Epistolae ad diversos

Greek source: src/greek/dorotheus-epistolae-ad-diversos/ — our own 600 dpi vision-OCR of the plate (raw/pg088/epistolae/out/leaf970–972.txt); this work has no Calfa text. Twin: src/pg-latin/dorotheus-epistolae-ad-diversos/, from raw/scans/pg088/copyA_088_djvu.xml (archive.org OCR of the facing Latin — a witness, not an authority). Apparatus: data/pg-notes/pg088-epistolae.json, 11 notes, APPARATUS.md.

Translated 2026-08-08. One chunk, 812 Greek words, Epistles I–VIII.

Work-wide conventions

• Section numerals follow the plate's own Latin column, which sets ROMAN (J. ll. 111. IV. V. Vi. VII. VIH., all OCR mangling of I–VIII). So Α΄ Β΄ Γ΄ Δ΄ Ε΄ Ϛ΄ Ζ΄ Η΄ → I. II. III. IV. V. VI. VII. VIII. The translation-style.md dialogue rule ("arabic, matching the plate's own Latin column") is a rule about deferring to the Latin column, not about arabic: here the column is Roman, so the English is Roman.
• τέκνον = "child" throughout, never "son" (the Latin's fili). Dorotheus addresses ἀδελφός in the headings and τέκνον in the body, and the two are a real distinction in the letters.
• λογισμός = "reasoning" everywhere — including in the fixed monastic sense at Ep. VIII's heading (λαμβάνοντα διαφόρους λογισμούς). One English word, because the whole argument of Epp. I and VII is that ἀνθρώπινοι λογισμοί are the thing to be let go of, and "thought"/"reasoning"/"consideration" by turns would dissolve it.
• ἀνάπαυσις = "rest" throughout (Ep. VIII turns on it three times in six lines); ἀμεριμνία and cognates = "freedom from care".
• Guillemets: the plate prints them once, round the Antony logion in Ep. V, and they are reproduced. Every other quotation mark on the page is ours (" ", Pattern 6) — Migne marks his scripture with bold/italic type, not with marks.

⛔ THE READING THAT PAID FOR THE SESSION — a dropped negation, in the TRANSCRIPTION

Col 1840 (leaf971 line 10). The transcription read ἀλλ' ᾧ θέλω ἀντιστῆναί σοι and did not flag it. Re-read at 600 dpi on leaf971_p972_seg2.png, zoomed: the glyph is an οὐ ligature with its breathing, not a ῳ with iota subscript. The facing Latin prints sed nolo tecum pugnare. Corrected in the raw transcription and re-chunked; the English reads "but I do not wish to contend with you."

Uncorrected, the page would have had Dorotheus say he wished to contend with a brother he is writing to console, immediately before "nor with myself either" — translation-style.md 7a exactly, arriving one step upstream of where 7a looks for it. The rulebook hunts the silent repair in the translation; this one was already in the Greek before the translator saw it, and no reading of the English against the Greek would ever have found it, because the two would have agreed.

⚑ The general lesson, and it bears on the 87-leaf fleet: on an our-OCR work the transcriber's unflagged readings are the dangerous ones. This word carried no UNCERTAIN entry while three weaker readings on the same leaf did. Rulebook 11 ("the transcriber wins until the plate says otherwise") is intact — the plate said otherwise — but its corollary needs saying: a confident transcription of a negation is still a negation, and polarity-bearing particles deserve a plate look whether or not they were flagged.

Other corrections made from the plate in the same pass

Three, all verified at 600 dpi, all applied to the raw transcription and re-chunked; each is recorded in a PLATE RE-READ block at the foot of its leaf file.

1. Col 1837 κοπούσαι → κοποῦσαι (leaf970_p971_seg8.png). The transcriber flagged the word as unparsable and was right to; the fault was the accent, not the letters. The plate prints a circumflex over the ου, which makes it an ordinary Byzantine 2 sg. mid./pass. of κοπόω — "you are worn out", which is what the Latin laborabis valde says. Rendered normally; no marker.
2. Col 1837 ἐπιρρίπτειν → ἐπιῤῥίπτειν (same crop). This fount prints breathings on the double rho and prints them here.
3. Col 1841 ἐπιρρίψωμεν → ἐπιῤῥίψωμεν (leaf972_p973_seg2.png).

⭐ Item 3 closes a standing open flag carried in next-session-resume.md ("leaf972 ἐπιρρίψωμεν breathing on the ῥ, against the clearly-printed ἐπίῤῥιψον two lines later — do not harmonise them"). They do not need harmonising: the plate prints both with the breathings, and always did. The apparent divergence was ours. leaf970's ἐπιῤῥίπτειν, corrected in the same pass, is the third instance and settles the fount's habit. Nothing was harmonised; the plate was read.

