Epistola II ad Timotheum

Translated 2026-08-18, one translator, 3 chunks (2,148 Latin words → 3,240 English, 1.51×; per chunk 1.57 / 1.46 / 1.54). verify-english.mjs clean.

Apparatus: 3 [var:] · 2 [sic:] · 2 [cj:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 4 [n:] (exact parity with the Latin twin: 1 + 1 + 2). 24 column anchors (6 + 11 + 7), 5 heads, 0 guillemets, 2 question marks EN vs 2 LA, 0 em-dashes.

Conventions are the Ruth pilot's (src/english/8968/cruces.md §1) and the NT agent brief's, and are not re-opened: VERS. n.-- passes verbatim, the lemma is Englished, a fragmentary lemma is rendered as the phrase it stands for, usque ad → as far as (8978 §205), the attribution sigla ride as ordinary [n: …] with contents untranslated.

§0 · PLATE STATUS — foot-of-page apparatus is a CHECKED ZERO

Read this session by the orchestrator, merged here. Method: archive.org patrologiaecurs06saingoog, leaf n = (column + 9) / 2, verified against the running head (n321 = cols 633/634, head GLOSSA ORDINARIA. — EPIST. II AD TIM.). Leaves n321, n322, n323 read at full page width across the whole foot.

No foot rule, no numbered notes: the columns simply end. notes_recovered: 0, tier 3, coverage full. Two of the three leaves carry a mid-page work division and were read there — n321 (1 Timothy ends; EPISTOLA II AD TIMOTHEUM opens below the rule, where this work begins) and n323 (this work ends; EPIST. AD TIT. begins). That mid-page case is what lost three notes on PL 202 and it is clean here. ⚑ n323 was already read last batch for Titus and returned the same: a second independent read agreeing with the first.

Consequence for the apparatus below: both [cj:] (0634C–D, 0637A) are ours, not conjectures Migne had already answered at the foot of his own page.

Two incidental corroborations from the same plate read, used but not substituted for reading: 0634D Sed verbum Dei non est alligatum. is confirmed on the plate (n321, right column) together with the gloss that repeats the negation (Licet ergo sim alligatus corpore, tamen verbum Dei non est alligatum) — both negations stand in the English; and 0636A Servum autem Domini non oportet litigare, sed mansuetum esse ad omnes. is confirmed (n322, right column) with the negation as printed.

§1 · THE LEMMA DIVERGENCE LIST — ALL 79 ITALIC SPANS CHECKED (7a″)

The dispatched list named 76 spans; the Latin actually prints 79 (22 + 38 + 19). The three the list did not carry are usque ad ×2 (0638B, both Migne's abridgement formula, not lemmata) and Misericordiam consecutus sum, at 0637D — a real scripture lemma, checked below. Recorded so the next reader knows the gap was found, not inherited.

Every span was collated word by word against the Clementine Vulgate. Agreements are listed too, so this cannot be told apart from a report by an agent that never looked.

Chapter I (chunk 0)

Chapter II (chunk 1)

Chapter III (chunk 1)

Chapter IV (chunk 2)

The three [var: …] markers

0633B · accipiens etiam recordationem fidei — 2 Tim I, 5 reads recordationem accipiens ejus fidei. Migne inverts the first two words and, more materially, prints etiam where the Vulgate has ejus. The change is load-bearing for the gloss, which closes cum etiam in feminis fuerit firma — the gloss has picked up the etiam the lemma introduced, and lost the ejus that in the Vulgate points at Timothy's own faith. Marker: [var: Vulg. recordationem accipiens ejus fidei, "receiving the recollection of that faith"].

0633D · Formam habens — 2 Tim I, 13 reads Formam habe, an imperative addressed to Timothy. Migne prints the participle at both occurrences (0633D and the VERS. 13 lemma at 0634B) and his gloss builds on the participle: similiter ergo et tu formam habens, etc. So this is his text throughout, not a slip he ignores. Rendered "having a form" both times. Marker fired once, at the first occurrence, per Pattern 14's "at the point of divergence"; a second copy on the same words two paragraphs later would be noise.

0637B · Increpa in omnibus — 2 Tim IV, 2 reads increpa in omni patientia, et doctrina. This is the work's clearest 7a″ case and it runs the mechanism backwards: the gloss's whole sentence is Timotheo acriorem suadet patientiam, "to Timothy he urges a sharper patience" — the exposition rests on a word the lemma above no longer prints. A translator conforming the lemma to the familiar English ("rebuke in all patience") would have made the page read perfectly and destroyed the evidence. Rendered "Rebuke in all things." with [var: Vulg. increpa in omni patientia, "rebuke in all patience"; the gloss turns on patientia].

§2 · THE DIVERGENCES DECLINED, each with its reason

Pattern 14 fires on divergence "in a way that changes the sense." These do not, and marking them would bury the three that do.

