# Codex H NT Overlap Report

Date: 2026-06-03

## Verdict

Codex H does not produce a critical change to our understanding of the overlapped New Testament text.

It does produce useful manuscript evidence: recovered folios, early Pauline chapter-list/paratactic material, scribal corrections, and a set of meaningful variants in the Pauline/Hebrews material it preserves. The important variants are major in the textual-apparatus sense, not critical in the doctrine/base-text sense.

The strongest unresolved textual row is Heb 4:15. The TEI extraction shows the verse ending before the "without sin" clause and Heb 4:16 resuming on the next folio with no explicit 4:15 gap marker. I classify it as a major unresolved candidate pending image/apparatus confirmation, not as a safe "fragmentary noise" row.

## Sources And Scope

Primary source:
- Glasgow Research Data deposit: `https://researchdata.gla.ac.uk/2105/`
- DOI: `10.5525/gla.researchdata.2105`
- Downloaded file: `outputs/audits/2026-06-03-codex-h/source/2105_data.zip`
- Extracted TEI/XML: `outputs/audits/2026-06-03-codex-h/source/combined-xml.xml`
- XML SHA256: `096b2bc5800033aa70fcadefb1e844a236f6267820c3dd4820ec394e1e2e2be3`

Public context:
- Codex H digital edition: `https://codexh.arts.gla.ac.uk/`
- Glasgow announcement, 2026-04-24: `https://www.gla.ac.uk/news/headline_1263245_en.html`

Open comparator:
- SBLGNT, downloaded locally as `outputs/audits/2026-06-03-codex-h/source/SBLGNT-master.zip`
- SBLGNT SHA256: `329253319cbe4cb0873a2289c92dcc97e9b4b9307c9f2bbc156027da66137991`

Accepted Greek baseline:
- The relevant scholarly baseline is NA28/UBS5.
- German Bible Society / INTF context: `https://www.die-bibel.de/en/en/bible-society-and-biblical-studies/greek-new-testament/comparison-na28-ubs5`
- I used SBLGNT only because it is open and machine-readable. It is not NA28/UBS5 and differs from that tradition in many places.
- I did not ingest the proprietary NA28/UBS5 apparatus. Rows that need apparatus verification are marked accordingly.

## Coverage

The TEI extraction covered 338 verse references across 125 TEI page/folio units:

| Book | Extracted refs |
| --- | ---: |
| 1 Corinthians | 28 |
| 2 Corinthians | 56 |
| Galatians | 42 |
| Colossians | 38 |
| 1 Thessalonians | 12 |
| 1 Timothy | 48 |
| 2 Timothy | 16 |
| Titus | 26 |
| Philemon | 1 |
| Hebrews | 71 |

Pauline material not present in this extraction includes Romans, Ephesians, Philippians, and 2 Thessalonians, plus large internal gaps within preserved books. That means famous variants in those absent spans were not evaluated as Codex H overlap.

Extraction flags:
- `unclear`: 1042
- `supplied`: 689
- `nomSac`: 206
- `expansion`: 88
- `gap`: 4
- app/correction readings: 64

The automated SBLGNT comparison found 257 `variant_candidate` rows and 7 `near_match` rows. That raw number is noisy. Most candidates collapse into nomina sacra, spelling, line breaks, supplied editorial restorations, or fragment boundaries.

## Classification

Tangential:
- Nomina sacra versus expanded divine/christological names.
- Accent, breathing, punctuation, capitalization, word split, line-break, and final-n differences.
- Spelling/itacism that does not change the word being read.
- Paratext and chapter/folio apparatus unless it affects biblical wording.
- Tokens supplied from NA28 or another editor without visible support.

Minor:
- Conjunction/article/function-word differences.
- Small grammatical or order shifts with little translation impact.
- Formulaic additions with low exegetical impact.

Major:
- Lexical substitutions, pronoun changes, phrase additions/omissions, or known apparatus variants that affect translation or commentary.

Critical:
- Unknown NT material, a recovered reading that changes core doctrine, a variant that compels replacing the accepted critical Greek base text, or a direct product/base-text policy change.

## Critical Findings

None found.

The recovered material contains known Pauline text, not a newly discovered canonical passage. No row below requires changing Biblex's current-Greek baseline. The major variants are real and should be footnoted or apparatus-aware, but they do not create or erase a central doctrine.