Rendered as printed — Pattern 7, no emendation

• Col 1841, Ep. VIII: the missing οὐχ. Migne prints παρακληθῆναι· καινὸν οὐρανὸν, καὶ καινὴν γῆν ἐὰν ποιήσῃ, εὑρίσκει ἀνάπαυσιν, and the Latin column prints novum cœlum et novam terram si faciet, inveniet requiem. Both columns print the affirmative, and the sentence contradicts itself: the protasis is a man who is not worthy of rest or for whom consolation is not expedient, and the apodosis grants him rest. The Greek transmitted elsewhere reads οὐχ εὑρίσκει ἀνάπαυσιν, which is certainly what is meant. Rendered as printed — "he finds rest" — and not repaired. This is the clearest kind of plate defect the edition exists to show, and the concurrence of the two independent witnesses (our plate read, and archive.org's OCR of the facing Latin) is what lets it be attributed to Migne rather than to us. No [sic:]: the words are ordinary Greek, correctly rendered; only the sense is broken, and Pattern 12 marks carried type, not a wrong reading.
• Col 1840 εὑρίθησεν for εὑρέθησαν/εὑρέθη. The transcriber verified at high zoom that the vowel after ρ is an accented iota, not an epsilon. The word is transparently the aorist passive of εὑρίσκω with a vowel misprint, so it is rendered as ordinary English ("are found") under Pattern 7's real-word-merely- wrong boundary; no [sic:]. (The singular ending is not a defect — πράγματα is a neuter plural and takes a singular verb.)
• Col 1841 ἀξιοῦσαι (Ep. V, ὅτι ὅλως ἀξιοῦσαι). Flagged by the transcriber as not parsing cleanly; ἀξιοῦσθαι would be expected. Rendered "you are counted worthy", which is the sense the form points at and what the Latin dignus reputaris gives. Recorded, not emended.
• Col 1841, Ep. V punctuation. The plate sets πειρασμοὺς εὑρίσκῃς· καὶ θλίψεις, — a middle point before "and afflictions", where the sense wants the two nouns joined. Carried through (Pattern 8): "you find temptations; and afflictions, but endure them undisturbed".
• Col 1837 συν-ετὸς is a line-break hyphenation in the plate at an unusual point, not a word division defect; the chunker rejoins it. No action.

[var:] — two

1. Col 1840, Ep. II, Acts 14:22. Migne's Greek gives εἰς τὴν βασιλείαν τῶν οὐρανῶν; the received text reads τοῦ Θεοῦ. Marked. Note this is the same divergence, in the same volume-neighbourhood, as the one already marked in antiochus-epistula-ad-eustathium (PG 89, col 1425) — two Palestinian monastic authors quoting Acts 14:22 with "of the heavens". Worth watching as a possible transmission fact rather than two independent slips; not asserted here.
2. Col 1840, Ep. IV, Phil. 4:7. Migne's Greek gives ἡ εἰρήνη τοῦ Χριστοῦ … φρουρήσει τὴν καρδίαν ὑμῶν; the received text has τοῦ Θεοῦ and the plural τὰς καρδίας ὑμῶν. The Latin column agrees with the Greek (Pax Christi … custodiet corda vestra) on the divine name, so this is not our OCR. Both points marked in the one note.

[lat:] — none fired, and four candidates declined

Allatius translates freely and Pattern 16 says looseness is not divergence. Each of these was looked at and let go; recorded so a later reader does not re-open them as discoveries.

1. Ep. I, λογισμῶν δαιμονικῶν ‖ diaboli insidiæ. "Demonic thoughts" vs "the devil's snares", and the Greek's active infinitives against the Latin's passive solent superari. Free rendering of the same claim.
2. Ep. III close, Ὁ Θεὸς μετὰ σοῦ ‖ Dominus sit tecum; Ep. VII, Ὁ Κύριος βοηθῆσαι σοι ‖ Deus adjuvet te. The columns swap God and Lord, in both directions, in the two formulaic closings — which is what makes it look like Allatius's formulary habit rather than a divergent exemplar. A reader of either column takes away the same fact. Declined; recorded because the pattern (not the instance) would be worth re-checking if the same swap turns up in the Doctrinae.
3. Ep. IV opening, Πεῖσον τὴν καρδίαν σου ("persuade your heart") ‖ Vince animum tuum ("conquer your mind"). Different verb, same instruction.
4. Ep. VIII, ὁμολογούμενον δικαίωμά ἐστιν ‖ aperte justificari vis. The Greek is a flat verbless predication ("it is an avowed claim of right"); the Latin supplies a second person and a verb of wanting ("you plainly wish to justify yourself"). Compression/expansion, not a different fact — but this is the loosest of the four and the one to revisit if the Doctrinae give more evidence about how far Allatius rewrites.

