These are the edition's working notes, published as they were written. They record, work by work, where Migne's plate is defective or the reading uncertain: the printed form, what the English does with it, and why. They are not a groomed apparatus. Where a note argues with itself, withdraws an earlier decision, or leaves a question open, that is the record of how the reading was actually reached, and it is left standing — a conjecture that is visible can be checked, and one that has been tidied away cannot.
The English never silently repairs the page. A construable but wrong reading is rendered literally and the conjecture is recorded here, never applied to the text. How the English is produced and checked →
Translated 2026-08-23, one translator, 7 chunks. Repaired 2026-08-23 after an independent blind read and a plate read at 900–1400 dpi; the work was re-chunked when data/tei-patches/8992.json was written (manifest sourceWords 7,513 → 7,511). Author of record Anselmus Laudunensis et schola; Migne prints it under Walafrid Strabo and the running head over these columns reads WALAFRIDI STRABI FULD. MON. OPP. PARS I. — THEOLOGICA. The honesty label on the work page carries that.
verify-english.mjs clean: 7 chunks, columns/notes/sections aligned, no duplicate paragraphs. Apparatus after the repair pass: 23 [var:] · 4 [sic:] · 1 [cj:] · 1 [d:] · 0 [ed:] · 0 [nt:] · 30 [n:] (before: 24 · 5 · 1 · 1 · 0 · 0 · 30) (exact parity with the Latin twins). Structure: ARGUMENTUM, CAPUT PRIMUM, CAPUT II–VI, with CAPUT IV split across chunks 3 and 4 (CAPUT IV. (cont.)). Zero guillemets in either language. Zero [cn:] — see §2 on plate status. Em-dashes: 0.
⭐ Start at §4 (the blind read) and §5 (the plate). Those two sections record what changed on 2026-08-23 and why; §§1–4 have been corrected in place to agree with them.
The five Glossa conventions settled on Liber Ruth (8968 §1) are followed and were not re-opened. Collation throughout is against sources/vulgate/clementine-flat.txt, cited by book chapter:verse.
§1 · THE LEMMA DIVERGENCE TABLE
Every italic span in data/briefs/8992-lemmata.txt is accounted for below, including the ones checked and found to agree with the Clementine exactly. A ✓ agrees line is a checked negative, not a skip. Where Migne's own note misattributes a quotation, that is recorded separately from the question of the wording, because it is a different fact.
1a. The divergences that took a [var:] (23)
⛔ One row was withdrawn on 2026-08-23. The 0573D row (Ut nos ingentes against Gal 2:9 ut nos in gentes) is no longer a scripture divergence at all — the plate read settled it as Migne's own fused type, Pattern 10, and a compositor's word-division is not a variant in his copy-text of Galatians. The [var:] was removed and a [sic: ingentes] put in its place. Reasoning in §2 and §5.
| col | Migne prints | Clementine | why it earned a marker |
|---|---|---|---|
| 0571D | Gustate, et eritis sicut dii (Gen. III) | Gn 3:5 in quocumque die comederitis ex eo, aperientur oculi vestri, et eritis sicut dii | Migne turns the serpent's conditional into a bare imperative. The gloss's whole point is that the angel spoke de proprio — the imperative is the thing being exhibited. |
| 0572A | Et anathematizavit eum et civitates ejus, et vocatum est nomen loci illius Anathema (Num. XXI) | Nm 21:3 et vocavit nomen loci illius Horma, id est, anathema | The received text has no anathematizing of him and his cities, and names the place Horma. The gloss is an etymology of the word anathema, so the wording is load-bearing. |
| 0572B | Placete omnibus per omnia, sicut et ego omnibus placeo (II Cor. X) | 1Cor 10:33 sicut et ego per omnia omnibus placeo | An imperative not in the received text at all, prefixed to an indicative clause whose word order also differs. |
| 0572B | Sine offensione estote Judaeis et Graecis Ecclesiae Dei (II Cor. VIII) | 1Cor 10:32 Sine offensione estote Judaeis, et gentibus, et ecclesiae Dei | Graecis for gentibus — a different word for the second party. |
| 0572C | Luceant opera vestra coram hominibus, etc. | Mt 5:16 Sic luceat lux vestra coram hominibus : ut videant opera vestra bona | Migne makes the works shine; the received text makes the light shine so that the works may be seen. Different subject. |
| 0572C | Nolite facere justitiam vestram coram hominibus, | Mt 6:1 Attendite ne justitiam vestram faciatis coram hominibus | Prohibition recast from attendite ne to nolite facere. |
| 0575A | Fascinatio malignitatis obscurat bona (Sap. IV) | Sap 4:12 Fascinatio enim nugacitatis obscurat bona | malignitatis for nugacitatis — and the whole paragraph is about invidia and malice, so Migne's word is the one the argument wants. Exactly the shape of 7914 @1079C's castimoniam: the keyword of the argument is not the keyword of the received text. |
| 0575A | Jesus Christus proscriptus est. | Gal 3:1 Jesus Christus praescriptus est | proscriptus for praescriptus, and the gloss immediately defines it: Id est, exhaeredatus — "that is, disinherited". The exposition rests on the divergent word. A translator half-remembering "set forth before your eyes" loses the entire gloss. |
| 0575D | Maledictus omnis homo, etc. | Gal 3:10 Maledictus omnis qui non permanserit in omnibus… | Migne inserts homo. See the 0576B entry — the insertion is glossed. |
| 0576B | Omnis homo. | Gal 3:13 Maledictus omnis qui pendet in ligno | The same intruded homo, and here it is the lemma being glossed: Etiam Christus, ne futuri haeretici negent ejus veram mortem. The gloss argues from a word the received text does not print. |
| 0577B | Sed concludit Scriptura. | Gal 3:22 Sed conclusit Scriptura omnia sub peccato | Present concludit for perfect conclusit. Marked because English cannot render the two tenses alike, so a reader would otherwise take our present for the Vulgate's — while the gloss's own verb is the perfect (clausit). |
| 0577B | Conclusit Deus omnia in justitia, ut omnium misereatur. | Rom 11:32 Conclusit enim Deus omnia in incredulitate, ut omnium misereatur | in justitia for in incredulitate. Adduced to prove men shut up under sin; Migne's word says the opposite kind of thing. |
| 0577D | Induite vos Dominum nostrum Jesum Christum. | Rom 13:14 induamini Dominum Jesum Christum | Imperative induite vos for passive induamini, plus nostrum. |
| 0578C | Coelum et terram ego impleo (Jer. XXIII) | Jr 23:24 Numquid non caelum et terram ego impleo? | ⚠ The received text is a rhetorical question containing a negative (numquid non); Migne prints it as a flat assertion, which is what the gloss needs. Two things change at once: sentence type and polarity. |
| 0579B | Conversi estis servire Deo vivo et vero (I Petr. II) | 1Thes 1:9 conversi estis ad Deum a simulacris, servire Deo vivo, et vero | Compressed; ad Deum a simulacris dropped. ⚠ Migne's note reads (I Petr. II) and the verse is 1 Thess 1:9 — see 1c. |
| 0580D | quos iterum parturis, | Gal 4:19 quos iterum parturio | parturis (2nd sg.) for parturio (1st sg.) — the labour is transferred from Paul to the Galatians. Rendered "whom thou bringest forth again" per Pattern 17 (singular tu → thou). Treated under 14 and not 12, because the form occupies its slot and can be rendered, and under 14 and not 18, because it sits inside a quoted scripture. |
| 0582A | Dilata locum tabernaculi tui, et protende funiculos tuos (Isa. LIV) | Is 54:2 Dilata locum tentorii tui, et pelles tabernaculorum tuorum extende : ne parcas : longos fac funiculos tuos | tabernaculi for tentorii; a whole clause dropped; protende funiculos for longos fac funiculos. |
| 0583D | Vos enim vocati estis a Deo in libertatem. | Gal 5:13 Vos enim in libertatem vocati estis, fratres | Migne adds a Deo and drops fratres. The agent is supplied by the plate, not by Paul. |
| 0584A | Qui enim non diligit proximum suum, quem videt, Deum, quem non videt, quomodo potest diligere? (I Joan. IV.) | 1Jo 4:20 Qui enim non diligit fratrem suum quem vidit, Deum, quem non vidit, quomodo potest diligere? | proximum for fratrem, and present videt for perfect vidit twice. The proof-text is adduced precisely for the neighbour (sine proximo non diligitur Deus), so Migne's substitution is the argument's hinge. |
| 0584B | Dico autem in Christo, etc. | Gal 5:16 Dico autem : Spiritu ambulate | Migne's in Christo replaces the imperative. The gloss then expounds it: Hoc autem dico posse esse vobis in Christo, et nunquam aliter. |
| 0584D | Regnum non Dei consequentur. | Gal 5:21 regnum Dei non consequentur | ⭐⭐ The negative has moved. In the plate non stands between Regnum and Dei, so the printed words say they shall obtain a kingdom that is not God's. Rendered as printed ("The kingdom not of God shall they obtain") per 7a — carrying a negative where the plate puts it, including when carrying it makes the sentence contradict its own gloss, which continues Nullus medius locus est… on the assumption that they obtain nothing. The [var:] gives the received order. See §3 for why no silent transposition was made. |
| 0585A | Non est impiis gaudere, dicit Dominus (Isa. XLVIII) | Is 48:22 Non est pax impiis, dicit Dominus | gaudere for pax — and the lemma being glossed is Gaudium. Migne's word is the keyword; the Vulgate's is not. Another pure 7a″ site. |
| 0587A | Operamini dum dies est: veniet nox quando jam nullus poterit operari (Joan. IX) | Jo 9:4 Me oportet operari opera ejus qui misit me, donec dies est : venit nox, quando nemo potest operari | Recast to a plural imperative, dum for donec, future veniet/poterit for present, jam nullus for nemo. |
1b. Divergences found, logged, and declined a marker (13)
Each is a real difference from the Clementine. None was marked, because in each the sense the gloss builds on is untouched and a [var:] would bury the twenty-four above in noise.
