Epistola ad Philemonem

One chunk, 591 Latin words → 952 English (ratio 1.61). Apparatus: 1 [var:] · 1 [cj:] · 0 [sic:] · 0 [d:] · 0 [ed:] · 0 [nt:] · 1 [n:] (parity with the Latin twin). Structure: ARGUMENTUM + CAPUT UNICUM (→ THE ARGUMENT / THE ONLY CHAPTER, per the Glossa brief). 6 column anchors, 24 italic spans, 0 question marks in either language. Guillemets: none.

Column band note: the run is 0641B · 0641C · 0641D · 0642B · 0642C · 0642D — 0642A is not printed, which is the known-false-positive class (bands are positional, not four per column). No text breaks mid-word or mid-construction at any anchor. Not reported as a gap.

§1 · THE LEMMA DIVERGENCE LIST — all eleven VERS. lemmata plus the six second lemmata

Every lemma was collated against the Vulgate of Philemon. Ten of the eleven VERS. lemmata agree with the Vulgate exactly, as fragments of their verse; the divergences are all in the second lemmata the gloss picks up mid-paragraph. Agreements are listed because a list of findings alone cannot be told apart from a report by an agent that never looked.

Second lemmata caught mid-paragraph (six, all found and all rendered as italic spans in place):

§2 · THE CRUCES

2.1 · 0641D · in agnitione quae est — the relative changes antecedent. [var:] TAKEN.

Migne's gloss re-quotes Philem. 6 as evidens fiat in agnitione quae est. In Christo Jesu. The Vulgate reads in agnitione omnis operis boni, quod est in vobis in Christo Iesu — a neuter relative referring back to operis boni, "of every good work which is in you." Migne's quae is feminine and can only attach to agnitione, and the phrase omnis operis boni and in vobis are both gone. The gloss then builds on exactly that: In Christo Jesu. Id est, cum cognitione boni aeterni — it takes the in Christo Jesu as qualifying the acknowledgment, not the good work. Rendered as printed; [var:] beside it with the received reading. No speculation offered as to the cause (compression by the glossator, a copy-text, or the compositor — that is a research question).

⚠ Related but NOT marked: two lines earlier the gloss paraphrases omne bonum esse in te per Christum where the Vulgate has in vobis (plural). That sits in the gloss's own prose, not in an italic lemma, so it is the glossator addressing Philemon, not a divergent citation. Logged, not marked.

2.2 · ⭐ VISCERA — ONE rendering, "bowels", held rigid across both lemmata

This is the word the work turns on. It carries two VERS. lemmata (7 Viscera sanctorum, 12 Ut mea viscera) plus a third occurrence at v. 20 (Refice viscera), and Paul's whole argument is that Onesimus goes back as Paul's own viscera. The two members meet here, so the rendering is fixed and rigid: bowels, everywhere.

Why "bowels" and not "heart(s)" or "inmost self": Migne's own first gloss on VERS. 7 is Id est filii sanctorum charissimi — viscera read as offspring. The second gloss in the same sentence reads it as the seat of affection (sanctorum affectus quievit in te). VERS. 12's gloss reads it as Paul's self (suum animum in illo significet recipi). A single English word has to be able to mean all three, and "bowels" is the only one in the register that does — it carries the affective sense and the archaic sense of offspring ("the bowels of thy body"), which is precisely what the filii gloss requires. "Hearts of the saints" cannot mean children and would make Migne's first gloss unintelligible; "inmost parts" cannot either. The devotional/anatomical split the brief warned against is therefore not merely inconsistent, it destroys the link the gloss is built on. Ruling: viscera = bowels, all three sites, no exceptions.

2.3 · ⭐ ONESIMUS'S NAME — Migne does NOT draw the pun out; the utilis chain is its only trace

The epistle puns on Onesimus = "useful" (qui tibi aliquando inutilis fuit, nunc autem et mihi et tibi utilis). Migne's gloss never prints inutilis, and never mentions the name's meaning or the Greek. The lemma at VERS. 11 stops at aliquando, before the negative, so the lemma itself carries no polarity at all. What survives is a four-link utilis chain, and it is held rigid:

• VERS. 11 Nunc tibi utilis erit → "Now he will be useful to thee"
• VERS. 11 Si enim Paulo futurus est utilis → "if he is going to be useful to Paul"
• VERS. 12 videns in illo utilitatis spem → "seeing in him the hope of usefulness"
• VERS. 16 ut, oblitteratis peccatis, utilis reverteretur → "that he might return useful"

"Profitable/unprofitable" (Douay, AV) was declined: it is the familiar English Bible wording, which §7a″ forbids importing, and it breaks the tie to utilitatis. Nothing was added to make the pun visible that Migne does not print.

