Sermo I in Nativitatem B. Mariam Virginem

Merged 2026-08-01 from cruces-0000.md (Fable anchor) and cruces-0001.md (chunks 0001–0003, completing the work). Source files follow verbatim; this header is the authority where they disagree.

WORK-WIDE CONVENTIONS (authoritative)

No conflicts to rule. The second agent explicitly held the anchor's conventions binding and extended the tables rather than redefining them — the merge outcome the anchor discipline is designed to produce. Recorded here so a later sweep does not churn it.

Structure: ΛΟΓΟΣ = "Sermon" (per registry + the plate's own SERMO) · Εὐλόγησον, Πάτερ = "Give the blessing, Father" · section letter-numerals as arabic + period · rhetorical pile-ups 1:1, exclamations and rhetorical questions kept · coinages rendered boldly, never normalized (ταφόζωος = tomb-bound · βρεφουργηθέντες = wrought as babes · ἔξαλμα = outleaping · προδιεκωδώνισε = sounded the bell beforehand · ἐξωβελικότες = struck out with the obelus · καινοτοκεῖν = new-bearing · ἱστουργία = weaving · θεουργικώτατον/πλαστουργικώτατον = most God-working / most fashion-working).

The Marian epithet table — one English per Greek epithet, never interchanged (anchor + extension, consolidated): πάναγνος = all-pure · πανυπέραγνος = all-surpassingly pure · Πανάχραντος = All-immaculate · Πάντιμος = All-honored · πολυύμνητος = much-hymned · πανύμνητος = all-hymned · Θεομήτωρ = Mother of God · Θεοτόκος = Theotokos · Δέσποινα = Lady · Δεσπότης = the Master · ἀειπάρθενος = ever-virgin · ἡ μακαρία = the blessed one · πανάμωμος = All-blameless · ἀμώμητος = unblemished · παναγέστατον = all-holiest · παμβασιλίς = all-queen · βασιλίς = Queen · θεόπαις = God-child · Μητρόπαις = Mother-Child · θεοπάτορες = God-ancestors · καλλίπαις = fair-childed · πολύπαις = many-childed · καλλίτοκος = fair-bearing · πολύτοκος = many-bearing. Maker-words: Κτίστης = Creator · Δημιουργός = Maker · πλάστης = Fashioner · πλάσμα = moulded work · δημιούργημα = created work.

⚑ Why this table is the work's spine — the evidence, not the theory

πολυύμνητος and πανύμνητος stand three lines apart in the same section (@0040), and Allatius gives ONE Latin word for both (laudatissima / laudatissima illa). Likewise καλλίπαις/πολύπαις, where the contrast is the sentence's argument, and Allatius flattens καλλίπαις to venustissima puella; and θεουργικώτατον/πλαστουργικώτατον, collapsed the same way. A translator following the Latin as a guide rather than a witness would have silently destroyed three arguments. This is the homily's analogue of 7a″ lemma-conformation, and it is why the epithet rule was written into translation-style.md before this batch ran.

Marker totals after merge

1 [lat:] (0024 καινὴν vs Latin communem, Greek plate-verified) · 8 [var:] scripture divergences + 6 logged unmarked · 2 [ed:] (the lost 1 Pet 2:22 clause carrying a negation, restored from the plate; and an unrepairable lacuna where both columns fail and the arbitrating leaf is missing) · 1 [lat:] candidate correctly rejected (Ἀχάβ vs Achaz) because it sits on the leaf the primary scan lacks — firing would have attributed to Migne's plate from our files alone.

⚑ ADDED 2026-08-01 by the blind polarity read — a printed self-contradiction that was NOT logged

0003 @0032/0033 — Migne's Greek prints:

Ἀλλ᾽ ὃ μήτ᾽ ἐκ γῆς πρῶτον συνέστη, μήτε τι τῶν μελῶν ἀφήρηται, εἴη δ᾽ οὐδ᾽ ἂν ἕν, οὐχ ἕν τι τῶν ἁπάντων πρὸς ταῦτ᾽ οἰκειότερον ἢ ἡ Παρθένος

The clause denies that any such thing could exist — and then, in the same breath, names the Virgin as that thing. Our English carries it literally ("and there could not be so much as one such — no, there is not one thing out of them all more suited to these ends than the Virgin"), which is the required treatment (Pattern 7): no repair, no substitution.

Plate-verified by the blind reader against the scan's own Greek column, independently of Calfa — the plate prints it exactly so. The Latin column simply omits the run (nihil ex omnibus ad id magis…).

No [lat:] fired, deliberately. A short omission is inside Allatius's normal compression, and Pattern 16 is explicit that his looseness is not a divergence. Firing here would be marker spam of the kind the pattern warns against. Logged instead — which is what Pattern 7 asks for and what was missing.

Why this entry exists at all: the translating agent rendered the passage correctly but never logged it, so the crux was invisible to any later reader. A correct rendering with no record is indistinguishable from an unexamined one — the same argument that makes sites: 0 a recordable result. The blind read found no defect in this work; it did find this hole in the record.

Two items deliberately left open (do not "tidy" them)

• §14 numeral. Calfa reads ΙΑʹ and so does the scan's own Greek OCR, but the Latin numbers it XIV and ΙΑʹ=11 is already used. Rendered "14." — but both independent OCRs agree on Α, so a leaf render showing the plate genuinely prints ΙΑʹ turns this into a [lat:], not an OCR fix.
• The Ἀχάβ/Achaz candidate, pending a leaf for Calfa p. 21.

Verified after merge

verify-english-pg.mjs isidore-glabas-sermo-1 → 4 chunks, clean.

Work-wide conventions (anchor session, 2026-07-31 — merge into the work-wide section)

• ΛΟΓΟΣ as a work-division head renders by the work's established English division term — here "SERMON" (registry + the plate's own Latin column: SERMO I) — consistent within the work. (Nicetas uses "Discourse" for the same word; per-work, not per-corpus.)
• Εὐλόγησον, Πάτερ — the liturgical formula before the sermon renders "Give the blessing, Father."
• Mariological epithets — fixed equivalences, one per Greek epithet, never interchanged: πάναγνος = all-pure · πανυπέραγνος = all-surpassingly pure · Πανάχραντος = All-immaculate · Πάντιμος = All-honored · πολυύμνητος = much-hymned · Θεομήτωρ = Mother of God · Δέσποινα = Lady · ἀειπάρθενος = ever-virgin · ἡ μακαρία = the blessed one. Glabas builds arguments on the differences between epithets; collapsing two into one English word is the homily analogue of a lemma-conformation.
• Maker-vocabulary distinctions, fixed within the work: Κτίστης = Creator · Δημιουργός = Maker · πλάστης = Fashioner · πλάσμα = moulded work · δημιούργημα = created work.
• Rhetorical pile-ups are structure, not padding: doublets/triplets (γυμνὸς, ἐλεεινὸς, αἰχμάλωτος) render 1:1, never compressed; exclamations kept (οἴμοι! = alas!); rhetorical questions stay questions.
• Bold coinages are rendered boldly, not normalized: ταφόζωος = tomb-bound (existence) · βρεφουργηθέντες = wrought as babes · ἔξαλμα = outleaping · προέτεκε = bore before. Allatius routinely flattens these (sepulcralem vitam, in quadam infantia); his paraphrase is NOT a license to flatten the Greek.

Latin-twin pass (pilot §6): result for chunk 0000 — FINDINGS (not clean)

Read against src/pg-latin/isidore-glabas-sermo-1/0000.md (cols 0011–0018; NOTE: the Latinited for cols 0011–0014 comes from the ALT witness patrologicursus73migngoog, per the twin's provenance.altSourcePages). This batch is chunk 0000 only; it was not clean. Polarity/negation collation found NO divergence between the columns (all of οὐ πρότερον/μηδ᾽ ἐγγὺς/οὐκ ᾤετο/ οὐδενὸς/οὐδ᾽ ἂν ἐμέλησε/εἰ μὴ ἀποπηδήσειεν carried in both). No numerals in the chunk beyond section marks. No [lat:] fired.

Ladder row 1 — Calfa OCR damage, corrected silently, Latin/context cited

• 0012 (title) Θεοχοήτορος → Θεομήτορος (Lat. Dei Genitricis).
• 0012 Section numeral: Calfa's Ηʹ → Βʹ (the Latin column numbers it II; sequence runs Αʹ–ϛʹ). Rendered "2."
• 0012 κςσιῳ δογματίζεσυες → κόσμῳ δογματίζεσθε (Col 2:20 as printed; Lat. in mundo decernitis).
• 0012 μέχρι; οὗ → μέχρις οὗ · ὡφέλιμον → ὠφέλιμον · οἶ → οἵ (routine).
• 0013 Παρβένου ποοῖλθε → Παρθένου προῆλθε · λόγμον → λόγιον (Lat. oraculum) · ὕανεῖται → φανεῖται.
• 0016 πανάννου → πανάγνου · υγχωρήσειε → συγχωρήσειε · λγαλμα → ἄγαλμα · ὑπερόολῆς → ὑπερβολῆς · ἰσοθεὶαν → ἰσοθεΐαν · χοῖκῷ → χοϊκῷ.
• 0017 θά λασσα → θάλασσα (split type) · πρὸς ανον → πρὸς αἶνον (Lat. ad laudandum) · τοῦ Κκροτεῖν → τοῦ κροτεῖν (intrusive-Κ class again — third instance in PG 139's Calfa file, cf. nicetas cruces-0000).
• 0017 σιγὴν ἀπήνει — the finite verb is OCR-damaged past recovery from our files (σιγὴν ἦγε? ἐσίγα?); the SENSE "kept silence" is certain from the coordinated participle μὴ τὸν ὕμνον ἀναβαλλόμενα and Lat. jam silebant, neque referebant hymnos. Rendered "kept silence," exact Greek wording unresolved — plate check would settle the letter.

Construal notes (not defects)

• 0017 ἑωρακότα is neuter plural nominative with τὰ καλὰ πάντα — the CREATURES "having seen their master pinioned"; agrees with Allatius's cernerent. Calfa's καθάπερ ἑωρακότα is sound; no emendation.
• 0016 ἔμελλεν / ἐμέλησε wordplay ("had this not been to come… would not have taken thought") — untranslatable pun, sense rendered, logged here.
• 0012 εἶκε ("yielded") vs Lat. paruit ("obeyed") — Allatius freedom, not a divergence.