- 0571A · Et Domino Jesu Christo. — Gal 1:3 has et Domino nostro Jesu Christo. A dropped possessive.
- 0571A · Praesenti saeculo nequam. / Saeculo nequam. — Gal 1:4 verbatim; the second is the first clipped.
- 0571A · Mundus in maligno positus. — 1Jo 5:19 mundus totus in maligno positus est. Clipped tag.
- 0571A · Tempora periculosa. — 2Tim 3:1 instabunt tempora periculosa. Clipped tag. (Dies mali sunt = Eph 5:16 verbatim.)
- 0572B · Providemus bona, non solum coram Deo, sed etiam coram hominibus — 2Cor 8:21 drops only enim.
- 0572D · Segregavit ex utero matris. — Gal 1:15 qui me segregavit ex utero matris meae. Both possessives clipped. Rendered as printed ("He separated from the womb of the mother") and NOT expanded, because the printed words construe as they stand; convention 3 governs fragments that are not English, and this one is. See §3.
- 0573C · nihil interest, — Gal 2:6 nihil mea interest.
- 0573C · ab his, — Gal 2:6 ab iis; an orthographic doublet, Pattern 9.
- 0578B · Ego a Patre exivi, et veni in hunc mundum — Jo 16:28 Exivi a Patre, et veni in mundum: word order plus the deictic hunc. No proposition of the gloss turns on either.
- 0579B · Imo cogniti estis. — Gal 4:9 immo cogniti sitis: indicative for subjunctive, plus the imo/immo doublet.
- 0584A · Qui enim diligit iniquitatem… — Ps 10:6 has qui autem diligit; Migne enim. A connective only. ⚑ This is the class 8960's reader leaned on and got wrong from memory; checked on disk, and the divergence is real but trivial.
- 0586C · Cum mali sitis, nostis bona dare filiis vestris; — Lc 11:13 / Mt 7:11 Si ergo vos, cum sitis mali, nostis bona data dare. Word order plus a dropped data.
- 0587B · Benefacite his qui oderunt vos. Diligite inimicos vestros — Lc 6:27 prints the two clauses in the opposite order. Declined because nothing in the gloss turns on the order; contrast 7914 @1079B, where cinis et pulvis was marked precisely because the gloss turned on ashes.
1c. Migne's citation notes, checked against the verses (a separate question from the wording)
⭐ A systematic displacement at 0572B–C. Four quotations run in series, each followed by a note, and every note names the reference of the FOLLOWING quotation:
| the quotation | its actual source | the note Migne prints after it |
|---|---|---|
| Placete omnibus per omnia… | 1Cor 10:33 | (II Cor. X) |
| Sine offensione estote… | 1Cor 10:32 | (II Cor. VIII) |
| Providemus bona… | 2Cor 8:21 | (Matth. V) |
| Luceant opera vestra… | Mt 5:16 | — none — |
So (II Cor. VIII) belongs to Providemus bona (2Cor 8), (Matth. V) to Luceant opera (Mt 5), and the series ends one note short. (II Cor. X) on the first is the odd man out and would be (I Cor. X) if the pattern held throughout. The notes pass through verbatim under convention 4 — they are Migne's, and passing a siglum through is not endorsing it.
Two further note-referent observations, both left unmarked:
- 0575D · (Deut. XXI) stands on Maledictus omnis homo — which is Dt 27:26, quoted at Gal 3:10. Dt 21:23 is the source of the other curse, Maledictus qui pendet, quoted three columns later at 0576B, where Migne prints no note at all. Same displacement shape.
- 0584A · (Levit. XIX) stands on Diliges proximum. The lemma quotes Gal 5:14 and agrees with it verbatim; but Lv 19:18 itself reads Diliges amicum tuum sicut teipsum. The note is right about the book and the gloss (Hoc scriptum est in Levitico) is right about the fact; the Leviticus wording simply is not the wording quoted. No marker: the lemma is a faithful quotation of Galatians.
- 0579B · (I Petr. II) on Conversi estis servire Deo vivo et vero — the verse is 1 Thess 1:9. Recorded in 1a with the wording divergence; the misattribution is Migne's and rides as printed.
1d. Checked and agreeing with the Clementine — the negatives (the rest of the 276)
Recorded because a list of findings alone cannot be told from a report by an agent that never looked. Each of these was read word against word at the cited verse.
Gal 1 (0569D–0572D). Paulus, apostolus. (1:1; Migne lower-cases apostolus, Clem. Apostolus) · Non ab hominibus. · Neque per hominem. · Qui mecum. (1:2, clipped from et qui mecum sunt omnes fratres) · Qui dedit. (1:4, clipped) · Semetipsum. (1:4) · De saeculo. (1:4, clipped from de praesenti saeculo nequam — the ⚠ in the pre-scan is the clip, not a divergence) · Cui est gloria. (1:5) · Miror. (1:6) · Sed licet nos, (1:8) · Praeterquam quod. (1:8) · Anathema sit. (1:8) · Anathema. · Sicut praediximus. (1:9) · Modo. (1:10) · Si adhuc. (1:10) · Et proficiebam in Judaismo. (1:14) · Cum autem placuit. (1:15) · Ex utero. (1:15) · Priusquam te formarem in utero, novi te (Jr 1:5, verbatim) · salvatione · translatio (gloss words, not scripture).
Gal 2 (0572D–0574C). Deinde post annos quatuordecim. (2:1) · Et contuli cum illis. (2:2) · Seorsum, (2:2) · Subintroductos. (2:4) · Neque ad horam. (2:5) · Quales aliquando. (2:6) · Qui videbantur aliquid esse. (2:2) · Columnae. (2:9) · Dextras dederunt. (2:9) · Ut nos ingentes. (2:9 — ⭐ moved here from 1a on 2026-08-23: read as Migne's fused type for the Clementine's ut nos in gentes, the reading therefore AGREES with the Clementine and the [var:] was withdrawn; the defect is word-division, marked [sic: ingentes], see §2 and §4b) · Restiti. (2:11) · Prius enim quam venirent, (2:12) · Quomodo gentes cogis. (2:14) · Nos. (2:15) · Nisi per fidem, (2:16) · Quod si quaerentes. (2:17) · Ego enim per legem. (2:19) · Mortuus sum. (2:19) · Ut Deo vivam. (2:19) · Confixus. (2:19) · Vivo. (2:20) · Jam non ego. (2:20) · Vivit vero in me Christus. (2:20).