Notes — placement

All 11 notes are Migne's, keyed in his Latin column, and are placed in the English at the close of the quotation each cites, per APPARATUS.md. The Greek carries no marker anywhere in this work and none was added to it. Series l–q (Latin col 1839, serving Greek col 1840) and r–v (Latin col 1842, serving Greek col 1841) are both complete and unbroken.

⚠ Note u (Matth. vi, 34.) rests on a moderate-confidence letter-reading — the in-text superscript at Ep. VII's de crastino prints as a small blot, and the transcriber said plainly that the sequence and the quoted content, not the glyph, are what made it a u. The letter is confirmed at 5× per APPARATUS.md; the placement follows the citation, which is the method the apparatus is built on. Unchanged, recorded.

Blind polarity read — 2026-08-08, TWO independent readers, reconciled

Runbook 4a. Two Opus readers, launched together, each blind to this file, to next-session-resume.md, to raw/…/out/ and to git history; one on the negation/scope lens, one on quotation fidelity and markers. 10 sites, 7 ours. Everything below was formed blind and reconciled against this file afterwards, never before.

Both readers independently confirmed the Ep. VIII missing οὐχ (above), one of them at the plate, and both reached the same verdict the translation had: Migne's, both columns concurring, English correct as printed. One sharpening worth keeping — the concurrence adjudicates the ATTRIBUTION, not the CONJECTURE: both columns descend from the same defect, so they cannot testify that οὐχ εὑρίσκει is what was meant, only that the affirmative is what Migne printed.

Repaired — ours

1. Col 1840D, Ep. V — our OCR invented an ano teleia. The chunk read πειρασμοὺς εὑρίσκῃς· καὶ θλίψεις. The stop is not on the plate: at 6× the mark is the word's own final sigma, glyph-identical to the ς of πειρασμοὺς two words earlier on the same line, and the Latin prints one pair (tentationes habes et afflictiones). καὶ θλίψεις is the second object of εὑρίσκῃς inside the ἐὰν clause; the invented stop detached it, and the English had faithfully preserved the break under Pattern 8, as though it were the plate's own punctuation. Transcription corrected, re-chunked, English rejoined. ⚑ This is a defect class the corpus had not met: our OCR inserting Greek PUNCTUATION the plate does not print. It is invisible to every verifier, and Pattern 8 actively protects it — a conscientious translator will carry it through. This fount's final sigma is a middle-point lookalike; that belongs in the fleet's transcriber prompt. ⚠ The two readers disagreed here: the second called the mark ambiguous at 3× and leaned to "printed". The higher-resolution look decides it, and the disagreement is recorded because it is the useful part.
2. Col 1840B, Ep. II — an Allatius leak. The plate prints τοιῶνδε once; the English read "as of this sort or of that", reproducing the Latin's paired per has aut illas. Both readers caught it independently; plate-verified. Now "as of such a sort". The Latin's doublet may be evidence that the plate lost a second τοιῶνδε by haplography — which is a note, not a licence to restore it.
3. Col 1840, Ep. I, Eph 3:20 — 7a″ lemma conformation. Migne prints ἐκ περισσοῦ; the English gave "above and beyond", importing the ὑπέρ of the received ὑπὲρ πάντα / AV "exceeding abundantly above" — a word not on the page. Now "more abundantly than", which renders the comparative force the genitive ὧν carries and nothing more. The tell was exactly the one 7a″ names: a familiar cadence arriving ahead of the printed words.
4. Col 1840, Ep. I — tense. εὑρίθησεν is an aorist-passive form however its disputed letters are read; the English had "are found", pulled to the present by the Latin's inveniuntur. Now "were found".
5. Two [var:] markers were in the wrong house form. Pattern 14 requires the content to open by naming the witness (Vulg. · LXX · Gk · Heb. · Douay · Vet. Lat.); both opened "the received text of … reads". Re-formed.
6. Col 1841C — an em-dash for an ano teleia (Pattern 8), where the same mark is a semicolon elsewhere in the same file. Now a semicolon.