1. 0633C Eum timete qui potest corpus et animam perdere in gehennam (Matth. X, 28, potius timete eum qui potest et animam et corpus perdere). Two changes: the opening is recast for the Unde: frame, and the doublet body/soul is reversed. Neither changes what the verse asserts, and the gloss turns on who is to be feared, not on the order of the pair (Spiritus … dat timorem non mundi, sed Dei). ⚠ Recorded rather than waved through, because 7914 @1079B is a case where a reversed doublet mattered; there the gloss was about the second member (cinericeum colorem). Here nothing hangs on it.
2. 0634C Quia oportet, etc. — quia is not in II, 6. It is the citation-frame conjunction at the head of a fragment lifted out of its sentence, exactly the class Titus declined for a dropped autem / -que (8997 §1).
3. 0634C Primum percipere. / De fructibus. — every word agrees; Migne simply glosses the two halves in the reverse of the printed order (II, 6 runs primum de fructibus percipere). That is his lemmatization, not a textual variant. Reproduced in his order, never reordered.
4. 0635B / 0635C, the resurrectio lemma split. The VERS. 18 lemma prints Dicentes resurrectionem jam factam (Vulgate: esse jam factam, so esse is dropped); the re-quotation five lines later prints Dicentes resurrectionem, … esse factam — which restores esse and drops jam. Neither omission changes the assertion (the elliptical accusative-and-infinitive is ordinary, and jam survives in the first form). ⚑ The inconsistency between the two forms is the interesting datum, and it is exactly the phenomenon the brief warns not to harmonize: the English renders the lemma "is already past" and the re-quotation "hath been made," because that is what each prints.
5. 0636A Nequando det illis Deus poenitentiam — II, 25 has Deus det illis. Word order only, and Migne is self-consistent: he re-quotes it the same way at 0636B. English renders either identically.
6. 0637A praepara animam tuam ad tentationes (Eccli. II, 1: ad tentationem). Singular to plural. Declined because Migne's own gloss two sentences earlier already reads qui non praeparant ad tentationes — the plural is his throughout, and the verse enjoins the same thing in either number.
7. 0638A ut fidelis essem (I Cor. VII, 25: ut sim fidelis). This is not a quotation presented as one: it sits inside Augustine's own antithesis non quia fidelis eram, sed ut fidelis essem, where the imperfect subjunctive is required by the sequence of eram. Recast, not divergent.
8. 0638B Multa mihi mala ostendit (IV, 14: multa mala mihi ostendit). mihi moved one word. Nothing turns on it; the gloss is about Alexander, not about the phrase.
9. Non-lemma scripture, checked and left alone (not italic, therefore outside Pattern 14 by the runbook's own boundary, but collated anyway): 0634C fodere non possim, mendicare confundar against Luc. XVI, 3 fodere non valeo, mendicare erubesco — a paraphrase put in Timothy's mouth, not a citation; 0636C charitas de corde puro, conscientia bona, et fide non ficta against I Tim. I, 5, which has et before conscientia; 0635C–D conformes imaginis Filii sui … praedestinavit, vocavit, justificavit, glorificavit against Rom. VIII, 29–30, which has conformes fieri imaginis. All three are the gloss quoting loosely in its own prose. No [var:] on any of them, deliberately — the marker is for lemmata and for scripture quoted as scripture.

§3 · THE UNEXPRESSED SUBJECT — EVERY SUPPLIED PRONOUN, NAMED

The brief's dominant defect class, and this work is dense with it: Paul, Timothy, Christ, the false teachers and the faithful hearer all move through one paragraph. Every "he / they / it" the English supplies is listed with what fixed it.

0633A, the ARGUMENTUM · scribit … scribens ei — "He writes to Timothy." Paul. Fixed by the closing participle scribens ei ab urbe Roma, which only the letter's author can govern, and by the form (the standard Glossa prologue to a Pauline epistle).

0633A · gratias agit de bono quod habet — ⚠ two different subjects in one clause. agit = Paul; habet = Timothy. Fixed by the VERS. 3 gloss below, which repeats it with the possessor spelled out: Agit gratias de bono ejus.

0633A · ubi suum videndi eum desiderium ostendit — "his own desire of seeing him." ⚑ The reflexive/non-reflexive pair decides it and nothing else does: suum is reflexive, so the desire is the subject's — Paul's; eum is non-reflexive, so the one seen is not the subject — Timothy. (2 Tim I, 4 desiderans te videre.) This is the exact class that produced last batch's Philemon failure, and here the morphology is unambiguous.

0633B · Agit gratias de bono ejus — ⭐ REPAIRED after the blind read. It first shipped as "for his good," which names the class in this file and then loses it on the page: English "his" defaults to the subject of "he gives thanks," i.e. Paul. ejus, not suo, is Migne's printed instruction that the possessor is NOT the subject — he spent a word saying so. Now reads "for his [Timothy's] good." Aggravating the original: the preceding sentence at 0633A (gratias agit de bono quod habet) also carries an ambiguous "he," so nothing anywhere cued the reader. ⚑ Same ejus/sua class that bit Philemon last batch.

0633B · Quod ergo praesens habuit, absens teneat … quae te commoneat — habuit and teneat = Timothy (the sentence's subject is Timothy, named in the previous clause, Flevit enim Timotheus). ⚠ Migne then shifts to the second person mid-sentence (te commoneat) and the English shifts with him; see §5.

0633B · cum etiam in feminis fuerit firma — "since it was firm even in women." The subject is fides, from the lemma recordationem fidei immediately before; the women are Lois and Eunice (2 Tim I, 5), whom the gloss does not name.

0633B · Commendata fide illius, ad eam … exhortatur — illius = Timothy's faith (non-reflexive again); exhortatur = Paul; eam = fidem.

0633C · Hujus timoris non accepimus spiritum — first person plural, expressed. No supply.

0633D–0634A · Ostendit Deus … ostendit, inquam, per praedicationem Evangelii — subject expressed (Deus). Named here because the neighbouring sentences are all Paul, and a reader skimming would carry Paul into it.

0634A · persuadere quippe vult — Paul, from the question immediately above (cum seipsum … gentium magistrum vocet), whose subject is Paul by seipsum.