Gal 3 (0574D–0578A). O insensati. (3:1) · Fascinavit. (3:1) · Ante quorum oculos. (3:1) · in vobis, (3:1) · Hoc solum a vobis. (3:2) · hoc volo a vobis discere, (3:2, the gloss's own re-ordering of its lemma) · Sine causa. (3:4) · Qui ergo tribuit. (3:5) · Ad justitiam. (3:6) · Providens. (3:8) · In te benedicentur, (3:8, clipped and re-ordered from benedicentur in te omnes gentes) · Quia ex fide. (3:8) · Praenuntiavit, ×2 (3:8) · benedicentur · In te. · Quicunque enim, (3:10; quicunque/quicumque is Pattern 9) · Scriptum est. (3:10) · Justificatur apud Deum. (3:11) · Lex autem, (3:12) · Vivet in illis. (3:12) · lex non est ex fide; (3:12) · Qui ergo fecerit ea, vivet in illis: (3:12; the ergo is the gloss's resumption) · Factus pro nobis, (3:13) · maledictus, (3:13) · Maledictus qui pendet. (3:13, clipped) · Hoc autem dico. (3:17) · hoc autem testamentum confirmatum · dico · lex non facit irritum, / Lex non facit irritum. (the gloss's own restatement of 3:17 non irritum facit) · Propter transgressiones. (3:19) · Ordinata per angelos. (3:19) · ordinata: · In manu. (3:19) · Mediator. (3:20) · Lex ergo, (3:21) · Priusquam. (3:23) · Veniret fides. (3:23) · Custodiebamur. (3:23) · Conclusi. (3:23) · In eam. (3:23) · Quae. (3:23) · Revelanda. (3:23) · Paedagogus noster fuit in Christo Jesu. (3:24; Migne adds Jesu, and the gloss's own second lemma in Christo, drops it again) · in Christo, · Omnes filii Dei estis per fidem. (3:26; enim dropped) · In Christo Jesu. (3:26) · Induistis. (3:27) · Ergo Abrahae semen estis. (3:29; Abrahae semen for semen Abrahae).
Gal 4 (0578A–0582B). Dico autem, (4:1) · Sub elementis. (4:3) · Mundi. (4:3) · Venit plenitudo temporis. (4:4) · Plenitudo temporis. · Misit Deus Filium. (4:4) · In mundo erat…non cognovit (Jo 1:10, verbatim) · In propria venit. (Jo 1:11) · Ex muliere. (4:4) · Factum sub lege. (4:4) · Ut adoptionem. (4:5) · Misit Deus. (4:6) · Spiritum Filii. (4:6) · usque ad (the abridgment formula) · Clamantem. (4:6) · Abba pater. (4:6; Clem. Abba, Pater) · Qui natura non sunt, (4:8) · Sic in sancto apparui tibi… (Ps 62:3, verbatim) · Infirma. / Egena. / Egena elementa (4:9) · Dies. / Menses. / Tempora. / Et annos. (4:10) · Timeo vos, ne forte, (4:11) · Estote sicut ego. (4:12) · Quia, / sicut vos, (4:12) · Fratres, obsecro. (4:12) · Nihil me. (4:12, clipped from Nihil me laesistis) · Tentationem vestram. (4:13) · Sed sicut angelum Dei, (4:14) · Esurivi, et dedistis mihi manducare (Mt 25:35; only enim dropped) · Ubi est ergo, (4:15) · fieri posset. (4:15) · Verum dicens vobis. (4:16) · Aemulantur. (4:17) · Donec Christus. (4:19, clipped from donec formetur Christus in vobis) · Vellem autem. (4:20) · Dicite, (4:21) · Legem non legistis. (4:21) · Scriptum est, (4:22) · Unum de ancilla et unum, (4:22, clipped) · Per allegoriam. (4:24) · Abba / am, / gentium. In monte Sina. (4:24 — see §2 on the italic run) · Mons. (4:25) · conjunctus, (4:25) · Arabia, (4:25) · Quae sursum est. (4:26) · Quae est mater. (4:26; nostra clipped) · Scriptum est. (4:27) · Laetare. ×2 (4:27) · Non paris. (4:27) · Non parturis, (4:27) · Erumpe. (4:27) · Clama. (4:27) · Quia multi filii. (4:27) · Desertae. (4:27) · Quae habet virum. (4:27) · Isaac. (4:28) · Persequebatur. (4:29) · Qua libertate. (4:31).
Gal 5 (0582B–0585C). Si circumcidamini. (5:2) · Debitor est. (5:3) · A gratia excidistis. (5:4) · Nos enim. (5:5) · Spem justitiae. (5:5) · Nam in Christo. (5:6) · Sed fides. (5:6) · Nam et daemones credunt et contremiscunt (Jac 2:19; Migne prefixes Nam) · Currebatis. (5:7) · Confido. (5:10) · Conturbat. (5:10) · Quicunque. (5:10) · Ego autem. (5:11) · Quid adhuc persecutionem. (5:11) · Scandalum crucis. (5:11) · Utinam abscindantur. (5:12; et dropped) · Ne libertatem. (5:13) · Sed per charitatem Spiritus. (5:13) · Omnis enim lex. (5:14) · Diliges proximum. (5:14) · Sicut teipsum. (5:14) · Comeditis. (5:15) · criminamini, (see §2) · Ne abinvicem consumamini. (5:15; abinvicem fused, but attested late-Latin — Pattern 9, no marker) · Non perficietis, (5:16) · Caro enim concupiscit, (5:17) · Quod si Spiritu ducimini, (5:18) · Non estis sub lege. (5:18) · Charitas. (5:22) · Gaudium. (5:22) · Qui autem. ×2 (5:24) · carnem suam crucifixerunt, (5:24) · Si spiritu vivimus. (5:25) · Inanis gloriae. (5:26) · Invidentes. (5:26).
Gal 6 (0585C–0588B). Fratres, et si. (6:1; et si for etsi, Pattern 9) · In aliquo delicto. (6:1) · Lenitatis. (6:1) · Considerans teipsum. (6:1) · Et sic adimplebitis legem Christi. (6:2) · Nihil. (6:3) · Probet. (6:4) · Opus autem suum. (6:4) · Non sum sicut caeteri hominum (Lc 18:11; caeteri/ceteri only) · In semetipso (6:4) · Gloriam habebit. (6:4) · Et non in altero. (6:4) · Et sic, / in semetipso, / Et sic in se. (6:4) · Unusquisque enim. ×2 (6:5) · Onus. (6:5) · Alter alterius onera portate; (6:2) · Invicem onera portate. (the gloss's own paraphrase of 6:2, not a quotation) · Unusquisque, / portabit onus suum, (6:5; Migne's word order for onus suum portabit) · Communicet. (6:6) · Nolite errare. (6:7) · Qui seminat in carne / sua. / in spiritu, / suo, (6:8 — ⭐ checked with care, and the plate is right: the Vulgate does read in carne sua and in spiritu with no possessive, so the gloss's whole observation stands on the received text) · In carne, in spiritu. (composite) · Qui autem seminat. (6:8) · Non deficiamus. (6:9) · Tempore enim. (6:9) · Dum tempus. (6:10) · Ad omnes. (6:10) · Videte autem. (6:11; Migne adds autem) · Hi cogunt. (6:12) · Ut crucis Christi persecutionem, (6:12) · Mihi mundus crucifixus est. (6:14) · Et ego mundo. (6:14) · Venit enim princeps mundi hujus, et in me non habet quidquam (Jo 14:30, verbatim) · De caetero. (6:17) · Stigmata. (6:17) · Gratia Domini nostri Jesu Christi. (6:18).
§2 · THIS BOOK'S CRUCES
0573D · Ut nos ingentes — MIGNE'S OWN FUSED TYPE. [sic:], and the [var:] withdrawn
⭐ Settled at the plate, 2026-08-23, and re-classified. The plate prints ingentes as ONE italic word, exactly as our file has it — so this was never a transcription question and no patch is owed. The question left was which convention it belongs to, and the answer is Pattern 10 (fused type, sedet for sed et), not Pattern 14.
Why fused type and not a scripture variant — four reasons, and they are cumulative:
- The printed lemma has no predicate. Ut nos ingentes is a purpose clause with a subject accusative and an adjective and nothing to construe them with. The Vulgate's ut nos in gentes is the familiar ellipsis (ut nos in gentes [iremus]), completed by ipsi autem in circumcisionem. Only the split reading yields a sentence.
- The gloss under it never touches greatness. It reads Ut essem primus in praedicatione gentium, sicut Petrus in circumcisione — wholly about the Gentiles. A lemma Englished as "That we great ones" bears no relation to its own gloss, which is the tell.
- Migne's own next clause requires the preposition. sicut Petrus in circumcisione is one half of an antithesis whose other half must be in gentes. The plate itself supplies the missing in, four words later, in the parallel member.
- A
[var:]is a public claim about another text. It asserts that Migne's copy-text of Galatians read ingentes. Nothing supports that; what the plate shows is a compositor setting two words as one, which is the defect Pattern 10 names and which Corpus Corporum simply reproduced.
⚠ The reasoning that lost, stated so a reader can weigh it. Test 4 of the agent brief (the digitization alters Migne's spelling in both directions) cuts both ways — Corpus Corporum fuses and splits as readily as Migne does — so before the plate read the honest position was that the fusion's author could not be determined from the Latin, and that ambiguity was why the original entry hedged. The plate removed exactly that doubt: the fusion is Migne's. What it did not decide, and what remained mine, is whether Migne's fused word is a reading or a misprint; reasons 1–3 decide that, and they are all internal to his own page.
