Liber Josue Ben Nun

Translated 2026-08-23, one translator, all 6 chunks (5,659 Latin words → 7,887 English, 1.39×). Repaired 2026-08-23 after four plate reads and an independent blind read (§6, §7). node scripts/verify-english.mjs 8958: clean — columns, notes and sections aligned, no duplicate paragraphs. Apparatus: 16 [var:] · 0 [sic:] · 1 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 144 [n:] (exact parity with the Latin twins).

⭐⭐ THE HEADLINE FINDING: all four [sic:] markers in this work have been WITHDRAWN, because the plate went against OUR TRANSCRIPTION and not against Migne. The translator marked four defects at 0510C, 0511D and 0514B (twice). Every one of them was read at the plate, and Migne prints the sound word in all four places: interpretatur, cum, per-|verse, mortales. The defects were introduced downstream of Migne — in the Corpus Corporum digitization we chunk from — and are now repaired in data/tei-patches/8958.json; the work has been re-chunked and the four markers deleted from the English. Four for four against our own text is the useful number here, and a reader of this apparatus should meet it before anything else: a [sic:] in this corpus is a claim about Migne's type, and on this work it would have been wrong every time it was made. Structure: PRAEFATIO + CAPUT PRIMUM–CAPUT XXIV. Zero guillemets in either language. Column anchors 57, 1:1 and position-checked.

The Glossa conventions are those settled on Liber Ruth (8968) §1 and are not re-opened here. What follows is this book's own collation and its own trouble.

§0 · WHAT IS SPECIFIC TO THIS BOOK, AND HOW IT WAS HANDLED

(a) Jesus and Josue are kept apart exactly as Migne prints them

This is the single most consequential decision in the work, and it is not a style choice. The commentary is built almost entirely out of Origen/Adamantius, whose whole argument turns on the son of Nave and the Son of God bearing one name. Migne prints Jesus in the gloss prose and Josue in the Vulgate lemmata, and the two are never harmonized here:

• Jesus → "Jesus", everywhere, including where the referent is plainly the son of Nave (Mittuntur a Jesu exploratores in Jericho; In Jericho est Jesus; fecit ergo Jesus sicut praeceperat ei Dominus). Rendering these "Joshua" would destroy the argument on every page — 0505D says in as many words that God gave the name above every name to Jesus Christ, hoc autem est Jesus, and that no man before was ever so named.
• Josue → "Josue", in the lemmata that print it (Praecepitque Josue, Cecidit Josue pronus, Ascenditque Josue) and at 0516B where the gloss itself writes Josue sorte dividit.

A reader therefore sees the same alternation Migne set. ⚠ Both words are rendered "Jesus"/"Josue" without comment where the referent is ambiguous — the ambiguity is the author's instrument.

(b) usque ad → as far as, in its own italic span — 8970 §6(b), reused not re-derived

97 occurrences, by a wide margin the heaviest in the Glossa block so far. Not one was varied, and no verb was supplied to any of them. Where the resumption lands mid-clause, the English lands mid-clause. The tails that read worst are the formula working, not defects:

• 0512D …etc. as far as thou hear not the words of him that calls thee back to thy former liberty, thou become the servant of sin for ever — the head of the conditional (si…) is what the abridgment removed.
• 0512D …as far as if he had not done what he had sworn when troubled by anger, than if he had performed it — a stranded quam whose comparative was cut.
• 0520A …It is asked, since Israel not only unto the day of the death of Jesus, etc. — a non solum whose answering sed was cut. The negative rides on usque ad diem mortis Jesu, which is the phrase it stands before; nothing was moved onto a verb, because the abridgment left no verb there to move it onto.
• 0513D …as far as which is Christ, and, at his own bidding, to be slaughtered — a feminine plural gerundive whose antecedent was cut. No subject and no verb supplied.

⚑ Each of these was parsed as ordinary prose before it was accepted, per the brief's rule. The one place a tail looked odd for a different reason is 0511D, and it is logged in §2 rather than excused as abridgment.