0634A–B · Spe et magnificentia Salvatoris securus est … quod commendat illi … cum coeperit judicare … pronuntiet — ⚠ the densest site in the work; four verbs, two subjects, none expressed. securus est and both instances of commendat = Paul; coeperit judicare, puniens and pronuntiet = Christ, i.e. the Salvator named at the head of the sentence and the illi that Paul commends his salvation to. istum ("this man") = Paul, the one who suffers hic pro illo. The chain is fixed by the lemma's own verse (I, 12, potens est depositum meum servare in illum diem): the depositor is Paul, the keeper is Christ, the day is the judgment.

0634B · Hic ostendit quo de se praedicta spectent — ostendit = Paul, de se = Paul. Then Ita laboro … et tu is first and second person, expressed.

0634B · Simulate hi cum Apostolo fuerunt … recesserunt ab eo — hi = Phigellus and Hermogenes, expressed in the lemma; eo = the Apostle, expressed in the sentence.

0634B–C · quod ne iste faceret … prohibet — iste = Timothy (and infirmus laborare non poterat is the standing Glossa reference to I Tim V, 23, Timothy's weak stomach); prohibet = Paul. ⚠ Read the other way round the sentence says the Apostle was too weak to work, which is both false to the passage and incoherent with prohibet.

0634C · jubet eum sumere qui primus est … Primo enim debet ille suam necessitatem implere — jubet = the Apostle, expressed; eum … qui primus est and ille = the labourer in the word, the man of the simile's agricola vel vinitor, i.e. Timothy and his like. The next sentence's ex his qui accepit settles it: the one who takes first is the one who received.

0634D · Haec autem non propter Timotheum memorat — memorat = Paul; praevidit = Paul. Timothy is expressed and is the person the recollection is not for.

0634D · Hic contumeliam notat quae ei ingerebatur — ⚠ notat = Paul and ei = Paul as well, though ei is non-reflexive. The lemma decides it: quasi male operans is II, 9's description of Paul in chains, and the gloss that follows says sim alligatus corpore in the first person. Recorded because it is the one place in this work where the reflexive/non-reflexive test would give the wrong answer if applied mechanically.

0635A · Justificati estis ut hic consequamini — ⚠ plural, and it is the only plural address in the work. Rendered "you" per Pattern 17, against the singular "thou" used for Timothy everywhere else. The referent is the electi of II, 10, the lemma's own verse, not Timothy.

0635A · Est iste, quod consequentur salutem — "they shall obtain": the same electi. sumus mortui and sustinentes are first person / participial, expressed.

0635B · ut intus in animo perdat, et foris victor accedat — "he loses … he comes off victor." The subject is the contender, generalized out of contentio in the same sentence and confirmed by the next sentence's Nemo enim patitur se vinci. Not contentio itself (which is feminine and would give "she"), and not the fratres (plural).

0635B · Horum prodidit nomina, ut ab his spiritualiter caveat — ⭐ REPAIRED after the blind read. prodidit and designat = Paul; caveat = Timothy, the person addressed by Haec commone and by the whole chapter. This file named the right subject and the English still shipped a bare "that he might beware," which attaches to the nearest antecedent, the subject of prodidit — Paul. Read that way the sentence says Paul published the names so that he himself might beware, which makes the purpose clause pointless: the point of naming Hymenaeus and Philetus is that Timothy should beware. Now reads "that [Timothy] might beware of them spiritually." ⚑ The idiom is already in this work at 0633B, where absens teneat is third person for Timothy.

0635C · Volentes ergo credi sibi omnes … non potuerunt — subject expressed (omnes … qui instituerunt).

0635D · quos praescivit … praedestinavit, vocavit, justificavit, glorificavit — God, from Cognovit Dominus qui sunt ejus in the lemma; Filii sui inside the clause confirms it. See §5 for the anacoluthon.

0636A · Non deesse voluit nos fratrum correptioni — voluit = the Apostle, whose lemma (Servum autem Domini non oportet litigare) the sentence is expounding; nos is expressed and is us.

0636A–B · nisi supernus medicus respiciat, id est nisi faciat — subject expressed (supernus medicus). And respexit eum Dominus two sentences later is expressed both ways.

0636B · Unde et Dominus ei oves pascendas credidit — expressed; ei = Peter, named in the previous clause.

0636C · Ingrediebantur enim domos mulierum, et decipiebant eas — "they entered": the qui penetrant of the lemma (III, 6), the false teachers. eas = the women, expressed.

0636D · Horum nomina non invenit Apostolus — expressed. Hi autem restiterunt = Jannes and Mambres, expressed.

0636D · ita hi scilicet haeretici mihi resistunt — ⚑ mihi, first person: this is the gloss speaking in Paul's voice, and it is where the gloss departs from its own lemma, which reads ita et hi resistunt veritati (III, 8). Not a [var:] — it is the gloss's prose, not a quotation — but worth a blind reader's eye.

0636D–0637A · Tales sunt qui timent hos laedere coram quibus loquuntur qui non praeparant — two relative clauses on one antecedent (Tales, i.e. the mali homines of the lemma), and hos … coram quibus loquuntur = their hearers. No supply; the English keeps both relatives.

0637A · Ille praedicit … Ille … venit pati … Ille confortans infirmum — Ille is expressed three times and is Christ each time (the passion catalogue — spittle, thorns, gall and vinegar, the wood — admits no other referent). The contrasted tu is the false teacher of the preceding sentence, not Timothy: the antithesis is Ille … labores : tu prospera, and it is the promisers of worldly felicity who are being answered.

0637B · Tu opportune agis … Tu tamen sciens — expressed, and singular: Timothy.

