2 Esdras (4 Ezra) — scope and source strategy

This document covers the People's Open Bible's dedicated sub-phase for 2 Esdras (called 4 Esdras or 4 Ezra in the Vulgate tradition). It complements DEUTEROCANONICAL.md (which covers the 13 LXX-based deuterocanonical books), 2BARUCH.md (the sibling Syriac-primary apocalypse track), and applies the same three-zone scholarly-source policy defined in REFERENCE_SOURCES.md.

Status: source-acquisition phase. The pipeline is architecturally distinct from our LXX infrastructure and is being set up fresh. Translation drafting will begin after Phase 9 (LXX deuterocanon) completes; no English verses from 2 Esdras ship before that.

Why 2 Esdras needs its own pipeline

The 13 books in DEUTEROCANONICAL.md all have Greek source texts preserved in the Septuagint tradition. Our entire Phase 8 infrastructure — Swete OCR, 4-source Greek triangulation (ours / First1KGreek / Rahlfs / Amicarelli), scan-grounded adjudication against Codex Vaticanus — assumes a Greek source-text layer.

2 Esdras has no Greek source text. The textual tradition is:

• Original Hebrew (c. 100 CE): lost entirely.
• Greek translation (from the Hebrew): lost. Only fragments survive (quotations in patristic authors, a handful of papyrus scraps).
• Latin translation (from the lost Greek): the most complete surviving witness. This is what every English translation works from as primary.
• Daughter translations from the same lost Greek: Syriac, Ethiopic, Arabic (two recensions), Armenian, Georgian, Slavonic, and Coptic. Each preserves readings the others may have lost.

A responsible 2 Esdras translation therefore:

1. Takes Latin as primary source text.
2. Uses Syriac + Ethiopic + Arabic + Armenian + Georgian as independent witnesses to the lost Greek. Where they disagree with the Latin, the disagreement is textually significant and footnote-worthy.
3. Does not pretend there is a Greek source to reconstruct at the word-choice level — the Greek is genuinely lost.

This is not a variant of the LXX pipeline. It is a parallel effort with its own source acquisition, OCR, and multi-witness adjudication logic.

Structure of the book

2 Esdras as printed in the KJV 1611 Apocrypha and modern editions is a composite text of three independent works:

The "core" apocalypse (chs 3–14, the Seven Visions of Ezra) is what most scholarly attention focuses on. We translate all 16 chapters as a single work per the traditional English ordering, with the compositional history noted in the front matter.

Canonicity

Following the People's Open Bible policy (see DEUTEROCANONICAL.md), we take no position on canonicity. We translate and publish with transparent labeling.

Source editions — per-witness

Primary: Latin

The Latin text survives in multiple manuscript families. The standard critical editions:

Secondary witnesses (daughter translations from the lost Greek)

Vendored via Violet 1910 (parallel columns) and where possible from independent PD editions:

All of the above are either directly PD or printed in Violet 1910 which is PD.

What we do NOT include

• Slavonic 4 Ezra: preserved text but textually close to the Latin; low incremental value for a translation project and not in Violet 1910.
• Modern reconstructions of the lost Greek (e.g., Stone's speculative retroversions): Zone 2 reference only, never reproduced.

Translation strategy

Adapted from METHODOLOGY.md and REFERENCE_SOURCES.md:

1. Latin is the primary source text. English renderings anchor in the Bensly 1895 critical Latin.
2. Daughter translations provide per-verse textual-critical context. Where Syriac / Ethiopic / Arabic / Armenian / Georgian disagree with the Latin in ways that affect meaning, the disagreement is flagged in the per-verse YAML and footnoted.
3. The adjudicator does NOT try to reconstruct the lost Greek. Reconstruction is scholarly speculation and is not the translator's job. Witness disagreements are preserved as disagreements.
4. Three-zone policy applies. Stuttgart Vulgate and Metzger OTP are Zone 2 consult. NIV-family English translations are Zone 3 forbidden.
5. Per-verse provenance YAML follows the same schema as the rest of POB, with additional fields for each witness's reading.