## Major And Major-Candidate Rows

| Ref | Class | Codex H evidence | Comparator / accepted-text caveat | Judgment |
| --- | --- | --- | --- | --- |
| 1Cor 10:28 | Major | A firsthand correction/app adds the Ps 24:1-style quotation after "conscience." | Omitted by SBLGNT and modern critical text; familiar later-tradition/TR-style expansion. | Major for apparatus; not critical. |
| 2Cor 4:4 | Major | `καταυγάσαι` where SBLGNT has `αὐγάσαι`. | NA28 status unchecked in this audit. | Lexical variant; sense remains illumination/shining. |
| 2Cor 11:3 | Major candidate | Visible line lacks SBLGNT's "and purity" phrase after "simplicity." | Verse has unclear/supplied flags; apparatus check needed. | Material if complete at that point; not critical. |
| Gal 1:6 | Major candidate | Reads `ἡμᾶς` where SBLGNT has `ὑμᾶς`. | Pronoun row has surrounding orthographic/unclear flags. | Meaningful if secure; not critical. |
| Col 1:24 | Major | Correction reads `παθήμασι μου`; Codex also has `ἀναπληρῶ` vs SBLGNT `ἀνταναπληρῶ`. | `μου` correction is firsthand in the extracted app list. | Possessive/lexical nuance; not doctrinally disruptive. |
| Col 2:2 | Major | Visible Codex ends "mystery of God" with no visible `Χριστοῦ`. | Col 2:2 has a famously complex apparatus; NA28 alignment is not resolved here. | Major Christological apparatus row; no new doctrine. |
| Col 2:3 | Major | Reads `τῆς ἐπιγνώσεως` where SBLGNT has `γνώσεως`. | NA28 status unchecked. | Translation nuance: "knowledge/recognition." |
| Col 2:11 | Major | Firsthand correction adds "of sins"; extracted line lacks SBLGNT's final "circumcision of Christ" phrase. | Known later-tradition style expansion around "body of flesh/sins." | Major for translation notes; not critical. |
| Col 3:4 | Major | Reads "our life" where SBLGNT has "your life." | NA28 status unchecked. | Meaningful pronoun variant; not critical. |
| 1Tim 1:17 | Major/minor border | Firsthand correction adds `σοφῷ` before God. | Known doxological expansion in later tradition; SBLGNT omits. | Apparatus-worthy, low doctrinal risk. |
| 1Tim 2:7 | Major/minor border | Adds `ἐν Χριστῷ` after "I speak truth." | SBLGNT omits. | Formulaic phrase addition; not critical. |
| 1Tim 6:10 | Major | `ποικίλαις` vs SBLGNT `πολλαῖς`. | NA28 status unchecked. | Lexical nuance: "various" vs "many" pains. |
| Titus 2:7 | Major/minor border | Adds `ἁγνίαν` in the teaching list. | SBLGNT omits. | Meaningful list expansion; not critical. |
| Heb 1:3 | Major | Visible Codex covers only the back half of the verse. Firsthand corrections include `δι' ἑαυτοῦ` and `τοῦ θρόνου`; Codex also has `ἡμῶν` after sins. | SBLGNT includes `δι' αὑτοῦ`; NA28-style text commonly omits it. SBLGNT omits `ἡμῶν` and `τοῦ θρόνου`. | Major against NA28/UBS5 at least for `δι' ἑαυτοῦ`; not critical. |
| Heb 2:9 | Major | Reads grace-of-God wording rather than SBLGNT's "apart from God" wording. | NA28/UBS5 main text is commonly the grace reading; SBLGNT differs here. | Important witness supporting mainstream reading, not a disruptive discovery. |
| Heb 4:15 | Major unresolved | TEI has 4:15 ending on folio `57v` after "our weaknesses"; 4:16 resumes on `58r` at "grace." No explicit 4:15 `<gap>` marker appears in the extracted `<ab>`. | Missing phrase includes the "without sin" clause; image/apparatus check required. | Treat as unresolved major candidate; not critical unless confirmed in context, and sinlessness is secure elsewhere. |
| Heb 10:34 | Major | Firsthand correction adds `ἐν οὐρανοῖς`; row also has `δεσμοῖς`/`δεσμίοις` and `κρείττονα` spelling/orthography issues. | Heavy supplied/unclear flags; `ἐν οὐρανοῖς` is app-derived. | Heavenly-location expansion is major; the bonds/prisoners row should be noted but not overclaimed. |
| Heb 10:38 | Major | `μου` is present in the original app reading and struck/deleted by a corrector; extracted line lacks SBLGNT's final "in him." | Later correction hand differs from firsthand evidence. | Apparatus-significant; not critical. |

## Minor Rows

Representative minor rows:
- 1Cor 10:23: extra `μοι` twice.
- 1Cor 10:31: repeated `-ται` forms where SBLGNT has imperatives.
- Gal 1:4: `ἀπό` / order / `θέλησιν` versus SBLGNT phrasing.
- Gal 1:11: `δέ` versus `γάρ`.
- Gal 5:1: word-order/ending differences, with `χς` present as a nomSac in a different position.
- Heb 2:16: non-repetition of the final verb in the extracted line.
- Titus 3:15 and Heb 13:25: final `ἀμήν`.