Rendered That we unto the Gentiles. [sic: ingentes] — the lemma Englished per convention 2, the plate's own form marked beside it. Strip the marker and the lemma still reads. Per Pattern 10 the carried run takes no second italic marker inside the already-italic lemma; the marker is set outside the italic span so no nesting is required.
0574A · ⭐⭐ aemulatores LEGES — THE NORMALIZATION NOBODY ASKED ABOUT. [sic:]
This entry replaces the former Qui dam entry, which is now void: the plate hyphenates Qui- / dam at the line end, so Quidam is one word on the page and the split was Corpus Corporum's. The Latin is patched, the [sic: Qui dam] is withdrawn, and "Certain" now stands unmarked.
But the same four words carried a second defect, running the other way, and it changes the English. The plate prints aemulatores leges; our Latin file read legis. Confirmed at 1400 dpi: the final e carries no dot, where qui and timens on the same and the following line show clear ones.
This is the NORMALIZATION class (8948 convivum, 8990 concilio): aemulator governs the genitive, so aemulatores leges leaves an accusative plural with nothing to govern it — Migne printed a form that does not construe, and the digitization tidied it into the grammatical legis.
⛔ Why nobody flagged it, and the general lesson. A corrected form reads perfectly. The translator met legis, rendered "zealots of the law" smoothly, and had no occasion to doubt a word that construes, agrees with its governor, and says what the paragraph wants. Every test this edition runs — parity, counts, negation attachment, band comparison, even the fourth test's "does it assert what he asserts?" — takes the Latin file as its input, so a defect that has already been repaired upstream is invisible to all of them by construction. It was found only because the plate was open at that line for a different word. A normalization can be detected from the Latin alone only by luck; it is a plate-read finding or it is nothing.
Rendered Certain emulators of the law [sic: leges]. Strip the marker and the clause keeps its subject, its verb, its object and its relative — the swallow test passes, and the fourth question (does the clause still have the thing it acts on?) is answered by "of the law" standing in plain English outside the bracket. The [sic:] is right rather than a [cj:] because the printed word has no grammatical slot — Pattern 12's first failure mode — not because it merely means the wrong thing.
0574D · quorumdam — the marker WITHDRAWN, the plate read settled it
The former entry marked [sic: quorum dam]. The plate hyphenates quo- / rumdam at the line end: one word on the page. Corpus Corporum dropped the hyphen and split at a different point than the plate does, so its "quorum dam" is not even evidence of where the break fell. Latin patched, marker removed, English now reads "the eyes of certain men".
⛔ The caveat that outlives the patch: a newline in the TEI is not reliable evidence of where the plate's line ends. It agrees with the plate at this work's Qui\ndam and disagrees here. Corroboration once the plate has spoken; never a substitute for reading it.
0581B · quibusdam — the marker WITHDRAWN, and the earlier column reference corrected
The former entry marked [sic: quisbusdam] at "0581C". The plate prints quibusdam plainly, one word, in an unbroken line — no hyphenation involved, so this was an ordinary misread by the digitization and not the line-break class at all. Latin patched, marker removed, English now reads "although in certain persons whom she brings forth here". ⚑ The column is 0581B, not 0581C as the first draft of this file said.
⚑ The former entry's own argument was wrong in an instructive way. It read three deformations of quidam/quibusdam in one short work as evidence about the plate. All three were the digitization. The frequency check that raised them (536 quidam / 37 quorumdam / 278 quibusdam against the split and metathesized forms) did its job perfectly — it is what got the plate opened — but it identifies a suspect, never a culprit.
0584A · digit for diligit — non-word, and it is the clause's only verb. [sic:]
Se autem non diligit qui digit iniquitatem. The marked run is the predicate of its own clause, so under 7a⁗ the English supplies it in plain words beside the marker: But he does not love himself who loves [sic: digit] iniquity. Strip the marker and the clause still has a verb, its negation and its subject. The conjecture is diligit, near-certain: the previous clause prints non diligit and the proof-text that follows prints diligit iniquitatem in full.
0584C · car nali for carnali — split type. [sic:]
when the soul by carnal [sic: car nali] concupiscence struggles against the spirit. The fourth word-division defect in the work.
0583A · Qui qui autem non credunt — dittography. [d:]
The two runs are identical, so Pattern 11 (not 11a) governs and the marker wraps the second: But they who [d: who] do not believe are slower and worse than the demons. It reads as a compositor's doubling, not a textual one — there is no construction in the sentence that could want two relatives.
0582C · nolunt where the argument needs volunt. [cj:], the one in the work
Gentiles vero ad talia non sunt cogendi. Sed si nolunt eis congruere, ut Timotheus, non sunt prohibendi. The printed word is a perfectly good Latin form, so Pattern 12 has nothing to wrap and Pattern 7 renders it — and the English then asserts something the gloss is arguing against, which is Pattern 18 exactly. The passage is the standard balance (do not compel, do not forbid), and its example is Timothy, who did conform; with nolunt the second clause forbids nothing and the example contradicts it. The word that went wrong is a negation (ne+volunt), which is the class 18a's 2026-08-15 extension explicitly admits even where the sentence is merely incoherent. Rendered are unwilling [cj: nolunt; read volunt, "are willing"] — additive, Migne's word kept.
0584D · Regnum non Dei consequentur — the negative in the wrong place. NO transposition
Two readings were weighed. The marker took the printed order and renders it, "The kingdom not of God shall they obtain", with a [var:] carrying Gal 5:21's regnum Dei non consequentur. The rival — reading the non as governing the verb and printing "shall not obtain the kingdom of God" — lost because it is a silent repair of exactly the class 7a exists to expose: it would produce smooth prose, satisfy every count-based check (the negative is present either way), and destroy the only evidence that the plate is defective here. A [cj:] was also weighed and declined: Pattern 18 does not reach inside a quoted scripture, and 14 does.
0571B · devotio / devotare — the word-play, and why no italics were added
et vulgo dicitur devotio … hoc erat anathematizare, quod vulgo dicitur devotare. Migne sets neither word in italics, and every italic span in the English must answer to one in the Latin, so the register's licence to keep a technical Latin word in italics could not be used. Rendered "a devoting" and "to devote" throughout, which carries the Roman sense (consecrating to the infernal gods) that the paragraph is defining, and keeps the pun with anathema legible.
0581A · gentium. In monte Sina. — one italic run across a lemma boundary
Migne's italics open at gentium (the tail of the previous gloss, completing pater multarum [gentium]) and close after In monte Sina. — which is the next verse lemma. The italic run therefore straddles the boundary between a gloss and the lemma that follows it, and there is no other place in the work where that happens. Reproduced 1:1 as one span, because span structure is sacred; but it is very likely two spans on the plate. ⬜ p. 296 was read for quibusdam at 0581B and this question was NOT put to it; it remains open, and it is safe open because it fires no marker. See §5e.
0583B · VERS. 11. printed without the .--
Every other verse address in the work is VERS. n.-- ; this one is VERS. 11. — the same class as Ruth's VERS.9, and reproduced verbatim rather than normalized, because the address passes through untranslated under convention 1 and the Latin side shows the same thing.
0583A · a sentence that simply stops
…quae nihil valent in Christo [0583A] Sed fides. — no terminal stop before the next lemma. Pattern 8: never supply. The English stops too.
0586A · In semetipso with no closing stop
In semetipso Id est intus in conscientia sua pura. The italic lemma carries no period where all its neighbours do. Reproduced as printed.
0586D · sua / suo — a gender distinction English cannot show
Qui seminat in carne addit sua. Cum vero dicit, in spiritu, non addit suo. The whole observation is that the possessive agrees with carne and would have had to be suo for spiritu — and English "his own" is the same word for both. Pattern 9: no emendation to undo, no marker to add, the facing Latin column carries it. Logged so a later sweep does not churn it. ⚑ The gloss's own next sentence supplies the point for an English reader (the carnal seed is from man, the spiritual from the Spirit of God), so nothing is actually lost.
0586D · dat offensum ex fide — ⭐ PLATE-CONFIRMED AS MIGNE'S. Rendered literally, no marker
qui spiritui dat offensum ex fide, cum charitate serviendo justitiae…
Plate read run 2026-08-23 at 1200 dpi: the plate prints dat offensum. The reading is Migne's own, not the digitization's, and there is nothing to patch. That is the fact this entry exists to carry, because it is the one a reader cannot recover: without it, "he who gives to the spirit an offence out of faith" looks like either our mistranslation or a transcription slip, and it is neither.
offensum is a real noun ("an offence"), so Pattern 12 has nothing to wrap and Pattern 7 renders it as printed. No marker fires, which is why this crux has to say plainly that the page is broken here rather than leave a reader to take the sentence at face value: the clause as printed cannot mean what the paragraph is arguing, which is that the man who serves the spirit reaps of the spirit.
