Brightman 1903 as a collation witness — pilot, method, and the anchor table
Pilot run 2026-08-04. Five probes, four hits, one crux solved. Nothing applied.
Source: archive.org/details/bwb_KP-665-232 — F. E. Brightman, The Preces Privatae of Lancelot Andrewes, Methuen, London, 1903, 472 pp. Full text public. Working copy of the OCR: bwb_KP-665-232_hocr_searchtext.txt.gz (the _djvu.txt endpoint 500s; the hOCR searchtext gz works).
1. What Brightman is, and is not, as a witness
He is NOT a witness to the Greek or Latin text. He prints none of it — "no attempt has been made to mark the distinction between the Greek and the Latin of the original" — confirmed mechanically: zero Greek runs and zero Hebrew runs in 755 KB of OCR across all 472 pages. He cannot correct a transcription.
He IS a strong witness to four things we actually need:
- Construal. An independent English of the same book. Where his reading and ours diverge, one of us has misread.
- Scripture references. Marginal refs throughout, with
heb./sept./vulg.added where the original departs from the English Bible. ⚠ This bears directly on our worst problem — the volume's dense bad-reference stretches, and the three places where it proves its own misprints. - Non-scriptural sources. He identifies them, and his index names the edition Andrewes used. This is what the pre-print attribution questions need.
- The manuscripts. He collates O (Oxford 1675) against H (Harleian), L (Laudian), W/Wy, and S (Stokes). Our Part III is the Harley recension, so H-vs-O variants in his notes speak directly to Part III against Part II.
2. The pilot — four findings
✅ 2.1 §35 Bradwardine — confirmed, and sourced exactly
Our §35 (printed 370) identified Bradwardine. Brightman's note: "P. 192. Thomas Bradwardine was archbishop of Canterbury for five weeks in 1349. His work de causa Dei contra Pelagium et de virtute causarum, from which the present passage is taken…" — with the marginal ref T. Bradwardine de virtute causarum ii cor. 39 and, in his index, the edition: Saville, London 1618.
Gain: the work, the exact corollary, and the edition Andrewes had.
✅ 2.2 §39's Fulgentius attribution — CORROBORATED; our Augustine doubt answered
Our notes flagged the 1853's ex Fulgentii l. i. ad Mon. as UNVERIFIED and "looks like Augustine" (De Doctrina Christiana IV), and kept it as printed. Brightman, working independently from the sources, also gives Fulgentius — index: Fulgentius of Ruspe, S. (Migne Patr. lat. lxv.), 257, and a note at his p. 257. 9 identifying him as bishop of Ruspe, 508–533. His introduction paraphrases the passage as ours reads it — "it is rather by the piety of his prayers for himself and his flock than by fluency of speech…"
Gain: a pre-print item moves from doubted to independently corroborated. (PL 65 is Fulgentius's volume and contains Ad Monimum.) The transcript is unchanged either way; what changes is the note.
✅✅ 2.3 Part III §1's mundam — confirmed as a HARLEIAN-vs-OXFORD variant
Our Part III §1 finding recorded, among five substantive divergences from §23, that the faith-catalogue adds mundam and drops vincentem mundum. Brightman's note:
P. 189. 6. H reads mundam 'pure'; O, vincentem mundum.
This is an independent confirmation of one of our own five divergences — and it names the mechanism. H = Harleian, O = Oxford 1675. ⚠ It corroborates the structural claim that Part III is the Harley recension, from a witness who had the Harleian manuscript itself.
✅✅ 2.4 THE Isaac CRUX AT PRINTED 425 IS SOLVED — the word is Mosis
Our §4 note (printed 425): "Cantemus Domino Isaac et Israelitarum (Exod. xv. 2) — Isaac has no part in the song at the Red Sea and no construal makes him fit. Translated straight, difficulty left visible, on the pre-print list."
