Extra-canonical scripture in the People's Open Bible

This document catalogs the texts outside the 66-book Protestant canon

• 13 LXX deuterocanonical books that the People's Open Bible is committed to translating and releasing under CC-BY 4.0. It complements DEUTEROCANONICAL.md (the 13 LXX-based deuterocanonical books), 2ESDRAS.md (the dedicated Latin-primary sub-phase for 2 Esdras), and REFERENCE_SOURCES.md (the three-zone scholarly-source policy applied throughout).

Status: active provisional expansion (updated 2026-08-29). The shared catalog now tracks 29 early-Christian works: Gospel of Philip and Gospel of Mary, the earlier 23-work expansion, and four ancient church orders. Source acquisition, bridge drafting, grounding review, metadata, and client publication are independently gated. A prepared or published bridge is not a claim of final direct source-language translation.

The four church-order witness bridges are Didascalia Apostolorum → Apostolic Tradition → Apostolic Church Order → Apostolic Constitutions. Their sources, cautions, and direct-review plan are recorded in docs/church_orders_source_plan.md.

Pseudo-Clementine source-only addition (2026-10-11)

The shared catalog additionally registers four unpublished Greek-source works: the Letter of Peter to James, Contestation, Letter of Clement to James, and Homilies. This brings the registry to 33 works while retaining 29 published catalog entries. These are historical literature, not additional canonical Scripture. English must be freshly generated from Greek and separately checked against the source before reader publication. The imported sources retain CC BY-SA 4.0, an exception to the general CC BY output goal above; output rights must be resolved before translation publication. The supplied Homilies stop at 19.14 and must not be presented as complete. See the source inventory and mandatory checks.

Why we include these

The People's Open Bible's mission is transparent translation of the biblical textual tradition in its fullest historical form, letting readers and their faith communities evaluate. That mission does not end at Protestant or Catholic canon borders. Three concrete reasons for expanding beyond the deuterocanon:

1. The New Testament quotes some of these texts explicitly. Jude 14-15 directly quotes 1 Enoch 1:9. Responsible engagement with Jude benefits from having a readable, auditable 1 Enoch.

2. Other Christian traditions receive some of these as canonical. The Ethiopian Orthodox Church's list includes Enoch and Jubilees. This is evidence of that community's reception, not a universal Christian canon or a conclusion derived from finding a manuscript. Other communities' lists require their own references and identification of the work or recension.

3. Early Christian writings shaped the formation of the NT canon itself. The Codex Sinaiticus project identifies Barnabas and the Shepherd of Hermas alongside its NT. The British Library's Alexandrinus record identifies 1 Clement at ff. 134r–143r and part of 2 Clement at ff. 143r–144v. Inclusion establishes a manuscript's contents, not necessarily equal canonical rank. The Didache merits separate reception study; the preceding examples do not attest its inclusion.

Terminology: "Extra-canonical"

We use "extra-canonical" as a deliberately neutral umbrella: extra meaning beyond / outside of, not extra as in additional Scripture authorized to modify doctrine. These texts have varying reception across Christian traditions: some appear on a named church's canon list, some in historical Apocrypha sections or appendices, and others are presented here only as historical writings. Printing or binding a work with a Bible does not itself establish canonical acceptance. Every release needs a tradition-specific reception label distinct from POB's decision to include it in a historical library.

For a scholar-facing label, "pseudepigrapha" is often used for the Jewish apocalyptic/testamentary works, "Apostolic Fathers" for the early Christian writings, and "Nag Hammadi library" for the 4th-c. Coptic codices. Our umbrella term lets one repository house all of them without forcing a single scholarly-tradition vocabulary on readers who won't know those conventions.

Scope

The three tiers below organize the existing library roadmap, not quality, inspiration or a universal ranking of canonical status. Reception must be recorded for each work and community; a tier cannot supply that evidence.

Tier 1 — Canonical reception or inclusion in historical Bible editions

This group combines different kinds of reception. Enoch and Jubilees are named on the Ethiopian list above. An Apocrypha section, appendix or table-of-contents entry is a different claim and does not prove that every work in this tier was received as Scripture. Keep the exact title, textual form, community and evidence separate.