Twin apparatus carried for later harvest (not ours to mark)

Scripture identifications printed in the Latin column's notes for this span: Exod. XIV,31 · Exod. XVII,1 seqq. · Exod. XVI,1 seqq. · Gen. 1,2 (the In principio quote) · Col. 2,20 · Ps. 66,7 (Terra dedit fructum suum) · Rom. 6,5 (complantati) · Gen. 1,31 · Ps. 18,2 (caeli enarrant). Plus Ballerini's editorial notes (2)–(4) and the long immaculist annotations — plate apparatus, in the twin, verifier-only.

Alt-witness note

Cols 0011–0014 of the twin (sermon title + §1) come from the second physical copy (patrologicursus73migngoog, tesseract render+OCR) — the primary scan has no leaves for Calfa pp. 13–14. The title and §1 Latin were still usable as verifier (Θεομήτορος and the Colossians quote both confirmed against it).

Continues cruces-0000.md (the Fable anchor), whose work-wide conventions section is binding here and is not restated. This file records (a) new epithets and maker-words met in 0001–0003 and the English fixed for each, (b) the scripture-divergence list, (c) the Latin-twin pass per chunk, (d) every silent OCR correction, (e) the one [lat:] fired and the candidates rejected with reasons.

Witnesses used. Greek = sources/pg/calfa as chunked. Latin twin = src/pg-latin/isidore-glabas-sermo-1/0001–0003.md. Plate = the Paris scan raw/scans/pg139/patrologiaecurs63migngoog_djvu.txt (leaves 10–19), consulted directly for this batch and quoted below where it settles a reading. Known gap: the leaf carrying PG 139 cols 27/28→29/30 (Calfa p. 21) is absent from the primary scan, so the stretch of chunk 0002 from Δάκνει τοῦτο τὸ ῥῆμα to ΙΒʹ has no plate witness at all and its Latin comes from the alt witness patrologicursus73migngoog (tesseract render + OCR, per the twin's provenance.altSourcePages). Every claim resting on that stretch says so.

1. NEW epithets and maker-words — English fixed here, to be carried forward

Extending the anchor table (πάναγνος = all-pure · πανυπέραγνος = all-surpassingly pure · Πανάχραντος = All-immaculate · Πάντιμος = All-honored · πολυύμνητος = much-hymned · Θεομήτωρ = Mother of God · Δέσποινα = Lady · ἀειπάρθενος = ever-virgin · ἡ μακαρία = the blessed one · Κτίστης = Creator · Δημιουργός = Maker · πλάστης = Fashioner · πλάσμα = moulded work · δημιούργημα = created work):

Coinages rendered boldly (never normalized; Allatius flattens all of them): προδιεκωδώνισε = sounded the bell beforehand (Lat. merely praenuntiasse) · ἱστουργία = weaving (Lat. drops the metaphor: gloria circumvestirentur) · καινοτοκεῖν = new-bearing · ἐξωβελικότες = having struck out with the obelus (Lat. repudiassent) · ἐκαινούργησε / ἐνεούργησεν = wrought newly / made new · θεουργικώτατον = most God-working and πλαστουργικώτατον = most fashion-working (Lat. collapses both into sacratissimum hoc ac plane dignissimum … germen, losing the pair outright) · κύημα = conception (Lat. filiam, "daughter", at 0001 — a flattening) · θεάνθρωπος = God-man (in the clause restored at 0037, below).

Register note carried from the anchor: where Calfa capitalizes a substantivized epithet (τὴν Πάναγνον, τῇ Πανάγνῳ) the English capitalizes ("the All-pure one"); where it is lower-case and adjectival (τὴν πάναγνον) it is not. This follows the plate's own practice and is applied consistently across 0001–0003.

2. Scripture-divergence list — where Migne's Greek departs from the received text

This is the deliverable. Every place in 0001–0003 where the printed Greek differs from the received LXX/NT is named here with the specific words. Markers [var: …] were fired only where the sense turns on it; the rest are recorded here alone.

Marked [var:] in the English:

1. 0001 @0020 — Ps 88:7 (LXX). Migne prints Τίς ἐν νεφέλαις ἰσωθήσεται … τῷ Κυρίῳ, ὁμοιωθήσεται τῷ Κυρίῳ ἐν υἱοῖς Θεοῦ; — one interrogative. LXX repeats it: … καὶ τίς ὁμοιωθήσεται τῷ Κυρίῳ …. Plate-verified in the Greek column, and adjudicated on the plate itself: Ballerini's note (9) — Isidorus hic magis cum Vulgata Latina quam cum Graeca LXX editione consentit; haec enim ita bis repetit τὸ τίς … quae tamen repetitio abest etiam ab Complutensi ac Aldina, et ab Alexandrino codice. Rendered as printed; [var:] names the LXX reading.
2. 0001 @0020 — Hab 3:3. Migne prints Ἐκάλυψεν οὐρανοὺς ἡ ἀρετὴ αὐτῆς, καὶ τῆς συνέσεως αὐτῆς πλήρης ἡ γῆ — both pronouns feminine, and σύνεσις ("understanding") where LXX has αἴνεσις ("praise"). Plate-verified; Ballerini's note (10) says outright that Isidore agrees neither with the Vulgate (gloria ejus … laudis ejus) nor with the LXX (ἡ ἀρετὴ αὐτοῦ, καὶ αἰνέσεως αὐτοῦ). The Latin twin's sapientiae ejus plena est terra concurs with σύνεσις — two-witness concurrence (ladder row 3), so the reading is attributable to the plate pair, not to our file.
3. 0001 (end) — John 1:14. Ἐθεασάμεθα τὴν δόξαν αὐτῆς, δόξαν ὡς μονογενοῦς Μητρὸς Θεοῦ for δόξαν ὡς μονογενοῦς παρὰ πατρός. Plate-verified; the Latin concurs (quasi gloriam unigenitae Dei Matris) and Ballerini annotates the substitution at note (15). Rendered as printed.
4. 0002 @0025 — Col 3:3–4. Ἡ ζωὴ ἡμῶν ἐκέκρυπτο σὺν αὐτῇ … Ὅτε αὕτη πεφανέρωται … καὶ ἡμεῖς σὺν αὐτῇ ἐφανερώθημεν for the received ἡ ζωὴ ὑμῶν κέκρυπται σὺν τῷ Χριστῷ … ὅταν ὁ Χριστὸς φανερωθῇ … ὑμεῖς σὺν αὐτῷ φανερωθήσεσθε. The author announces this himself (μικρὸν παρῳδήσαντας), so the change of person and referent is authorial, not textual — but it is material and the English must not drift back toward the received verse. On the ἡμεῖς/ὑμεῖς itacism class: both columns read first person (Lat. Vita nostra … et nos apparuimus), so no [lat:].
5. 0002 @0025 — Hab 3:13. Migne's Greek adds ἐλήλυθας ("thou art come"), which the LXX lacks. Plate-verified, and Migne's own Ballerini note (16) raises exactly this: Utrum Isidorus τὸ ἐλήλυθας, quod a LXX abest, ideo addiderit, quod ita in Codice legeret, incertum. The Latin column concurs (venisti). Note also Ἐξῆλθες … τοῦ σῶσαι τοὺς χριστούς σου is addressed to the Virgin here, feminine in the Latin (Egressa es) — the refitting the author calls μεθαρμόσειεν.
6. 0002 @0025 — Luke 1:78. Migne prints the aorist Ἐπεσκέψατο ἡμᾶς ἀνατολὴ ἐξ ὕψους; the received NT has the future ἐπισκέψεται. Plate-verified; Latin concurs (Visitavit). Two-witness concurrence.
7. 0002 @0029 — Isa 26:18 (LXX). Migne prints … ἐτέκομεν πνεῦμα σωτηρίας σου, ὃ ἐκυήσαμεν ἐπὶ τῆς γῆς; LXX has πνεῦμα σωτηρίας σου ἐποιήσαμεν ἐπὶ τῆς γῆς. Latin concurs (quod utero gessimus super terram). Polarity note, and it is the one worth flagging: Ballerini's note (24) prints the Vulgate of the same verse — salutem non fecimus in terra — which is negative where both of Migne's columns are positive. That is a Vulgate/LXX difference, adjudicated on the plate by Migne's own note; the two columns agree with each other, so no [lat:], and the English follows the Greek.
8. 0002 @0029 — Col 2:3. Ἐν αὐτῇ γάρ εἰσι … πάντες οἱ θησαυροὶ τῆς σοφίας καὶ τῆς γνώσεως ἀπόκρυφοι for ἐν ᾧ εἰσιν … σοφίας καὶ γνώσεως. The whole point is the transfer from Christ to the Virgin; the added article on γνώσεως is minor and is not in the marker.