The conjectures weighed, and why each lost:
- obsequium ("obedience, service") — the strongest, and the one this crux would take if it took any: it yields "renders to the spirit obedience out of faith" and matches serviendo justitiae in the same clause. It lost the marker, not the argument.
- assensum ("assent") — rejected. It construes and is graphically near, but dat assensum is a formula for agreeing with a proposition, while the clause's other members (charitate serviendo, legem spiritualiter intelligendo) are all acts of service, not acts of assent.
- affectum ("affection") — rejected. Furthest from the printed letters of the three, and it duplicates cum charitate, which already carries the affective member; the clause would say the same thing twice and leave serviendo unanswered.
[cj:] declined, and the reason is 18a rather than the merits: the census is CLOSED and Wilson ruled against re-opening it. Independently, the word that went wrong is neither a negation nor a correlative, and the resulting English is incoherent rather than false — a reader meets a sentence that does not work and knows something is wrong, which is not the Pattern 18 population (a sentence that works smoothly and asserts the opposite).
0582D · externa beatitudo — DECLINED, and ⚠ THE PLATE WAS NOT READ
quia ex fide justitia, et spes, id est, externa beatitudo.
The work's own control is against the printed word: 0577B and 0576A both print aeterna beatitudo in the identical phrase, glossing the identical idea. "External blessedness" is not a thing this gloss anywhere means.
Rendered "external blessedness" under Pattern 7 — externa is a real word occupying its slot — and the conjecture aeterna lives here and not in the text.
Does it clear Pattern 18's bar? I agree with the blind reader that it does not, and the reason is worth stating rather than asserting. The bar is that the printed word must make the sentence say something the author is arguing against. Externa beatitudo does not: it says something slightly odd but not contrary — hope's object as what lies outside the present is a reading a sympathetic reader can construct — whereas 11081's munde made the angels announce peace purely in a sentence whose next clause is in terra pax hominibus. The distance between "odd" and "opposite" is the whole width of Pattern 18. ⛔ And in any case 18a closes the census.
⚠ PLATE NOT READ HERE, recorded deliberately. This site sits on p. 296 (cols 581/582), a page opened for quibusdam at 0581B — so the read was available and was not extended to it. It is safe to ship unread because no marker fires on it: the English renders the printed word, nothing in the apparatus makes a claim about the plate, and if a later read finds aeterna the repair is a one-word Latin patch plus a one-word English change, with no marker to withdraw and no public claim to retract. That is the general test — an unread candidate is safe to ship if and only if it fires no marker.
0573B · non remanerent — a plural verb against a singular subject. ⚠ PLATE NOT READ
Quod si esset veritas Evangelii apud gentes, non remanerent, sed omnes judaizarent.
The reading taken makes veritas Evangelii the subject: "if it were so, the truth of the Gospel would not remain among the Gentiles, but all would judaize" — the sense the paragraph plainly wants, since the false brethren's aim was to bring the Gentiles back into the servitude of the law. The rival, making gentes the subject of remanerent ("the Gentiles would not remain"), lost because it strands veritas with no verb of its own; the plural is best explained as attraction to the gentes immediately before it.
⭐ What the blind reader added, and it is right: the English REPAIRS the disagreement in the act of rendering it. remanerent is plural against singular veritas, and English "would not remain" is number-neutral — so the defect passes through our column leaving nothing a reader could see. That is the 11632 @0304A shape (ut signa prius ardeat, "that signs may burn"), where the ruling was that where the English has no exponent for the defect, rendering it silently is a repair.
⚠ Why no marker fires here even so, and why 11632 does not govern. 11632's signa was a noun whose number IS visible in English (sign/signs), so the plural in our column was ours and concealed his. Here the disagreement lives entirely in a Latin verb ending that an English subjunctive does not carry at all — Pattern 9's territory, a systemic feature with no English exponent — and the facing Latin column prints remanerent where any reader can see it. The honest treatment is this crux, and the crux is the only place the fact can live.
⚠ PLATE NOT READ. p. 292 (cols 573/574) was opened, for Qui dam and ingentes; this site was not put to it. Safe to ship unread by the same test as 0582D: no marker fires, so nothing in the apparatus asserts anything about the plate that a later read could falsify.
0577B · non quia ex lege non est vita — a double negative that has to ride
ergo est ita contra promissa, ut per eam impleatur aliter quam promissum est: non quia ex lege non est vita… Both negatives carried ("not because from the law there is not life"), per 7a and against the temptation to smooth one away. ⚠ Note also that Migne prints no question mark on Lex ergo…ergo est ita contra promissa although the construction is the Vulgate's Lex ergo adversus promissa Dei? Absit. Pattern 8 forbids supplying one, so the English is a statement too.
0572B · ⭐ THE MISSING NEGATION IS MIGNE'S OWN — plate-confirmed, and NO marker fires
Dicit modo quia olim dum dixit hominem ex lege justificari, hominis gloriam inutilem, gratiam Dei praedicavit.
Plate read run 2026-08-23: there is NO non on the plate. The line prints dixit hominem ex lege justificari, hominis gloriam inutilem, gratiam Dei praedicavit, exactly as our file has it. So the missing negation is Migne's own, not the digitization's — what Paul is described as having preached formerly (olim) requires hominem non ex lege justificari, and the plate does not supply it.
The English is faithful and stands unchanged: "when he said that a man is justified out of the law, he preached the glory of man to be unprofitable, and the grace of God." Note also that praedicavit governs two objects of which only the first has its predicate (inutilem); the ellipsis is rendered intact.
⛔ No marker fires, by ruling and not by oversight. A dropped negative in the author's own prose is the Pattern 18 shape, and Pattern 18's census is CLOSED (18a); Wilson ruled against re-opening it. The crux carries it instead, with the plate as witness — which is the strongest form this finding can take, because a reader now knows the negative is absent from the page, and not from our transcription or our English.
0582B–C · aut non causa promittenti Christi, instituta, aut adhuc promittere — ⚠ PLATE NOT READ
Gentibus vero non sunt imponenda, ne videantur aut non causa promittenti Christi, instituta, aut adhuc promittere, sed cessando jam patent signa futuri fuisse.
The Latin does not construe. causa + the genitive Christi is right; the intervening dative participle promittenti agrees with nothing in the sentence.
Rendered as closely as the printed words allow — "lest they seem either not to have been instituted for the cause of Christ the promiser, or to be promising still" — with no marker, because the run is real Latin words in real forms.
⛔ And that is exactly the danger the blind reader named, which this entry now carries. The English is smooth where the plate is broken, so a reader meets a tidy sentence with no signal at all that anything is wrong. Worse, the smoothing lands on the sentence's pivot: the aut … aut turns on who promises, and the argument is that the rites were instituted in order to promise Christ and are no longer promising (sed cessando jam patent signa futuri fuisse). Our English makes Christ the promiser, which breaks the pivot — the second limb is then left with no one to be contrasted with.
The rival, named and weighed: causa promittendi Christum — "for the cause of promising Christ", a gerund with an accusative object, which restores the pivot exactly and is a small palaeographical step from promittenti Christi. It was not taken into the running text, under Pattern 7: it is a conjecture, and conjecture belongs here. ⛔ Nor did it take a [cj:] — 18a's census is closed, and in any case Pattern 18 requires a real word in its slot to make the sentence assert something false, whereas here the words do not construe at all, which is a different defect.
⚠ PLATE NOT READ, and of the three unread sites this is the one I would read first if a fourth read were run — because if the plate prints promittendi Christum the whole difficulty is Corpus Corporum's and dissolves, and the shape (a defective form tidied into something that half-works) is the normalization class that 0574A proves this very page-opening is capable of. Safe to ship unread only because no marker fires; the apparatus asserts nothing here.
0571D · Media aqua posita erat omnibus praeceptum Dei
media aqua nominative, praeceptum Dei in apposition. Rendered "Water in the midst had been set before all, the commandment of God" rather than smoothing the apposition into a copula.
0580C · Quem deformatis formosus sit in vobis — the construal I took, and the one I rejected
Taken: deformatis is the 2nd person plural of deformare — "Whom you disfigure, let him be beautiful in you." This construes completely: Quem is the object of deformatis, and the following sentence (ut formosus aliis per vos appareat) and the closing comparison (gravius dictum, quam si diceret eos formari in Christo) both depend on Christ being the one formed and disfigured in them, which is the graver saying. Rejected: deformatis as the dative/ablative plural of deformis ("to you who are deformed"), which was the first reading tried; it lost because it leaves Quem accusative with no verb to govern it, and would have required reading Quem as Qui — a silent emendation.