These can affect apparatus or translation notes, but they do not change the base conclusion.

## Tangential Inventory

The largest divergence class is tangential:
- Nomina sacra: 206 flags.
- Itacism/spelling: frequent `ει/ι`, `αι/ε`, movable/final nu, and similar forms.
- Word segmentation: e.g. split words across line breaks.
- Punctuation and paratext: folio titles, chapter numerals, lectional headings, and early chapter-list apparatus.
- Editorial supply: 689 supplied flags. Any `source="na28"` token is restoration, not independent Codex H evidence.

This is why the raw `variant_candidate` count should not be read as "257 real variants."

## Non-Evidentiary Or Fragmentary Major-Looking Rows

I do not treat these as secure Codex H omissions without image/apparatus review:
- 1Cor 11:6, 11:8, 11:9: folio/line starts create SBLGNT insertions in the diff.
- 2Cor 4:7; 2Cor 11:9, 11:14, 11:16, 11:23: damaged/supplied lines and sequence noise.
- Gal 2:7-9 and 2:20: long displaced/missing clauses in fragmentary spans.
- 1Tim 3:2: line stops after `κόσμιον`; likely physical/extraction cutoff before "hospitable, able to teach."
- Titus 2:5: explicit TEI gap of three illegible lines between the visible list and the final verb.
- Titus 2:8: already marked too fragmentary and has a gap.
- Heb 2:10, 2:14, 3:1, 3:5, 3:10, 3:12, 3:15, 4:16, 12:9, 12:18, 13:21: raw diff shows long SBLGNT phrases absent, but the local evidence points to gaps, page truncation, or fragment boundaries.

Heb 4:15 is intentionally not in this downgraded list after subreview; it remains a major unresolved candidate.

## Subreview

Adapted subreview artifacts:
- Synthesis: `outputs/audits/2026-06-03-codex-h/SUBREVIEW_SYNTHESIS.md`
- Prompt: `outputs/audits/2026-06-03-codex-h/subreview/prompt.md`
- Claude receipt: `outputs/audits/2026-06-03-codex-h/subreview/claude.md`
- Gemini receipt: `outputs/audits/2026-06-03-codex-h/subreview/gemini.md`

Result:
- Claude completed and supported the headline conclusion: "major variants, no critical findings."
- Claude required auditability fixes: disclose hand provenance, make supplied-vs-visible limits explicit, row-level NA28/SBLGNT caveats, Heb 10:34 note, Heb 1:3 coverage note, and Heb 4:15 escalation.
- Gemini failed before review because this local Gemini CLI rejects the wrapper's `--skip-trust` argument.

## Answer To The User Question

Does Codex H have major findings that change understanding of NT overlaps?

Yes, but the change is mostly manuscript-historical, not doctrinal or base-text-changing.

It strengthens knowledge of Codex H as a sixth-century Pauline witness, its scribal corrections, its later/firsthand correction layers, and its early chapter/reading apparatus. It also gives useful evidence in known variant units, especially in 1 Corinthians, Colossians, 1 Timothy, and Hebrews.

It does not reveal an unknown NT passage, does not overturn accepted current Greek, and does not require changing Biblex's translation base. The correct product posture is: use Codex H as an apparatus/study witness and note major variants, but do not treat it as a new base text.

## Follow-Up Work

1. Collate the major rows against licensed NA28/UBS5/ECM apparatus, especially Heb 1:3, Heb 2:9, Heb 4:15, Heb 10:34, Col 2:2, and Col 2:11.
2. Pull folio images or the digital edition view for Heb 4:15/4:16 to decide whether the missing final clause is a genuine omission, damage, or edition/extraction boundary.
3. Re-emit the comparison JSON with token-level `visible`, `supplied`, `source`, `unclear`, `gap`, `hand`, and `folio` metadata.
4. Keep Codex H notes in Biblex as study/apparatus annotations, not as a translation-base replacement.