2.4 · ⚠ 0642B, 0642C · Dominus = the MASTER, not the Lord — Migne capitalizes both

ut eo ipso exstinguat iracundiam Domini and quanto amplius Domino are printed with a capital D. Both are rendered "the master" / "his master", not "the Lord":

1. The argument is Chrysostom's on Philem. 11 — the fault is confessed so as to defuse the owner's anger; the person who might be angry with Onesimus is Philemon.
2. The same word is printed lowercase twice in the VERS. 16 gloss for the same referent (offenso proprio domino, domino suo aequalis), where "master" is not in doubt.
3. quanto amplius Domino closes a comparison whose other term is Paulo — a man.

Capitalization is a 19th-century typographic habit and carries no lexical force in Latin; English must choose a word, and dominus in this context means master. This is a rendering of the printed word, not an emendation of it. ⚠ Distinguished from in Domino at VERS. 16 and v. 20, which IS the Lord (it is the scripture lemma, opposed to in carne). Fixed vocabulary for the work: dominus = master · magister = teacher · servus = slave · custodia = custody · carcer = prison · vinctus = prisoner · vincula = bonds.

2.5 · 0642B · Senex. is verse 9 printed under VERS. 10 — NOT reordered

Senex is Philem. 9 (ut Paulus senex), glossed inside the VERS. 10 paragraph. Migne's ordering; reproduced as printed, per the brief's rule on out-of-sequence verse numbers. The same thing happens on a larger scale at VERS. 18, whose paragraph run glosses vv. 20 and 21 (§2.7). Nothing renumbered, nothing moved.

2.6 · 0642C · Fratrem et in carne, — a conflated re-quotation. [var:] DECLINED.

Migne: humiliat eum dicens: Fratrem et in carne, quia ex uno Adam omnes, et in Domino per fidem. The Vulgate of v. 16 has carissimum fratrem, maxime mihi: quanto autem magis tibi et in carne, et in Domino? — so every word of Migne's italics is in the verse, but not in that order and not as one phrase: the glossator has stitched fratrem to et in carne … et in Domino.

Declined, with the reason, because a blind reader will look here. Pattern 14 fires where the divergence changes the sense, and it does not: the Vulgate already makes Onesimus a brother in the flesh and in the Lord, and the gloss's stitched phrase asserts nothing more. Marking every compressed re-quotation in a lemma-and-gloss commentary would be spam on the page. Recorded here instead.

2.7 · 0642D · Ita frater ego te fruar in Domino. — the Vulgate's FULL STOP is gone

The Vulgate punctuates v. 20 Ita frater. Ego te fruar in Domino — two sentences. Migne prints it as a single unstopped run. Words identical; punctuation only. Under Pattern 8 the plate's marks ride as printed, so the English is one clause: "So brother may I enjoy thee in the Lord." No comma was supplied around the vocative — that would be the same smoothing in miniature, and the divergence is exactly what this line records. No [var:]: the sense is unaffected and the pattern is for wording, not for pointing.

2.8 · ⚠⚠ 0642D · ut NON amplius faciat quam postulatur ab eo — the negative reverses the argument. [cj:] TAKEN, negative CARRIED.

Migne: Provocat eum sic blandiendo, ut non amplius faciat, quam postulatur ab eo. Solet enim fieri, ut qui de se videt bene sentiri, meliorem se praebeat.

Rendered literally, with the non: "so that he may not do more than is asked of him." The negative contradicts the paragraph it stands in, and the verse it glosses:

• its own next sentence gives the mechanism — a man well thought of meliorem se praebeat, shows himself better, i.e. does more;
• the lemma is Philem. 21, whose Vulgate continues sciens quoniam et super id, quod dico, facies — "knowing that thou wilt do beyond what I say";
• provocat … blandiendo (coaxing him forward) is not a device for capping performance.

Negation fidelity is absolute, so the non stays and no word was translated as its opposite. But non is a real word, so [sic:] cannot reach it and Pattern 7's literal English then asserts what the glossator did not — which is exactly Pattern 18's gap. Marker placed: [cj: non amplius; read amplius, "that he may do more than is asked"]. The printed run non amplius is verbatim in the Latin twin (the verifier's substring test passes), so the conjecture is offered against words Migne actually prints. The English is unchanged; the gloss is additive. ⚠ The alternative reading — that flattery is meant to keep Philemon from exceeding the request — was considered and rejected: it leaves meliorem se praebeat and Philem. 21 both stranded.