0577A · cui mediat per humanam — the antecedent
tamen est Deus unus cum Deo Patre per divinam naturam, cui mediat per humanam. Taken as Deo Patre, the only dative-eligible antecedent in the sentence: "to whom he mediates by the human." The alternative (cui reaching the two extrema of the previous sentence) lost on number — cui is singular.
0579B · quia perfecte non justificat — where the negative sits
Vere lex est egenum elementum: quia perfecte non justificat. Test 2a's trigger fired (a negative on a verb with an adverb adjacent). The negative was placed on perfecte — "because it does not justify perfectly" — proved from the structure of the next gloss, not from what reads well: Egena. Quia quantum juvat, non per se, sed respectu gratiae facit concedes that the law does help, so the law is egenum because its justification is incomplete, not because there is none.
0574C · a question mark on what is not a question
Non mihi, sed quod Deo jam possum? quia cum Christo sum. The mark rides as printed and the sentence continues in lower case after it, exactly as Migne sets it. Pattern 8, and not 8b — the clause is not an indirect question, so no colon recasting was used.
0585B · neuter neutrum videt
Two negatives in Latin idiom that English cannot double. Rendered "neither of the two beholds the other". Logged because it is a place where a negation count will read as one short.
0572B · pseudo as a bare noun
magis pseudo excommunicandi sunt (0572B), Imponebatur a pseudo (0583B), Isti erant pseudo (0587B), quia pseudo dixerant Paulum Galatis inimicum (0579D). Migne uses the Greek stem alone as an indeclinable noun for the false apostles, and the ARGUMENTUM's own falsis apostolis fixes the referent. Rendered "the false apostles" throughout. Not a defect — it is a usage, and the work is internally consistent in it.
0584A · criminamini — Migne records a variant reading, and it is his to record
Comeditis. Vel criminamini, — the gloss offers a second reading of the text of Gal 5:15 (the Clementine has only comeditis). Both are set in italics as lemmata and both are rendered. No [var:]: the divergence is one Migne himself is reporting, and marking it would tell the reader twice — the same principle as the rule against stacking our [sic:] on his ( sic ).
Second person throughout
Pattern 17 governs and the plate decides. Plural vos → you in the gloss prose (as in the shipped NT gloss books 8991 and 8998); ye kept only where a plural scripture imperative is being quoted or glossed (Gustate → "Taste, and ye shall be as gods"; Placete → "Please ye all men"; Nolite facere → "Do not ye your justice"; Estote → "Be ye as I"; Induite vos → "Put ye on"; Dicite → "Say ye"). Singular tu → thou (Quomodo gentes cogis → "How dost thou compel the Gentiles"; ut intelligas → "that thou mayest understand"; Considerans teipsum → "Considering thyself"; quos iterum parturis → "whom thou bringest forth again").
0576C–D · ⛔ dignus is MASCULINE — it is the PEOPLE made worthy, not the law. REPAIRED
Posita est in medio, inter promissionem et semen … ut populum Dei erudiret sub timore Dei: ut dignus fieret [0576D] excipere promissionem quae est Christus.
The English rendered both clauses with "it" (= lex), and they cannot both be the law. Lex is feminine and would require digna. Dignus is masculine and agrees with populus — the one made worthy to receive the promise is the people the law instructs, which is also what the sentence itself argues (ut populum… erudiret, then ut dignus fieret).
As it stood, the English made the law worthy to receive Christ instead of the people the law was given to school — a different proposition, and a doctrinally loaded one.
Latin morphology settles this outright; no plate read is needed and none was sought. Corrected to "that they might be made worthy to receive the promise which is Christ" — the plural pronoun because English "it" can only be read back onto lex, while "they" can only be read onto the people; the collective populus takes an English plural naturally and no word is supplied.
⚑ This is the unexpressed-subject class the brief names as dominant, in its purest form: both readings construe in English and only the Latin morphology decides, so nothing but reading the ending catches it.
0579A · sed scientiae … et dogmatum magnitudo — a mis-parse that produced plausible English. REPAIRED
Notandum quod clamor in Scripturis non magnae vocis emissio, sed scientiae intelligatur, et dogmatum magnitudo.
Was: "but as knowledge, and the greatness of doctrines." Now: "but as the greatness of knowledge and of doctrines."
scientiae is genitive (or dative) and cannot stand parallel to the nominative emissio. The single nominative in the sed limb is magnitudo, and it governs both genitives — scientiae across the interposed verb, dogmatum after it. The old rendering asserted a two-member predicate where Migne has one.
⚑ This is the fourth test's class exactly — Judith's ferias. Nothing was missing, everything was at parity, the negation sat on the right limb (non magnae vocis emissio), and the sentence read perfectly well. Only the question "does it ASSERT what he asserts?" reaches it.
0586B · Gal 6:2 quoted TWICE, DIFFERENTLY, five lines apart — the variation was erased. REPAIRED
Migne prints, within one gloss:
- Alia sunt opera participandae infirmitatis, de quibus supra: Alter alterius onera portate;
- Contra negligentes … dicit: Invicem onera portate.
Both were rendered "Bear ye one another's burdens." The Clementine reads Alter alterius onera portate (Gal 6:2), so the first is the verse and the second is Migne's own loose re-quotation — and rendering them identically erases the fact that he varies, which is precisely the datum this edition exists to carry.
Corrected: the second now reads "Bear ye burdens one for another." — differentiated, and differentiated in the direction the Latin points (invicem, reciprocal adverb, no alter alterius).
⚑ Note what class this is, because the apparatus does not cover it: a re-quotation INSIDE the gloss, not a lemma. The named-lemma list in the launch brief covers lemmata, and §1's divergence table is built from that list; neither reaches a phrase the gloss quotes to itself in passing. 8987 warned about exactly this ("your lemma list does its job; it does not cover the sentences between the lemmata") and here it is again, one book later. No [var:] fires — the second form is not a scripture citation diverging from the received text but Migne paraphrasing himself, and §1d already records it as such.
0583A · daemones — one word, one voice. REPAIRED
Nam et daemones credunt et contremiscunt [n: (Jac. II)] … Nam ipsum Christum esse daemones crediderunt.
The quotation was rendered "the devils also believe and tremble" while the four sentences expounding it all said "the demons". Migne prints daemon- throughout and agrees with the Clementine at Jac 2:19, so the traditional English echo was permitted — nothing in the scripture policy forbids it.
But the gloss's argument runs straight through the quoted word (Nam ipsum Christum esse daemones crediderunt, nec tamen in Christum crediderunt), so the reader has to carry the word from the quotation into the exposition, and our page changed it under him.
⚑ The tell was the cadence, not the lexicon. "the devils also believe, and tremble" is the AV/Douay line, and it had arrived at exactly the point where the translator's own settled word elsewhere in the work is "demons" — the 7a″ conformation shape, in its mildest form: not a wrong rendering, but the familiar Bible's word displacing our own at the one place where consistency was load-bearing. Corrected to "the demons also believe and tremble."
0575C · De eo quod sub lege fuerat — a person read as a thing. REPAIRED
Was: "Concerning that which had been under the law there could be doubt." Now: "Concerning the fact that he had been under the law there could be doubt."
The sentence before is Paul on himself: ego qui habui opera legis, non per ea, sed per fidem accepi potestatem. The argument is that Paul's own case could be doubted — he had the works of the law, so a sceptic could attribute his powers to them — while the Galatians' case, who never had the works, settles the matter (sed per similitudinem istorum totum liquet).
Both readings construe — de eo quod + neuter ("concerning that which") does parse — and the English chose one silently, which is the defect whether or not the choice was right. The chosen reading also happened to be the one that dissolves the argument: "that which had been under the law" has no referent in the paragraph at all.
⚑ De eo quod + indicative is the ordinary Latin for "as to the fact that", and it takes its subject from the preceding clause — here Paul, in the third person (fuerat), as the gloss shifts out of his voice. The rival ("concerning him who had been under the law") would require de eo qui; it is named here because it yields the same sense and a reader may prefer it.
0574D–0575A · fascinatio — one word, rendered two ways inside the passage that DEFINES it. REPAIRED
dum et hic eorum actus fascinatio dicitur … unde Salomon: Fascinatio malignitatis obscurat bona.
The first was "a fascination", the second "bewitching". The paragraph's whole point is that one word covers both the evil eye (fascinus … qui nocet infantibus, eyes that burn with their look) and envy (Similiter invidia tanquam fascinus urit) — and the Solomon text is adduced as the proof of that identity. Splitting the word breaks the link between the definition and its proof-text: an English reader met "fascination" in the definition and "bewitching" in the proof and had no way to see they were the same claim.
