Epistola III B. Joannis

Translated 2026-08-18, one chunk, 373 Latin body words → 593 English (1.59×). Structure: ARGUMENTUM. + CAPUT UNICUM (no closing period in Migne — none supplied in the English). Apparatus taken: 0 [sic:] · 0 [var:] · 0 [d:] · 0 [ed:] · 0 [cj:] · 0 [nt:]. noteCount: 0 — Migne prints no inline citation notes in this work. Guillemets: 0 in both columns. Question marks: 0 in both columns. Column anchors 6, identical in order. verify-english.mjs: clean.

Conventions followed without re-opening them: the Glossa block's settled rules (GLOSSA-NT-AGENT-BRIEF.md, src/english/8968/cruces.md §1) — VERS. n.-- passes through verbatim, the lemma is Englished, a fragmentary lemma is rendered as the phrase it stands for with its printed comma/etc. carried.

§1 · THE LEMMA DIVERGENCE LIST (7a″) — every lemma checked against the Vulgate

Checked against the Clementine Vulgate of 3 John. Eight VERS. lemmata plus six second lemmata picked up mid-paragraph. Where a lemma agrees, that is recorded as a checked agreement, not passed over in silence.

Result: no scripture divergence anywhere in this work. Every lemma, including every mid-paragraph pick-up, reproduces the Vulgate word for word. No [var: …] is therefore taken, and the zero is a checked zero, not an unexamined one.

⚠ The one near-candidate, declined: Charissime (VERS. 5). Migne prints Charissime where the Clementine prints Carissime. This is the standard nineteenth-century ch- orthography of the word, printed throughout PL, not a textual variant: Pattern 9 governs (attested spelling, no English exponent, translate normally). No [var:], no [sic:]. Rendered "Dearly beloved", fixed at first occurrence per the launch instruction. ⚑ It in fact occurs once in this work — the recurrence warned of does not materialize here, so the fixed rendering was never put under strain. (Vulgate 3 John has it three times, vv. 1, 5, 11; Migne lemmatizes only v. 5.)

Fragment expansions — what each stump was rendered as, and why

Rule 2 of the settled conventions (a fragment is rendered as the phrase it stands for). Five lemmata here are fragments; the judgment differs per stump and each is recorded:

• VERS. 2 De omnibus orationem, etc. → "Concerning all things I make prayer, etc." The verse's own verb (facio) supplied, exactly as Ruth's Non te. took its verb. "Concerning all things prayer" is not English.
• VERS. 4 Majorem horum. → "I have no greater grace than these." The bare stump ("Greater than these") is not English and, worse, is unintelligible as to what is greater. ⚠ This expansion carries a printed non into the lemma line (non habeo gratiam) — checked against 7a: the negation is the Vulgate's own and the gloss re-quotes it four words later (non habeo majorem gratiam), so it is carried, not supplied.
• VERS. 7 Pro nomine enim ejus, etc. → "For on account of his name, etc." enim = "for" (conjunction), pro = "on account of"; the doubled English "for … for" was avoided by taking pro as "on account of", not by dropping either word.
• VERS. 9 Scripsissem forsitan, etc. → "I would perhaps have written, etc." ⭐ The mood is the gloss's whole subject and is kept. Scripsissem is pluperfect subjunctive and forsitan is the potential particle: the point of the gloss is that John did not write. Flattening to "I perhaps wrote" would destroy the paragraph, which turns on the unfulfilled condition (sed hac necessitate dimisi). The same form is rendered identically at both re-quotations in the gloss below (scripsissem ×2 → "I would have written"), held fixed.
• VERS. 12 Demetrio testimonium. → "Testimony to Demetrius." ⚠ Deliberately NOT expanded, unlike the four above: unlike "Not thee" or "Greater than these", this stump is idiomatic English on its own. Supplying redditur ab omnibus would have added a verb and an agent Migne does not print in the lemma, with no gain in intelligibility. The declining reason is stated because it is the one place this work departs from the expansion the other fragments took.
• VERS. 10 Commonebo. (not a fragment but a one-word lemma) → "I will admonish." Left minimal for the same reason. Note the gloss immediately re-reads it as publication (in omnium notitiam … producam), which is a construal of the lemma, not a variant of it.