Brightman's text, in the same Benedictus-catena and in the same position:
"Blessed be the Lord (of Jethro) Ex. xviii 10" "I will sing unto the Lord (of Moses and the Israelites) Ex. xv 1"
The catena tags each text with whose blessing it is, and the reading is Mosis et Israelitarum — Exod. xv. 1, "Then sang Moses and the children of Israel this song unto the LORD." ⚠ Isaac is an error in the printed line, and the reference is off by one as well (xv. 1, not xv. 2).
Kept as printed in the transcript, per standing discipline. What changes is that the English's note can now say what the reading must be instead of leaving the difficulty merely visible. The crux comes off the pre-print list.
⊘ 2.5 The garbled Hebrew at printed 386 — NOT testable from the OCR
Brightman marks Hebrew by printing its translation in italic (preface §4). ⚠ Italics do not survive OCR into plain text. So חזתלת → תוחלת cannot be checked from the text dump; it needs his page images. Recorded as a method limit, not a negative result.
But the same probe paid anyway. His p. 106 is our §41 (printed 386), and his margin supplies the source identifications our transcript lacked: Cic. de senect. 11 (Cato's daily audit) · Auson. Idyll. · Ps. lxxvii 6 · Cp. Lucian Hermotimus 64 · S. Aug. Serm. XX 2 · Ecclus. xxiii 2 · 1 Cor. xi 31 · S. Greg. Nyss. de orat. dom. 1.
✅ THE SENECA DIVERGENCE IS CLOSED (Wilson located it, 2026-08-04). Seneca. is printed in the plate at printed 385, and the attribution is correct: sentence 1, Deprehendas te oportet, antequam emendes, is Seneca, Ep. Mor. 28. 10. So the plate is right and Brightman's margin is simply short an authority — his silence never indicated an error, and nothing here was ever ours to correct. ⚠ Seneca's own lead-in credits «Initium est salutis notitia peccati» to EPICURUS, which makes the §41 catena's chain Epicurus → Seneca → Andrewes — an introduction asset. ⚠⚠ Sentence 2 (Ignotum ulcus in deterius abit, et curatione destituitur) is not in Ep. 28 and, on Wilson's search, occurs ONLY IN ANDREWES — so the one *Seneca.* tag covers two sentences of which only the first is Seneca's, the second amplifying his metaphor (the sore never found festers and goes uncured) and quite possibly being Andrewes' own. Stated at the strength the evidence bears: unattested elsewhere, NOT "Andrewes' own" — a florilegium or commonplace source is not excluded. The tag is kept as printed (§39 Fulgentius rule: an over-reaching attribution is flagged, never trimmed to fit); the note carries what the tag does not.
3. Method — why a linear pass is the wrong shape
Brightman rearranged the book on principle (preface §3), and his notes and index are keyed to his own pagination. So there is no mechanical crosswalk from his page to our printed page. The anchors below make the point:
| Brightman p. | our printed p. | section |
|---|---|---|
| 106 | 386 | Part II §41 (Monita / evening examination) |
| 189 | ~403 | Part III §1 (Confessio Fidei) |
| 192 | 370 | Part II §35 (Bradwardine) |
| 257 | 380 | Part II §39 (Ante Concionem, Fulgentius) |
His 189 < 192 < 257 maps to our 403 > 370 < 380. The order is genuinely scrambled; a page-by-page collation would mean building a 472-page crosswalk first.
The right shape is a QUESTION-DRIVEN pass. Every hit above came from grepping a proper name or a distinctive phrase — cheap, and aimed at items that already matter. Run our existing lists through him:
- the pre-print verification list
- the running list of wrong and doubtful references (Part III §2 alone has 4 demonstrably wrong + 6 doubtful — the densest in the volume)
- the attributions (§39 done; §35 done)
- the cruces (425 done;
p. 207;Tuâ.at 421; the unnumbered witness at 423) - the five plate defects in the watch list — ⚠ especially printed 370/371, where a heading and an opening are MISSING from the 1853. Brightman had the manuscripts; he may simply have the missing text.