0637B · Timotheo acriorem suadet patientiam. Tito … persuadet — suadet / persuadet = Paul; both addressees expressed. ⚑ The cross-reference to Titus is the gloss's own, and it is accurate to the sister epistle (Tit. II, 15 argue cum omni imperio), which is a small confirmation that the Glossa block was compiled as a block.

0638A · Qui sua jam propinquante passione bona merita sua commemorat — Paul; sua reflexive twice over, and the passion drawing near is IV, 6's tempus resolutionis meae instat.

0638A · videtur … Deus vitam aeternam … reddere cui debet fideli — expressed both ways.

0638B · Iste est quem supra dixit reversum ad apostasiam — dixit = Paul (I Tim I, 20, where Alexander is named with Hymenaeus); reversum, fabricabat, commovit = Alexander, the lemma's own subject.

0638C · Non orat pro Alexandro … orat ut eis ignoscatur — orat twice = Paul, whose prayer the lemma Non illis imputetur is.

0638C–D · liberatus est … quando venit Romam … mansit … transivit — ⚠ the Philemon trap, and it is live here. All four are Paul, not Nero and not the men who brought him: liberatus est a Domino has Paul as the one delivered (the lemma is de ore leonis, out of which Paul was delivered, IV, 17); venit Romam adductus ab his qui praeerant Judaeae, ob hoc quod Caesarem appellasset is Acts XXV, 11–12 and only Paul appealed; duobus annis mansit in libera custodia is Acts XXVIII, 30, verbatim in substance. The nearest preceding noun is Neronis, and a bare "he" attached to it would say Nero appealed to Caesar and lived two years in free custody in his own capital. Named here because word order alone would strand it, which is exactly how Philemon failed.

§4 · THE CRUCES PROPER

⭐ 0637A · quam Deus ipse saeculo nos promisit — [cj:] FIRED

Migne prints, of the teachers who promise worldly happiness:

sed promittunt felicitatem hujus saeculi, quam Deus ipse saeculo nos promisit.

Rendered as printed: "but promise the felicity of this world, which God himself promised us to the world." That is the opposite of what the paragraph argues, and the next two sentences say so on the same plate: Ille praedicit labores usque in finem venturos saeculo: tu prospera ("He foretells labours coming upon the world unto the end: thou, prosperous things"), and ut cum crediderit, non speret prospera hujus saeculi.

The conjecture: nos is the plate's non. One letter, a real word in a real slot, and the restored sentence is precisely the argument: quam Deus ipse saeculo non promisit, "which God himself did not promise to the world." This is the founding shape of Pattern 18 (11081's munde for mundo): nothing is defective type, so [sic:] cannot reach it; Pattern 7 renders it and the English then asserts what the author is refuting.

Marker: [cj: nos; read non, "which God himself did not promise to the world"], additive — Migne's word keeps its English and the conjecture stands beside it. Not swallowed into a marker, not silently repaired. §0 records that the foot of leaf n322 carries no note, so Migne had not answered this himself.

A second conjecture considered and rejected: that nos is sound and a separate non has dropped out before promisit. It yields the same sense but glosses no printed word, so Pattern 18's safety check (the marker must open with a word verbatim in the Latin twin) could not be satisfied, and it is the less economical account.

⭐ 0634A · non confundor, id est deficio — DECIDED: RENDERED AS PRINTED, AND THE ASYMMETRY IS MIGNE'S

Closed by the blind polarity read of 2026-08-18 (§9). This entry originally stood as an open flag and the reader resolved it in favour of the rendering; the hedging is withdrawn.

Migne prints:

Ideo non confundor, id est deficio, vel non frustra patior…

The English reads "Therefore I am not confounded, that is, I fail, or I do not suffer in vain." The asymmetry between the two appositives is real, and it is in the Latin, not in our English. They are not parallel glosses:

• id est deficio glosses the bare verb confundor — a one-word lexical equation, the commonest move in the Glossa — and therefore carries no negation of its own.
• vel non frustra patior glosses the whole negated phrase non confundor, which is why it has to print its own non.

Once the two are seen to be doing different jobs, nothing is missing from the plate. Rendering it "that is, [I do not] fail" would have been the silent repair — it would have supplied a negation Migne does not print (7a) and flattened a distinction his sentence actually makes. The English preserves the asymmetry exactly, and the lemma's own non stands three words away where the reader meets it.

No [cj:], no [sic:], and no residual doubt: deficio is the right word in the right slot under the reading the sentence itself supplies.

0635D · sed nutiles et sine disciplina — [sic: nutiles], WITH THE FORCE KEPT OUTSIDE

nutiles is not a Latin word: it is inutiles with its initial in- lost. Carried untranslated in italics per Patterns 10 and 12.

⚑ 7a⁗ applies at full strength, because the quarantined run contains the negation. The whole point of the sentence is that not all questions are to be shunned, only the unhelpful ones — so the English supplies the sense in plain words beside the marker: "but the unprofitable [sic: nutiles] ones and those without discipline." Wrapping nutiles and letting the English go without a negative would have made the gloss forbid profitable questions, which is the mechanism 7a⁗ was written to stop.

0636A · aurursus dormire — [sic: aurursus], A FUSED WORD-DIVISION DEFECT

Multi enim homines cum a somno excitantur, litigare volunt, aurursus dormire, cum litigare prohibentur.

aurursus is not a word. It is a Pattern 10 fusion, and the most economical account is aut rursus, "or again" (au + rursus); ac rursus is the alternative and yields the same connective force. Carried whole, in italics, inside the marker.