Reader-facing presentation

Because 2 Esdras is not canonical in the Catholic, Orthodox, or Protestant traditions and its compositional layers matter to an honest reading, the book ships with explicit reader-facing labeling rather than as an unmarked translation.

Appendix designation

2 Esdras is presented as an Appendix book in the Cartha Open Bible, distinct from the 13 LXX-based deuterocanonical books. It appears after the last Apocrypha book in the reading order with a visible "Appendix" tag. It is not counted toward the Apocrypha completion metric in status.json; it is tracked as its own Appendix track.

Book-level headnote (rendered above chapter 1)

2 Esdras (Appendix)

Also called 4 Ezra or the Ezra Apocalypse. Not canonical in modern Catholic, Orthodox, or Protestant Bibles. Canonical in the Ethiopian Orthodox tradition and preserved in the appendix of the Latin Vulgate. Included in the Apocrypha of the 1611 King James Bible. Presented here because of its long reception history in the Christian tradition and its profound wrestling with evil, judgment, resurrection, and Messiah.

The book is a composite. The core (chapters 3–14) is a Jewish apocalypse written in Hebrew or Aramaic around 100 AD, after the fall of the Second Temple under Rome. The Semitic original is lost; the text survives through Latin, Syriac, Ethiopic, Arabic, Armenian, and Georgian translations of a lost Greek intermediary. Chapters 1–2 and 15–16 are later Christian additions preserved only in Latin. All three layers are translated and labeled rather than blended.

This translation is based on Bensly's 1895 critical Latin edition (Cambridge Texts and Studies III.2), with the daughter- language witnesses consulted via Violet's 1910 parallel-column edition for chapters 3–14. The "lost fragment" of 7:36–105, missing from the KJV 1611 and from most medieval Latin manuscripts, is restored to its place from Bensly 1875.

Section headers inside the book

Three visible section breaks mark the compositional layers.

Before chapter 1 — "5 Ezra":

Christian addition preserved only in Latin, c. 2nd–3rd century AD. These two chapters are distinct in origin from the Jewish apocalypse that follows; they recast Ezra as a prophet rejected by Israel, with God turning to the Gentiles. Included here because the KJV 1611 Apocrypha prints them in this position.

Before chapter 3 — "4 Ezra", the core Jewish apocalypse:

The original Jewish apocalypse, c. 100 AD. Composed in Hebrew or Aramaic after the fall of the Second Temple under Rome. The Semitic original is lost; the book survives through Latin, Syriac, Ethiopic, Arabic, Armenian, and Georgian translations of a lost Greek intermediary. Seven visions of the seer Ezra, answered by the angel Uriel — wrestling with God's justice in the aftermath of catastrophe.

Before chapter 15 — "6 Ezra":

Christian addition preserved only in Latin, c. 3rd century AD. Prophetic oracles against the nations, appended to the Latin tradition of 4 Ezra. Distinct in authorship and register from the core apocalypse. Included here because the KJV 1611 Apocrypha includes them.

Inline footnotes

At 7:28 ("my son the Messiah"):

The Latin reads filius meus Christus*: "my son the Messiah." The passage describes the Messiah reigning with the righteous for 400 years and then dying, followed by seven days of primeval silence and then resurrection. Among the most striking Jewish messianic passages from roughly the era of the New Testament. The Syriac and Ethiopic witnesses preserve similar readings, though later Latin manuscripts sometimes adjusted the numeral.*

At 7:36 ("the recovered section"):

Verses 7:36–105 are Bensly's "Missing Fragment." Cut from most medieval Latin manuscripts — almost certainly because this section teaches that intercession for the damned is refused, which later Christian piety found difficult — and therefore absent from the Vulgate and from the KJV 1611, where verse 7:35 was followed directly by what older editions numbered as verse 7:36 some seventy verses later. Robert Bensly recovered the fragment from a Spanish manuscript in 1875 and it is now restored in all modern critical editions.