2.9 · 0641B–0641C · anacolutha rendered as printed, not repaired

Three sentences do not complete their construction. All are rendered as closely as the printed words allow and none is smoothed:

• VERS. 5: Quia enim in Christum credebat … et ita … serviebat. A quia-clause with no apodosis — the gloss never says what the because explains. English keeps the bare "For because…" opening.
• 0641C, VERS. 6: ut fides tua communicans fidei nostrae — no verb; the parallel clause beside it has evidens sit. An English copula was supplied ("that thy faith be communicating"), which is elision of a Latin sit, not an emendation; recorded so the choice is visible.
• 0641D, VERS. 6: ita evidens sit, hoc merito cognoscatur — the connective ut before the second subjunctive is unexpressed; rendered as the two jussives Migne prints, in asyndeton, with his comma.

2.10 · 0641D · hoc merito — two constructions, both construe. NOT adjudicated.

ita evidens sit, hoc merito cognoscatur omne bonum esse in te per Christum. Either (a) hoc merito = ablative, "by this desert/merit", or (b) hoc = "hereby" with adverbial merito = "deservedly". Rendered (b) — "let all good hereby be deservedly known to be in thee" — because (a) would introduce a meritum the gloss nowhere else discusses, while (b) needs nothing outside the sentence. Both readings are recorded rather than one chosen silently; no marker, since nothing here is defective.

2.11 · communicatio — fixed English across the four occurrences

communicatio / communicans / communicas all sit in one sentence and are one argument, so: communication · communicating with · thou communicatest to. "Fellowship" and "partaking" were declined because they break the tie to the verb, and "communication" is what the lemma has to be for the twice-quoted communicatio fidei tuae to read the same both times.

⚠ Two different Latin words land on one English word and it is harmless: refrigeratus est (VERS. 7) → "has been refreshed" and Refice (v. 20 lemma) → "Refresh". They are 60 lines apart, no argument links them, and the Vulgate's own requieverunt / refice pairing is what the glossator is following. Recorded so a later sweep does not read it as a collapse.

§3 · THE UNEXPRESSED SUBJECT — every supplied pronoun, named

Paul, Philemon and Onesimus move under bare third-person verbs throughout. Every "he" the English supplies is listed with what fixed it. Nine sites; two are ambiguous and are logged as ambiguous, not resolved silently.

⚠ A note on laesit (0642D): the gloss paraphrases v. 18's nocuit as laesit. It is the glossator's own prose, not an italic lemma, so no [var:] — recorded here only.

§4 · READ CLEAN — and the candidates declined

Paragraphs read clean (no crux, no candidate): the ARGUMENTUM; VERS. 7; VERS. 11 (apart from Domini, §2.4); VERS. 12; VERS. 15; VERS. 22. Negation was checked word by word through the whole chunk — the printed negatives are Non dicit · non imperat · non crederet (inside ne) · non solum · ne … inflaretur · non amplius · non autem retributionem · nihil (none) — all eight carried, none invented. The double negative ne ille non crederet is carried as a double negative ("lest that man should not believe it") and not resolved into a positive.

Candidates examined and declined, with reasons:

1. [f:] inline citation tails — declined categorically. This is a lemma-and-gloss commentary, not a florilegium; runbook false-positive 4 governs. No tag taken.
2. [sic:] anywhere — nothing on this plate is broken type. Every word is a real Latin word; there is no split or run-together type, no non-word. The one place the argument breaks (§2.8) breaks on a real word and took [cj:], which is the marker for exactly that.
3. [nt:] on the single note — declined. [n: (CHRYS.)] is a bare attribution siglum, i.e. a citation, not editorial prose. Passed through verbatim and unanglicized, per the Glossa convention (and passing it through is not endorsing the attribution).
4. [var:] on Fratrem et in carne — declined, §2.6.
5. [var:] on the punctuation of Ita frater ego te fruar — declined, §2.7.
6. [ed:] — nothing is missing from the digitization; the TEI and the printed run agree and no data/tei-patches/8994.json was needed.
7. A supplied ? at quanto amplius Domino. — declined, and this is the punctuation site worth naming. Si enim Paulo futurus est utilis … quanto amplius Domino. is a verbless comparative that reads as a rhetorical question, and Migne prints a full stop. Pattern 8: never supply a mark the plate lacks. The English ends "how much more to his master." with a period, and the chunk's question-mark count stays at 0 = 0.
8. An em-dash anywhere — none used; the register prefers the comma, colon and stop.