(c) The attribution sigla vary and are passed through verbatim

ADAM. (44) · ADAMANT. (5) · ADAMAN. (5) · ADAMANTIUS (1) · ORIG. · ID. · AUG. · ISID. · BEDA · HIER./HIERON. · GREG. · JUSTIN. — none normalized, Ruth §1.3 and Glossa convention 4. Passing them through is not endorsing them; Migne's Josue attributions are in fact tangled (see §2 (h)). Two sigla sit inside an italic lemma span and keep their place and their italics: Moses servus (ADAMANT., hom. 2 in [0507A] Josue.) Moses famulus meus mortuus est. at 0506D (with the column anchor inside the parenthesis) and (AUG. quaest. [0513B] 14.) at 0513A, plus [n: (ORIG. ubi supra.)] Jerusalem at 0518C and Exstructo. [n: (AUG., quaest. 32 in Jos.)] Et in sacrificiis salutarium nostrorum. at 0520B.

(d) The LXX and Alia littera readings take NO [var:]

Migne names the witness himself six times — LXX: Sicut in diebus messis tritici. (0508B), LXX: Linguam, (0511C), LXX: In gemino ligno, (0512B), Alia littera: Et terra relicta est multa valde. (0515B), Juxta LXX… (0519B), LXX: Vespas. (0520C). A divergence the plate already labels is adjudicated on the page; ours would tell the reader twice. Same reasoning as Pattern 12's rule about Migne's own ( sic ).

(e) Fixed equivalences, never varied inside the work

dogmata = doctrines · anathema = anathema · pascha = the pasch · saeculum = the world where it is moral/spatial (0511D in saeculo), the age where it is temporal (0518C saeculi aeternitatem, quandiu durat hoc saeculum) · regula = rule (0511C, where the whole gloss is the pun auream regulam / LXX linguam) · tubae ductiles = beaten trumpets (0510D, where inde duc sensus puns on it) · minister = minister (0507A, so famulus/servus could not take it).

§1 · THE LEMMA DIVERGENCE TABLE — EVERY SPAN IN data/briefs/8958-lemmata.txt

Every italic span in the work, in order, checked against sources/vulgate/clementine-flat.txt at its own verse, not merely somewhere in the canon. The 97 usque ad spans and the two Et infra: / (AUG. quaest. …) spans are the abridgment formula and Migne's own apparatus, not scripture, and are not collated — they are covered by §0(b) and §0(c). What follows is the remaining population, all of it.

Legend. ✅ = agrees with the Clementine at its own verse (spelling and Migne's Moses/Moyses, coelestium/caelestium orthography set aside). ⚠ = diverges, marker fired. ○ = diverges, marker declined, reason given. ◇ = labelled LXX/alia littera, §0(d).

PRAEFATIO — 0505D–0506D

CAPUT PRIMUM–IV — 0506D–0509A

CAPUT V–VII — 0509B–0511D

CAPUT VIII–X — 0512A–0513C

CAPUT XI–XIII — 0513C–0516B

CAPUT XIV–XVII — 0516B–0519A

CAPUT XVIII–XXIV — 0519B–0520D

Totals for §1: 108 substantive spans walked. 74 ✅ · 6 ◇ · 16 ⚠ (marked) · 12 ○ (declined, reasons above).

§1a · Divergences re-raised by the blind read — where each already stands

The blind read reported the divergence list incomplete and named six sites. ⚑ Five of the six were already in the table above, each as a recorded ○ with its reason; the sixth is not a divergence at all. They are gathered here because a reader hunting a particular divergence should not have to walk 108 rows to learn it was seen, and because the objection is worth answering in the open rather than silently.

Nothing in §1 changed as a result of this item. The five ○ rows keep their declines: each was declined on the ground that the divergence changes no assertion, and re-reading them against the flat file did not disturb that. What the item is right about is findability, and this sub-table is the fix.

§2 · THIS BOOK'S OWN CRUCES

(a) 0510C · interpetratur — WITHDRAWN. The plate reads interpretatur.

The translator read Jericho luna interpetratur in our chunk, judged it a metathesis, and fired [sic: interpetratur] — correctly, on the evidence then in hand (PL 113–114 carry interpretatur in quantity and this was the only interpetr- in the work), and correctly asked for the plate, because a two-letter transposition is exactly the shape the digitization is known to produce on its own (the Nephtlali case, 8970 @0725C). The plate was read (p. 260, cols 509/510) and Migne prints interpretatur. The defect was ours. data/tei-patches/8958.json carries the repair, the work was re-chunked, and the English now reads plainly "Jericho is interpreted moon, etc." No marker.