What the translator prompt must honor

The build_translation_prompt.py builder enforces these reader-facing decisions at draft time:

• The book-level headnote is emitted as the book's intro field in the publishing JSON, not as chapter 1 text.
• Section headers are emitted as structural markers ahead of verses 1:1, 3:1, and 15:1.
• The 7:28 and 7:36 footnotes are emitted as footnotes[] on the corresponding verse YAML, not inlined into the verse body.
• The harder-teaching passages (7:45–61, 7:102–115, 8:38–41, 9:13–22) are rendered faithfully without smoothing; the translator prompt explicitly forbids softening Uriel's answers.

Pipeline components (to be built)

Timeline (indicative)

• Phase 8b (source acquisition): this week — vendor PDFs, build OCR pipeline. ≈ 3–4 days. OCR scaffold now in place.
• Phase 8c (transcription): OCR + hand-verify Latin + primary witnesses. ≈ 1 week. Bensly 1895 main-text raw OCR (pages 97–178) and Bensly 1875 Missing Fragment raw OCR (pages 65–83) now complete. The canonical loader path is now live across all 16 chapters, with chapter 7 specifically strengthened by the Missing Fragment (36–105). The remaining publication gaps were then completed from the public-domain digital Latin 4 Esdras text at vulgate.org, with the source URL recorded per chapter header. The final coverage and QC reports now live at:
  • sources/2esdras/latin/transcribed/COVERAGE.md
  • sources/2esdras/latin/transcribed/QUALITY_CHECK.md
  • sources/2esdras/latin/transcribed/SCAN_GROUNDING_QUEUE.md

The current follow-up pass has begun replacing supplemented text with manual Bensly-grounded segmentation. So far:

• Chapter 1 manual rescue complete for verses 1–11, 23, 25–26
• Chapter 2 manual rescue complete for verses 1–17, 44–45
• Chapter 3 manual rescue complete for verses 2–12
• Chapter 4 manual rescue complete for verses 1–9
• Chapter 8 manual rescue complete for verses 1–8
• Chapter 11 manual rescue complete for verses 1–13
• Chapter 12 manual rescue complete for verses 49–51
• Chapter 14 manual rescue complete for verses 1–14

The remaining queue is explicitly tracked in SCAN_GROUNDING_QUEUE.md.

• Phase 10 (drafting): after Phase 9 completes — translate ~400 verses across 16 chapters with multi-witness context. ≈ 2 weeks.
• Phase 10 (revision): Claude Opus reviser per REVISION_METHODOLOGY.md. ≈ 1 week.

Total ≈ 4-5 weeks of focused effort, largely parallelizable against Phase 9 LXX deuterocanon translation work once source acquisition completes.

Why this is worth doing

• No existing English 2 Esdras is under a clean open license. NRSV and NABRE are copyrighted. Brenton did not include 2 Esdras. Bensly's own 1895 English is Victorian in register and limited to his critical reconstruction. An open, modern, multi-witness-grounded 2 Esdras in English is a real contribution.
• Multi-witness auditability as our Phase 8 pipeline delivered for LXX, applied to 2 Esdras, would give each verse's textual-critical footing publicly per verse — which no existing English translation of 2 Esdras documents at that granularity.
• Ethiopian Orthodox readers have 2 Esdras as canonical but have had to use English translations produced for Protestant or Catholic audiences. An open translation is a direct service to that community.

Deferred: 1 Enoch

The Book of Enoch (1 Enoch) is structurally similar to 2 Esdras: Aramaic original lost (except Qumran fragments), Greek translation partially lost, Ethiopic the only complete witness. It is canonical in the Ethiopian and Eritrean Orthodox Tewahedo Churches and is explicitly quoted in Jude 14-15. PD source editions exist (Charles 1906 Ethiopic, Charles 1912 English, Flemming 1902).

1 Enoch is considered for a future phase after 2 Esdras, using a similar Ethiopic-primary + Greek-fragments + Qumran-Aramaic-fragments (Zone 2 only via Milik 1976) approach. Scope is larger than 2 Esdras (~108 chapters vs 16), and the Ethiopic-primary OCR pipeline is different from either LXX or Latin work.

See ENOCH.md (to be written) for the full scope document when we commit to that phase.