Recorded here only (divergence real but the sense does not turn on it — no marker, to avoid marker spam):

• 0001 @0020 — John 3:31. Ὁ ἐξ οὐρανοῦ ἐρχόμενος ἐπάνω πάντων ἐστίν; received ὁ ἐκ τοῦ οὐρανοῦ ἐρχόμενος. Plate-verified. Article only.
• 0001 @0021 — Ps 32:13–14 (LXX), conflated. Ἐξ οὐρανοῦ ἐπέβλεψεν ὁ Κύριος· εἶδε πάντας τοὺς υἱοὺς τῶν ἀνθρώπων ἐξ ἑτοίμου κατοικητηρίου αὐτοῦ welds v. 13 to v. 14, dropping the second ἐπέβλεψεν. Migne flags the conflation himself by citing Psal. xxxi, 13, 14 on the plate. Latin concurs. Rendered as the weld stands.
• 0001 @0021 — Ps 67:9 (LXX). Γῆ ἐσείσθη, καὶ γὰρ οἱ οὐρανοὶ ἔσταξαν — quoted without ἀπὸ προσώπου τοῦ Θεοῦ; a truncation, not a variant.
• 0001 @0020 — Ps 67:17 (τὸ μετεωρότατον ὄρος, ὃ εὐδόκησεν ὁ Θεὸς κατοικεῖν ἐν αὐτῷ) and Ps 10:4 / 102:19 (Κύριος ἐν ναῷ ἁγίῳ αὐτοῦ· Κύριος ἐν οὐρανῷ ὁ θρόνος αὐτοῦ) are woven into the homilist's own syntax but verbally as printed in LXX.
• 0002 @0028 — the priest's rebuke. Οὐκ ἔξεστι … ὑμῖν δῶρα Θεῷ προσενεγκεῖν, καθότι ἐν τῷ Ἰσραὴλ σπέρμα οὐκ ἐποιήσατε is not canonical scripture: it is Protevangelium Jacobi 1:2, there singular and in a different order (οὐκ ἐξεστίν σοι πρῶτον ἐνεγκεῖν τὰ δῶρά σου, καθότι σπέρμα οὐκ ἐποίησας ἐν τῷ Ἰσραήλ). Rendered as Migne prints it, in the plural. Migne prints no citation for it; do not index it as scripture.
• 0002 @0028 — Matt 3:9 woven, not quoted (τὸν δυνάμενον ἐκ λίθων ἐγεῖραι τέκνα, without ἐκ τῶν λίθων τούτων … τῷ Ἀβραάμ).
• 0003 @0037 — 1 Pet 2:22 / Isa 53:9. See §4: the first half of the quotation is missing from our Greek and is restored from the plate in an [ed:].
• 0003 @0040 — Luke 1:28. Χαῖρε κεχαριτωμένη, δι᾽ ἧς χαίρει πᾶσα ἡ κτίσις· ὁ Κύριος μετὰ σοῦ — the middle clause is the homilist's insertion into the angelic salutation (Latin concurs: per quam omnis creatura gaudet), and it turns on the χαῖρε/χαίρει pun. Rendering decision: χαῖρε is englished as Rejoice rather than Hail precisely so the pun survives ("Rejoice … through whom all creation rejoices"); Hail would have silently destroyed the figure the insertion exists to make.

3. The Latin-twin pass, chunk by chunk (pilot §6) — including the chunks that were clean

Method as specified: draft first, then one deliberate pass against the twin hunting (1) OCR damage in our Greek, (2) different asserted facts, (3) polarity disagreement.

• Chunk 0001 — NOT clean. One [lat:] fired (§5). Polarity collation found no disagreement: οὐδεὶς … ἀντίῤῥοπος / nemo … idoneus repertus est; οὐδὲ γάρ ἐστιν … οὐδ᾽ ἂν εἷς … οὐδεὶς / Nulla enim … Sed neque in posterum … ullus; οὐ θανασίμους … ἀνοιγνύμενοι / non quidem apertos ut olim mortiferas … pluvias demisisse; οὐκ ἄδηλον / satis manifestum sit; οὐκ ἄν τις ἁμάρτοι / ne minimum quidem a veritate aberrat; εἰ καὶ μὴ πρὸς οὐδὲν ἴσως ἕτερον / et sin minus ad aliud quidpiam; οὐ παρ᾽ ἑαυτοῦ / neque enim aurum … praebet. All carried both sides.
• Chunk 0002 — NOT clean. One [ed:] fired for a lacuna (§4). No [lat:]. One [lat:] candidate rejected (§5). Polarity: οὔτε λαμβάνει τὰ δῶρα / dona non acciperet; σπέρμα οὐκ ἐποιήσατε / semen in Israel non fecistis; μὴ παρὰ γνώμην τι καινοτομῆσαι / non quidem ad aliquid praeter Dei voluntatem moliendum; ὥσπερ οὐκ ἀνασχομένη, μὴ … μεταδοῦναι / quod ferre nempe ipsa non posset, ut ceteri … in partem non vocarentur — the double negative is carried in both columns and is carried in the English ("could not endure not to impart"); οὔτε … οὔτ᾽ εἶχε / Neque enim … aliam vero gignere … jam non poterat; οὐδὲ γὰρ οἷόν τε / Neque enim fieri poterat; οὔτε ἄνθρωποι, οὔτε ἄγγελοι, οὔτε τις ἑτέρα κτίσις / neque homines, neque angeli, neque ulla creatura; οὐδένα … πεποίηται λόγον / nihil prorsus curavit; τῷ μὴ παῖδα … τεκεῖν ἕτερον / aliam prolem amplius non genuerunt. No disagreement.
• Chunk 0003 — NOT clean. One [ed:] fired (§4). No [lat:]. Polarity, the densest stretch in the work and checked word by word: οὔτε γῆς αὖθις ἐδεήθη … οὔτε μέλους ἀνδρός / neque terram iterum assumpsit … neque membrum aliquod hominis; ὃ μήτ᾽ ἐκ γῆς … μήτε τι τῶν μελῶν ἀφήρηται / quod neque e terra primo compactum esset, neque membris deficeret; μὴ πρὸς τέλος ὁρᾷν ἐχαρίσατο / ut nullum unquam finem attingerent; οὐδενὸς τῶν οὐκ ἐπαινουμένων … κοινωνός / prorsus exsors eorum omnium, quae in hominibus non censentur laude digna — the double negative stands in both and is carried; μόνην ἐχρῆν μὴ … κοινωνεῖν δυσκληρίας / sola communis … miseriae expers esset; οὐκ ἂν ἦν οὕτω θαυμαστόν / neque tanta admiratione digna forent; οὐκ ἀνήλωκεν / non … absumpsit; οὐ παρακατέχειν ἔσχεν ἐπιπολύ / non diu apud se retinere illud potuit; μὴ … διεφθορέναι / corruptioni reipsa obnoxium non fuit; οὐκ ἀντιλέγοντας ἕξει καὶ ἡμᾶς / nos utique contradicentes non habebit — note the Greek prints the litotes and the English keeps it ("he will not have us contradicting him"), not the smoothed positive; πρὸς ἃ μὴ πέφυκε / quibus perfruendis nata non est; οὐδεὶς … ἄψαυστος γέγονεν / nemo hominum exstitit, qui … integer fuerit; κατὰ … τὸ μηδὲν ἁμαρτεῖν` / ad hanc ab omni peccato immunitatem*. No polarity disagreement anywhere in the batch.

Chunks on which the twin pass found nothing at all: none. Each of the three yielded at least one marker. The pass is therefore not reportable as "clean" for any chunk, and the absence of a clean chunk is itself the honest result.

Pronoun-itacism check (ἡμεῖς/ὑμεῖς), the class the paired pilot flagged: four sites in this batch — Ἐπεσκέψατο ἡμᾶς (0002, Lat. nos), ἡ ζωὴ ἡμῶν … καὶ ἡμεῖς (0002, Lat. nostra … nos), τεχθήσεται ὑμῖν παιδίον (0002, Lat. vobis), ὡς τὰ καθ᾽ ἡμᾶς and οὐκ ἀντιλέγοντας ἕξει καὶ ἡμᾶς (0003, Lat. corpora nostra, nos). All four agree across the columns. No [lat:] in this class; no plate check needed.

Numerals: the only numerals in the batch are the section marks (§6) and τεσσαράκοντα ὅλας ἡμέρας / quadraginta solidos dies (0002) — which agree. No numeral [lat:].

4. Source loss — the two [ed:] markers, and why each is not a [sic:]

(a) 0003, before the anchor [0037] — a whole clause of 1 Pet 2:22 lost from our Greek. Calfa runs … τὸν καινόν φημι τῆς μακαρίας τόκον, [0037] οὐδὲ εὑρέθη δόλος ἐν τῷ στόματι αὐτοῦ — where the οὐδέ dangles with no preceding negative clause to answer. The Latin twin has more: praeter illum novum, inquam, hujus Beatae Filium, qui solus, utpote … Deus homo, peccatum non fecit, neque dolus inventus est in ore ejus (twin 0003, Latin col. 0037/0038). The plate settles it. At the foot of the Greek column 35 the Paris scan reads:

ὃς μόνος οἷα θεάνθρωπος ἁμαρτίαν οὐκ ἐποίησεν,

immediately before the page turn at whose head the scan's Greek resumes οὐδὲ εὑρέθη δόλος ἐν τῷ στόματι αὐτοῦ. So the clause is on Migne's plate and is missing from our source: a digitization loss, Pattern 13, not a plate defect — hence [ed:], not [sic:]. The marker quotes the plate (the PG precedent set by the Nicetas anchor's restored ΛΟΓΟΣ ΠΡΩΤΟΣ heading at col. 0173). Note that the lost clause carries a negation (ἁμαρτίαν οὐκ ἐποίησεν): without it our English asserted, of the one exception to universal sin, only the second half of the proof-text. Recommended follow-up (not done here, since it changes stored Greek): a data/calfa-patches/ entry restoring the clause. The ladder requires a plate check for that, and the plate check is in hand — cited above.

(b) 0002, §11 tail — a lacuna at a page boundary, unrecoverable. Calfa runs … ἐκεῖ μὲν γὰρ κατάρα, χαρὰν δὲ ὁ νέος ΙΒʹ. Τῇ οὖν τοιαύτῃ …: ὁ νέος has no verb and no noun, and the section number follows immediately. The Latin twin breaks off at the same place (Sed et locus, qui Annae precum testis fuit, sedi, in qua versabatur prima parens, then the column anchor). Both witnesses fail together — ladder row 4, no attribution — and the plate cannot arbitrate because this is exactly the leaf the Paris scan lacks. The [ed:] therefore states the loss and states that it cannot be repaired from our files; it supplies nothing. This is not a Pattern-16 case and not a defect of Migne's.

5. [lat:] — one fired, one candidate rejected

FIRED — 0001, Greek col. 0024 / Latin col. 0023: καινήν vs communem. Greek: … τὴν καινὴν δὴ ταύτην ἐζήτουν καὶ τελεσιουργικωτάτην καὶ φωτοδότιδα μητέρα. Latin twin, col. 0023: … communem hanc matrem quaerebant, quae ad perfectam formam proli tribuendam, et ad lucem impertiendam maxime par foret. The Latin translator read κοινήν, "common"; the Greek prints καινήν, "new". Greek plate-verified — the Paris scan's Greek column reads τὴν καινὴν δὴ ταύτην ἐζήτουν, independently of Calfa. The divergence is material: the Greek's argument is that every father, having rejected his own mother, sought a new one; communem makes her a mother shared, which is a different claim. English follows the Greek; marker records the Latin. (This is the οι/αι confusion class — the prose analogue of the itacism pairs, and the second live [lat:] of the paired pilots.)