- class A's H/L variants — his notes are the second opinion on the Wright apograph's readings, from different manuscripts.
Accumulate the anchor table as a by-product. Each lookup fixes one (his page ↔ our printed page) pair at zero marginal cost. If enough structure emerges, a fuller pass becomes cheap later. If it never does, we lost nothing.
⚠ Two standing limits:
- Anything depending on italics — i.e. ALL of the Hebrew — needs his page images, not the OCR. This folds into the 600-dpi digitization job that already gates both books.
- Brightman is a witness, not an authority. He is emending from manuscripts we have not seen, and he says himself that where he adds words to complete a sense he brackets them. The transcripts still record the 1853 plate; nothing here writes back to them.
3a. ✅ THE FLAG-LIST PASS — run 2026-08-04 on Opus, Wilson's go.
⚠ The old header here read "Nothing applied." That was true on the day and is no longer true (corrected 2026-08-09). Every finding below has since been carried into the relevant section's English flags — §23, §26, §35, §41, Part III §1 and §4 all state their results in place. What is still true, and is the rule: nothing was applied to a TRANSCRIPT. The 1853 plate stands as printed everywhere; the repairs live in the English layer's notes.
⚠⚠ The method finding, which is worth more than any single repair
THE VOLUME REPAIRS ITS OWN DEFECTS ACROSS THE RECENSION BOUNDARY. Part III is the Harley recension of Part II's penitential matter (§1 : §23+§38 :: §2 : §17 :: §4 : §24). So where Part II's plate is defective, Part III may carry the same passage intact — a free, internal, entirely unused repair route. It resolved watch item #4 below before Brightman was even needed. Check the recension parallel FIRST on any Part II defect; it costs nothing.
⚠ And the hard limit on Brightman
He prints no Greek and no Latin, so he can NEVER repair a lost Latin or Greek word — only a reference, an attribution, a construal, or a sense. Watch items 1 (printed 239, cut-off Latin recto) and 2 (printed 210, ink damage) therefore stay open for the digitization regardless, and no amount of collation will close them. ⚠ His headings are HIS OWN — preface §3: "For titles, etc., which are printed in thick type… I am responsible." Never import a Brightman heading as recovered text.
Watch items
| # | item | outcome |
|---|---|---|
| 3 | printed 370/371 — "a heading AND an opening are missing" | ⚠ DOWNGRADED. Brightman's litany also begins at "For the wounded in spirit" — so the opening is not lost; the litany does start there. What is missing is a heading — and his ("FOR THOSE IN AFFLICTION AND PERIL") is his own and cannot be used. Restate the watch item as heading only. |
| 4 | printed 339/340 §23 — missing antecedent + wanted non tantum | ✅✅ RESOLVED, from TWO independent witnesses that agree verbatim. Our own Part III §1 reads 3. In Spiritum Sanctum, … Extrinsecus et invisibiliter, sed / Efficaciter et manifeste in / Nos operantem, / Illuminatione justitiæ, / Infusione gratiæ, / In Reprehensione, Doctrina, Tolerantia, Subventione…; Brightman renders the same, "in the Holy Ghost… from without and invisibly, but effectuously and manifestly operating upon us by illumination of righteousness, infusion of grace, in reproof, teaching, bearing with, help, witnessing with." The lost antecedent is the Holy Ghost (our conjecture, confirmed) and the lost hinge is Extrinsecus et invisibiliter, sed — not a non tantum. ⚠ in Nos operantem is accusative: our supplied [in us] should be [upon us]. Recommended correction, NOT applied. |
| 5 | printed 353 §26 V.7 — omnia m broken mid-word, book name lost | ✅✅ CONFIRMED EXACTLY. Brightman: "Send not a cruel messenger unto us: Prov. xvii 11 / but let all evils be put far away from our houses. 13" — our reconstruction omnia m[ala. Prov.] xvii. 11, 13 is right in every part. |
| 1, 2 | printed 239, printed 210 | ⊘ Not closable by Brightman — see the hard limit above. Digitization only. |
| 0 | printed 386 garbled Hebrew | ⊘ Needs his page images; italics do not survive OCR. |
References and attributions
- ✅ §28
[Dan. vi. 4.]→Dan. i. 8. Brightman: "If not wholly from all; Dan. i 8 — yet from dainties?" Daniel refusing the king's meat. Identified; kept as printed. - ✅ §28
tricesimam partem— CONFIRMED as the reading, with Deut. xxvi. 12, in a second witness who also declined to mend it. Our note said the choice between Andrewes' own reduction and a corruption of tertiam "cannot be settled from this witness." It is now settled: it is the text. The oddity is Andrewes'. - ✅ Part III §2
Jo. ix. 30→ 31 CONFIRMED. Brightman: "But I am a sinner: S. Jo. ix 31 and God heareth not sinners." - ✅ Printed 425
Isaac→Mosis, Exod. xv. 1 (§2.4 above). - ✅ §39 Fulgentius corroborated; ✅ §35 Bradwardine sourced exactly (§2.1–2.2 above).