Tier 2 — Apostolic Fathers, Jewish apocalyptic, and pseudepigrapha

These works warrant study of early Christian or Second Temple Jewish literature. Some have the specific codex inclusion described above; that does not apply to every row. POB presents this group as historical comparanda, not additional NT books. Claims about current canonical reception require community-specific evidence and recension checks; neither universal exclusion nor scholarly unanimity is established here.

Tier 3 — Early Christian mystical, apocryphal, and Nag Hammadi texts (with framing)

These works are included because they are historically important, not because POB endorses their theology. We identify the date of the work separately from the date of its surviving manuscript, preserve damaged or reconstructed passages, and state the work's literary and theological character. We do not present them as Scripture in the canonical sense.

Expansion order

The authoritative ordered backlog now lives in sources/early_christian_texts/catalog.json. Source availability and responsible framing determine when a work can ship; theological disagreement by itself does not remove a text from consideration.

23-work provisional expansion

The following works are being prepared as a single catalog-backed expansion. All are labeled non-canonical in POB. “Public-domain bridge” means that POB may produce a fresh, openly licensed provisional rendering from an identified public-domain witness; it does not mean that the draft has completed direct review against Coptic or Greek.

What “provisional” requires

1. No canonical ambiguity. Each work is presented as a historical early Christian writing, not as an additional New Testament book.
2. No authorship overclaim. Traditional titles such as “Gospel of Peter” or “Epistle of Barnabas” are retained for discovery, while metadata states when authorship is anonymous or pseudonymous.
3. No source-language overclaim. A public-domain English bridge is labeled provisional_source_bridge; it may be useful and readable, but it is not called a direct Coptic or Greek translation.
4. Independent grounding review. Every bridge unit receives a separate review before publication, with source-language review remaining pending until completed by the appropriate pipeline.
5. Stable editorial divisions. Reader “chapters” may be editorial sections, letters, fragments, or ancient chapters. Those divisions are identified as navigation aids and are not represented as manuscript chapter numbers when the witness has none.

Manuscript damage and editorial punctuation

Fragmentary witnesses must look readable without pretending that the missing text survives. The source record and translation record retain the distinction among:

• lacuna / illegible text — a genuine manuscript gap, represented in source data as a lacuna marker and rendered by clients as a labeled gap rather than unexplained raw [...];
• restored text — letters or words reconstructed by an editor, visibly marked as a restoration and available in provenance;
• supplied words — words added for intelligible English, marked as a translator supplement rather than manuscript text;
• ordinary parentheses — reserved for parenthetical content that is actually present in the witness, not used as a generic uncertainty marker.

No drafting or display layer may silently invent wording across a lacuna. A smooth reading may explain the surrounding sense, but the translation text and provenance must preserve the boundary of what is lost, uncertain, restored, or supplied.

Pipeline groupings

Implementation cost depends on which source-text pipeline a text reuses. Grouping by pipeline makes the phased plan natural.

Group A — reuses LXX-style Greek pipeline (Phase 8 infra)

• Didache
• 1 Clement
• Shepherd of Hermas
• Psalms of Solomon (already in Swete vol III; full page span now transcribed)
• Testaments of the Twelve Patriarchs

Shared infra: vision OCR of Greek typeset text, 4-source triangulation against independent transcriptions where they exist, scan-grounded adjudication against the scan image. For the non-Swete Group A books, see GREEK_EXTRA_CANONICAL.md and tools/greek_extra_pdf_ocr.py.

Group B — reuses 2 Esdras multi-witness pipeline (Phase 8b infra)

• 2 Esdras itself (Phase 10)
• 2 Baruch (same Semitic-apocalypse family as 2 Esdras, but with a Syriac-primary base text; the currently rehydrated Violet 1924 copy is a German-form GCS volume rather than the 1910-style witness parallels, so it reuses the surrounding tooling but not the exact OCR prompt)

Group C — new Ethiopic (Ge'ez) pipeline

• 1 Enoch
• Jubilees

What's new: vision OCR for Ethiopic script (Unicode block U+1200–U+137F), Ge'ez-specific transcription prompts, Ethiopic verse-index data model. Both texts amortize the pipeline investment.

Group D — new Coptic pipeline (with careful license handling)

• Gospel of Thomas
• Gospel of Truth
• Thunder, Perfect Mind

What's new: vision OCR for Coptic script (Unicode block U+2C80–U+2CFF) plus substantive licensing research, because the Coptic primary-source situation is fundamentally different from Groups A/B/C. The 4th-century codices are themselves public domain by age, but we don't have physical access to them; every scholarly transcription is a copyrighted reconstruction.

