From the repository

Jubilees (Mashafa Kufale)

Scope and source strategy for the Book of Jubilees, paired with Enoch on the Ethiopic track.

Source: JUBILEES.md on GitHub

Jubilees (Mashafa Kufale) — scope and source strategy

This document covers the People's Open Bible's dedicated sub-phase for the Book of Jubilees (Mashafa Kufale / መጽሐፈ፡ ኩፋሌ / "Book of Division"). It sits alongside ENOCH.md under the shared Ethiopic pipeline (see DEUTEROCANONICAL.md's Pseudepigrapha section), and applies the three-zone scholarly-source policy in REFERENCE_SOURCES.md.

Current qualification, 2026-10-05: chapter and reader drafts exist; manuscript-specific source review is incomplete. Three byte-identical OCR runs are not a measured character-accuracy benchmark, and an English cross-check cannot certify a Geʿez transcription. Source PDFs are hash-bound in the manifest, not committed to Git. The Jubilees 1:27 comparison retains the working source/English and holds the recording-role claim until primary apparatus/plate evidence is available. Do not bulk regenerate reader verses from chapter drafts: later per-verse revisions can be overwritten. The historical plan below is not a completed all-witness collation.

Why Jubilees

  1. Canonical in the Ethiopian Orthodox Tewahedo Church — part of the 81-book canon (alongside Enoch, Meqabyan, and others). Eritrean Orthodox also receives it. No permissively-licensed modern English exists for these communities.
  2. Deeply influential on Second Temple Judaism and early Christianity. Jubilees is quoted in the Damascus Document, by Tertullian, Syncellus, and indirectly by Epiphanius. Its calendrical theology shaped Qumran sectarian thought.
  3. Hebrew witnesses survive at Qumran, including 4Q216. Identify the composition and surviving content of each siglum before comparison; the former blanket 4Q216–228 equivalence was inaccurate. In particular, 4Q225 is discussed as Pseudo-Jubilees, not automatically another copy of the same work. Early attestation makes these valuable evidence for Jewish interpretation of Genesis–Exodus, without by itself fixing a composition date or the earliest local wording.
  4. Textual situation parallels Enoch. Hebrew original → Greek translation (now largely lost, preserved in citations) → Ethiopic (complete, only full witness). Latin fragments survive for chapters 13-49 (Rönsch 1874), Syriac citations exist. The translation pipeline is nearly identical in structure to Enoch, which is why we group them in the same Ethiopic phase.

Canonicity

TraditionStatus
Ethiopian Orthodox TewahedoCanonical (81-book canon)
Eritrean OrthodoxCanonical
Beta Israel (Ethiopian Jewish)Sacred
Qumran communityQuoted authoritatively (Damascus Document)
Roman CatholicNot canonical
Eastern Orthodox (Greek/Russian)Not canonical
ProtestantNot canonical

POB policy: no position on canonicity; transparent labeling.

Compositional overview

Jubilees retells Genesis 1 through Exodus 14 as a revelation to Moses on Mt. Sinai from the "Angel of the Presence." Its distinctive feature is the division of history into jubilees (49-year units), weeks (of years), and years — from creation through the exodus.

50 chapters, ~1,350 verses in Charles's numbering. No compositional subdivisions like Enoch's 5 books; Jubilees is a single unified composition.

Source editions — vendored

Ethiopic (primary)

EditionYearRoleLicense
Charles, The Ethiopic Version of the Hebrew Book of Jubilees (Oxford, Anecdota Oxoniensia Semitic Series I.viii)1895Zone 1 primary — critical Ethiopic based on 4 MSS (A, B, C, D from Paris/British Museum/Tübingen/Abbadie) with full apparatusPublic Domain
Dillmann & Rönsch, Das Buch der Jubiläen oder die kleine Genesis (Leipzig)1874Zone 1 secondary — Dillmann's German translation + Rönsch's recovered Latin fragments (chs 13-49) edited criticallyPublic Domain

Latin (secondary witness, chs 13-49)

EditionYearCoverageLicense
Rönsch (in Dillmann-Rönsch 1874)1874~half the book; independent witness to the lost GreekPublic Domain (Rönsch d. 1888)

English reference (not reproduced; verse numbering + cross-check)

EditionYearLicense
Charles, The Book of Jubilees, or The Little Genesis (London: Black)1902Public Domain
Charles in Apocrypha and Pseudepigrapha Vol. 21913Public Domain

See sources/jubilees/MANIFEST.md for SHA-256 hashes and rehydration commands.