Added

1. [var:] at Ezek 33:11 (col 1841, Ep. VI). Migne prints τοῦ ἁμαρτωλοῦ for the LXX's τοῦ ἀσεβοῦς, and omits the whole clause ἀποστρέψαι τὸν ἀσεβῆ ἀπὸ τῆς ὁδοῦ αὐτοῦ. Material, and it was the best-known verse in the work going out unmarked while two lesser divergences carried markers.
2. [var:] at Eph 3:20, alongside repair 3 — no δυναμένῳ, no ὑπὲρ πάντα, ἐκ περισσοῦ for ὑπερεκπερισσοῦ.
3. ⭐ [lat:] at the close of Ep. VII (Greek 1841 ‖ Latin 1842) — the work's only fired Pattern 16 marker, and BOTH readers arrived at it independently after I had declined it. Greek μὴ ἔχοντας θάρσος περὶ τῆς αὔριον ζωῆς = having no confidence of living to see tomorrow; Latin ne habeamus curam vitæ crastinæ = having no care for tomorrow. The Greek grounds detachment in mortality; the Latin assimilates the clause to Matt 6:34 and makes it a warning against anxiety. A reader of the Latin column takes away the opposite counsel from a reader of the Greek — a different fact, not Allatian looseness, which is precisely Pattern 16's test. My own first pass had filed this under "free rendering" in the declined list below; two independent readers overturning it is the argument for running the read blind and in duplicate.

⛔ NOT changed, and why — εὑρίθησεν stands

The quotation-lens reader reported the plate as printing εὑρέθησαν (epsilon, -σαν) and recommended correcting both the Greek chunk and the English. Declined. The original transcriber recorded an explicit high-zoom examination asserting an accented iota and the ending -σεν; the reader's look was at 2×; and re-examination at 8× finds the ink of that word merged past legibility at 600 dpi, in both the vowel and the ending.

Rulebook 11: the transcriber wins until the plate says otherwise — and here the plate has said nothing. An adjudicator must not prefer the later reader merely because it is later; that is the 7a′ error the adjudication step already committed once on this work, and it would be committing it a second time on weaker evidence. A higher-resolution render is what settles it. The English change that does NOT depend on the disputed letters (the tense, repair 4) was made.

Declined, with reasons

• [var:] on Matt 6:34 (cited twice, Epp. I and VII; infinitive for the prohibitive subjunctive, περὶ τῆς for εἰς τήν, τινας added in VII). The English diverges correctly at both. Declined as authorial citational adaptation, not a divergent reading — the infinitive is governed by εἴρηται in indirect statement. This is the same test the reader itself used to reject a [var:] on Ps 54:23, whose 2 sg → 1 pl shift is proved authorial by Ep. VII quoting the same verse in the LXX's exact form. Applying it to one and not the other would be inconsistent.
• [lat:] on Phil 4:7's singular/plural heart. Allatius conformed to the Vulgate's plural — normal assimilation, and the existing [var:] already tells the reader the received text is plural.
• [var:] on Matt 8:26 (singular ἀνέμῳ, active ποιήσας): the reshaping an allusion requires, and already visible in the English.
• Act. xiv, 21 beside a quotation of v. 22. Plate-verified as 21 in the printed note block. Migne's citation, reproduced verbatim as policy requires.
• Sap. xi, 27: Vulgate versification of LXX Wis 11:26. Not an error.
• Band [b: C] in Ep. VIII sits one to two lines lower on the plate than the chunk places it. Within transcription tolerance for a positional quarter-guide, and not a translation defect. Recorded for a future band pass.

✅ Second person — RULED 2026-08-08, and this work is the worked case

The blind read flagged that this work used archaic second person only in address to God, while the neighbouring antiochus-epistula-ad-eustathium used none at all. Surveying the corpus settled it the other way from where I expected: 52 of 99 works already use thou for a singular addressee of any kind — God, a saint, a bishop, a monk, the reader — and you for a plural.

Ruled and written up as translation-style.md Pattern 17: thou is the SINGULAR, you is the PLURAL. A number marker, not a reverence marker. Greek and Latin have no reverential second person; σύ to God is the same σύ as to a novice, so reserving thou for God invents a distinction the source does not make — the same fault as latinizing a Greek divine name. What the archaism earns its place by is recovering a distinction the sources mark and modern English cannot.

This work is converted throughout. All eight letters address one brother, so the body is now thou — and the one place you survives is the payoff: Ep. IV quotes Phil 4:7, φρουρήσει τὴν καρδίαν ὑμῶν, plural. The English runs thou for nine paragraphs and then "shall guard your heart" inside the quotation, so a reader sees Dorotheus move from the man in front of him to the church, with no apparatus. Under a flat "you" that is invisible; under a flat "thou" it is falsified.

⚠ antiochus-epistula-ad-eustathium is now the known outlier — 21 tokens, every one a singular address to Eustathius rendered "you", and it is shipped and live. Convert before its next deploy, or rule explicitly that it stays.

## Still open

• The genre — terse ascetic letters of spiritual direction — has no PG anchor pair. It was translated against the antiochus-epistula-ad-eustathium precedent (Palestinian monastic prose, the nearest thing the corpus holds) and the general PG register rules, not against a calibrated anchor. If the Doctrinae are to run as a fleet, that is the genre a Fable anchor session should be built on, and these eight letters are the natural material for it.