§2 · CRUCES AND JUDGMENT CALLS

2.1 · 0705B, the ARGUMENTUM — the unexpressed subject of the whole paragraph

Gaium … extollit … exhortatur … objurgat … perhibet. Four third-person verbs, no subject printed anywhere in the paragraph. Supplied as "he" = the Apostle John, the epistle's author: the argumentum is a summary of the letter, and the three objects (Gaius, Diotrephes, Demetrius) are the letter's three named parties. Nothing else in reach could be the subject. Recorded because the brief requires every supplied pronoun to be named.

Ambiguity declined, and it is real: Demetrio autem bonum testimonium perhibet cum fratribus universis. Two construals both parse — (a) John bears witness together with all the brethren, (b) the good testimony is borne to Demetrius and to all the brethren. Taken as (a), on v. 12's own testimonium redditur ab omnibus … sed et nos testimonium perhibemus, where John explicitly joins his witness to everyone else's. Rendered "together with all the brethren." Reading (b) would need fratribus to be a second dative governed by perhibet, which cum forbids. Logged as adjudicated, not as open.

2.2 · 0706B–0706C — Diotrephes carried by bare third persons for six clauses

The dominant defect class, and this work's densest site. After Diotrephes is named at 0706B, the gloss carries him in five finite verbs with no subject expressed: usurpabat · signet · habet · facit · invenit, plus utens. Each supplied "he" is Diotrephes, fixed by the naming at 0706B and by v. 10's ejus opera, quae facit. ⚠ Note that facit at 0706C is italic, i.e. it is a lemma pick-up from the verse, whose subject in the Vulgate is also Diotrephes — the two chains agree, which is what makes the reading safe.

One genuinely open subject, and it is flagged rather than decided silently: ut perfidiam cordis etiam nomine signet. The subject may be (a) Diotrephes — the nominative of the immediately preceding clause, the default under continuity — or (b) the name / the interpretation itself, or Scripture as the name-giver: "so that it may betoken the perfidy of his heart by his name also." Rendered (a) — "so that he may betoken the perfidy of his heart by his name also" — on grammatical continuity alone. Both readings construe and neither is provable from the page; (b) is arguably the better sense, since a man does not deliberately betoken his own perfidy, and a passive English ("so that the perfidy of his heart might be betokened") would have concealed the choice altogether, which is why it was not used. This is the site in this work most worth a second reader.

2.3 · 0706B — the onomastic gloss speciose insulsus / decor insaniens

Migne prints two italic etymologies of the name Diotrephes. They are not scripture and take no [var:]. Rendered "speciously foolish," and "comeliness gone mad,": speciose is the adverb ("showily, plausibly"), insulsus "unsalted, insipid, silly"; decor insaniens is literally "raving comeliness". Both are rendered as the coinages they are rather than smoothed into one idea — the pair is a contrast (fair outside, mad inside) and collapsing them would be the lemma-conformation fault in its onomastic form. ⚠ Neither corresponds to the actual Greek Διοτρεφής ("Zeus-nourished"); that is Migne's tradition and is not ours to correct — no marker, no note in the text.

2.4 · 0706C — quae forsitan, the gloss re-construing its own lemma

Illi Ecclesiae, quae forsitan, id est dubitative est Ecclesia … scripsissem. The italic run is not a Vulgate phrase: v. 9 has Scripsissem forsitan Ecclesiae, with forsitan modifying the verb. The glossator detaches it and re-attaches it to Ecclesiae — "the Church which perhaps [is one]" — and then says so explicitly (dubitative est Ecclesia). This is a construal, not a variant, so no [var:] fires; but it is the interpretative move the whole paragraph exists to make, and it is recorded here because a reader meeting an italic phrase that is in no Bible will otherwise think our lemma is wrong. Rendered "which perhaps, that is, it is doubtfully a Church".