Zone 2 consult (scholarly, not reproduced)

  • VanderKam, The Book of Jubilees (CSCO 510-511, Scriptores Aethiopici 87-88, Leuven 1989) — modern critical Ethiopic edition with English translation
  • VanderKam & Milik, Qumran Cave 4 VIII: Parabiblical Texts Part 1 (DJD XIII, 1994) — the Qumran Hebrew Jubilees fragments 4Q216-228
  • Segal, The Book of Jubilees: Rewritten Bible, Redaction, Ideology and Theology (Brill, 2007) — modern critical commentary
  • Wintermute, "Jubilees" in OTP vol. 2 (Charlesworth ed.) (1985) — standard English translation with introduction

Translation strategy

Same three-zone discipline, adapted for Jubilees:

  1. Primary source: our own Gemini 3.1 Pro OCR of Charles 1895 Ge'ez, cross-validated against Dillmann/Rönsch 1874 as a second witness. No Beta maṣāḥǝft-style digital Ge'ez oracle exists for Jubilees (unlike Enoch), so our OCR cross-check relies on the two PD scholarly editions alone.
  2. Secondary textual apparatus for chapters 13-49: the Latin fragments via Rönsch 1874 preserve the earliest textual stratum for that material and are shown per-verse alongside the Ethiopic.
  3. Qumran Hebrew fragments: Zone 2 consult via VanderKam-Milik DJD XIII. We footnote fact-level agreements ("4Q219 attests the reading here") without reproducing VanderKam's reconstructions.
  4. Per-verse YAML: standard POB schema with an enoch_witnesses analog (jubilees_witnesses) listing which languages attest each verse.

Pipeline components (planned — shared with Enoch)

The Ethiopic pipeline is built once and used by both Enoch and Jubilees:

ComponentPurposeShared / book-specific
tools/ethiopic/ocr_geez.pyGemini 3.1 Pro plaintext-mode OCR of Ge'ez scansShared
tools/ethiopic/run_parallel_ocr.pyMulti-key / multi-worker batch runner over ocr_geez.pyShared
tools/ethiopic/verse_parser.pyParse Ge'ez numeral verse markers (፩–፼)Shared
tools/enoch/multi_witness.pyEnoch-specific witness aggregatorEnoch-specific
tools/jubilees/multi_witness.pyJubilees-specific aggregator (Ge'ez + Latin + Zone 2 Qumran Hebrew)Jubilees-specific
tools/jubilees/detect_chapters.pyDetect chapter-start pages across Charles 1895Jubilees-specific
tools/jubilees/build_page_map.pyBuild/update sources/jubilees/page_map.json from detection cacheJubilees-specific
tools/*/build_translation_prompt.pyBook-specific translator prompt assemblyBook-specific

OCR note — same as Enoch

Critical findings:

  • 2026-04-21: Azure GPT-5 fails on Ge'ez; Gemini 2.5 Flash hallucinates; Gemini 2.5 Pro plaintext mode succeeds on Enoch.
  • 2026-04-22: Gemini 3.1 Pro preview is the regime-shift winner for Jubilees: chapter 1 benchmark closed with 100% byte-identical 3-run consistency and correct prologue/verse-1 structure.

Relative costs per page (rough):

  • Azure GPT-5: ~2K tokens, FAILS on Ge'ez
  • Gemini 2.5 Flash: ~1K tokens, produces Ge'ez but hallucinates content
  • Gemini 2.5 Pro plaintext: workable but less stable on Jubilees
  • Gemini 3.1 Pro plaintext: current production default for Enoch/Jubilees

Phased timeline

PhaseWorkEffortDependencies
12a — source acquisition✓ PDFs vendored, MANIFEST writtenDone 2026-04-21—
12b — transcriptionGemini 3.1 Pro OCR of Charles 1895 Ethiopic (~180 Ge'ez pages). Build chapter map. Cross-validate against Dillmann-Rönsch 1874 where possible. OCR Rönsch Latin material separately.In progressPhase 11 (Enoch) transcription complete — gives us shared tooling
12c — translation~1,350 verses across 50 chapters, multi-witness context, three-zone prompt2-3 weeks12b complete
12d — revisionReviser pass per REVISION_METHODOLOGY.md1 week12c complete
12e — releaseTagged release, status.json updated1 day12d complete

Total ≈ 4-6 weeks once Phase 11 (Enoch) establishes the shared Ethiopic tooling. Much of the per-book effort is just running the shared pipeline against a different book's sources.

Unique strengths of this translation

  • First CC-BY 4.0 modern English Jubilees. VanderKam 1989, Segal 2007, Wintermute OTP 1985, and Charlesworth collections are all copyrighted. Charles 1902 is PD but 123 years old and Victorian-register.
  • Ethiopian Orthodox / Eritrean Orthodox readers gain a permissive modern English.
  • Transparent textual provenance per verse — Ethiopic reading, Latin fragment (where applicable, chs 13-49), Qumran Hebrew consult (Zone 2), all documented in the per-verse YAML.
  • Shared infrastructure with Enoch keeps the pipeline lean and makes both works benefit from any improvements.

Future extensions in the Ethiopian broader canon

If and when Jubilees completes successfully, candidates for further Pseudepigrapha phases remain:

  • Meqabyan 1-3 — Ge'ez-only, Ethiopian Orthodox canonical
  • 4 Baruch / Paraleipomena Jeremiou — Greek + Ge'ez
  • 2 Baruch — Syriac primary

These are not committed to. Each would get its own scope doc if we pursue it.