(b) 0514B · per verse and nortales — BOTH WITHDRAWN. The plate reads perverse and mortales.

Our chunk carried …tam per verse de operibus Dei quam de peccatis hominum judicant… cum casuri dejiciuntur et nortales moriuntur, and the translator fired a [sic:] on each: per verse as a word-division defect of the dangerous class (7a⁗-b, the quarantine the size of the adverb the correlative tam … quam hangs on), nortales as a broken m in the clause's only subject. Both were read at the plate (p. 262, cols 513/514). Migne prints perverse — hyphenated across his line as per-/verse, which is precisely how the false word-division entered our text — and he prints mortales, with a sound m. Both defects were ours. Patched, re-chunked, and the English now reads "judge as perversely of the works of God as of the sins of men" and "and mortals die" with no markers.

⚑ The per-|verse case is worth keeping: a line-break hyphen swallowed in transcription looks exactly like split type, and it is the one [sic:] shape in this work that no corpus-frequency argument could have caught, because per and verse are both real words.

(c) 0511D · Quaeritur eum Dominus… — WITHDRAWN. The plate reads cum. ⭐ And this is the work's best argument for declining a conjecture.

Our chunk printed eum twice: Quaeritur eum Dominus eum qui furtum fecerat, etc. The second eum is the object of the abridged sentence (Achan) and construes; the first had no grammatical slot at all, since quaeritur is impersonal and governs nothing in the accusative. The translator carried it under Pattern 12 and marked it [sic: eum].

The plate (p. 261, cols 511/512) reads cum. Migne's line is Quaeritur cum Dominus eum qui furtum fecerat, etc., usque ad peccato suo secum consumptis mori — an ordinary causal cum, and with it the sentence construes on the first reading. Patched, re-chunked; the English now runs "It is asked, since the Lord him who had committed the theft, etc.", which is the same idiom this work already uses for Quaeritur cum at 0520A ("It is asked, since Israel not only unto the day of the death of Jesus, etc."), and no verb is supplied — the abridgment cut it, per §0(b).

⭐ THE POINT, AND IT IS THE METHODOLOGICAL FINDING OF THE WORK. The translator weighed a conjecture — Quaeritur cur — named it in this crux, and declined to fire a [cj:] on it, on the reasoning that Pattern 18's bar is a printed word that makes the sentence assert something false, whereas this sentence asserted nothing at all. That decline was right, and the conjecture it declined was wrong. cur was a good guess — every other Augustinian excerpt in the work opens Quaeritur cur / quomodo / utrum (0507B, 0509B, 0512A, 0513A, 0514B, 0520A), and the argument for it was as strong as such arguments get. The plate gives cum. Had the [cj:] fired, this apparatus would now be printing, in our own voice, a reading Migne does not have and never had. This is the case for declining rather than guessing, stated from the one direction that settles it: not that the guess was weak, but that a strong guess still lost to a plate.

(d) 0511A · si non modo blandus remissionem, modo truculentus iracundia — a case that will not pair

remissionem is accusative; the answering iracundia is ablative; there is no verb in either member. The construction the sentence's own parallel demands is an instrumental, and English cannot leave an accusative floating. Rendered "and if thou be not now bland, [offering] remission, now truculent with wrath" — the supplied word is bracketed, per corpus practice for supplied English, so the reader can see it is ours. No case was silently changed and no [sic:] was fired: remissionem is a perfectly good form, it is only unpaired.

(e) 0514C · aut certe nulla earum capta, sed earum quae in regionibus supra memoratis fuerunt — elliptical, and the ellipsis decides the sense

Two constructions are possible and they say opposite things. The one taken: nulla earum [capta est], sed earum quae… — "none of them was taken, but [only] those which were in the regions mentioned above" — i.e. Augustine restricts the scope of Jos 11:19's conclusion. The one rejected: nulla earum capta as an ablative absolute ("none of them being taken"), which is grammatically available but leaves the genitive earum quae… in the sed clause with nothing to depend on. The following sentence settles it: Enumeratae sunt enim regiones in quibus fuerunt civitates de quibus facta est ista conclusio, et omnes cepit bello — the conclusion is about the cities of the enumerated regions, and all of those he took. Nothing was supplied in the English beyond the copula.