Subject of mereretur in the same sentence: the Church (feminine), supplied as "she" — fixed by perfecta … Ecclesia in the same clause.

2.5 · 0706C — in nos doing double duty (italic lemma inside a gerundive)

falsis rationibus utens in nos confutandos. The italic span is the two words in nos, which are simultaneously (a) the lemma pick-up from v. 10 garriens in nos and (b) the governing preposition of the gerundive confutandos ("for confuting us"). English cannot let one phrase serve both without repeating the pronoun. Rendered "using false reasonings against us to confute us" — the italic span 1:1 on "against us", the gerundive rendered after it. The repetition of "us" is a cost of keeping the italic where Migne puts it; the alternative (moving or widening the italic) would have broken the 1:1 rule.

Pattern 17 datum, and it is visible on this page. The same sentence has in nos / ad vos against the letter's tibi throughout: Gaius is addressed singular ("thou … thee … thy") in VERS. 2, 5, 9 and 14, while ad vos … retrahendos at 0706C is plural ("to draw you back from the faith") — Diotrephes' target is the congregation, not Gaius. Rendered accordingly; a flat "you" would have erased the shift.

2.6 · 0706C — the supplied object of aliquando compescere

debemus aequanimiter tolerare … aliquando compescere. The second infinitive has no object printed; the object is linguas … excitatas from the first limb. Rendered "sometimes to check them". ⚑ The asyndeton is kept: Migne prints no connective before aliquando, and an "and" was drafted and then removed rather than supply a conjunction the plate lacks.

2.7 · 0706C — the semicolon before sed facit ea

quorum exemplar nusquam habet; sed facit ea. An earlier draft rendered the semicolon as a comma; restored, per Pattern 8 (never delete a mark Migne prints). Recorded because it was an actual near-miss in this chunk, not a hypothetical.

2.8 · 0706D — velit imitari bonum: two construals, one taken

Quod illum velit imitari bonum, aperit ut imitetur Demetrium. Either (a) bonum is the object of imitari — "that he would have him imitate the good" — or (b) bonum agrees with illum — "that he would imitate that good man". Taken as (a), because v. 11, which this gloss follows directly, is noli imitari malum, sed quod bonum est: the neuter good/evil contrast is the verse's own, and Demetrius is then adduced as the instance of it. Reading (b) is not impossible and is recorded. Supplied subjects here: velit / aperit = the Apostle, illum and possit et ipse = Gaius.

2.9 · 0706D — gratiam pacis et salutis / a salute: greeting or salvation?

Amicis gratiam pacis et salutis mandat, ut veritatis inimicos a salute monstret extraneos esse. Latin salus is both "greeting" and "salvation"; English must choose, and the choice is invisible to a reader once made, which is why it is logged. Read as "salvation" in both places, for consistency within the sentence and because gratia pacis et salutis is a settled Latin idiom and extraneus a salute is a theological predicate the Glossa uses freely. The competing reading is decent: the lemmatized verse continues Salutant te amici. Saluta amicos nominatim, so a gloss explaining why John greets only friends would want "greeting" — "strangers to the greeting". Declined, not overlooked.

§3 · CANDIDATES DECLINED, AND WHAT WAS READ CLEAN

Declined markers, with reasons:

• [var: …] — none. Every lemma agrees with the Vulgate (§1). Charissime is orthography, Pattern 9.
• [sic: …] — none. No broken, split, fused or non-word type anywhere in the chunk. Every word is a well-formed Latin form occupying a grammatical slot. Checked specifically for the fused/split class of 7a⁗-b, which no mechanical test finds: none.
• [cj: …] — none. Pattern 18's bar is that the printed word makes the English assert something the author is arguing against. Nothing here does: the two hard sentences (2.2's signet, 2.8's bonum) are ambiguous, not false, and ambiguity stays in the crux.
• [d: …] — none. Diotrephes stands three times in three clauses at 0706B and scripsissem twice in one paragraph; both are rhetorical repetition, not dittography — each occurrence has its own grammatical slot and its own clause, which is the test.
• [ed: …] — none. No hole in the digitization; the chunk construes end to end.
• [nt: …]/[n: …] — none exist. noteCount: 0; Migne prints no inline citations in this work, so there is nothing to pass through and nothing to translate.
• [f: …] — not applicable. Lemma-and-gloss commentary; runbook false positive 4.
• Band letters. The chunk runs 0705B · 0705C · 0705D · 0706B · 0706C · 0706D. Checked against runbook false positive 1 and not reported as a finding: no text breaks mid-word or mid-construction at any anchor.
• Foot-of-page conjecture notes. Not recovered — PL 114 has not been plate-read ([cn: …] is injected per volume at chunking). Recorded so a later reader can tell an unread volume from a checked zero: this is unread, unlike Ruth's PL 113 pages.

Read clean (negation, polarity and sense checked clause by clause): the ARGUMENTUM (0705B); VERS. 2 (0705B–C); VERS. 5 (0705C); VERS. 7 (0705D); VERS. 14 (0706D). Every printed negation was carried and counted: non habeo (VERS. 4, twice — lemma and gloss), in nullo facto vel verbo (VERS. 4), non debes deficere and non pro vana gloria (VERS. 5), non enim pro suis sceleribus (VERS. 7), non solum … sed etiam and non curat (VERS. 9), nusquam and non debemus excitare and ne ipsi pereant and ne … corrumpant (VERS. 10) — 14 negation sites, all present in the English, none supplied.

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

A fenced reader (Latin twin + English only, never this file) worked the chunk three times — once for polarity, once lemma-by-lemma against the verse each fragment opens, once for every unexpressed subject. 5 sites raised · 3 confirmed and repaired · 2 declined.

The headline is negative and belongs on the record as such: zero polarity defects, zero silent repairs. All fifteen italic lemmata were checked word by word. Where Migne and the Vulgate genuinely part company the English followed Migne: he prints the future indicative Commonebo where the Vulgate has the subjunctive commoneam, and the English gives "I will admonish." Majorem horum likewise gives grace (Vulgate gratiam) not the Authorized Version's "joy", and than these (horum, plural) not Douay's "than this". Neither Bible was reached for. This file did not fall for the class that beat Ruth.

✅✅ 0705C + 0706B — CONFIRMED, REPAIRED. A defect that nets to zero.

Two Pattern 8 sites, and they must be read together:

• 0705C …et facultas opum; ita semper… — Migne's semicolon closes the protasis of the Sicut … ita period. It had been softened to a comma, and "and the means of wealth, so mayest thou" reads as a run-on with the as/so architecture gone flat. Restored.
• 0706B …de commendatione eleemosynae, sed hac necessitate dimisi… — Migne prints a comma; a semicolon had been supplied. Restored.

⭐⭐ This is the finding to carry forward. Migne prints three semicolons; the English printed three semicolons. Any count-based or parity-based check passes — one mark was taken away in one place and a new one put in another, and the totals cancel. It was caught only because the reader located each mark instead of tallying them. verify-english's question-mark parity would never have seen it, and neither would a diff of counts.

⚑ Method worth copying into the next punctuation check: match the marks in ORDER and by position, not by total — the same move as sentence-mark-sites.mjs for the ?/! sweep, applied to the marks nothing scans for.

✅ 0706C · VERS. 12 lemma — CONFIRMED, REPAIRED

Demetrio testimonium. had been rendered "Testimony to Demetrius." — a verbless caption. Now "Testimony is borne to Demetrius." (3 John 12, Demetrio testimonium redditur ab omnibus).