- ⊘ No hit: Part III §2's
Marc. iv. 28and1 Cor. xi. 30— a phrasing mismatch, not a negative result. Retry with our English wording.
✅ Printed 398 — the epigraph leaf, previously carrying NO sources at all, is now sourced
| our line | Brightman's identification |
|---|---|
ORATIONIS fundamentum et basis est fides. *Theophylact.* | Theophylact, in S. Lucam xviii (ed. i. 433): πάσης προσευχῆς βάθρον καὶ κρηπὶς ἡ πίστις — βάθρον καὶ κρηπίς = fundamentum et basis, exact. |
Sursum corda. | first in the Hippolytean canons 3 (2nd–3rd cent.) and Cyprian de Orat. dom. 31 |
Brightman also notes a lost wordplay in this neighbourhood — fundit / fundatur, closest parallel Aug. serm. CXV 1, fides fundit orationem, fusa oratio fidei impetrat firmitatem.
✅ Printed 386 (§41) — the authorities catena, sourced
Cic. de senect. 11 (Cato's audit) · Auson. Idyll. · Ps. lxxvii 6 · Cp. Lucian Hermotimus 64 · S. Aug. Serm. XX 2 · Ecclus. xxiii 2 · 1 Cor. xi 31 · S. Greg. Nyss. de orat. dom. 1.
✅ THE SENECA DIVERGENCE IS CLOSED (Wilson located it, 2026-08-04). Seneca. is printed in the plate at printed 385, and the attribution is correct: sentence 1, Deprehendas te oportet, antequam emendes, is Seneca, Ep. Mor. 28. 10. So the plate is right and Brightman's margin is simply short an authority — his silence never indicated an error, and nothing here was ever ours to correct. ⚠ Seneca's own lead-in credits «Initium est salutis notitia peccati» to EPICURUS, which makes the §41 catena's chain Epicurus → Seneca → Andrewes — an introduction asset. ⚠⚠ Sentence 2 (Ignotum ulcus in deterius abit, et curatione destituitur) is not in Ep. 28 and, on Wilson's search, occurs ONLY IN ANDREWES — so the one *Seneca.* tag covers two sentences of which only the first is Seneca's, the second amplifying his metaphor (the sore never found festers and goes uncured) and quite possibly being Andrewes' own. Stated at the strength the evidence bears: unattested elsewhere, NOT "Andrewes' own" — a florilegium or commonplace source is not excluded. The tag is kept as printed (§39 Fulgentius rule: an over-reaching attribution is flagged, never trimmed to fit); the note carries what the tag does not.
✅ Two of our Part III §1 divergences independently confirmed
mundam— BrightmanP. 189. 6: H reads mundam; O, vincentem mundum(§2.3).- FIVE Spirit-operations, no Consolatione — Brightman's Part III text likewise gives reproof, teaching, bearing with, help, witnessing with. Our note recorded five against §23's six; confirmed.