(f) 0519C · et sunt alia quae nos, si ignoranter admittimus, decernitur nobis… — a relative with no verb

quae nos opens a relative clause that never receives a predicate; the sentence then restarts impersonally with decernitur nobis. Rendered as printed — "and there are others which, if we admit them unknowingly, there is decreed for us, I believe, and prepared by the commandment of God, some place…" — and no verb supplied. This is Migne's anacoluthon, not the abridgment's: there is no usque ad anywhere in the paragraph.

(g) 0514D–0515A · in supercilio Sodomorum quos interfecit Chodorlaomor — an ambiguity kept, and named

quos is masculine plural and can attach either to Sodomorum or to the gigantes two words earlier. Gn 14:5 settles the fact — Chodorlahomor smote the Raphaim in Astarothcarnaim, the giants, not the Sodomites — but Latin and English are ambiguous in exactly the same way, so the English keeps the ambiguity ("upon the brow of the Sodomites whom Chodorlaomor slew") and nothing is resolved in the running text. Recorded so a reader knows it was seen.

⚠ Re-raised by the blind read (item (e)) and re-affirmed, with one correction to the record. The objection was that English word order does force a choice the Latin leaves open, because "the Sodomites whom Chodorlaomor slew" binds the relative to the nearest noun, whereas Gn 14:5 has him smiting the giants. That is true, and it is the reason this entry exists; but the remedy — rewriting to bind quos to gigantum — would force the other choice just as hard, and in the author's voice. English cannot host this ambiguity, so the only honest options are to pick one and say so, or to pick one and say so. The picked reading is Jerome's word order; the fact that Genesis is against it is stated here, which is where a reader can act on it. No edit.

(h) Migne's own apparatus is not tidied

• 0513C attributes the CAPUT XI opening to ADAMANT., hom. 24 in Jos. and then the very next gloss to ADAMAN., hom. 14 in Jos. — the homily numbers run backwards across two adjacent paragraphs, and hom. 24 turns up again at 0519B where it belongs. Reproduced as printed; never reordered.
• 0518B attributes Jos 15:63 to ADAM., hom. 11 in Jos. in the middle of the hom. 20 sequence.
• 0519A prints [n: (ADAM., nom. 22.)] for hom. — inside an [n:], which passes through verbatim by the locked register. No [sic:] is stacked on a note.
• 0520B prints [n: (AUG., quaest 27.)], the point dropped. Same treatment.
• 0509B, 0509B, 0520A print in Job. for in Jos. three times (AUG., quaest. 6. in Job.; ISID. in Job.; ADAMAN., hom. 26 in Job.). Verbatim.
• 0511A cites Gal 6:14 as (Gal. V); 0512C cites 1 Cor 2:12 as (II Cor. II); 0509D cites 1 Cor 5:7–8 as (I Cr. XVI). All rendered as printed. Citation repair belongs to data/citation-corrections.json at index time — the standing rule, and this book gives three clean instances of it.

(i) Sentences Migne leaves without a terminal point — four, all preserved

0509A (…ut multas perfectiones ad unum diem pendere significent), 0509D (…opprobrium Aegypti a vobis), 0510C (Sitque civitas), 0519A (…quae est figura coelestis), 0520B (Solus enim Salvator est ipse). Pattern 8: never supply. The English stops where he stops.

(k) 0507A · Et factus est Christi minister circumcisionis — a genitive, rendered as a genitive, with the conjecture beside it. The work's only [cj:]

Migne prints Et factus est Christi minister circumcisionis ad confirmandas promissiones Patrum, and the subject carried down from the preceding sentence is Jesus Filius Dei (Vide quomodo minister fuerit Mosi Jesus Filius Dei). Christi is genitive. The English read "And Christ was made a minister of the circumcision" — which is smooth, and is the received English of Rom 15:8, and is only available by silently emending the printed genitive to a nominative. It is also inconsistent within its own sentence: nine words earlier the identical minister Mosi is correctly "the minister of Moses".

Repaired 2026-08-23 (blind-read item 1). Pattern 7 settles the text: the printed genitive is rendered as a genitive — "And he was made Christ's minister of the circumcision, to confirm the promises of the Fathers."