The argument that decided it is internal consistency, not taste. Migne terminates four lemmata in this work with a full stop — Majorem horum. · Commonebo. · Demetrio testimonium. · Pax tibi. — and three of the four had been expanded into complete clauses under Glossa convention 1.2 (a fragment is rendered as the phrase it stands for). Only this one was left a stump, and the gloss below it does not construe testimonium grammatically, so nothing downstream repaired the gap.

⛔ 0706C–D · VERS. 12 — DECLINED, and the reader's stated ground is mistaken

Quod illum velit imitari bonum, aperit ut imitetur Demetrium, ut simili laude omnium possit et ipse dignus existere. The reader proposed reading illum … bonum attributively ("that good man") with both ut clauses final, and objected that our rendering "leaves aperit with no object at all."

It does not. The fronted Quod clause is the object of aperit — "That he would have him imitate the good, he discloses" — so the strongest plank of the alternative reading fails. The supplied "namely" for the first ut is explicative and defensible on the printed words. The reader itself did not claim its reading was certainly right. Left as it stands, and named here so a later eye knows the ambiguity was seen and weighed rather than missed.

⛔ 0705C · VERS. 5 lemma — DECLINED, with the boundary recorded

Charissime, fideliter, etc. → "Dearly beloved, faithfully, etc." The reader noted that 3 John 5 reads fideliter facis and that the gloss immediately re-quotes the verb (id est sicut decet fidelem facis), so our lemma drops the word the exposition turns on. Rated low confidence by the reader itself, and declined here: a comma-plus-etc. fragment is Migne's own signal that he is abbreviating, and the brief carries the comma and the etc. through as printed. "Dearly beloved, faithfully," is readable English, unlike Ruth's rejected "Not thee". The reader drew the same line itself when it declined Pro nomine enim ejus, on the ground that the Latin fragment is a bare prepositional phrase.

⚑ The convention's working boundary, now stated twice across this cadence and worth promoting into the brief: expand a fragment when the English stump is not a usable phrase (a bare Non te, a verbless caption); leave it when the stump reads as ordinary abbreviated English and Migne's own etc. marks the abbreviation. The defect is never expansion or its absence — it is conformation, an English Bible's wording arriving where Migne's Latin differs from it.

Read clean, and named

All 15 italic lemmata matched 1:1 in order with terminal punctuation carried; ten negations in both directions with zero defects; 18 unexpressed-subject sites named and checked — John (extollit, exhortatur, objurgat, perhibet, producam, mandat, monstret), the brethren (sunt honorandi, expulsi sunt, profecti sunt, annuntiant), Diotrephes (habet, facit, invenit), the detractors (pereant, disseminant, corrumpant, with suam malitiam correctly reflexive to them and not to "us"); all eight Pattern 17 sites correct, including the one plural (ad vos a fide retrahendos); ? 0 : 0; six column anchors identical in set and position with no text broken across any.

Rejected with reasons: ut perfidiam cordis etiam nomine signet (three subjects construe; etiam decides it for a personal subject — "even by his name" presupposes his deeds, and a name that betokens by its own name is circular); principatum → "primacy" (checked against Douay's "pre-eminence" — renders the Latin, not a Bible import); garriens → "prating" (means what it means; coinciding with Douay is not conforming to it); mereretur … Ecclesia → "she … a perfect Church" (the demotion is the better reading, since the antecedent is dubitative est Ecclesia); abundat tibi → "abounds in thee" (idiom); Charissime for Carissime (Pattern 9); the absent 0706A band (false positive 1).

⚑ Plate: checked zero

The full-width foot bands of leaves n356 and n357 of archive.org patrologiaecurs06saingoog (PL 114; leaf n = (column + 9) / 2, calibrated against the running head at n330 = cols 651–652) cover this work's columns, including n357, where the 3 John / Jude division falls mid-page. Migne prints no foot-of-page apparatus there — no foot rule, no numbered notes. Recorded in data/plate-notes/coverage.json as tier 3, coverage full, notes_recovered: 0.