⚑ Again 7a⁗-b: the fused run is exactly the size of a lexeme, and the lexeme is a conjunction. Quarantining it silently would leave "wish to wrangle, to sleep, when they are forbidden to wrangle" — a list where the Latin has an alternative. The English carries the force outside the marker: "wish to wrangle, [sic: aurursus] or again to sleep." The conjecture is here; the connective is on the page.

⭐⭐ 0638C · sed pro his qui abruperant amorem, sed timore succubuerunt — DECIDED: THE TEXT CONSTRUES, AND THE CONJECTURE IS POSITIVELY WRONG

Closed by the blind polarity read of 2026-08-18 (§9). This entry originally proposed that a non had dropped before abruperant. That conjecture is wrong, and recording why is more use to the next reader than an open flag.

Migne prints:

Non orat pro Alexandro, qui invidentia fraternitatem oppugnando peccaverat, sed pro his qui abruperant amorem, sed timore succubuerunt; orat ut eis ignoscatur.

Rendered as printed: "but for those who had broken off love, but yielded through fear."

The decisive evidence is two lines below, in Migne's own text:

Multum enim interest inter eos qui hoc modo et eos qui illo modo peccant.

"There is a great difference between those who sin in this manner and those who sin in that." Both groups sin. Had the second group not broken off love at all, they would not be sinners, the comparison would have nothing to compare, and the closing sentence of the paragraph would be empty.

So the second sed is not adversative against abruperant; it is adversative against invidentia in the clause before. The contrast the paragraph is drawing is over the cause of the sin, not its existence: Alexander sinned by envy, assailing the brotherhood; these sinned too, but through fear. That is exactly why Paul prays for the second group and not the first, and why ut eis ignoscatur — pardon, which presupposes a fault — is the thing asked for.

Nothing is missing from the plate, both seds are carried, and the English as it stands is sound. The earlier reasoning — that [cj:] had no printed word to open on — reached the right result for a weaker reason; the real ground is that there is no defect to conjecture about.

⚠ Recorded as a rejected conjecture, deliberately. A later reader meeting the doubled sed will have the same instinct, and this entry exists so that he does not spend the session on it.

0638A · reddentibus bona pro malis — TWO READINGS, ONE TAKEN

Apostolum enim invenimus sine ullis bonis meritis, imo cum multis vitiis Dei gratiam consecutum, reddentibus bona pro malis.

An ablative absolute whose participle is active and whose subject is not expressed. (a) The agent is God, the sentence's own theme (Paul had vices; God gave grace: good for evil) — so the clause is reddente Deo in sense, with the plural either loose or attracted to vitiis. (b) reddentibus is a dative/ablative of persons, "to those rendering good for evil," which leaves it hanging on nothing.

(a) taken, on the argument's own logic and on what the next two sentences say (qui post mala merita consecutus est gratiam). The English leaves the agent unexpressed as the Latin does: "good things being rendered for evil." No marker: nothing is defective and no word is mistranslated, only a number is loose, which Pattern 9 exempts.

0637C · retorqueatur os turturis ad ascellas — RENDERED LITERALLY

Lev. I, 15 and V, 8 have the priest twist back the head (caput) of the turtle-dove ad collum / ad pennulas. Migne prints os (mouth) and ascellas. Both are rendered as printed: "that the mouth of the turtle-dove be twisted back to its wings."

• ascellas takes no marker: it is the well-attested medieval form of axillas (whence Italian ascella), and Pattern 9 rules attested medieval spellings correct Latin for their period. "Wings" rather than "armpits" because the subject is a bird.
• os for caput is a real word in its slot and the sentence construes; it is a divergence from Levitical wording, not from a quoted lemma, so Pattern 14 does not reach it either. ⚑ And there may be a reason it is os: the same sentence opposes opere to ore ("what thou preachest with the mouth") and closes on in extremis oris (the hem of the tunic, ora). The gloss is playing on the word, and emending it would flatten the play.

0637C · libare / delibare / delibatio — THE ETYMOLOGY KEPT IN LATIN

The gloss is the etymology (Libare dicimus, degustare, fundere et immolare. Inde delibare … passionem enim suam delibationem appellat), so the three root-words stay Latin, in italics, under the register rule for technical terms with no clean English equivalent — the shipped precedent is 9741's precaria / emphyteusis, which are likewise italicized in the English where the Latin twin has no italics. The lemma itself is Englished as required: Delibor. → "I am poured out as a libation," which keeps the root visible to a reader who does not read Latin.

Same treatment, same reason, at 0638B: Pallium hic penulam dicit → "He here calls the pallium a penula." The gloss's whole content is the identification of two garment-words; rendering both "cloak" would make it say nothing. The lemma is Englished (Penulam. → "The cloak.").

0635A · Contentio minus stabilis saepe generat scrupulum — RENDERED AS PRINTED

"Contention, being less stable, often begets scruple." minus stabilis is odd (one expects a noun for the mind, or mentis), but it is real Latin in a real slot and yields sense. No conjecture is economical enough to offer; recorded for a reader with the plate.

0638B · aedes Dianae apud Athenas — MIGNE'S OWN ERROR OF FACT, REPRODUCED

The shrine-makers of Acts XIX are at Ephesus, not Athens, and their leader is Demetrius, not Alexander. The gloss has conflated two chapters of Acts with I Tim I, 20. Rendered exactly as printed ("who used to make shrines of Diana at Athens"). This is a mistake in Migne's content, not in his type, and Pattern 7 covers it precisely: our business is to reproduce what he printed, not to correct what the Glossa believed.