REJECTED — 0002, §11: Ἀχάβ vs Achaz. Greek: ἢ ὡς Ἰεζάβελ θύειν τῇ Ἀστάρτῃ τὸν Ἀχάβ. Alt-witness Latin: aut sicut Jezabel ad sacrificandum Astartae impulit Achaz. Ahab was Jezebel's husband; Ahaz is a different king — on its face a textbook Pattern 16 name divergence. Not fired, and here is why: this passage falls in the stretch for which the primary scan has no leaf, so the only witness to the Latin reading is uncorrected tesseract OCR of a second physical copy, whose output on the same lines gives sacriflcandum and As- tart;. A b→z misread is well inside that OCR's demonstrated error rate, and firing [lat:] would be a public claim about what Migne's plate prints resting on our file alone — the exact attribution the PG standing rule forbids. Logged as a live candidate: if the leaf is ever obtained, this is the first thing to check. (For what it is worth, the primary scan renders Ἀχὰβ / Achabum correctly a column earlier, in the Ahab-and-Gideon comparison; that is the same name in the same work, and it tells against Achaz being Migne's.)

Not divergences (Allatius's freedom, per Pattern 16's own boundary) — named so a later pass does not re-raise them: inaccessibilis splendor for ἄδυτον … φέγγος ("unsetting") · coelorum filiam for οὐρανῶν κύημα ("conception of the heavens") · venustissima puella for ἡ καλλίπαις · nascetur vobis puella for τεχθήσεται ὑμῖν παιδίον ("a child") · the whole added clause cum scilicet divino huic ac mystico corpori ex hominibus tanquam membris compingendo ipsa Dominum tanquam caput imposuit (Lat. col. 0032) expanding τῶν γὰρ αὐτοῦ μελῶν κεφαλὴν ἐπεκάθισε τὸν Δεσπότην · Quaquaversus enim resonat maximi illius theologi vox supplied where the Greek has only ἡ θεολογικωτάτη … φάσκει φωνή. Scripture in the Latin column conformed to the Vulgate is convention, not divergence and fires nothing: similis erit Deo in filiis Dei for ὁμοιωθήσεται τῷ Κυρίῳ (Ps 88:7) is the worked instance here, and Ballerini's note (9) says as much on the page.

6. Section numerals — the sequence, and one unresolved

Calfa's numerals in this stretch are heavily damaged; the Latin column's Roman numerals and the plate settle them. Rendered as arabic + period, per the anchor.

7. Silent OCR corrections — logged as classes, then per site

Class 1 — the intrusive Κ, already known in Calfa's PG 139 file (Κκροτεῖν, ΠνενύΚματος in the anchor and in Nicetas). Four more instances in this batch, all corrected silently: Κὁοὐρανὸς → ὁ οὐρανὸς (0001) · Κσταξαν → ἔσταξαν (0001, Ps 67:9) · ἐδημιούρΚγησεν → ἐδημιούργησεν (0003) · Κοὐδὲ → οὐδὲ (0003 @0037). The class is now seven attested sites across three works in this Calfa volume — worth a corpus note, not worth per-site marking.

Class 2 — split and run-together type (Pattern 10 territory, but these are our source's, not the plate's, so they are corrected rather than carried): θά λασσα → θάλασσα (anchor) · ὑύη λὸν → ὑψηλὸν (0001) · ἐπιμ ησθέντες → ἐπιμνησθέντες (0002) · ἀδι ήσαμεν → ὠδινήσαμεν (0002) · ἐπι.σπεύδειν → ἐπισπεύδειν (0003) · διασώσω. μεν → διασώσωμεν (0003).

Per site, chunk 0001 (plate-verified where marked ✱): Ὁψὲ → Ὀψὲ · ὅρόσος → δρόσος · λύτται → λύεται ✱ · εὖρε → εὗρε · ἶνα συνελών εἴπω → ἵνα συνελὼν εἴπω ✱ · ὡραῖσμοῦ → ὡραϊσμοῦ · ὄψοςἰ χὗ → ὕψος! Ζʹ. ✱ (§6) · ὁξίαν → ἀξίαν · Δαβὲ → Δαβὶδ ✱ · τῷ Κυρίφ → τῷ Κυρίῳ · ἀλλ’οὺδ᾽ ἂν εἶς → ἀλλ᾽ οὐδ᾽ ἂν εἷς ✱ · τοσοῦταν → τοσοῦτος ✱ (plate) · Ἐκάλυύεν → Ἐκάλυψεν ✱ · θώηλότατον → ὑψηλότατον ✱ · ἐξ οὐανοῦ → ἐξ οὐρανοῦ ✱ · τὴν νοσον → τὴν νόσον · χεῖρα ὁρέξαι → ὀρέξαι · ἁν εἰκότως → ἂν εἰκότως · τοῦ Οείου Δαβίδ → τοῦ θείου Δαβίδ · ἀπιβλέψαι → ἐπιβλέψαι · οἷθε κάντας → εἶδε πάντας ✱ · κατκητηρίου → κατοικητηρίου ✱ · ναῷ ἀγίῳ → ναῷ ἁγίῳ · στιν ὃς → ἔστιν ὃς ✱ · Ἀννα → Ἄννα · ἡ προσευνῆς → ἡ προσευχῆς ✱ (Lat. quae per preces mater ei fuit) · ἐρείψιμον ἐχούσης sound as printed.

Per site, chunk 0002: Ἐκεσκέψατο → Ἐπεσκέψατο ✱ · Ἐκιὰ γὰρ → Σκιὰ γὰρ ✱ · Τεδεών → Γεδεών ✱ · συνειτενεγκόντι → συνεισενεγκόντι · ποιἡσειν → ποιήσειν · ἐγεῆραι → ἐγεῖραι · γαστρόσ → γαστρός · Ἰοραὴλ → Ἰσραὴλ · συντόνω → συντόνῳ · Ἐκόαίνει → Ἐκβαίνει ✱ · ἡ δέσσις → ἡ δέησις ✱ · τεχόήσεται → τεχθήσεται ✱ · Διὰ τὀν ρόβον → Διὰ τὸν φόβον ✱ · ἐν γεστρὶ → ἐν γαστρὶ ✱ · ἐτένομεν → ἐτέκομεν ✱ · Κυνθείσης → Κυηθείσης ✱ · τῆς μονογενους → τῆς μονογενοῦς · τοῦ τκτειν ἔξῆς → τοῦ τίκτειν ἑξῆς ✱ · ὁ´ ἀφθονίᾳ → δ᾽ ἀφθονίᾳ · παρά [0032] τοὺς λοιπούς sound.

Per site, chunk 0003: παυορᾶται → παρορᾶται ✱ · τὸν λεῖπον → τὸ λεῖπον · φανενωθῇ → φανερωθῇ ✱ · ἐνῆν sound ✱ · δισκληρίας → δυσκληρίας ✱ · γένεσιω → γένεσιν ✱ · ἀθλου δημιουργίας → ἀΰλου δημιουργίας ✱ (Lat. immaterialis esse naturae) · διειμγασμένης → διειργασμένης ✱ · πανἀννου → πανάγνου · ἐν πῷ περὶ → ἐν τῷ περὶ · εἰρήαι → εἴρηται · ἀφθαροτέρα τῳ μέρει → ἀφθαρτοτέρα τῷ μέρει ✱ · πανὰμνητος → πανύμνητος ✱ (§1 — the load-bearing one) · θερυργικώτατον/Calfa θεουργικώτατον sound · Εθεν → ὅθεν ✱ · κάκείνους → κἀκείνους · Χαῆρε → Χαῖρε · ἀπελαύσαμεν sound (Calfa ο´ ἀπελαύσαμεν → δ᾽ ἀπελαύσαμεν) · ἐνεούργησεν sound · Θεοτόκῳ sound.

8. Construal cruces — printed reading kept, conjecture here

• 0001 — καὶ τὸ πρῶτον ἀναιρεῖται ἀξίωμα. ἀναιρέω in the active means destroy; the middle here means takes up for himself, and the Latin confirms it (pristinam resumit dignitatem). Rendered "takes up again his first dignity." Flagged because the naive rendering reverses the sentence — man would be stripped of his dignity in the very clause that says he recovers it. This is a Pattern 7a mechanism-5 trap (agency/voice), and it leaves no lexical trace.
• 0001 — the unclosed guillemet. Calfa prints καὶ ὁ « καινὸς ἄνθρωπος, … with no closing ». Pattern 8: never supply, never delete. The « is carried into the English at the same point and left unclosed. The plate's own OCR shows no guillemet, so this may be Calfa's artifact rather than Migne's mark; it is not corrected either way, and a leaf render would settle it.
• 0001 — ἵν᾽ ὅτον ἡ Παρθένος Θεῷ προσωμοίωται. Calfa's ὅτον is not a word; the plate OCR gives ὅτου. Read as ὅσον ("how far") with the Latin (quam similis Deo esset Virgo). Rendered "how far"; the letter is unresolved.
• 0001 — καὶ τῷ προπάτορι ταύτης ἐδέησεν, ὅσαπερ οἱονεὶ μητρὶ χρήσασθαι. ὅσαπερ is what both Calfa and the plate print, and the clause will not construe smoothly with it (one expects ὥσπερ, or ταύτῃ ὥσπερ). Rendered as closely as the printed words allow — "had need — so far as to use her, as it were, as a mother" — rather than smoothed to just as. Two-witness concurrence: the difficulty is the plate's, not ours.
• 0001 — τῆς ἔσταξαν λέξεως, a word mentioned as a word. The Nicetas rule keeps a mentioned particle in Greek script; that rule was written for disputed credal particles. Here the word discussed is one the English has just translated, and keeping ἔσταξαν in Greek would make the sentence opaque. Rendered as italic English — "the word dropped down" — matching the quotation four words earlier. Decision recorded so it is not churned; if the convention is later tightened, this is the site.
• 0002 — τῶν μὲν ἄλλων ἕκαστος … προσῆγον. Singular subject, plural verb. Pattern 9: no English exponent, no emendation to make or undo. Logged so a later sweep does not churn it.
• 0002 — ὁ δὲ … Ῥουβὶμ … οὔτε λαμβάνει τὰ δῶρα, καὶ ὀνείδεσι βάλλει. The correlative is broken on the plate (οὔτε … καί for οὔτε … οὔτε or οὐ … ἀλλά); the Latin has non acciperet, sed praetereaacerbis eos conviciis impeteret, i.e. it repairs the correlation. Rendered with the break kept — "neither receives the gifts, and pelts …" — under Pattern 7. Plate-verified: the Paris scan's Greek reads οὔτε λαμβάνει τὰ δῶρα, καὶ ὀνεί-. Not a [lat:]: Allatius is smoothing syntax, not asserting a different fact.
• 0002 — Ῥουβίμ. Kept as printed (transliterated Rubim, not normalized to Ruben). Migne's own note (18) observes that Isidore appears to have taken Rubim seu Ruben for the priest's proper name — the plate has already adjudicated it, so no marker of ours.
• 0003 — ὁ πάντων τῶν τῆς ἐκείνου μανίας … ἐπέκεινα Θεοῦ χωρισθεὶς … ἀσώματος. The word order permits binding ἐπέκεινα either to χωρισθείς or to the genitive plural; the Latin binds it to the devil's pre-eminence (cunctis insanis ipsius sociis … supereminens). Rendered with the Latin's binding; the choice is recorded rather than silently made, per the "never let the Latin settle a reading without saying so" rule.
• 0003 — ἔμελλεν/ἐμέλησε and Χαῖρε/χαίρει. Two untranslatable figures; the second is preserved by rendering χαῖρε as Rejoice (§2).