⚑ And a [cj:] was fired, which is the judgment call this crux exists to record. Pattern 18's bar is that the printed word is a real form and the faithful English then asserts something the author did not — and both halves are met here as cleanly as anywhere in the corpus. Christi is a perfectly good genitive, so Pattern 12 has nothing to wrap; and the faithful English makes the Son his own minister, which is not what Origen is saying and not what the verse he is quoting says. Rom 15:8 prints Dico enim Christum Jesum ministrum fuisse circumcisionis… ad confirmandas promissiones patrum — verbatim in its second half with Migne's line, which is the warrant that Christus (or Christum) is what the sentence was built out of. The English therefore reads:

And he was made Christ's [cj: Christi; read Christus, "Christ was made a minister"] minister of the circumcision

The gloss is additive: Migne's word keeps its English, our conjecture stands beside it, and verify-english confirms Christi appears verbatim in the Latin twin — the check that makes it impossible to gloss a word the plate does not carry. Contrast §2(c), where a [cj:] was declined and the decline was vindicated by the plate. The difference is not confidence: it is that at 0511D the printed word asserted nothing, while here it asserts something false about the Son.

(l) 0515D · Sicut umbrae exemplari deserviunt coelestium — two parses, one taken, the rival named

Migne prints Sicut umbrae exemplari deserviunt coelestium, etc. Two readings are available and the English took one silently. It is named here because the crux must carry the rival.

• Taken: umbrae nominative plural, subject; exemplari dative under deservire; coelestium genitive under exemplari. → "As shadows serve for a pattern of heavenly things" — a general simile.
• The rival: Hbr 8:5 reads qui exemplari, et umbrae deserviunt caelestium, where exemplari and umbrae are both datives under deservire — "who serve the pattern and the shadow of heavenly things." Migne's line is that clause with et dropped and the two words swapped, which is exactly what an abridger's eye produces.

Why the rival lost, and it lost narrowly. Under the dative reading the sentence has no subject at all — Migne's Sicut clause would need one supplied, and §0(b)'s rule against supplying into an abridged span is the strictest rule in this work. Under the nominative reading every word has a slot as printed. What the reading costs is real and is recorded: it dissolves a verbatim scriptural allusion into a general simile, and a reader who knows Hbr 8:5 will hear the loss. No marker: [var:] fires on a divergence in a quoted scripture, and Migne does not set this out as a lemma — he writes it into his own sentence.

(m) 0512B · pro palma victoriae et virtutis merito coronari — an ablative that can go two ways

Taken: pro palma victoriae as a prepositional phrase, virtutis merito as a bare instrumental ablative coordinated with it → "to be crowned for the palm of victory and by the merit of virtue." The rival, which is the cleaner Latin: victoriae et virtutis as twin genitives both under palma, and merito as the common adverb → "deservedly to be crowned for the palm of victory and virtue." Named and not adopted: it is a resumption tail (usque ad), the choice changes no assertion the gloss makes, and the reading taken keeps every word in the case Migne prints it in. A reader collating the tail should know the sentence does not decide between them.

(j) 0517B · ab Africa beside Africum — rendered literally, both

Hi sunt fines eorum ab Africa … in occidentis partibus Africum Aegyptum et occasum nominat. The second is the south-west wind used as a quarter of the compass (Africus); the first is the ablative of Africa, the country. Reading ab Africo for ab Africa would make the two agree and is very likely right — and it is exactly the silent repair Pattern 7 forbids. Rendered "from Africa" and "Africus" respectively, unharmonized, and logged here.