0634C, 0635B, 0636A · MIGNE'S PUNCTUATION KEPT WHERE ENGLISH WOULD MEND IT

• 0636A · the semicolon inside a comparative: dum plus delectat hominem aliorum existimatio; quam justitia, qua se quisque humiliat poenitendo. English wants no mark at all before quam. Pattern 8: never delete one he prints. Kept: "…delights a man more; than justice, by which each man humbles himself in repenting."
• 0638A · the semicolon after a protasis: Si enim fides gratia est, et vita aeterna quasi merces est fidei; videtur quidem Deus… Kept.
• 0633D · the capital after a colon: Vicem reddere liberanti convenit: Cujus cum beneficiis… Kept as a capital ("…to him who delivers: And since we are not able…"), on the same reasoning Titus used in the other direction at 0637D: normalizing his case is our repair of his punctuation.
• ⭐ Two supplied "and"s REMOVED after the blind read. At 0633C Migne joins the two tamen limbs with no connective and no mark at all (gestis tamen apparuit Deus etsi crucifixus est, resurrexit tamen…); the English had supplied both a semicolon and the word "and", tidying his asyndeton into a compound sentence. At 0638A an "and" had been supplied after his colon (consecutus est gratiam: and unless…). Both words are gone; the marks stay (see the declined note below). ⚑ A connective the Latin does not print is the class that concealed Titus's sollicite — a smooth join is exactly what stops a defective page from failing to construe.
• The five added semicolons as such are DECLINED and stay (0633C, 0634A, 0635C, 0636C, 0638A). The corpus-wide semicolon measurement closed last batch as a rate question, not an absence, expressly "do not sweep it," and the blind read confirms all eight of Migne's own semicolons are matched in place with nothing softened. Fix the supplied words; leave the marks.
• Lower case after both question marks (0634A …vocet? persuadere quippe vult; 0634C …confundar? subdit) — kept lower case, so the reader sees his sentence division.

⭐ 0634C–D (VERS. 6) · ex his qui accepit — [cj: qui; read quae], AFTER A SILENT EMENDATION WAS CAUGHT

Primo enim debet ille suam necessitatem implere ex his qui accepit, et post, aliis de his quae supersunt ministrare.

This entry replaces an earlier one that got the class wrong, and the correction is instructive enough to keep in view. The first reading treated qui as a Pattern 9 gender mismatch — qui for quae, neuter plural agreeing with his — and concluded that English relatives carry no gender, so there was nothing to mark. That was a silent emendation dressed as an exemption.

The printed qui construes without changing a letter: it is a nominative attracted to ille, giving "he who has received ought first to fulfil his own need out of these things." Both readings work; the English took the one that requires reading qui as quae, and took it with no marker at all — two paragraphs from a correctly flagged [sic: nutiles] and one from a correctly flagged [cj: nos; read non]. ⛔ Consistent apparatus elsewhere in the chunk is not evidence the apparatus was used here — 7a″'s standing warning, and this is a live instance of it.

The English sense is kept (the quae reading is the likelier one, since de his quae supersunt in the very next clause shows Migne using the neuter plural relative for exactly this referent), and the conjecture is now declared additively: [cj: qui; read quae, "those things which"; the printed qui construes instead as "he who has received"].

0635C–D · qua quos praescivit … id est praedestinavit — ANACOLUTHON, RENDERED AS PRINTED

Migne compresses Rom. VIII, 29–30 so that the relative quos is left with no verb of its own before the list of four finite verbs begins. Rendered with the break visible: "by which those whom he foreknew would be conformable to the image of his Son, that is, he predestinated, called, justified, glorified." No marker; the sentence is his abridgement of the Apostle, not broken type.

0633B · THE PERSON SHIFT — teneat … te commoneat

Within one sentence Migne moves from third person about Timothy (absens teneat) to second person addressing him (quae te commoneat). The English shifts with him ("let him hold while absent, receiving also the recollection of the faith which may admonish thee"). Pattern 17: te is singular, so thou. Not a defect; the Glossa slides between speaking of the addressee and to him throughout, and smoothing it would be our voice.

§5 · MIGNE'S OWN ORDERING AND HIS NOTES — REPRODUCED, NOT TIDIED

The verse numbering, exactly as the brief anticipated. Chunk 1 runs VERS. 4, 6, 8, 9, 10, 11, 14, 17, 18, 19, 20, 23, 24, 25, then CAPUT III. restarts at VERS. 2, and chunk 2 opens CAPUT IV. at VERS. 2. Both restarts are correct — they are new chapters, and the Glossa simply does not gloss verse 1 of either. Within each chapter the sequence is monotonically increasing with gaps only (the gaps are verses he passes over). ⚑ Checked and found regular: there is no out-of-sequence number anywhere in this work, unlike Jude. Reproduced as printed regardless; recorded because the brief asked for it and a blind reader should not have to re-derive it.

Column anchors. 24 in all, running 0633A–0634B, 0634C–0637A, 0637B–0638D, unbroken and in order, one band per step with no gaps. ⚑ Per the runbook's false-positive 1, band arithmetic is not run on these; the only signal would be text broken mid-word or mid-construction across an anchor, and there is none. Two anchors fall inside a sentence (0634C in colligendae cura pecuniae [0634C] sine labore) and one falls inside an italic lemma (0637D in Misericordiam [0637D] consecutus sum) — reproduced in the same position inside the English span. Checked, deliberately not reported.