9. Ratio and shape

Greek 1009 / 1010 / 1060 words → English ≈ 1.5× on each chunk, tracking the anchor's 1.49× homily ratio rather than the chronicle 1.25×.

Where the sermon ends. Chunk 0003's tail is the end of Sermon I: the doxology … ᾗ πρέπει δόξα εἰς ἀπεράντους αἰῶνας τῶν αἰώνων. Ἀμήν. falls inside column 40, and the plate confirms that ΛΟΓΟΣ Βʹ. Εἰς τὴν εἰς τὰ Ἅγια τῶν ἁγίων εἴσοδον … (Sermon II) begins in the same column 40, immediately below. So the work ends mid-column 40, with Sermon II starting on the same column — which is why chunk 0003 carries a [0040] anchor and stops without a further one. The three remaining sermons are not part of this pilot; when they are taken up, the epithet table above and the anchor's must travel with them, since Glabas reuses the whole vocabulary across the set.

Work-wide conventions (anchor session, 2026-07-31 — merge into the work-wide section)

• ΛΟΓΟΣ as a work-division head renders by the work's established English division term — here "SERMON" (registry + the plate's own Latin column: SERMO I) — consistent within the work. (Nicetas uses "Discourse" for the same word; per-work, not per-corpus.)
• Εὐλόγησον, Πάτερ — the liturgical formula before the sermon renders "Give the blessing, Father."
• Mariological epithets — fixed equivalences, one per Greek epithet, never interchanged: πάναγνος = all-pure · πανυπέραγνος = all-surpassingly pure · Πανάχραντος = All-immaculate · Πάντιμος = All-honored · πολυύμνητος = much-hymned · Θεομήτωρ = Mother of God · Δέσποινα = Lady · ἀειπάρθενος = ever-virgin · ἡ μακαρία = the blessed one. Glabas builds arguments on the differences between epithets; collapsing two into one English word is the homily analogue of a lemma-conformation.
• Maker-vocabulary distinctions, fixed within the work: Κτίστης = Creator · Δημιουργός = Maker · πλάστης = Fashioner · πλάσμα = moulded work · δημιούργημα = created work.
• Rhetorical pile-ups are structure, not padding: doublets/triplets (γυμνὸς, ἐλεεινὸς, αἰχμάλωτος) render 1:1, never compressed; exclamations kept (οἴμοι! = alas!); rhetorical questions stay questions.
• Bold coinages are rendered boldly, not normalized: ταφόζωος = tomb-bound (existence) · βρεφουργηθέντες = wrought as babes · ἔξαλμα = outleaping · προέτεκε = bore before. Allatius routinely flattens these (sepulcralem vitam, in quadam infantia); his paraphrase is NOT a license to flatten the Greek.

Latin-twin pass (pilot §6): result for chunk 0000 — FINDINGS (not clean)

Read against src/pg-latin/isidore-glabas-sermo-1/0000.md (cols 0011–0018; NOTE: the Latinited for cols 0011–0014 comes from the ALT witness patrologicursus73migngoog, per the twin's provenance.altSourcePages). This batch is chunk 0000 only; it was not clean. Polarity/negation collation found NO divergence between the columns (all of οὐ πρότερον/μηδ᾽ ἐγγὺς/οὐκ ᾤετο/ οὐδενὸς/οὐδ᾽ ἂν ἐμέλησε/εἰ μὴ ἀποπηδήσειεν carried in both). No numerals in the chunk beyond section marks. No [lat:] fired.

Ladder row 1 — Calfa OCR damage, corrected silently, Latin/context cited

• 0012 (title) Θεοχοήτορος → Θεομήτορος (Lat. Dei Genitricis).
• 0012 Section numeral: Calfa's Ηʹ → Βʹ (the Latin column numbers it II; sequence runs Αʹ–ϛʹ). Rendered "2."
• 0012 κςσιῳ δογματίζεσυες → κόσμῳ δογματίζεσθε (Col 2:20 as printed; Lat. in mundo decernitis).
• 0012 μέχρι; οὗ → μέχρις οὗ · ὡφέλιμον → ὠφέλιμον · οἶ → οἵ (routine).
• 0013 Παρβένου ποοῖλθε → Παρθένου προῆλθε · λόγμον → λόγιον (Lat. oraculum) · ὕανεῖται → φανεῖται.
• 0016 πανάννου → πανάγνου · υγχωρήσειε → συγχωρήσειε · λγαλμα → ἄγαλμα · ὑπερόολῆς → ὑπερβολῆς · ἰσοθεὶαν → ἰσοθεΐαν · χοῖκῷ → χοϊκῷ.
• 0017 θά λασσα → θάλασσα (split type) · πρὸς ανον → πρὸς αἶνον (Lat. ad laudandum) · τοῦ Κκροτεῖν → τοῦ κροτεῖν (intrusive-Κ class again — third instance in PG 139's Calfa file, cf. nicetas cruces-0000).
• 0017 σιγὴν ἀπήνει — the finite verb is OCR-damaged past recovery from our files (σιγὴν ἦγε? ἐσίγα?); the SENSE "kept silence" is certain from the coordinated participle μὴ τὸν ὕμνον ἀναβαλλόμενα and Lat. jam silebant, neque referebant hymnos. Rendered "kept silence," exact Greek wording unresolved — plate check would settle the letter.

Construal notes (not defects)

• 0017 ἑωρακότα is neuter plural nominative with τὰ καλὰ πάντα — the CREATURES "having seen their master pinioned"; agrees with Allatius's cernerent. Calfa's καθάπερ ἑωρακότα is sound; no emendation.
• 0016 ἔμελλεν / ἐμέλησε wordplay ("had this not been to come… would not have taken thought") — untranslatable pun, sense rendered, logged here.
• 0012 εἶκε ("yielded") vs Lat. paruit ("obeyed") — Allatius freedom, not a divergence.

Twin apparatus carried for later harvest (not ours to mark)

Scripture identifications printed in the Latin column's notes for this span: Exod. XIV,31 · Exod. XVII,1 seqq. · Exod. XVI,1 seqq. · Gen. 1,2 (the In principio quote) · Col. 2,20 · Ps. 66,7 (Terra dedit fructum suum) · Rom. 6,5 (complantati) · Gen. 1,31 · Ps. 18,2 (caeli enarrant). Plus Ballerini's editorial notes (2)–(4) and the long immaculist annotations — plate apparatus, in the twin, verifier-only.

Alt-witness note

Cols 0011–0014 of the twin (sermon title + §1) come from the second physical copy (patrologicursus73migngoog, tesseract render+OCR) — the primary scan has no leaves for Calfa pp. 13–14. The title and §1 Latin were still usable as verifier (Θεομήτορος and the Colossians quote both confirmed against it).

Continues cruces-0000.md (the Fable anchor), whose work-wide conventions section is binding here and is not restated. This file records (a) new epithets and maker-words met in 0001–0003 and the English fixed for each, (b) the scripture-divergence list, (c) the Latin-twin pass per chunk, (d) every silent OCR correction, (e) the one [lat:] fired and the candidates rejected with reasons.

Witnesses used. Greek = sources/pg/calfa as chunked. Latin twin = src/pg-latin/isidore-glabas-sermo-1/0001–0003.md. Plate = the Paris scan raw/scans/pg139/patrologiaecurs63migngoog_djvu.txt (leaves 10–19), consulted directly for this batch and quoted below where it settles a reading. Known gap: the leaf carrying PG 139 cols 27/28→29/30 (Calfa p. 21) is absent from the primary scan, so the stretch of chunk 0002 from Δάκνει τοῦτο τὸ ῥῆμα to ΙΒʹ has no plate witness at all and its Latin comes from the alt witness patrologicursus73migngoog (tesseract render + OCR, per the twin's provenance.altSourcePages). Every claim resting on that stretch says so.