Corrected to "bewitching" at the first site, matching the settled word already used for fascinus three times, for Fascinavit, for fascinati and for fascinantes in the same paragraph. One voice within one work; the direction of the fix follows the majority usage rather than imposing a new word.
0578A · a patre substitutus — the cognate is the false friend. REPAIRED
Was: "one substituted by his father." Now: "one placed under [tutelage] by his father."
The gloss is expounding Gal 4:1–2, where the heir sub tutoribus et actoribus est usque ad praefinitum tempus a patre. substitutus here is sub- + statutus, "set under [guardianship]" — the state the verse describes — not "replaced", which asserts a different relation entirely (an heir put in someone else's place, which is nowhere in Galatians and would contradict haeres futurus in the same breath).
⚑ The English cognate "substituted" is the false friend, and it is the kind that reads fluently: "like an heir that is to be, and one substituted by his father" is grammatical, plausible, and about a different subject. The supplied "[tutelage]" is in square brackets per the corpus convention for our own supplied words; the Latin has no noun there, and a patre is retained as the agent Migne prints.
0575A · ille tantus qui est — CRUX, both readings named, neither taken silently
Vere fascinati, quia ante oculos, id est, in vestra praesentia ille tantus qui est. Jesus Christus proscriptus est. Id est, exhaeredatus.
Taken: "he who is so great" — tantus predicative, qui est an ordinary relative, the whole phrase a periphrasis for Christ, named outright in the lemma that follows.
The rival, and it is strong: ille tantus, qui EST — "that great one, HE WHO IS" — taking qui est as the Exodus 3:14 divine title (Ego sum qui sum; Qui est misit me ad vos), which is a commonplace of exactly this kind of gloss, with the following lemma supplying the predicate: the one who IS was disinherited. That reading makes the sentence's point rather than merely restating the subject, and it explains why tantus stands where it does.
Both construe, and the English chose one silently — which is the defect this crux exists to undo. The rendering is left as it stands (Pattern 7: the printed words support it, and re-casting to "HE WHO IS" would import a capitalised divine title Migne does not typographically mark), but a reader is now told there are two, and which was taken. No marker fires: nothing is defective, and [cj:] is closed by 18a in any case.
0572B · Judaeis et Graecis Ecclesiae Dei — a supplied "and" removed. REPAIRED
Sine offensione estote Judaeis et Graecis Ecclesiae Dei (Migne's note: II Cor. VIII; the verse is 1 Cor 10:32).
Was: "to the Jews and to the Greeks and to the Church of God." Now: "to the Jews and to the Greeks, to the Church of God."
Migne prints two members joined by et and the third in asyndeton. The Clementine has all three joined (Judaeis, et gentibus, et ecclesiae Dei), so the supplied "and" drifted the English toward the received text — in the same breath as a [var:] flagging a divergence from it, which is what makes a very small thing worth fixing rather than merely logging: the marker tells the reader the citation differs, and the prose beside it quietly made it differ less.
⚑ The [var:] on Graecis / gentibus is unaffected and stands; §1a's row is unchanged.
§3 · CANDIDATES RAISED AND REJECTED
- Contracting Qui dedit. / expanding it to "Who gave himself". Rejected, and this is the class the brief warns about. Qui dedit. (Gal 1:4, clipped before its object) and Semetipsum. (a complete one-word lemma, glossed separately in the very next sentence) are different classes. Expanding the first would pre-empt the second and manufacture an inconsistency that is not on the page. Rendered "Who gave." and "Himself." — and an early draft that read "he offered himself of his own accord" in the gloss on Qui dedit was withdrawn, because it imported the next lemma's word into the first lemma's exposition.
- Expanding Segregavit ex utero matris. Rejected. Convention 3 expands a fragment that is not English; this one construes as printed. Supplying me and meae from Gal 1:15 would be conformation to the received text, which is the defect, not a cure for it.
- Contracting Nihil me. to a stump. Rejected in the other direction: "Nothing me" is not English, so convention 3 applies and it is rendered "You have injured me in nothing." Per the brief's ruling, where a work treats bare lemmata two ways the fix is to expand, never to contract.
- A
[sic:]on abinvicem (0584A). Rejected. Pattern 9: abinvicem is attested late Latin, not a word-division defect, and marking it would put this work's four genuine split runs into noise. - A
[var:]on the quicunque/quicumque, caeteri/ceteri, coelum/caelum, Hierusalem, imo/immo, etsi/et si doublets. Rejected wholesale — Pattern 9, orthography of the period. - A
[cj:]on externa beatitudo (0582D). Weighed against a two-instance control inside this same work and declined; reasoning in §2. - A
[cj:]on offensum (0586D). Declined; reasoning in §2. - A
[cj:]or a transposition on Regnum non Dei (0584D). Declined in favour of[var:]and the printed order; reasoning in §2. This is the site a blind reader is most likely to flag, so the reasoning is stated there in full rather than here. - A
[var:]on the note (Levit. XIX) at 0584A. Rejected: the lemma quotes Gal 5:14 and agrees with it exactly. The mismatch is between Migne's note and Leviticus, not between his text and the Clementine, and[var:]marks the second. Recorded in §1c instead. - Reordering Benefacite … Diligite (0587B) to the Vulgate's sequence. Rejected outright — Migne's order is Migne's, and nothing in the gloss turns on it.
- Splitting the italic run at 0581A into two spans. Rejected: span structure is 1:1 with the Latin and is not ours to tidy. Logged as a plate-read question instead.
- Normalizing
VERS. 11.toVERS. 11.--. Rejected. Convention 1 passes the address through untranslated; Pattern 9's licence to normalize applies to translated structural labels, and this one is not translated. - Supplying a
?at 0577B (Lex ergo…). Rejected. The Vulgate's Lex ergo adversus promissa Dei? is a question; Migne's gloss sentence is not punctuated as one, and Pattern 8 forbids supplying the mark. 8b was checked and does not apply — the clause is not an indirect question. - Supplying a terminal stop at 0583A and 0586A. Rejected, Pattern 8.
§4 · THE BLIND READ, AND WHAT IT CHANGED (2026-08-23)
An independent reader who had not seen this file read all seven chunks against the Latin. Fifteen findings; thirteen applied, two applied as cruces only, and one — finding 7 — handed back for me to settle with the plate. Every edit in the repair pass traces to a numbered finding; nothing else was touched and the work was not re-translated.
4a. What it found, and what was done
| # | col | the finding | outcome |
|---|---|---|---|
| 1 | 0576C–D | dignus is masculine: the PEOPLE are made worthy, not the law | repaired — "it" → "they" |
| 2 | 0579A | magnitudo is the only nominative and governs both genitives | repaired — one predicate, not two |
| 3 | 0586B | Gal 6:2 quoted twice, differently; both rendered alike | repaired — the second differentiated |
| 4 | 0583A | daemones "devils" in the quotation, "demons" in its gloss | repaired — one voice |
| 5 | 0582B–C | promittenti Christi smoothed; the aut…aut pivot lost | crux only, rival named, plate not read |
| 6 | 0575C | De eo quod read as a thing where the argument wants Paul | repaired, rival named |
| 7 | 0573D | Ut nos ingentes — Pattern 10 fused type, not a Pattern 14 variant | re-classified: [var:] → [sic:] |
| 8 | 0573B | plural remanerent silently repaired by number-neutral English | crux only, plate not read |
| 9 | 0572B | the missing non | crux only — plate says it is Migne's; 18a closes the marker |
| 10 | 0586D | dat offensum | crux only — plate-confirmed Migne's; conjectures named |
| 11 | 0582D | externa against the work's own aeterna ×2 | crux only, declined, plate not read |
| 12 | 0574D–0575A | fascinatio rendered two ways inside its own definition | repaired |
| 13 | 0578A | substitutus = "set under", not "replaced" | repaired |
| 14 | 0575A | ille tantus qui est — the Ex 3:14 title as rival | crux only, both named |
| 15 | 0572C | a supplied "and" in an asyndetic citation | repaired — "and" dropped |
4b. Where I think the reader was wrong, or only half right
⭐ Finding 7 was flagged rather than asserted, and the caution was right at the time but the conclusion is right now. The reader argued Pattern 10 on two grounds and declined to assert it because Test 4 of the agent brief (the digitization alters Migne's spelling in both directions) cuts both ways. I agree with the conclusion and think the case is stronger than the two grounds given: see §2 @0573D for four reasons, two of which (the missing predicate, and in circumcisione four words later demanding a parallel in gentes) are internal to Migne's own sentence and do not depend on the plate at all. The plate read settled the transcription question; it did not settle the classification, and the classification was decidable from the Latin all along.