§3 · CANDIDATES REJECTED, WITH THE REASON, AND WHICH READING LOST

• Et fecerunt. → "And they made." — rejected. A transitive verb with no object is not usable English, and convention 3 governs a clipped lemma. The expansion takes pascha from Migne's own next sentence (fecerunt filii Israel pascha), not the Clementine's Phase at Jos 5:10 — which is the whole point of the Ruth ruling that an expansion supplies Migne's printed verse and not a foreign one. Both occurrences (0509D, 0510A) render identically.
• Regulamque. → "And a golden rule." — rejected. Regulamque is a complete word with nothing elided; expanding it from auream would manufacture the inconsistency the brief warns about against the genuinely clipped lemmata beside it (Muri illico., Omnis autem.). "And a rule." stands, and rule is compulsory because the gloss is the pun.
• Quia ipse Deus est haereditas eorum. → [var:] — rejected, and this was the closest call in §1. The Vulgate's possessio and Migne's haereditas are near-synonyms; the sense the gloss builds (horum haereditas ipse Dominus est) is the same under either. A marker would have fired on vocabulary rather than on assertion, which is Pattern 14's own boundary.
• Novissimus inimicus forced to show the masculine in English ("the last enemy, he shall be destroyed, death") — rejected. That is not English, and it would have looked like our error rather than his reading. The [var:] is what carries the gender, and it is why the marker fired on a divergence English cannot show. ⚠ But the received-text cadence was ALSO rejected, on the second pass: "The last enemy that shall be destroyed is death" supplies a relative and a copula Migne does not print and reverses the gloss's own claim. The rendering is the literal one the work already used at 0515B — see the 0512B row in §1.
• Ne verearis a facie eorum rendered with a familiar "Fear not" cadence — kept, but the cadence is the literal rendering of Ne verearis, not a conformation; the divergence from Ne timeas eos is marked, so the reader is not left with the familiar verse only.
• et ab oriente mare salsum → a second [var:] — rejected. It belongs to the same quotation as the ad occidentem marker eleven words earlier, and Pattern 14 annotates a divergence, not every word of it.
• conteretur Satanas → declined because it is unitalicized — rejected. Migne attaches [n: (Rom. XVI)] to it, which makes it a scripture citation on the face of the page; italics are a typographic accident of the resumption tail. Marker fired.
• usque ad tails given a supplied verb — rejected everywhere, four sites named in §0(b). The ferias defect in Judith is exactly what that temptation produces.
• A [cj:] on 0511D's eum — rejected, and the plate has since proved the decline right and the declined conjecture (cur) wrong: Migne reads cum. §2(c). This is the entry to cite the next time a conjecture looks unarguable.
• A [cj:] on 0507A's Christi — FIRED, and it is the only [cj:] in the work. §2(k).
• The dative reading of 0515D's umbrae — weighed and not adopted; §2(l) carries the rival.
• Re-parsing 0512B's merito as an adverb — weighed and not adopted; §2(m).
• A [sic:] on ab Africa — rejected; §2(j). It is a real word in a real form.
• Migne's homily numbers reordered to run forwards — rejected. §2(h); his ordering is his.

§4 · SECTIONS READ CLEAN, BY COLUMN BAND

Read a second time against the Latin with the counts already known clean, asking only does it assert what he asserts (the brief's fourth test). Nothing found in:

• 0505D–0506D the whole PRAEFATIO (Jerome's headnote and Adamantius hom. 1).
• 0507A–0507C CAPUT PRIMUM and CAPUT II entire.
• 0508A–0509A CAPUT III and CAPUT IV entire, including the seven etymologies at 0508B and the two direct questions at 0508C, whose marks Migne prints and the English keeps.
• 0509B–0509C CAPUT V through VERS. 8.
• 0510A–0510B the Isidore and Justin paragraphs.
• 0510C–0511B CAPUT VI entire, including the long second-person chain at 0510D–0511A with its five negatives (each named to its host: separes, the aliquando…aliquando pair, the unpaired blandus/truculentus member of §2(d), concupiscat).
• 0511C the lingua aurea gloss.
• 0512A–0512D CAPUT VIII and CAPUT IX entire.
• 0513A–0513C CAPUT X entire.
• 0513D–0514A CAPUT XI through VERS. 12.
• 0514C–0515A CAPUT XI's close and CAPUT XII entire (Jerome's gazetteer).
• 0515B–0516B CAPUT XIII entire.
• 0516C–0517A CAPUT XIV entire.
• 0517A–0518A CAPUT XV's long boundary excursus — the densest prose in the work, read twice.
• 0518B–0518D CAPUT XV's close and CAPUT XVI entire.
• 0519A CAPUT XVII entire.
• 0519B–0520D CAPUT XVIII through CAPUT XXIV entire.