1. NEW epithets and maker-words — English fixed here, to be carried forward

Extending the anchor table (πάναγνος = all-pure · πανυπέραγνος = all-surpassingly pure · Πανάχραντος = All-immaculate · Πάντιμος = All-honored · πολυύμνητος = much-hymned · Θεομήτωρ = Mother of God · Δέσποινα = Lady · ἀειπάρθενος = ever-virgin · ἡ μακαρία = the blessed one · Κτίστης = Creator · Δημιουργός = Maker · πλάστης = Fashioner · πλάσμα = moulded work · δημιούργημα = created work):

Coinages rendered boldly (never normalized; Allatius flattens all of them): προδιεκωδώνισε = sounded the bell beforehand (Lat. merely praenuntiasse) · ἱστουργία = weaving (Lat. drops the metaphor: gloria circumvestirentur) · καινοτοκεῖν = new-bearing · ἐξωβελικότες = having struck out with the obelus (Lat. repudiassent) · ἐκαινούργησε / ἐνεούργησεν = wrought newly / made new · θεουργικώτατον = most God-working and πλαστουργικώτατον = most fashion-working (Lat. collapses both into sacratissimum hoc ac plane dignissimum … germen, losing the pair outright) · κύημα = conception (Lat. filiam, "daughter", at 0001 — a flattening) · θεάνθρωπος = God-man (in the clause restored at 0037, below).

Register note carried from the anchor: where Calfa capitalizes a substantivized epithet (τὴν Πάναγνον, τῇ Πανάγνῳ) the English capitalizes ("the All-pure one"); where it is lower-case and adjectival (τὴν πάναγνον) it is not. This follows the plate's own practice and is applied consistently across 0001–0003.

2. Scripture-divergence list — where Migne's Greek departs from the received text

This is the deliverable. Every place in 0001–0003 where the printed Greek differs from the received LXX/NT is named here with the specific words. Markers [var: …] were fired only where the sense turns on it; the rest are recorded here alone.

Marked [var:] in the English:

1. 0001 @0020 — Ps 88:7 (LXX). Migne prints Τίς ἐν νεφέλαις ἰσωθήσεται … τῷ Κυρίῳ, ὁμοιωθήσεται τῷ Κυρίῳ ἐν υἱοῖς Θεοῦ; — one interrogative. LXX repeats it: … καὶ τίς ὁμοιωθήσεται τῷ Κυρίῳ …. Plate-verified in the Greek column, and adjudicated on the plate itself: Ballerini's note (9) — Isidorus hic magis cum Vulgata Latina quam cum Graeca LXX editione consentit; haec enim ita bis repetit τὸ τίς … quae tamen repetitio abest etiam ab Complutensi ac Aldina, et ab Alexandrino codice. Rendered as printed; [var:] names the LXX reading.
2. 0001 @0020 — Hab 3:3. Migne prints Ἐκάλυψεν οὐρανοὺς ἡ ἀρετὴ αὐτῆς, καὶ τῆς συνέσεως αὐτῆς πλήρης ἡ γῆ — both pronouns feminine, and σύνεσις ("understanding") where LXX has αἴνεσις ("praise"). Plate-verified; Ballerini's note (10) says outright that Isidore agrees neither with the Vulgate (gloria ejus … laudis ejus) nor with the LXX (ἡ ἀρετὴ αὐτοῦ, καὶ αἰνέσεως αὐτοῦ). The Latin twin's sapientiae ejus plena est terra concurs with σύνεσις — two-witness concurrence (ladder row 3), so the reading is attributable to the plate pair, not to our file.
3. 0001 (end) — John 1:14. Ἐθεασάμεθα τὴν δόξαν αὐτῆς, δόξαν ὡς μονογενοῦς Μητρὸς Θεοῦ for δόξαν ὡς μονογενοῦς παρὰ πατρός. Plate-verified; the Latin concurs (quasi gloriam unigenitae Dei Matris) and Ballerini annotates the substitution at note (15). Rendered as printed.
4. 0002 @0025 — Col 3:3–4. Ἡ ζωὴ ἡμῶν ἐκέκρυπτο σὺν αὐτῇ … Ὅτε αὕτη πεφανέρωται … καὶ ἡμεῖς σὺν αὐτῇ ἐφανερώθημεν for the received ἡ ζωὴ ὑμῶν κέκρυπται σὺν τῷ Χριστῷ … ὅταν ὁ Χριστὸς φανερωθῇ … ὑμεῖς σὺν αὐτῷ φανερωθήσεσθε. The author announces this himself (μικρὸν παρῳδήσαντας), so the change of person and referent is authorial, not textual — but it is material and the English must not drift back toward the received verse. On the ἡμεῖς/ὑμεῖς itacism class: both columns read first person (Lat. Vita nostra … et nos apparuimus), so no [lat:].
5. 0002 @0025 — Hab 3:13. Migne's Greek adds ἐλήλυθας ("thou art come"), which the LXX lacks. Plate-verified, and Migne's own Ballerini note (16) raises exactly this: Utrum Isidorus τὸ ἐλήλυθας, quod a LXX abest, ideo addiderit, quod ita in Codice legeret, incertum. The Latin column concurs (venisti). Note also Ἐξῆλθες … τοῦ σῶσαι τοὺς χριστούς σου is addressed to the Virgin here, feminine in the Latin (Egressa es) — the refitting the author calls μεθαρμόσειεν.
6. 0002 @0025 — Luke 1:78. Migne prints the aorist Ἐπεσκέψατο ἡμᾶς ἀνατολὴ ἐξ ὕψους; the received NT has the future ἐπισκέψεται. Plate-verified; Latin concurs (Visitavit). Two-witness concurrence.
7. 0002 @0029 — Isa 26:18 (LXX). Migne prints … ἐτέκομεν πνεῦμα σωτηρίας σου, ὃ ἐκυήσαμεν ἐπὶ τῆς γῆς; LXX has πνεῦμα σωτηρίας σου ἐποιήσαμεν ἐπὶ τῆς γῆς. Latin concurs (quod utero gessimus super terram). Polarity note, and it is the one worth flagging: Ballerini's note (24) prints the Vulgate of the same verse — salutem non fecimus in terra — which is negative where both of Migne's columns are positive. That is a Vulgate/LXX difference, adjudicated on the plate by Migne's own note; the two columns agree with each other, so no [lat:], and the English follows the Greek.
8. 0002 @0029 — Col 2:3. Ἐν αὐτῇ γάρ εἰσι … πάντες οἱ θησαυροὶ τῆς σοφίας καὶ τῆς γνώσεως ἀπόκρυφοι for ἐν ᾧ εἰσιν … σοφίας καὶ γνώσεως. The whole point is the transfer from Christ to the Virgin; the added article on γνώσεως is minor and is not in the marker.

Recorded here only (divergence real but the sense does not turn on it — no marker, to avoid marker spam):

• 0001 @0020 — John 3:31. Ὁ ἐξ οὐρανοῦ ἐρχόμενος ἐπάνω πάντων ἐστίν; received ὁ ἐκ τοῦ οὐρανοῦ ἐρχόμενος. Plate-verified. Article only.
• 0001 @0021 — Ps 32:13–14 (LXX), conflated. Ἐξ οὐρανοῦ ἐπέβλεψεν ὁ Κύριος· εἶδε πάντας τοὺς υἱοὺς τῶν ἀνθρώπων ἐξ ἑτοίμου κατοικητηρίου αὐτοῦ welds v. 13 to v. 14, dropping the second ἐπέβλεψεν. Migne flags the conflation himself by citing Psal. xxxi, 13, 14 on the plate. Latin concurs. Rendered as the weld stands.
• 0001 @0021 — Ps 67:9 (LXX). Γῆ ἐσείσθη, καὶ γὰρ οἱ οὐρανοὶ ἔσταξαν — quoted without ἀπὸ προσώπου τοῦ Θεοῦ; a truncation, not a variant.
• 0001 @0020 — Ps 67:17 (τὸ μετεωρότατον ὄρος, ὃ εὐδόκησεν ὁ Θεὸς κατοικεῖν ἐν αὐτῷ) and Ps 10:4 / 102:19 (Κύριος ἐν ναῷ ἁγίῳ αὐτοῦ· Κύριος ἐν οὐρανῷ ὁ θρόνος αὐτοῦ) are woven into the homilist's own syntax but verbally as printed in LXX.
• 0002 @0028 — the priest's rebuke. Οὐκ ἔξεστι … ὑμῖν δῶρα Θεῷ προσενεγκεῖν, καθότι ἐν τῷ Ἰσραὴλ σπέρμα οὐκ ἐποιήσατε is not canonical scripture: it is Protevangelium Jacobi 1:2, there singular and in a different order (οὐκ ἐξεστίν σοι πρῶτον ἐνεγκεῖν τὰ δῶρά σου, καθότι σπέρμα οὐκ ἐποίησας ἐν τῷ Ἰσραήλ). Rendered as Migne prints it, in the plural. Migne prints no citation for it; do not index it as scripture.
• 0002 @0028 — Matt 3:9 woven, not quoted (τὸν δυνάμενον ἐκ λίθων ἐγεῖραι τέκνα, without ἐκ τῶν λίθων τούτων … τῷ Ἀβραάμ).
• 0003 @0037 — 1 Pet 2:22 / Isa 53:9. See §4: the first half of the quotation is missing from our Greek and is restored from the plate in an [ed:].
• 0003 @0040 — Luke 1:28. Χαῖρε κεχαριτωμένη, δι᾽ ἧς χαίρει πᾶσα ἡ κτίσις· ὁ Κύριος μετὰ σοῦ — the middle clause is the homilist's insertion into the angelic salutation (Latin concurs: per quam omnis creatura gaudet), and it turns on the χαῖρε/χαίρει pun. Rendering decision: χαῖρε is englished as Rejoice rather than Hail precisely so the pun survives ("Rejoice … through whom all creation rejoices"); Hail would have silently destroyed the figure the insertion exists to make.

3. The Latin-twin pass, chunk by chunk (pilot §6) — including the chunks that were clean

Method as specified: draft first, then one deliberate pass against the twin hunting (1) OCR damage in our Greek, (2) different asserted facts, (3) polarity disagreement.