The four [n:] notes are all citations, so all four stay [n:]. Pattern 15's [nt:] does not fire: (Matth. XXVI.), (Matth. XXVI), (CHRYS.), (CHRYS.) are locators, not prose. Contents untranslated, positions unmoved, counts 1 / 1 / 2 matching the Latin.

⚑ 0633C · [n: (Matth. XXVI.)] cites the wrong chapter. The verse quoted is Nolite timere eos qui occidunt corpus, Matth. X, 28, and the verse quoted next in the same paragraph (Eum timete qui potest…) is the second half of that same verse. Matthew XXVI has neither. Passed through verbatim under the marker rule; recorded here because a reader will notice. ⚑ Note that the other (Matth. XXVI) — at 0636B, on Ut enim Petrus amare fleret, respexit eum Dominus — is right about the chapter (Peter's denial and bitter weeping, Matth. XXVI, 75), though respexit eum Dominus is Luke XXII, 61. So the two notes are not one systematic error; the first is simply wrong.

Migne's ( sic ): none in this work. Nothing stacked.

§6 · FIXED RENDERINGS HELD ACROSS THE WHOLE WORK (one voice, three chunks)

• confundor / confundar — be confounded (0634A, the lemma) and be ashamed (0634C, Timothy's mendicare confundar, where the Luke XVI, 3 sense is begging-shame). ⚠ Deliberately not unified: the two occurrences are different senses of the verb and the glosses turn on different ones. Recorded because a merge-time reconciler would otherwise "fix" it.
• contentio / collatio / altercatio / certamen (0635A–B, 0636A) — contention / conference / altercation / strife, four words kept four, because the paragraph's whole point is Collatio ergo inter Dei servos esse debet, non altercatio.
• correptio / corrigentibus / increpa — correction / those who correct / rebuke, one root for the first two (0636A, 0636B), and increpa kept separate as the lemma's own word (0637B).
• opportune / importune / opportunum / importunum (0637B) — in season / out of season, rigid through all five occurrences; the gloss is built entirely on the antithesis.
• immolare / immolatio / delibare (0637C) — immolate / immolation, and the libare family kept Latin (§4). immolatur → "is immolated," not "is sacrificed," so that delibatio has room to be the libation.
• resurrectio mentium / resurrectio corporum (0635B–C) — resurrection of minds / resurrection of bodies, never "souls" for mentium even though the same paragraph says fit in anima, which is soul. The gloss distinguishes them; so does the English.
• praescitos / praescientia / praescivit (0635C–D) — foreknown ones / foreknowledge / foreknew, one root.
• vasa aurea et argentea / lignea et fictilia (0635D) — golden and silver / wooden and earthen, rigid at all four occurrences.
• gratia (0633A, 0637C–0638B) — grace throughout, including praedicandae gratiae ("the grace that is to be preached"); gratuita → freely given, on the same root.
• merita / merces / debitum / corona (0638A) — merits / wage / debt / crown, kept apart because the paragraph is an argument about their relation.
• Second person, Pattern 17. Thou everywhere Timothy is addressed in the singular (0633A tu, 0633B accepisti, 0634B tu, 0634C vis / te / implices, 0637A tuum / te, 0637B Tu … Tu); you at the one plural, 0635A Justificati estis ut … consequamini. That single switch is the whole of the distinction and it is visible on the page.

§7 · READ CLEAN — named, so a blind reader can tell checked ground from unexamined

• The ARGUMENTUM (0633A) — one sentence, one supplied subject (§3), no lemma, no divergence, no defect.
• CAPUT I, VERS. 1–10 (0633A–0634A) — read clean apart from the two [var:] (§1), the Matth. XXVI note (§5) and the capital after the colon at 0633D (§4).
• CAPUT I, VERS. 11–13 (0634A–B) — the question at 0634A is direct and keeps its mark with no recasting; Pattern 8b does not apply. The depositum chain at 0634A–B is the work's densest pronoun site and is fully named in §3. Otherwise clean.
• CAPUT II, VERS. 4–14 (0634B–0635B) — the question at 0634C is direct (Timothy's own, behind Quasi Timotheus quaereret:) and keeps its mark; no 8b recasting. Clean otherwise.
• CAPUT II, VERS. 17–20 (0635B–0635D) — the long duae sunt resurrectiones excursus read clean; every non / nullam / nisi carried. The correlative Sicut … sic at 0635D (Sicut enim vasa lignea … sic mali prosunt) is carried whole.
• CAPUT II, VERS. 23–25 (0635D–0636B) — two [sic:] here (§4); otherwise clean.
• CAPUT III, VERS. 2–8 (0636B–0636D) — read clean. usque ad tertium signum at 0636D is ordinary prose, not the abridgement formula (it is roman, not italic) and is rendered "as far as the third sign," which happens to coincide with the formula's English; recorded so nobody reads it as a stray marker.
• CAPUT III, VERS. 13–15 (0636D–0637A) — the [cj:] site (§4); otherwise clean.
• CAPUT IV, VERS. 2–8 (0637B–0638B) — read clean apart from the [var:] at 0637B, the reddentibus ambiguity and the turtle-dove (§4).
• CAPUT IV, VERS. 13–17 (0638B–0638D) — the abruperant amorem crux (§4) and the Athens error (§5); the Nero paragraph's pronouns are named in §3. Otherwise clean.