⚑ Finding 8's diagnosis is right and its proposed remedy would have been wrong. The reader called it "the 11632 @0304A ruling" and asked for a crux, which is what it got. But had the ruling been read as reaching a marker, it would have been a mistake: 11632's signa is a noun whose number English shows, so the silent repair was in a word the reader can see; remanerent is a verb ending English subjunctives do not carry at all. Same defect class, different exponent, and the exponent is what decides between a marker and a crux. Recorded at §2 @0573B.
⚑ Finding 15 is described as "very minor" and is not. The drift is small in itself; what makes it worth the edit is that it moved the citation toward the received text in the same sentence where a [var:] announces a divergence from it. The reader saw this and still hedged. The hedge was the only soft judgement in fifteen findings.
⚠ Finding 6 offers two paraphrases ("him who" and "the fact that he") without choosing. I chose the fact that he, because de eo quod + indicative is the ordinary Latin for it and "him who" would want de eo qui. The reader's instinct — a person, not a thing — was correct; the grammar picks which person-construction.
4c. What the reader checked and found CLEAN, and it is the most valuable part
Recorded because a list of findings alone cannot be told from a report by an agent that never looked.
- ⭐⭐ Zero polarity defects in a book made of nothing but negatives. Not one negative dropped, intruded, relocated or mis-hosted. Every high-risk site is correctly scoped: non modo sibi … sed et his · non solum tempora, sed etiam loca · non seipsum, sed Deum in seipso diligi volebat · non est tantum unius populi … sed et gentilium · nullam omnino generandi potentiam · non tam … quam · mulieres dicit non corruptas, sed feminas · and both double negatives. Note that four of those are the exact 2a trigger population (a negative on a verb with a quantifier or correlative adjacent) and all four hold.
- ⭐ 0584D Regnum non Dei consequentur — left exactly as it is. The negative is displaced onto Dei on the plate and the English renders it displaced, with a
[var:]giving the received order. That is Pattern 7 obeyed at the precise point where obeying it looks like an error, and it is the site a future reader is likeliest to "fix". Do not. - All
[var:]verified exact against the Clementine, including the real finds: proscriptus/praescriptus with exhaeredatus confirming Migne (plate-confirmed this session) · Rom 11:32 in justitia · Is 48:22 gaudere · Sap 4:12 malignitatis (plate-confirmed). - Every
[sic:]passes the swallow test, the fourth question (does the clause still have its object?) included. Nothing here is the Tobit failure. The two new markers written in this pass (ingentes, leges) were held to the same test. - The
[cj:]at 0582B is well aimed (nolunt → volunt) — and the plate confirms Migne prints nolunt, so it is anchored on a genuine reading and not on a digitization artifact. - All 71 column anchors checked by position, not merely presence and order (test 2b); zero mismatched punctuation bands across all 71.
- Qui dedit. / Semetipsum. are different classes and the contrast is not an inconsistency. Nisi per fidem → "But by the faith." declined correctly. Segregavit ex utero matris. left objectless and NOT conformed — correct.
§5 · THE PLATE (2026-08-23)
raw/scans/pl114/patrologiaecurs06saingoog.pdf, PDF page = (column + 11) / 2 — confirmed against the printed corner numbers (p. 292 heads "574", p. 293 heads "575"). Read at 900 dpi on cropped column bands, one word re-read at 1400 dpi and one at 1200 dpi. Every crop was confirmed by eye before data/tei-patches/8992.json was written.
5a. Four patches to the Latin — three word-boundary defects and ONE NORMALIZATION
| col | our old Latin | the plate | class |
|---|---|---|---|
| 0574A | Qui dam aemulatores | Quidam (plate hyphenates Qui- / dam) | unjoined line break — the digitization's |
| 0574A | aemulatores legis | ⭐⭐ aemulatores LEGES | NORMALIZATION — Migne's defective form, tidied upstream |
| 0574D | quorum dam oculi | quorumdam (hyphenated quo- / rumdam) | unjoined line break — the digitization's |
| 0581B | in quisbusdam | quibusdam, one word, unbroken line | ordinary misread — the digitization's |
Consequences in the English: three [sic:] withdrawn (Qui dam, quorum dam, quisbusdam — all three marked the digitization, not the plate, and every one of them was therefore a marker blaming Migne for someone else's error), and one new [sic: leges] written, with the evident sense standing beside it in plain words. Net [sic:] 5 → 4.
⭐⭐ The normalization is the finding of the session and nobody asked for it. It was found while the plate was open at that line for a different word. Full reasoning at §2 @0574A; the short form is that a form which reads perfectly may read perfectly because someone upstream fixed it, and no test that takes the Latin file as its input can see that. Two of the three prior instances in the corpus (8948 convivum, 8990 concilio) were found the same way — incidentally, with the plate already open.
⛔ A transcription newline is NOT evidence of where the plate's line ends. It agreed with the plate at Qui\ndam here and at 8958's tam per\nverse, and disagreed at 0574D, where the TEI breaks after quorum and the plate hyphenates at quo-. Corroboration once the plate has spoken; never a substitute for reading it.
5b. Three reads that came back CONFIRMING MIGNE — checked negatives, and they are worth as much
| col | suspicion | the plate |
|---|---|---|
| 0573D | Ut nos ingentes a transcription fusion | prints it as ONE italic word — Migne's own; re-classified Pattern 10, [var:] → [sic:] |
| 0572B | a non lost in digitization | there is NO non on the plate — the missing negation is Migne's; no marker (18a) |
| 0586D | dat offensum a misread | prints dat offensum at 1200 dpi — Migne's; rendered literally, no marker |
Also confirmed on the same pages, and already carried in §1a: 0575A proscriptus (against the Vulgate's praescriptus, with Migne's own exhaeredatus confirming it) and 0574D–0575A Sap 4:12 malignitatis (against nugacitatis). Both [var:] stand, plate-confirmed.
5c. ⚑⚑ THE BATCH PATTERN — non-words versus real words
Across three works this session, every suspicion aimed at a NON-WORD was confirmed by the plate as ours (the digitization's), and every suspicion aimed at a REAL, WELL-FORMED WORD was refuted, with Migne printing it: credulitatem, Homo, convenientem, dat offensum. This work contributes to both columns — quisbusdam and the two split runs to the first, dat offensum and ingentes to the second.
The operational consequence: corpus frequency is decisive for non-words and worthless for real words. "278 quibusdam against 2 quisbusdam" is a near-proof that the odd form is the digitization's, because Migne's compositor and Corpus Corporum's OCR do not produce non-words at comparable rates. "offensum is a real noun and occurs N times in the corpus" proves nothing at all about who set it. Raise a non-word on frequency and expect to be right; raise a real word on frequency and expect the plate to refuse you — and raise it anyway, because a refused suspicion that gets recorded is what makes the next one cheap.
⚠ And the class that beats both: a normalization, where the suspicion cannot be raised at all because the word is real, well-formed, frequent, and correct. Only an eye on the plate finds it.
5d. Three candidates NOT read, and why that is safe
0582B–C promittenti Christi · 0573B remanerent · 0582D externa beatitudo.
All three sit on pages that were opened for other words (p. 292 and p. 296), so the reads were available and were not extended. Each is recorded as unread at its own §2 entry.
⭐ The rule this rests on, stated once: an unread candidate is safe to ship if and only if it fires no marker. None of the three fires one. The English renders the printed word in each case, the apparatus makes no claim about the plate at any of them, and the cruces name the rival and say which reading was taken. If a later read overturns any of the three, the repair is a Latin patch plus a local English change — no marker to withdraw and no public claim about another text to retract. Contrast the three markers withdrawn in 5a: those did fire, they did blame the plate, and they were wrong for two months' worth of shipping had the read not been run.
If a fourth read is ever run, the order is: (1) 0582B–C promittenti Christi, because the normalization class is now proven live on this very page-opening and causa promittendi Christum would dissolve the whole difficulty; (2) 0582D externa, where the work's own control is two-to-one against the printed word; (3) 0573B remanerent, lowest value, since either outcome leaves the English and the apparatus exactly as they are.
5e. Standing plate coverage for this work
Coverage is "spot". Pages read: p. 290 (cols 569/570), p. 292 (cols 573/574), p. 293 (cols 575/576), p. 295 (cols 579/580), p. 296 (cols 581/582), and the band carrying 0586D. On p. 290 and p. 295, read full width at 300 dpi: no foot rule and no numbered Forte apparatus — which is why there are no [cn:] markers in this work. That is a checked zero for those two pages only and is not evidence that the work carries no apparatus elsewhere.
⬜ Still unread and still open: whether the italic run at 0581A is one span or two (gentium. closing one gloss and In monte Sina. opening the next lemma). Reproduced 1:1 as one span because span structure is sacred and is not ours to tidy; the question fires no marker either.