Licensing research — findings (2026-04-21)

Unlike for 2 Esdras (Bensly 1895 is cleanly PD) or 1 Enoch (Charles 1906 is cleanly PD), for Nag Hammadi there is no pre-1929 published Coptic transcription of any of the three target texts. The codices were discovered in 1945 and first published by Pahor Labib starting in 1956. Every Coptic transcription in existence is post-1929 and therefore potentially under copyright.

Full survey of accessible Coptic primary sources:

Pragmatic path forward for Phase 16

Because the cleanest candidate (Mattison/Zinner) is ambiguous on the Coptic specifically, Phase 16 begins with a licensing-clarification step rather than jumping to OCR:

1. Email Mark Mattison and Rachel Bousfield (Other Gospels) requesting a written CC0 or CC-BY 4.0 statement covering the Zinner Coptic transcriptions specifically (not only the translations). If granted, all three texts have a clean Zone 1 primary and Phase 16 proceeds normally.
2. Email Joseph Gebhardt-Klein (U. Nevada Reno) requesting CC-BY release of his 2024 Thomas transcription. Fallback if Mattison/ Bousfield can't grant.
3. For Gospel of Thomas regardless of (1)/(2): use the Grenfell & Hunt Greek P.Oxy. fragments as definitively-PD Zone 1 for the ~20% of sayings they cover.
4. If both (1) and (2) fail: pursue either (a) a formal SCA licensing application for direct manuscript photograph access, or (b) a Coptic Scriptorium partnership to accelerate their Nag Hammadi roadmap for the three specific texts we need. Both are months-scale processes.
5. Until Zone 1 is cleanly established, do not describe a bridge as a direct Coptic translation or begin Coptic OCR from a restricted transcription. An explicitly public-domain English witness may be used for a clearly labeled provisional bridge, but it does not satisfy the source-language review gate and cannot erase manuscript uncertainty inherited from that witness.

This explicit gate is why Group D is the last pipeline in the phased rollout. We do not want the project to ship Gnostic texts under muddy provenance.

Zone 2 consult registry for Group D (common across all three)

When Zone 1 is established and translation begins, the following copyrighted works will be named in the translator-prompt registry as consulted references (not reproduced, per the standard three-zone policy):

• Bentley Layton, Nag Hammadi Codex II,2-7 together with XIII,2, Brit. Lib. Or. 4926(1), P.Oxy. 1, 654, 655 (Brill NHS 20-21, 1989) — critical edition of Codex II including Gospel of Thomas.
• Harold W. Attridge (ed.), Nag Hammadi Codex I (The Jung Codex) (Brill NHS 22-23, 1985) — critical edition of Codex I including Gospel of Truth.
• Douglas M. Parrott (ed.), Nag Hammadi Codices V,2-5 and VI (Brill NHS 11, 1979) — critical edition of Codex VI including Thunder Perfect Mind.
• Marvin Meyer, The Gnostic Scriptures: Introductions, Translations, Annotations, Indexes (HarperOne, 2007) — English translations and introductions.
• April D. DeConick, The Original Gospel of Thomas in Translation (T&T Clark, 2007) — reconstruction-oriented scholarship on Thomas.
• Elaine Pagels, The Gnostic Gospels (Random House, 1979) — Pulitzer- winning framing/context, not a critical edition.
• Samuel Zinner's published essays on Thunder Perfect Mind.

Phased rollout

Source-acquisition and pipeline-setup work is already underway in parallel with Phase 9 prep. Translation drafting for each begins after Phase 9 (LXX deuterocanon) completes:

Phases after 10 are parallelizable by pipeline group. Phase 13 is where we can produce the largest volume of newly-translated extra-canonical material relative to effort — Greek is what our stack handles best.

Independent work sessions (operational note)

This section is operational rather than reader-facing and may be removed once the roadmap becomes stable. It documents which parallel work-streams can run without stepping on each other, so distinct sessions (whether separate people, separate Claude sessions, or separate worktrees) can proceed in parallel.