The five tests, as run

1. [sic:] swallowing — all four markers stripped and the remainder read aloud. Each leaves a clause with its verb, its object and its force: interpreted, perversely, mortals, and at 0511D a clause whose object (eum qui furtum fecerat) is outside the quarantine. The fourth question — does the clause still have the thing it acts on — was asked at every site. ⭐ This test is now retired for this work, because all four markers are gone: the plate read interpretatur, cum, perverse, mortales, and the English says exactly those things with nothing quarantined. §5. ⚑ Worth noticing that the test passed at all four sites — the swallow test proves a marker is safely placed, not that it is deserved, and nothing short of the plate could have told the difference.
2. Negatives named to their host — every non · nec · neque · nisi · ne · nullus · nihil · nemo · nunquam in all six chunks was located and its single host word named. Nine sites fired 2a's trigger condition (a negative with a quantifier or correlative adjacent) and each was proved from the sentence's own structure, not from what read well: 0507A non solum … sed (→ haeres), 0509C non filium Nave … sed Jesum (→ filium Nave), 0512A non tam auctor … quam minister (→ auctor), 0513C hodie omnes illos non possumus opprimere (→ scope over omnes, proved by the answering in crastino perimentur), 0514B nullum vivum … dimisit, 0514B nec … nec … omnes omnino … capere potuerint (→ scope over omnes omnino, proved by the two answers Augustine then gives), 0514C Non sub Jesu … sed sub Jesu Christo (→ sub Jesu filio Nave), 0518C sed non omnes simul ejici possunt (→ omnes), 0519C non omnes homicidae … sed qui ignoranter (→ omnes homicidae, proved by the sed), 0520C non omnia quae facta sunt esse scripta (→ omnia).
3. Column anchors — 57 anchors checked against the word each stands beside in the Latin, not merely for presence and order. Two are deliberately split mid-construction to match Migne: usque [0511C] ad → as far [0511C] as, and Dominus [0516C] pars haereditatis, → The Lord [0516C] is the portion of the inheritance,. Two more sit inside a parenthesis or an italic span and keep their place there (0507A, 0513B).
4. Punctuation compared band by band, never by total, using the anchors as the partition: colons, semicolons, question marks and guillemets, across all 63 bands of the work. Zero mismatched bands, with three accounted exceptions, all of them Pattern 8b: 0509B twice (Quaeritur cur dixerit… ? → "It is asked: why did he say…?" and Dicant Judaei quomodo… ? → "Let the Jews say: how can any man…?") and 0514B once (Quaeritur quomodo hoc verum sit… ? → "It is asked: how is this true…?"). Each adds exactly one colon and preserves the question mark Migne prints, which is what 8b exists to do. Question-mark parity is exact: 2·2·2·1·1·0 across the six chunks, matching the Latin twins.
5. The fourth test, run last — see §4's band list. The passages re-read hardest were 0511A (the negative chain), 0514C (§2(e)), 0517A–0518A (the boundary excursus, where every direction of the compass had to be checked against the sentence that comments on it), and 0512B (where the English cannot show the gender the argument turns on, which is why the marker is there).

The unexpressed subject — every supplied pronoun named

• 0510B Aliquando etiam Magistrum militiae se dicit → "he calls himself" = the rational power begotten out of God, the subject of the whole Justin paragraph, not God the Father.
• 0510B utrum angelo se prostraverit… → "he" = Josue, from the lemma Cecidit Josue pronus immediately above.
• 0513D …et ipso jubente jugulandae → "at his own bidding" = Christ, named in the same clause (quae est Christus).
• 0518D Subjecit Chananaeos filiis Ephrem → "He subjected" = God*, the agent of the whole hom. 21 excerpt; the alternative (Ephraim, or Josue) is excluded because the verse being glossed says the Chanaanite was not* subdued by Ephraim.
• 0520B …fit ergo unus grex… → no subject supplied; the impersonal is kept impersonal.
• 0516A et omnibus praeclaris virtutibus excolunt → object elided; supplied as bracketed "[it]", referring to mentem suam four words earlier.
• 0517D sanctorum tantum efficitur → predicate elided; supplied as bracketed "[the city]", from civitas Dei est in the next clause.

§5 · PLATE READS — SIX REQUESTED, SIX RUN, ALL SIX RESOLVED

PDF page = (column + 11) / 2, raw/scans/pl113/patrologiaecurs04migngoog.pdf (verified for this volume against the printed corner numbers).

⭐ Four patches out of seven reads, and every patch went against OUR TRANSCRIPTION rather than against Migne. No [sic:] survives in this work. See the headline note at the top of this file.

⚑ The two checked negatives are recorded because a negative plate read is evidence too, and because in both cases the argument for a defect was good and the plate said otherwise — which is the same lesson as 0511D from the other side.