• Chunk 0001 — NOT clean. One [lat:] fired (§5). Polarity collation found no disagreement: οὐδεὶς … ἀντίῤῥοπος / nemo … idoneus repertus est; οὐδὲ γάρ ἐστιν … οὐδ᾽ ἂν εἷς … οὐδεὶς / Nulla enim … Sed neque in posterum … ullus; οὐ θανασίμους … ἀνοιγνύμενοι / non quidem apertos ut olim mortiferas … pluvias demisisse; οὐκ ἄδηλον / satis manifestum sit; οὐκ ἄν τις ἁμάρτοι / ne minimum quidem a veritate aberrat; εἰ καὶ μὴ πρὸς οὐδὲν ἴσως ἕτερον / et sin minus ad aliud quidpiam; οὐ παρ᾽ ἑαυτοῦ / neque enim aurum … praebet. All carried both sides.
• Chunk 0002 — NOT clean. One [ed:] fired for a lacuna (§4). No [lat:]. One [lat:] candidate rejected (§5). Polarity: οὔτε λαμβάνει τὰ δῶρα / dona non acciperet; σπέρμα οὐκ ἐποιήσατε / semen in Israel non fecistis; μὴ παρὰ γνώμην τι καινοτομῆσαι / non quidem ad aliquid praeter Dei voluntatem moliendum; ὥσπερ οὐκ ἀνασχομένη, μὴ … μεταδοῦναι / quod ferre nempe ipsa non posset, ut ceteri … in partem non vocarentur — the double negative is carried in both columns and is carried in the English ("could not endure not to impart"); οὔτε … οὔτ᾽ εἶχε / Neque enim … aliam vero gignere … jam non poterat; οὐδὲ γὰρ οἷόν τε / Neque enim fieri poterat; οὔτε ἄνθρωποι, οὔτε ἄγγελοι, οὔτε τις ἑτέρα κτίσις / neque homines, neque angeli, neque ulla creatura; οὐδένα … πεποίηται λόγον / nihil prorsus curavit; τῷ μὴ παῖδα … τεκεῖν ἕτερον / aliam prolem amplius non genuerunt. No disagreement.
• Chunk 0003 — NOT clean. One [ed:] fired (§4). No [lat:]. Polarity, the densest stretch in the work and checked word by word: οὔτε γῆς αὖθις ἐδεήθη … οὔτε μέλους ἀνδρός / neque terram iterum assumpsit … neque membrum aliquod hominis; ὃ μήτ᾽ ἐκ γῆς … μήτε τι τῶν μελῶν ἀφήρηται / quod neque e terra primo compactum esset, neque membris deficeret; μὴ πρὸς τέλος ὁρᾷν ἐχαρίσατο / ut nullum unquam finem attingerent; οὐδενὸς τῶν οὐκ ἐπαινουμένων … κοινωνός / prorsus exsors eorum omnium, quae in hominibus non censentur laude digna — the double negative stands in both and is carried; μόνην ἐχρῆν μὴ … κοινωνεῖν δυσκληρίας / sola communis … miseriae expers esset; οὐκ ἂν ἦν οὕτω θαυμαστόν / neque tanta admiratione digna forent; οὐκ ἀνήλωκεν / non … absumpsit; οὐ παρακατέχειν ἔσχεν ἐπιπολύ / non diu apud se retinere illud potuit; μὴ … διεφθορέναι / corruptioni reipsa obnoxium non fuit; οὐκ ἀντιλέγοντας ἕξει καὶ ἡμᾶς / nos utique contradicentes non habebit — note the Greek prints the litotes and the English keeps it ("he will not have us contradicting him"), not the smoothed positive; πρὸς ἃ μὴ πέφυκε / quibus perfruendis nata non est; οὐδεὶς … ἄψαυστος γέγονεν / nemo hominum exstitit, qui … integer fuerit; κατὰ … τὸ μηδὲν ἁμαρτεῖν` / ad hanc ab omni peccato immunitatem*. No polarity disagreement anywhere in the batch.

Chunks on which the twin pass found nothing at all: none. Each of the three yielded at least one marker. The pass is therefore not reportable as "clean" for any chunk, and the absence of a clean chunk is itself the honest result.

Pronoun-itacism check (ἡμεῖς/ὑμεῖς), the class the paired pilot flagged: four sites in this batch — Ἐπεσκέψατο ἡμᾶς (0002, Lat. nos), ἡ ζωὴ ἡμῶν … καὶ ἡμεῖς (0002, Lat. nostra … nos), τεχθήσεται ὑμῖν παιδίον (0002, Lat. vobis), ὡς τὰ καθ᾽ ἡμᾶς and οὐκ ἀντιλέγοντας ἕξει καὶ ἡμᾶς (0003, Lat. corpora nostra, nos). All four agree across the columns. No [lat:] in this class; no plate check needed.

Numerals: the only numerals in the batch are the section marks (§6) and τεσσαράκοντα ὅλας ἡμέρας / quadraginta solidos dies (0002) — which agree. No numeral [lat:].

4. Source loss — the two [ed:] markers, and why each is not a [sic:]

(a) 0003, before the anchor [0037] — a whole clause of 1 Pet 2:22 lost from our Greek. Calfa runs … τὸν καινόν φημι τῆς μακαρίας τόκον, [0037] οὐδὲ εὑρέθη δόλος ἐν τῷ στόματι αὐτοῦ — where the οὐδέ dangles with no preceding negative clause to answer. The Latin twin has more: praeter illum novum, inquam, hujus Beatae Filium, qui solus, utpote … Deus homo, peccatum non fecit, neque dolus inventus est in ore ejus (twin 0003, Latin col. 0037/0038). The plate settles it. At the foot of the Greek column 35 the Paris scan reads:

ὃς μόνος οἷα θεάνθρωπος ἁμαρτίαν οὐκ ἐποίησεν,

immediately before the page turn at whose head the scan's Greek resumes οὐδὲ εὑρέθη δόλος ἐν τῷ στόματι αὐτοῦ. So the clause is on Migne's plate and is missing from our source: a digitization loss, Pattern 13, not a plate defect — hence [ed:], not [sic:]. The marker quotes the plate (the PG precedent set by the Nicetas anchor's restored ΛΟΓΟΣ ΠΡΩΤΟΣ heading at col. 0173). Note that the lost clause carries a negation (ἁμαρτίαν οὐκ ἐποίησεν): without it our English asserted, of the one exception to universal sin, only the second half of the proof-text. Recommended follow-up (not done here, since it changes stored Greek): a data/calfa-patches/ entry restoring the clause. The ladder requires a plate check for that, and the plate check is in hand — cited above.

(b) 0002, §11 tail — a lacuna at a page boundary, unrecoverable. Calfa runs … ἐκεῖ μὲν γὰρ κατάρα, χαρὰν δὲ ὁ νέος ΙΒʹ. Τῇ οὖν τοιαύτῃ …: ὁ νέος has no verb and no noun, and the section number follows immediately. The Latin twin breaks off at the same place (Sed et locus, qui Annae precum testis fuit, sedi, in qua versabatur prima parens, then the column anchor). Both witnesses fail together — ladder row 4, no attribution — and the plate cannot arbitrate because this is exactly the leaf the Paris scan lacks. The [ed:] therefore states the loss and states that it cannot be repaired from our files; it supplies nothing. This is not a Pattern-16 case and not a defect of Migne's.

5. [lat:] — one fired, one candidate rejected

FIRED — 0001, Greek col. 0024 / Latin col. 0023: καινήν vs communem. Greek: … τὴν καινὴν δὴ ταύτην ἐζήτουν καὶ τελεσιουργικωτάτην καὶ φωτοδότιδα μητέρα. Latin twin, col. 0023: … communem hanc matrem quaerebant, quae ad perfectam formam proli tribuendam, et ad lucem impertiendam maxime par foret. The Latin translator read κοινήν, "common"; the Greek prints καινήν, "new". Greek plate-verified — the Paris scan's Greek column reads τὴν καινὴν δὴ ταύτην ἐζήτουν, independently of Calfa. The divergence is material: the Greek's argument is that every father, having rejected his own mother, sought a new one; communem makes her a mother shared, which is a different claim. English follows the Greek; marker records the Latin. (This is the οι/αι confusion class — the prose analogue of the itacism pairs, and the second live [lat:] of the paired pilots.)

REJECTED — 0002, §11: Ἀχάβ vs Achaz. Greek: ἢ ὡς Ἰεζάβελ θύειν τῇ Ἀστάρτῃ τὸν Ἀχάβ. Alt-witness Latin: aut sicut Jezabel ad sacrificandum Astartae impulit Achaz. Ahab was Jezebel's husband; Ahaz is a different king — on its face a textbook Pattern 16 name divergence. Not fired, and here is why: this passage falls in the stretch for which the primary scan has no leaf, so the only witness to the Latin reading is uncorrected tesseract OCR of a second physical copy, whose output on the same lines gives sacriflcandum and As- tart;. A b→z misread is well inside that OCR's demonstrated error rate, and firing [lat:] would be a public claim about what Migne's plate prints resting on our file alone — the exact attribution the PG standing rule forbids. Logged as a live candidate: if the leaf is ever obtained, this is the first thing to check. (For what it is worth, the primary scan renders Ἀχὰβ / Achabum correctly a column earlier, in the Ahab-and-Gideon comparison; that is the same name in the same work, and it tells against Achaz being Migne's.)

Not divergences (Allatius's freedom, per Pattern 16's own boundary) — named so a later pass does not re-raise them: inaccessibilis splendor for ἄδυτον … φέγγος ("unsetting") · coelorum filiam for οὐρανῶν κύημα ("conception of the heavens") · venustissima puella for ἡ καλλίπαις · nascetur vobis puella for τεχθήσεται ὑμῖν παιδίον ("a child") · the whole added clause cum scilicet divino huic ac mystico corpori ex hominibus tanquam membris compingendo ipsa Dominum tanquam caput imposuit (Lat. col. 0032) expanding τῶν γὰρ αὐτοῦ μελῶν κεφαλὴν ἐπεκάθισε τὸν Δεσπότην · Quaquaversus enim resonat maximi illius theologi vox supplied where the Greek has only ἡ θεολογικωτάτη … φάσκει φωνή. Scripture in the Latin column conformed to the Vulgate is convention, not divergence and fires nothing: similis erit Deo in filiis Dei for ὁμοιωθήσεται τῷ Κυρίῳ (Ps 88:7) is the worked instance here, and Ballerini's note (9) says as much on the page.

6. Section numerals — the sequence, and one unresolved

Calfa's numerals in this stretch are heavily damaged; the Latin column's Roman numerals and the plate settle them. Rendered as arabic + period, per the anchor.