Because each pipeline group touches a different source tree and different tools, the groups can proceed independently. The mapping of pipeline groups to independent work sessions:

Why these groupings (and not finer splits)

• B stays together because 2 Esdras and 2 Baruch still share the same Semitic-apocalypse provenance model and a lot of the same local tooling (OCR orchestration, witness apparatus, provenance schema), even though Violet 1924 is not the same parallel-column page layout as Violet 1910.
• C stays together because the Ge'ez OCR backend decision and typesetting-prompt work applies equally to Enoch and Jubilees, both edited by Charles in similar editions.
• D stays together because every text in the group fits the same Greek-OCR + 4-source-adjudication pattern we already use for LXX. The Apostolic Fathers editions (Lightfoot, Bryennios, Funk) cover multiple texts per source acquisition.
• E is alone because Coptic is a new script, the OCR work is genuinely different, and the licensing research (documented above) is a blocker that must be resolved before any code or OCR work starts.

Which sessions are live right now

• A — Phase 9 drafting, separate dedicated session.
• B — 2 Esdras setup, this session or a dedicated 2 Esdras session.
• C — Enoch + Jubilees setup, separate dedicated session.
• D and E are not started. D can begin any time (lowest risk, highest volume per unit effort). E is gated on the Mattison/Zinner Coptic-licensing clarification step documented in the Group D research above.

When to remove this section

Once all sessions have run at least one full phase cycle, the grouping will have stabilized into normal Phase-N entries in the CHANGELOG. At that point, this operational notes section becomes redundant and can be deleted. Until then, it serves as a cross-session contract of who-owns-what.

Labeling and framing principles

Every extra-canonical text released by POB carries the following in its reader-facing front matter and per-verse provenance:

1. Plain-language description of what the text is and when it was composed.
2. Canonical-status table across traditions — same format as the deuterocanonical section. For texts canonical nowhere, this is stated explicitly.
3. Textual-situation summary — what the primary surviving witness is, what was lost, what languages the text passed through.
4. Reception-history note — how the early Church, Second Temple Jewish communities, or later scholars received it.
5. For Tier 3 especially: explicit framing — that the text is 4th-century Christian mystical/contemplative literature, not canonical Scripture in any historical Christian canon, and that its inclusion is for historical and theological-literary study, not doctrinal formation.

The reader should never be confused about whether a given book is "Scripture" in the same sense as Romans. Labels are information, not editorial judgment.

License handling

Same three-zone policy as every other POB text:

• Zone 1 (vendored, safe as source): Pre-1929 public-domain source editions (Charles 1895/1902/1906/1912, Bryennios 1883, Lightfoot, Bensly 1895, Ceriani 1871, Ryle & James 1891, Grenfell & Hunt 1897/1904, Violet 1910/1924, etc.)
• Zone 2 (consult, not reproduced): Modern critical editions, translations, and commentaries (Ehrman Loeb Apostolic Fathers, Milik Books of Enoch, Charlesworth OT Pseudepigrapha, Nickelsburg 1 Enoch Hermeneia, Layton/Meyer/DeConick on Nag Hammadi, Robinson Nag Hammadi Library, DJD volumes).
• Zone 3 (forbidden): modern commercial English translations of the Bible or of these specific texts. Not consulted during drafting, to prevent derivative-work exposure.

Each per-text translation ships with a per-verse YAML showing Zone 1 sources consulted at drafting time and Zone 2 scholars named at drafting time (never their specific words).

Why doing this at all matters

Three things happen when we open this work under CC-BY 4.0:

1. Every Christian tradition gets a reference point. An Ethiopian Orthodox reader has an open, modern 1 Enoch. A scholar teaching patristics has an open, modern Didache. A pastor who wants to understand Jude 14-15 doesn't have to choose between a Victorian English translation and a copyrighted one.

2. The formation of the canon becomes legible in one library. You can read the Didache next to James, 1 Clement next to Hebrews, Psalms of Solomon next to the Psalter, 2 Esdras next to Daniel. The historical conversation the canonical authors were embedded in becomes visible in a single browsable corpus.

3. Nothing about this threatens orthodoxy. Canonical labeling is transparent and preserved. Readers know what they are reading. Traditions that receive these texts as Scripture finally have a free modern version; traditions that don't receive them have them as historical context. Neither tradition is asked to change; both are served.

That is why this is on the roadmap, and why it is worth doing carefully.