Plate-notes status for this work remains "spot", checked zero (p. 258, the Judges/Josue division, which falls mid-page, and p. 262). The four patch reads and the three checks above were targeted line reads, not a foot-of-page sweep, and do not upgrade the coverage. There are therefore no [cn:] markers in this work and none were expected. ⚑ That is a checked zero on two pages, not a claim that the work has no foot-of-page notes anywhere.

§6 · CHECKED NEGATIVES — arguments for a defect that the plate refused

Three sites were argued for repair, the argument was good in each, and the plate said otherwise. They are recorded because a negative read is evidence, and because an unrecorded negative gets re-argued.

(a) 0513B · credulitatem — the plate prints creduli-|tatem. No change.

Jesus interficit inimicos non credulitatem docens → "Jesus slays the enemies, not teaching credulity." The objection was that crudelitatem would read more naturally in a work whose own crudelitas/crudelem sits two columns away at 0514B (Non putanda est crudelitas… ipsum Deum fuisse crudelem), and that credulitas/crudelitas is a classic single-transposition confusion. ⭐ The objection also said, correctly, that corpus frequency could not settle it and that the plate was required. That was the right instinct and it is the reason this entry is a clean negative rather than a silent emendation: the plate (p. 261) prints creduli-|tatem, hyphenated across Migne's line, exactly as our file has it. The English stands.

(b) 0514C · aut certe nulla earum capta — the plate prints exactly our text. No change.

The objection was that nulla earum capta ("none of them was taken") contradicts et omnes cepit bello four lines on. The plate (p. 262) prints our text without variation. The apparent contradiction is Migne's own ellipsis, and §2(e) already resolves it: nulla earum [capta est], sed earum quae in regionibus supra memoratis fuerunt restricts the scope of Jos 11:19's conclusion, and the omnes four lines on is of the cities within that restricted scope. The sentence is elliptical, not corrupt.

(c) 0513B · tradidit vobis omnes inimicos in manus — the plate carries no vestras. ⚠ This one WAS a defect, and it was ours.

The read was requested to test the opposite hypothesis — that Migne printed in manus vestras and we had dropped it. He did not: the plate (p. 261) reads in manus, bare. Which means the English's two possessives ("all your enemies into your hands") were not recovering a dropped word but reaching for the Clementine's in manus vestras at Jos 10:19 — a 7a″ conformation. Repaired; see the 0513A row in §1. A negative read that convicts our English instead of our Latin is the most useful shape a plate read has.

§7 · THE BLIND READ — what an independent reader found, and what was done

An independent reader who had not seen this file read the six chunks against the Latin twins. Six findings and five raised items. Six sites were repaired, three were affirmed as they stood with the rival named, one raised item was rejected as a non-divergence, and one was refuted by the plate before it could be applied.

Repaired

Affirmed, with the rival now on the record

• (6) 0515D umbrae — both parses are real; the nominative reading was kept because the dative reading has no subject and §0(b) forbids supplying one into an abridged span. The rival, and what the choice costs (a verbatim Hbr 8:5 allusion dissolved into a simile), is now named. §2(l)
• (e) 0512B merito — the twin-genitive/adverbial parse is the cleaner Latin and is named; not adopted, because it changes no assertion and the reading taken keeps every word in its printed case. §2(m)
• (e) 0514D quos — re-affirmed. English cannot host the ambiguity; the reading taken is Jerome's word order, and Gn 14:5's evidence for the giants is stated where a reader can act on it. §2(g)

Rejected

• (d) Jos 14:2 "dividentes for dividens" at 0516C — there is no such divergence. The Clementine reads dividentes, Migne reads dividentes, and the span is at 0516B. §1a
• (d) "the divergence list is incomplete" — the other five sites named were already in §1, each as a recorded ○ with its reason. The complaint that stands is findability, and §1a is the answer to it.

Affirmed by the blind read and left alone

• Jesus / Josue preserved 1:1 throughout, and the typological argument with it. §0(a). Not harmonized in either direction, and not to be.
• All 57 column anchors present, in order, and checked against the word each stands beside.
• Punctuation at band-by-band parity, with three +1 colons all licensed by Pattern 8b.
• Every printed negative located on its correct host, including confugiunt non omnes homicidae, sed qui… at 0519C.
• Nm 22:4's Old-Latin ablinget … vitulus rendered "the calf licks up" against the Vulgate's bos.