Governing Greek
Robinson–Pierpont 2018
Only the byztxt v3.3.2 RP2018 files determine the displayed Byzantine Greek. They are pinned to upstream commit 27a45ff1b7be6c17ccbfeac414f3f55732ae8e28.
RP2018 · four-Gospel edited textform
The Byzantine Textform on Urevangelium
The Byzantine column represents the Robinson–Pierpont 2018 Byzantine Textform electronic edition. It is an edited representation of the Byzantine Greek tradition—not one physical manuscript and not a reconstruction of the original Gospel text. Its distinctive readings and omissions are preserved rather than harmonized with another column.
Certification claim: all four Gospels passed Urevangelium’s reproducible RP2018 source-to-display and English-admission checks. This is internal source and process certification, not independent scholarly peer review.
Current certified corpus
3,778
RP2018 Gospel verse records
66,130
displayed RP2018 Greek tokens
66,130
admitted English placements
61,774
direct contextual placements
4,356
orange adjudicated placements
0
unresolved placements
Luke 17:36 is marked omitted because RP2018 contains no verse record there; it is not filled from another edition. English coverage is 100% under the declared admission rules.
Source hierarchy
Only RP2018 governs the Greek. Every other source has a narrower role in identity, alignment, lexical explanation, or verification.
Source identity
RP2018 Strong and parsing files
The matching v3.3.2 morphology files establish the identity and grammatical form of every displayed token.
Contextual English
STEPBible TAGNT
English is eligible only where TAGNT explicitly identifies the Byzantine reading and the token aligns exactly to RP2018. TAGNT never governs the Greek.
Lexical evidence
TBESG / Abbott-Smith
Extended Strong identities and published lexical evidence check the English bridge. Site-selected results display in orange.
Secondary verification
MorphGNT, PROIEL, and MorphGNT lexicon
These sources adjudicate legacy Strong-number and lemma conventions without importing a reading or English cell from another tradition.
Versions, paths, hashes, licenses, and governing rules also appear on the Sources & Rules page.
How the system works
1
Pin
Freeze RP2018, its upstream commit, and every local source hash.
2
Rebuild
Generate the column solely from the RP2018 Gospel records.
3
Verify
Match all 66,130 cells to RP2018 surface, order, Strong number, and parsing.
4
Align
Place readings in the shared grid without changing wording or source order.
5
Admit
Use explicit Byzantine context or the declared lexical chain for English.
6
Disclose
Color project-adjudicated English orange.
7
Certify
Require complete accounting, passing tests, and a production build.
What the English means
- Normal English
- Contextual English from an explicitly Byzantine TAGNT reading after RP2018 identity and lexical checks.
- Orange English
- A lexical result selected by Urevangelium’s declared evidence chain. Orange marks the project’s responsibility; it is not presented as a published continuous translation.
- No cross-column transfer
- Another tradition’s displayed English is never copied into this column.
Rules that cannot be broken
- No other Greek edition may replace RP2018.
- No absent word or verse may be filled from another tradition.
- No English is admitted without exact RP2018 token identity.
- No project-selected result appears as ordinary source English.
- No disagreement is removed merely to make columns agree.
- No internal certification is called independent peer review.
Certification results
All 66,130 RP2018 tokens and English decisions are accounted for with zero held placements. The reproducible decision set is fixed by SHA-256 2853943abfc8af3e78430178ae8dd65b91e61b1f6ab21755831789f0649b0dbb. All 184 automated tests and the production build passed at certification.
What remains for external scholarly certification
The next level is independent review by qualified Byzantine-text and Koine Greek specialists: rerunning the pinned corpus, auditing the rules and orange selections, examining every exceptional rule class, publishing corrections, and signing version-specific findings.