Migne's own ( sic ): none in this work, so nothing was stacked on.

Plate-notes coverage: Migne's foot-of-page conjecture apparatus was not consulted for this work — no plate render was made and none is recorded in data/plate-notes/. This is a gap, not a checked zero, and it matters at §2.8: if the plate carries a note on that sentence it would settle the non directly. Recorded so a later reader can tell an unchecked book from a checked one.

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

A fenced reader (Latin twin + English only, never this file). 5 sites raised · 3 confirmed and repaired · 2 recorded without repair. Zero polarity defects: all eight negations Migne prints are carried (Non dicit · non imperat · non autem retributionem · non de humili · non ait · ne ille non crederet · non solum … sed · ut non amplius faciat), and none is invented. Zero silent repairs — Douay's utilis → "profitable" was resisted at all four sites, and quievit in te … refrigeratus est renders Migne's in te rather than Douay's "refreshed by thee".

✅ 0642C · VERS. 12 — CONFIRMED, REPAIRED. Reference mis-bound by English word order.

Printed: Onesimum profugum [0642C] recurrentem ad divinum auxilium cum esset in custodia Apostolus baptizavit…

Was: "The Apostle baptized Onesimus, a fugitive [0642C] running back to the divine help, when he was in custody, seeing in him…" Now: "The Apostle, while he was in custody, baptized Onesimus, a fugitive [0642C] running back to the divine help, seeing in him…"

esset prints no subject, and the work has already named who is in custody twice — the ARGUMENTUM (scribens ei ab urbe Roma de carcere) and VERS. 1 (Eo ergo tempore in custodia erat. Ideo ait se vinctum). It is Paul. Our word order stranded the cum clause after "a fugitive running back to the divine help", so the nearest available antecedent became Onesimus — putting on a public page the claim that the runaway slave was in prison, which Migne does not print and which makes nonsense of profugum recurrentem: a man in custody is not a fugitive running for refuge. No word was wrong, only the binding — invisible to every automated check.

✅ 0641D · VERS. 6 — CONFIRMED, REPAIRED. ejus, not sua, and the morphology decides it.

Printed: Hoc enim orat Apostolus, ut operatio ejus cum cognitione fiat bonitatis, id est, ut tali mente operetur, ut in futurum inde exspectet praemium Jesum Christum…

Was: "…that his working may be done…" Now: "…that that man's working may be done…"

ejus is the genitive of is — by definition not the subject of orat. The working is Philemon's, and our own English had identified it correctly one clause earlier (operatio fidei qua servis Christo et sanctis → "the working of faith by which thou servest Christ and the saints"). But "the Apostle" then stood immediately before "his working", so every following pronoun read as Paul: the page said Paul prays that Paul's own working be done with knowledge and that Paul look for Christ as his reward. That is a different sentence from the one printed, and it detached the gloss from the lemma it expounds.

"That man's" is the literal value of ejus and matches this work's own register — VERS. 11 already renders ille as "that man" two paragraphs later. Migne's own shift from second person (tua, servis) to third (ejus, operetur, exspectet) is his, and is preserved.

✅ 0642B · VERS. 11 — CONFIRMED, REPAIRED, and the mark chosen deliberately

Migne prints …exstinguat iracundiam Domini, non ait: Nunc tibi utilis erit… — one sentence. Our English had supplied a terminal stop and made it two. Now a semicolon.

⚑ Why a semicolon and not Migne's comma, when the 3 John read (8989 §4) restored a comma in the same cadence — the two are not inconsistent. Pattern 8b's test is whether English refuses the mark. At 8989's 0705C English does not refuse Migne's semicolon, so it was restored exactly. Here English does refuse a bare comma between two finite main clauses (exstinguat … non ait), so the minimal faithful accommodation is the mark that keeps his sentence one sentence. The full stop was the over-correction; recasting his finite non ait … sed introduxit into participles to save the comma would have been a larger liberty than the punctuation it rescued.

⬜ 0641C · VERS. 7 — viscera: the rigidity is right, and there is one site where NO English word works

The fixed rendering "bowels" holds at all three sites — Viscera sanctorum (0641C), Ut mea viscera (0642C), Refice viscera (0642D) — with no drift, and the reader verified it independently. But at the one place where Migne glosses the word himself it fails: Viscera sanctorum. Id est filii sanctorum charissimi. Viscera in Latin covers offspring (viscera mea = "my own child"), which is why the gloss can say "that is, the most beloved sons of the saints" — and "bowels … that is, the most beloved sons" is an inference no English reader can follow.

