweeps/polarity-7914/. Unit boundaries deliberately STRADDLED the translation batches, so every batch seam fell inside some reader's range.

The finding that matters beyond this work

All seven ours-side defects were lemma conformations (5) or mis-bound references (2). Not one was a dropped negation. The same day's read of 11065, a homily collection, gave the opposite profile. The genre predicts the failure mode. In a verse-by-verse commentary the risk is not losing a non; it is the familiar English of a known verse arriving before the printed Latin is read. Written up as translation-style.md 7a″, with these cases as the worked examples.

Fixed (ours)

• @1079C — Heb 12:14. Migne prints et castimoniam, sine qua nemo videbit Deum; we gave the Vulgate's "holiness" (sanctimoniam). Nine words earlier our own English renders castimonia as "chastity", and the text is adduced precisely for chastity of the flesh — the argument's keyword had vanished from its own proof-text. → "and chastity".
• @1094A — Heb 11:10. Printed Exspectabat enim manentem civitatem; we gave "a city that hath foundations", importing a word the printed text does not contain. → "an abiding city".
• @1103A — Matt 7:25. Printed fuit enim aedificata supra petram; we gave "founded", six words after rendering aedificavit as "built". → "built".
• @1071D — Heb 2:9. Printed perfect Vidimus; we restored the Vulgate present "We see". → "We saw".
• @1079B — Gen 18:27. Printed cinis et pulvis; we gave Douay's "dust and ashes", reversing an order the gloss turns on (cinnamon's cinericeum colorem). → "ashes and dust".
• @1102A — speaker. Et bene arborem mali dicit rendered "does she say". The speaker of the lemma Sub arbore malo suscitavi te is the bridegroom (ipsius sponsus et dilectus), and our own next paragraph renders the following dicit as "he means". The subject does shift to the bride at comparavit — dilectus meus is hers — but that shift is the author's and belongs to the second verb only. → "does he say".
• @1095D — contra smoothed. Contra Damascum ergo est haec turris: quia … semper … resistit rendered "is toward Damascus", so our own "because … always resists" no longer followed. → "is against Damascus". The lemma at 7:4 keeps "which looks toward Damascus" — respicit contra is idiomatic facing; the author's gloss exploits the adversative sense, and flattening the two would destroy the point he is making.

Controls — both found blind

• @1069B was NOT named in any prompt and u2 found it anyway: the dropped first nec in quae circumcisione mundata, nec adhuc baptismo sanctificata, which leaves the Church of the Gentiles cleansed by circumcision against this work's own argument.
• @1082D (u3) was the one declared control and was found from the libero / occupationibus contradiction on the reading itself, before any cruces file was opened.
• @1097A (singular statura, plural assimilantur) was unannounced and u4 confirmed it had been carried correctly rather than reporting it as a defect — the right result.

## Source-side, rendered literally (new, previously unlogged) @1082C ×2 — nunquam where the lemma Lavi pedes meos, quomodo inquinabo illos? requires quondam; and exui for exuta, which makes the Church fear both the stripping she boasts of and the reclothing. @1084A — a dropped non against 1083C's nemo eum in praesenti vita, sicuti est, comprehendere potest.

## A structural gap worth naming The per-batch negation audits in this file cover chunks 0007–0012 only; 0013–0017 had no negation audit at all, and that is exactly where three of the new source-side findings sit. A per-batch audit is not a work-wide one.