The Latin Question — the class-A test, run
⚠ SURFACED, NOT ACTED ON. Nothing in CONVENTIONS, no English, and no part of the Prototype B decision has been changed on the strength of this file. It is Wilson's call and it is an introduction-level question, not a translation-level one.
The question
The apparatus preface (PDF 482, apparatus [i]) states:
Hoc apographum Græca tantum continet usque ad pag. 250 nostræ editionis, sine versione Latina, quam Editor ipse confecisse videtur. — "This apograph contains only Greek as far as p. 250 of our edition, without the Latin version, which the Editor himself seems to have made."
If the Latin of printed 1–250 is an editor's Latin rather than something with independent standing, then for that stretch the Prototype B mirror sets Andrewes beside his editor, and "construal witness" means less than it has meant all along.
The test the brief designed
Class A gives apograph-versus-printed-Greek divergences for exactly this stretch. For each one, check which reading the facing Latin supports.
- Latin consistently follows the printed Greek → it is downstream of the edited Greek, and the question is settled: editor's Latin.
- Latin ever supports the apograph against the printed Greek → it has a source independent of the printed Greek, and the preface's videtur is too modest.
The result: the second branch fires.
Ten divergences were resolved against our own transcripts. Five support the printed Greek, two support the apograph, three cannot decide.
| # | class A | printed Greek | apograph reads | facing Latin | supports |
|---|---|---|---|---|---|
| 1 | p. 6. 20 | τοῦ ἁγίου φροντιστηρίου τούτου | τῆς ἁγίας μονῆς ταύτης | hujus sacri Collegii (7) | printed |
| 2 | p. 12. 26 | Τέμενος Βασιλικόν | dele Βασιλικόν | Sacellum Regium (13) | printed |
| 3 | p. 26. 26 | Κύριε ὁ Θεὸς τῆς ἀληθείας | ὁ Θεὸς not in MS | Domine, Deus veritatis (27) | printed |
| 4 | p. 34. 27 | Τὰ ἔργα τῶν χειρῶν σου | τὸ πλάσμα | Figmenta manuum tuarum (35) | APOGRAPH |
| 5 | p. 50. 18 | ἀπὸ τῆς προσδοκίας | ἐλπίδος | ab hac mea exspectatione (51) | printed |
| 6 | p. 100. 7 | Βουλευτὰς | συμβουλευτάς | Consiliarios (101) | apograph (weak) |
| 7 | p. 162. 16 | ἡ ὑπερέχουσα πάντα νοῦν | ὑπερβάλλουσα | quæ exsuperat omnem sensum (163) | inconclusive |
| 8 | p. 170. 16 | τὰς ἀνομίας μου | ἁμαρτίας | iniquitates meas (171) | printed |
| 9 | p. 174. 13 | ἀπόλυσαι | ἀπόλουσαι | Ablue (175) | APOGRAPH |
| 10 | p. 224. 8 | ἡ θυσία ἑσπερινή | δειλινή | Sacrificium vespertinum (225) | inconclusive |
Why 7 and 10 cannot decide: both Latin words are Vulgate lemmata for the verses being quoted — exsuperat is the Vulgate's own word at Phil. iv. 7, and sacrificium vespertinum is the Vulgate's at Ps. cxl. 2. A translator would write them whichever Greek stood in front of him. (10 is further spoiled because the Latin renders the heading ΟΜΟΛΟΓΗΣΙΣ ΔΕΙΛΙΝΗ as CONFESSIO VESPERTINA on the same spread — so vespertinus covers both Greek words in this translator's usage.)
⚠ Case 4 is the one that matters, and it is very hard to explain away
Τὰ ἔργα τῶν χειρῶν σου μὴ παρίδῃς is Ps. cxxxviii. 8, and the Vulgate of that verse reads opera manuum tuarum ne despicias.
So a Latin made from the printed Greek would read opera. A Latin made by a translator who did not know the Greek and simply reached for his Vulgate would also read opera. A translator silently repairing a corruption by sense would reach for opera too, because that is the phrase every reader knows.
The 1853 Latin reads Figmenta.
Figmentum is the standard Latin for πλάσμα — the thing moulded — and it is exactly what the apograph reads. The Latin has departed from the printed Greek and from the Vulgate and from the reader's expectation, and landed on the manuscript. There is no route to Figmenta that does not pass through a Greek text reading πλάσμα.
Case 9 is the corroboration. ἀπόλυσαι ("release") stands in a column of cleansing verbs — Tege · Ablue · Dele · Purga answering κάλυψον · ἀπόλυσαι · ἐξάλειψον · καθάρισον, all under the head Τὸ ῥυπαρὸν / Macula, the stain. Ablue is "wash away" and cannot come from ἀπόλυσαι; it renders the apograph's ἀπόλουσαι, which differs by a single letter and is manifestly the true reading. This one could be a competent translator repairing an obvious corruption by sense — but it repairs it to precisely the manuscript's letter.
What this does and does not establish
Does: the 1853 Latin of printed 1–250 is not purely downstream of the printed Greek. At least once it witnesses a Greek reading found in the manuscript, absent from the printed text, and absent from the Vulgate. The preface's videtur is too modest, exactly as the brief anticipated it might be.
Does not: it does not show the Latin is Andrewes'. Five of ten cases follow the printed Greek, which is what one expects if the Latin and the printed Greek largely share a source. The likeliest reading of the whole pattern is that the Latin was made from a Greek text closer to the apograph than the printed text is — and that where the two now differ, the printed edition, not the Latin, is the one that has moved.
Still open, and not to be flattened:
- Which Editor? The brief's argument that Editor most naturally means the 1675 Sheldonian editor (contrasted with nostra editio = 1853) still stands and is untouched by this test.
- ⚠ The class A / class C collisions (see
CLASS-A-ledger.md§4) are a real complication and are not resolved here. At several lines — pp. 34. 27, 50. 18, 104. 3, 116. 9, 162. 16, 170. 16, 190. 13, 224. 8 — a class-A reading and a Drake marginal address the same words, and the direction of class C's]cannot be settled from the printed apparatus alone. Do not build an argument on the collisions until the manuscript itself has been seen.
What changes
Almost nothing, and that was true before the test. CONVENTIONS §6 already makes the Greek primary and the Latin the first witness to construal — a role this result strengthens rather than weakens. Prototype B reproduces the 1675 book, which is what it always claimed to do. The project's framing — "Greek original, historic Latin parallel, fresh English translation" — is accurate under any answer.
What genuinely changes is the introduction, and there it is now an asset: the edition can say, with a worked example, that its Latin column is an independent witness and not a shadow of its Greek. That is a stronger claim than the project has ever been able to make, and it is now evidenced.
✅ RE-CHECKED AND CONFIRMED 2026-08-10 — the argument's one load-bearing word is secure. Case 4 (Figmenta / τὸ πλάσμα, printed 34/35) was on the pre-print list because the whole case rests on reading a single word correctly. The apparatus entry was re-read on the University of Toronto scan (leaf 0480, apparatus ii): p. 34. lin. 27. pro τὰ ἔργα lege τὸ πλάσμα et dele punctum post παριδῇς, et in lin. 28. lege ὁμοίωμα puncto adjecto. — unambiguous at native resolution, and our transcript matches it exactly. The reading may now be printed in an introduction. Off the pre-print list.