7. Silent OCR corrections — logged as classes, then per site

Class 1 — the intrusive Κ, already known in Calfa's PG 139 file (Κκροτεῖν, ΠνενύΚματος in the anchor and in Nicetas). Four more instances in this batch, all corrected silently: Κὁοὐρανὸς → ὁ οὐρανὸς (0001) · Κσταξαν → ἔσταξαν (0001, Ps 67:9) · ἐδημιούρΚγησεν → ἐδημιούργησεν (0003) · Κοὐδὲ → οὐδὲ (0003 @0037). The class is now seven attested sites across three works in this Calfa volume — worth a corpus note, not worth per-site marking.

Class 2 — split and run-together type (Pattern 10 territory, but these are our source's, not the plate's, so they are corrected rather than carried): θά λασσα → θάλασσα (anchor) · ὑύη λὸν → ὑψηλὸν (0001) · ἐπιμ ησθέντες → ἐπιμνησθέντες (0002) · ἀδι ήσαμεν → ὠδινήσαμεν (0002) · ἐπι.σπεύδειν → ἐπισπεύδειν (0003) · διασώσω. μεν → διασώσωμεν (0003).

Per site, chunk 0001 (plate-verified where marked ✱): Ὁψὲ → Ὀψὲ · ὅρόσος → δρόσος · λύτται → λύεται ✱ · εὖρε → εὗρε · ἶνα συνελών εἴπω → ἵνα συνελὼν εἴπω ✱ · ὡραῖσμοῦ → ὡραϊσμοῦ · ὄψοςἰ χὗ → ὕψος! Ζʹ. ✱ (§6) · ὁξίαν → ἀξίαν · Δαβὲ → Δαβὶδ ✱ · τῷ Κυρίφ → τῷ Κυρίῳ · ἀλλ’οὺδ᾽ ἂν εἶς → ἀλλ᾽ οὐδ᾽ ἂν εἷς ✱ · τοσοῦταν → τοσοῦτος ✱ (plate) · Ἐκάλυύεν → Ἐκάλυψεν ✱ · θώηλότατον → ὑψηλότατον ✱ · ἐξ οὐανοῦ → ἐξ οὐρανοῦ ✱ · τὴν νοσον → τὴν νόσον · χεῖρα ὁρέξαι → ὀρέξαι · ἁν εἰκότως → ἂν εἰκότως · τοῦ Οείου Δαβίδ → τοῦ θείου Δαβίδ · ἀπιβλέψαι → ἐπιβλέψαι · οἷθε κάντας → εἶδε πάντας ✱ · κατκητηρίου → κατοικητηρίου ✱ · ναῷ ἀγίῳ → ναῷ ἁγίῳ · στιν ὃς → ἔστιν ὃς ✱ · Ἀννα → Ἄννα · ἡ προσευνῆς → ἡ προσευχῆς ✱ (Lat. quae per preces mater ei fuit) · ἐρείψιμον ἐχούσης sound as printed.

Per site, chunk 0002: Ἐκεσκέψατο → Ἐπεσκέψατο ✱ · Ἐκιὰ γὰρ → Σκιὰ γὰρ ✱ · Τεδεών → Γεδεών ✱ · συνειτενεγκόντι → συνεισενεγκόντι · ποιἡσειν → ποιήσειν · ἐγεῆραι → ἐγεῖραι · γαστρόσ → γαστρός · Ἰοραὴλ → Ἰσραὴλ · συντόνω → συντόνῳ · Ἐκόαίνει → Ἐκβαίνει ✱ · ἡ δέσσις → ἡ δέησις ✱ · τεχόήσεται → τεχθήσεται ✱ · Διὰ τὀν ρόβον → Διὰ τὸν φόβον ✱ · ἐν γεστρὶ → ἐν γαστρὶ ✱ · ἐτένομεν → ἐτέκομεν ✱ · Κυνθείσης → Κυηθείσης ✱ · τῆς μονογενους → τῆς μονογενοῦς · τοῦ τκτειν ἔξῆς → τοῦ τίκτειν ἑξῆς ✱ · ὁ´ ἀφθονίᾳ → δ᾽ ἀφθονίᾳ · παρά [0032] τοὺς λοιπούς sound.

Per site, chunk 0003: παυορᾶται → παρορᾶται ✱ · τὸν λεῖπον → τὸ λεῖπον · φανενωθῇ → φανερωθῇ ✱ · ἐνῆν sound ✱ · δισκληρίας → δυσκληρίας ✱ · γένεσιω → γένεσιν ✱ · ἀθλου δημιουργίας → ἀΰλου δημιουργίας ✱ (Lat. immaterialis esse naturae) · διειμγασμένης → διειργασμένης ✱ · πανἀννου → πανάγνου · ἐν πῷ περὶ → ἐν τῷ περὶ · εἰρήαι → εἴρηται · ἀφθαροτέρα τῳ μέρει → ἀφθαρτοτέρα τῷ μέρει ✱ · πανὰμνητος → πανύμνητος ✱ (§1 — the load-bearing one) · θερυργικώτατον/Calfa θεουργικώτατον sound · Εθεν → ὅθεν ✱ · κάκείνους → κἀκείνους · Χαῆρε → Χαῖρε · ἀπελαύσαμεν sound (Calfa ο´ ἀπελαύσαμεν → δ᾽ ἀπελαύσαμεν) · ἐνεούργησεν sound · Θεοτόκῳ sound.

8. Construal cruces — printed reading kept, conjecture here

• 0001 — καὶ τὸ πρῶτον ἀναιρεῖται ἀξίωμα. ἀναιρέω in the active means destroy; the middle here means takes up for himself, and the Latin confirms it (pristinam resumit dignitatem). Rendered "takes up again his first dignity." Flagged because the naive rendering reverses the sentence — man would be stripped of his dignity in the very clause that says he recovers it. This is a Pattern 7a mechanism-5 trap (agency/voice), and it leaves no lexical trace.
• 0001 — the unclosed guillemet. Calfa prints καὶ ὁ « καινὸς ἄνθρωπος, … with no closing ». Pattern 8: never supply, never delete. The « is carried into the English at the same point and left unclosed. The plate's own OCR shows no guillemet, so this may be Calfa's artifact rather than Migne's mark; it is not corrected either way, and a leaf render would settle it.
• 0001 — ἵν᾽ ὅτον ἡ Παρθένος Θεῷ προσωμοίωται. Calfa's ὅτον is not a word; the plate OCR gives ὅτου. Read as ὅσον ("how far") with the Latin (quam similis Deo esset Virgo). Rendered "how far"; the letter is unresolved.
• 0001 — καὶ τῷ προπάτορι ταύτης ἐδέησεν, ὅσαπερ οἱονεὶ μητρὶ χρήσασθαι. ὅσαπερ is what both Calfa and the plate print, and the clause will not construe smoothly with it (one expects ὥσπερ, or ταύτῃ ὥσπερ). Rendered as closely as the printed words allow — "had need — so far as to use her, as it were, as a mother" — rather than smoothed to just as. Two-witness concurrence: the difficulty is the plate's, not ours.
• 0001 — τῆς ἔσταξαν λέξεως, a word mentioned as a word. The Nicetas rule keeps a mentioned particle in Greek script; that rule was written for disputed credal particles. Here the word discussed is one the English has just translated, and keeping ἔσταξαν in Greek would make the sentence opaque. Rendered as italic English — "the word dropped down" — matching the quotation four words earlier. Decision recorded so it is not churned; if the convention is later tightened, this is the site.
• 0002 — τῶν μὲν ἄλλων ἕκαστος … προσῆγον. Singular subject, plural verb. Pattern 9: no English exponent, no emendation to make or undo. Logged so a later sweep does not churn it.
• 0002 — ὁ δὲ … Ῥουβὶμ … οὔτε λαμβάνει τὰ δῶρα, καὶ ὀνείδεσι βάλλει. The correlative is broken on the plate (οὔτε … καί for οὔτε … οὔτε or οὐ … ἀλλά); the Latin has non acciperet, sed praetereaacerbis eos conviciis impeteret, i.e. it repairs the correlation. Rendered with the break kept — "neither receives the gifts, and pelts …" — under Pattern 7. Plate-verified: the Paris scan's Greek reads οὔτε λαμβάνει τὰ δῶρα, καὶ ὀνεί-. Not a [lat:]: Allatius is smoothing syntax, not asserting a different fact.
• 0002 — Ῥουβίμ. Kept as printed (transliterated Rubim, not normalized to Ruben). Migne's own note (18) observes that Isidore appears to have taken Rubim seu Ruben for the priest's proper name — the plate has already adjudicated it, so no marker of ours.
• 0003 — ὁ πάντων τῶν τῆς ἐκείνου μανίας … ἐπέκεινα Θεοῦ χωρισθεὶς … ἀσώματος. The word order permits binding ἐπέκεινα either to χωρισθείς or to the genitive plural; the Latin binds it to the devil's pre-eminence (cunctis insanis ipsius sociis … supereminens). Rendered with the Latin's binding; the choice is recorded rather than silently made, per the "never let the Latin settle a reading without saying so" rule.
• 0003 — ἔμελλεν/ἐμέλησε and Χαῖρε/χαίρει. Two untranslatable figures; the second is preserved by rendering χαῖρε as Rejoice (§2).

9. Ratio and shape

Greek 1009 / 1010 / 1060 words → English ≈ 1.5× on each chunk, tracking the anchor's 1.49× homily ratio rather than the chronicle 1.25×.

Where the sermon ends. Chunk 0003's tail is the end of Sermon I: the doxology … ᾗ πρέπει δόξα εἰς ἀπεράντους αἰῶνας τῶν αἰώνων. Ἀμήν. falls inside column 40, and the plate confirms that ΛΟΓΟΣ Βʹ. Εἰς τὴν εἰς τὰ Ἅγια τῶν ἁγίων εἴσοδον … (Sermon II) begins in the same column 40, immediately below. So the work ends mid-column 40, with Sermon II starting on the same column — which is why chunk 0003 carries a [0040] anchor and stops without a further one. The three remaining sermons are not part of this pilot; when they are taken up, the epithet table above and the anchor's must travel with them, since Glabas reuses the whole vocabulary across the set.