✅ A class-B boundary corroborated
Class B marks two blank pages after printed 190. 4, at the close of προαιρέσεις, ἐπιχειρήσεις, ἔξοδον καὶ εἴσοδον, κάθεδραν καὶ ἔγερσιν. Brightman sets exactly that text as a discrete titled unit with the siglum L 134 — "I commend unto Thee, o Lord, impulses, occasions, purposes, endeavours, going out and coming in, downsitting and uprising." Independent corroboration of the class-B unit boundary at 190.4, from the Laudian MS rather than the Wright apograph.
Anchor table, extended
| Brightman p. | our printed p. | section |
|---|---|---|
| 106 | 386 | II §41 |
| ~131 | 357 | II §28 (the alms ladder) |
| ~147 | 409 ff. | III §2 |
| 189 | ~403 | III §1 |
| 192 | 370 | II §35 |
| ~196 | ~402 | III §1 (Creed, art. ii–iv) |
| 257 | 380 | II §39 |
| 289–290 (notes) | 398 | III §1 epigraph leaf |
| 337 | 371 | II §36 |
3b. ✅ THE RETRY PASS — 2026-08-09. Part III §2's reference list is now fully adjudicated: 8 of 8.
The 2026-08-04 pass closed on two no-hits and left six references "doubtful and unadjudicated." All eight are now settled or improved, from a single method change.
⚠ THE METHOD FINDING — search him by OUR ENGLISH, not by the Latin
Both 2026-08-04 no-hits were failures of the query, not of the witness. Brightman prints no Latin, so a Latin phrase can never match him; and his English is his own, so it will not match ours word for word either. What does match is the SCRIPTURE WORDING, because both of us are quoting the same English Bible — grep the AV/BCP phrase the line is built on ("carest thou not that we perish", "known in the dark", "for mine own sake") and he surfaces immediately, margin and all. Once one line of a section is located, his whole rendering of that section is contiguous and every neighbouring reference comes free. Our printed 409–416 is his pp. 146–149, a dense line-by-line parallel. This is much cheaper than the 2026-08-04 pass and should be the default shape for any further lookup.
| our reference, as printed | status before | status now |
|---|---|---|
[Marc. iv. 28.] (410) | no hit | ✅ Mark iv. 38 confirmed, three times |
1 Cor. xi. 30. (413) | no hit | ✅ 1 Cor. xi. 31 confirmed |
*Psal.* xl. 9. (409) | doubtful; guessed Ps. xxxi. 10 | ⚠⚠ REVERSED → Ps. lxxviii. 33 |
*Job.* xxx. 33. (412) | doubtful | ✅ Job xxxi. 33 confirmed |
liii. 13. (415) | doubtful | ✅ Ps. lxxxviii. 12 confirmed |
*Is.* xliii. 24. (415) | doubtful | ✅ Isaiah xliii. 25 confirmed |
*Num.* xiv. 7. (416) | doubtful | ✅ Num. xiv. 17–19 corroborated |
*Job.* xxv. 3. (414) | doubtful | ⚠ improved, NOT closed — needs page image |
Every one is kept as printed. The transcripts are untouched; the notes now name verses instead of raising questions.
⚠⚠ The one real correction: we were wrong at printed 409
*Psal.* xl. 9. heads Defecerunt in vanitate dies mei, et anni in anxietate cordis, and our note proposed Ps. xxxi. 10 as "closer." It is Ps. lxxviii. 33 — Brightman cites it twice in unrelated parts of his book, and the Latin proves it without him: Defecerunt in vanitate dies is verbatim Vulgate Ps. lxxvii. 33, and anni in anxietate renders that verse's own second half. Ps. xxxi. 10 supplies neither half. ⚠ A conjecture that reads plausibly is not a collation; this one survived two passes before a witness was actually asked.