Nothing was changed, and that is the finding. "Hearts" fails filii just as badly and fails Refice viscera; "children" fails the other two sites and would be a silent repair of the boldest kind. No English word carries both senses, so no rendering can be made to work at every site. "Bowels" is the literal choice, not a Douay reach (Douay merely agrees), and the polysemy stays visible in the facing Latin column.

⚑ The general lesson, against §2's own ruling: a fixed-vocabulary pair earns its rigidity where the two members meet — but rigidity cannot manufacture a sense the English word does not have. Where the author glosses his own word, the gloss is the test of the rendering, and a rendering that survives the test everywhere except at the author's own definition is the best available, not a correct one.

⬜ 0642B · VERS. 10 — debet obsequi, a silent choice, recorded

Seni magis aetatis similitudine debet obsequi → "One ought the more to comply with an old man." debet prints no subject and two readings construe: generic ("anyone"), or Philemon — the surrounding gloss is about what Philemon owes to Paul, who has just called himself senex and vinctus to press that very claim, and reading (b) gives aetatis similitudine something to do. §2.8 had already named this site as ambiguous. Left generic; named here rather than assumed right.

Read clean, and named

Every paragraph read against its twin: ARGUMENTUM · VERS. 1 · 5 · 6 · 7 · 10 · 11 · 12 · 15 · 16 · 18 · Ita frater · Confidens · 22. VERS. 5, 15, 16, 18, Ita frater and VERS. 22 are clean on every axis — negation, subject, lemma wording, punctuation, second person. 24 italic spans 1:1 counted per paragraph; ? 0 : 0, with quanto amplius Domino. — a rhetorical question Migne closes with a full stop — correctly riding the stop; twelve Pattern 17 sites all singular and correct, the only "you" being the plural inside the [var:]'s Vulgate quotation.

Three further Vulgate divergences found and judged sub-threshold, recorded because the divergence list is the deliverable: Confidens in obedientia (Vulg. oboedientia tua — preposition added, tua dropped; English does not supply "thy"), Fratrem et in carne (a compression, not a variant), and si vel laesit (Vulg. nocuit — the glossator's own paraphrase, outside Pattern 14's reach).

Rejected with reasons: In agnitione omnis → "of all" (the gloss itself requires it — omne bonum esse in te — so expanding to "of every good work" would import words the plate lacks); Domini → "master" at 0642B against "the Lord" at 0642C (Chrysostom's a fortiori only works if the second term is Philemon the dominus; capitalization is not a word); "because he believed" at VERS. 5 (as bare in Latin, no competing antecedent, nothing rearranged — unlike site 1); "Onesimus his slave" in the ARGUMENTUM (servo ejus equally terse); vinctus/vinciri/vincula split across "prisoner"/"bound"/"bonds" (not the 11551 stylus/epistylium problem, because the gloss makes no etymological claim); the absent 0642A band (false positive 1); [f:] (item 4).

⚑ The [cj:] at 0642D — independently confirmed

The reader formed its own verdict from the Latin before reading our marker and would have proposed the same conjecture: ut non amplius faciat, quam postulatur ab eo glosses Phlm 21, whose point is sciens quoniam et super id quod dico facies, and the gloss's own next sentence — qui de se videt bene sentiri, meliorem se praebeat — is a reason for exceeding the request. No reading construes the non with what follows it. It also confirmed that the English carries the non literally in the running text rather than translating around it, with the conjecture standing beside it additively per Pattern 18.

⚑ Correction to §8 — the plate WAS checked, so this conjecture stands as ours

§8 logged Migne's foot-of-page apparatus as a skip for PL 114. It is now a checked zero: the full-width foot bands of leaves n323–n325 of archive.org patrologiaecurs06saingoog (leaf n = (column + 9) / 2, calibrated against the running head at n330 = cols 651–652) cover this work's columns, including n325, where the Titus / Philemon division falls mid-page — the case that lost three notes on PL 202. Migne prints no foot-of-page apparatus there: no foot rule, no numbered notes. So the 0642D non is not a crux Migne already answered, and the [cj:] is properly ours. Recorded in data/plate-notes/coverage.json as tier 3, coverage full, notes_recovered: 0.