Every correlative pair in the work, checked (the Titus lesson)

Last batch's worst defect was a dropped ita. Every correlative and quasi-correlative in these three chunks was listed and confirmed present in the English:

Quasi dicat ×8 (0633A, 0634A, 0634B, 0634C ×2, 0634D, 0635A, 0635A) and Quasi Timotheus quaereret ×1 (0634C) — all nine carried as "As if he should say / As if Timothy should ask", never dropped, never compressed into the sentence. · sicut … tamen (0633C, the concessive pair on visus est … apparuit) · etsi … tamen (0633D) · Ita … similiter (0634B) · etsi … tamen (0634C) · ut … ita (0634C, ut agricola … ita Apostolus) · Licet … tamen (0634D) · Nec … quin (0635B) · licet (0635B) · non … sed (0633A, 0633C, 0634D, 0635B, 0635D, 0636B, 0636C, 0636D, 0637A, 0638A ×2, 0638C ×2) · Sicut … sic (0635D) · non solum … verum etiam (0636B) · nisi … nisi (0636A) · Quemadmodum … ita (0636D, split across the gloss's paraphrase sicut illi … ita hi) · et … et (0635A, 0636C, 0636D) · quia tamen (0638B, an odd pair kept as printed: "because nevertheless").

The 42 negation particles, checked

Counted in the Latin twin, case-insensitive, on word boundaries: non ×41 · nec ×4 · nisi ×4 · nemo ×4 · ne ×3 · nequando ×1 · nunquam ×1 · nusquam ×1 · nullam ×1 · nihil ×1 — 61 in all; neque and haud do not occur. (The brief's figure of 42 counts a narrower set; the wider count is given here so it can be re-derived.) Every one appears in the English and none appears that Migne does not print. The three sites where carrying a negation makes the passage read badly are the three logged above: 0634A (id est deficio, where the negation is absent and was not supplied), 0635D (nutiles, where the negation is inside a [sic:] and was carried outside it in plain words), 0638C (abruperant amorem, where the sed was carried and the contradiction left standing).

§8 · WHAT A LATER READER SHOULD STILL LOOK AT

The original version of this section listed six sites. The blind read (§9) has since closed the top two and repaired three others, so the list is rewritten to what actually remains open.

1. 0637A, the [cj: nos; read non] (§4). The load-bearing conjecture on the reading page, and the one place in this work where the printed sentence contradicts its own paragraph. It was confirmed clean by the blind read as a rendering, but the conjecture itself is a judgement and would be wrong in public if it is wrong.
2. 0634C–D, the [cj: qui; read quae] (§4). Newly added. Both readings construe and the marker declares the one not taken; confirm the direction is the useful way round.
3. 0637B, Increpa in omnibus (§1). The lemma whose missing word the gloss depends on. Confirm the [var:] says enough.
4. 0638A, reddentibus bona pro malis (§4). Two readings; (a) taken. Declined by the blind read as sound, but it is the one place where the agent of a clause is chosen rather than read.

Closed, and not to be churned: 0634A id est deficio and 0638C abruperant amorem (§4, both decided); the depositum pronoun chain at 0634A–B; the Nero paragraph at 0638C–D; 0635C's comma grouping; 0634C's prius placement; the five added semicolons.

§9 · THE BLIND POLARITY READ — 2026-08-18, and the adjudication

A reader that never saw this file read all three chunks against the Latin.

Result: no polarity reversal, no dropped or relocated negation, no scripture conformation, and no dropped correlative anywhere in the 2,118 words. The marquee classes were checked by name and returned clear:

• Ita laboro in isto officio at 0634B keeps the correlative that Titus lost — the failure the brief was built around did not recur.
• volunt / audiant at 0637B do not repeat the 2 Thessalonians velint failure.
• The depositum pronoun chain at 0634A–B: the referent switches at exactly the right joint, and istum → "this man" correctly keeps Paul as the object at the turn.
• The Nero paragraph, 0638C–D: the Philemon failure shape is absent, because Nero stands only inside an oblique relative and never competes for the subject slot.
• 0637A's participle promoted to a finite verb: sound.

Two cruces CLOSED by it

Both are written up in full at §4 and are summarized here so the closure is findable:

0634A non confundor, id est deficio — decided in favour of the shipped rendering. Migne's two appositives are doing different jobs (deficio glosses the bare verb; vel non frustra patior glosses the whole negated phrase), the asymmetry is his, and supplying "[I do not] fail" would have been the silent repair. The hedge is withdrawn.

0638C qui abruperant amorem — the proposed dropped non is positively wrong, on Migne's own next sentence: Multum enim interest inter eos qui hoc modo et eos qui illo modo peccant — both groups sin, so the second sed contrasts timore with invidentia, not with abruperant. Recorded as a rejected conjecture so the next reader does not re-derive it.

Three repairs made, and the one mechanism behind two of them

⭐ REPAIR 1 · 0635B — "that he might beware" → "that [Timothy] might beware." (§3) ⭐ REPAIR 2 · 0633B — "for his good" → "for his [Timothy's] good." (§3) ⭐ REPAIR 3 · 0634C–D — a silent qui → quae emendation now declared as [cj:]. (§4) ⭐ REPAIR 4 · 0633C, 0638A — two supplied "and"s removed. (§4)

The mechanism, and it is worth stating once because it is the whole of the class:

English cannot leave a possessor or a subject unmarked, so it supplies the nearest one — and Migne's ejus is the printed instruction that the nearest one is wrong.

That is why repairs 1 and 2 are the same defect wearing two grammatical costumes. In both, this cruces file had already named the correct referent in §3 and the English had shipped a bare pronoun anyway. ⛔ Naming a subject in the apparatus is not the same as marking it in the text — the reader meets the sentence, not the crux. Where the Latin's morphology makes the choice (ejus not sua; a purpose clause whose point is the addressee), the English has to spend a word on it, and a bracketed name is the cheapest word available.