✅ THE ONE THAT RESISTED — CLOSED 2026-08-17: *Job.* xxv. 3. at printed 414
Settled, and the page image was got. Brightman's margin, read on the plate (archive.org/details/bwb_KP-665-232, printed p. 148 = leaf n213; the image endpoint is archive.org/download/bwb_KP-665-232/page/nNNN.jpg, and printed page = n − 65 for this scan), reads Job xiv 4 vulg., iv 18. The OCR's av 18 was iv 18; there is no third figure, and he gives no reference at all for the stars clause. ⚠ His vulg. label disagrees with his own note at his p. 143. 3 sq., which quotes Andrewes' Pentecost sermon: infans unius diei super terram, as the Seventy read it (Job xiv 4). The substance is settled without him and against the label — the Clementine Vulgate of Job xiv. 4–5 has no day-old child at all (verified: Quis potest facere mundum de immundo conceptum semine? … Breves dies hominis sunt), while the LXX does (ἐὰν καὶ μία ἡμέρα ὁ βίος αὐτοῦ ἐπὶ τῆς γῆς), and Augustine gives it in that Latin at Conf. I. vii. 11. So the cento is Job xiv. 4 (Sept.) + xxv. 5 (Vulg.) + iv. 18 (not Vulg., which reads reperit pravitatem; the plate's insipientiam is the Hebrew's word, as AV). Brightman's bare iv 18, on his own version-flag convention, is itself the witness that the last clause follows the English Bible. Note applied at 414; nothing in any transcript touched.
A by-product: a third witness to the BCP Psalter
Brightman numbers "He considered that we are but flesh… a wind that passeth away" as Ps. lxxviii. 40 — in two margins and in his own printed psalm text — where the AV has 39. That is the Prayer Book Psalter's numbering, one ahead through this psalm. It joins printed 433's xlii. 10 and lxv. 8 as evidence that the BCP was the book physically to hand, and it strengthens CONVENTIONS §9's Coverdale-first rule. It does not unsettle our own note at 411, which reads the plate's lxviii. 39 on AV numbering with a dropped x.
⚠ And a plate pattern worth recording: Part III §2 misnumbers Psalm lxxviii twice in seven pages — at 409 (xl. 9) and 411 (lxviii. 39). Two corruptions of one psalm reads as a compositor's hand rather than an author's numbering, which is a small independent support for the standing "kept as printed, flagged, never mended" discipline.
4. Verdict on Wilson's question
Yes — but question-driven, not linear. Four hits from five probes, including one solved crux and one independent confirmation of a Part III structural finding, is a decisive pilot. A full linear collation is the wrong instrument and would cost many times more for less.
✅ The flag-list pass was RUN on Opus, 2026-08-04, on Wilson's go — see §3a. Three of the five watch items resolved or downgraded, four wrong references identified or confirmed, two Part III §1 divergences independently corroborated, one class-B boundary corroborated, and two previously unsourced pages (398, 386) sourced. Nothing applied.
Still open after BOTH passes (updated 2026-08-09): printed 210 and 239 (lost Latin/Greek words — digitization only, Brightman structurally cannot help) · printed 386's Hebrew (needs his page images) · printed 414's Job. xxv. 3 cento (same blocker — his page image) · p. 207 (an apparatus crux; Brightman cannot speak to it at all) · printed 370/371's missing heading, which he has but which is his own and unusable.
✅ Closed since: §41's Seneca (sentence 1 at Ep. Mor. 28. 10; sentence 2 unattested outside Andrewes) · all eight of Part III §2's open references (§3b).
⚠ What the residue has in common: it is now almost entirely a SOURCE-IMAGE problem, not a collation problem. Four of the five remaining items are blocked on the same thing — a non-Google digitization, plus Brightman's own page images for anything carried in italics. Further text-only lookups have low expected yield; the next real gain comes from images.
Related: CLASS-A-ledger.md · CLASS-B-unit-boundaries.md · THE-LATIN-QUESTION.md · ../EDITION-SHAPE.md §4a.