Governing manuscript text
INTF NTVMR transcription of GA 03
The published original-hand TEI transcription for NTVMR document 20003 controls what Greek is displayed. Its Gospel files are pinned by SHA-256 in the audit record.
CC BY 4.0
The Gospels across their earliest witnesses, and the tradition they carried forward
GA 03 · four-Gospel witness
The Vaticanus column represents one manuscript witness: Codex Vaticanus, Gregory–Aland 03. It does not display a reconstructed critical text, a blended Alexandrian tradition, or words borrowed from another column. If Vaticanus omits something, the column says omitted. If its physical text is unavailable, the site preserves that condition rather than filling it.
3,779
Gospel verses classified
63,511
INTF source tokens
63,546
displayed lexical words
63,543
words with certified English
134
orange system-generated glosses
3
red manuscript events without English
The token and lexical-word totals differ because 35 joined manuscript tokens contain two identifiable lexical units and are displayed as 70 cells without changing the archived diplomatic source record. English coverage is 99.995%.
Sources have different authority. Only the governing transcription decides the Vaticanus Greek; the remaining sources check, place, or explain it.
Governing manuscript text
The published original-hand TEI transcription for NTVMR document 20003 controls what Greek is displayed. Its Gospel files are pinned by SHA-256 in the audit record.
CC BY 4.0
Independent transcription check
A pinned CNTR revision is compared automatically with INTF. Agreement corroborates a reading; disagreement is recorded but never silently changes the governing INTF text.
CC BY-SA 4.0
Lexical and morphological checks
These sources identify lemmas, morphology, and lexical English. They explain a Vaticanus form; they do not become the Vaticanus text. Shared ancestry is not counted as independent agreement.
Pinned open or public-domain datasets
Alignment aid
TAGNT supplies contextual alignment and existing English candidates only after the Greek identity is fixed. It is never allowed to replace a Vaticanus omission or reading.
CC BY 4.0
The complete versions, local paths, licenses, and upstream records are listed on the Sources & Rules page.
Pin the INTF original-hand Gospel transcriptions and record their hashes and reading layer.
Preserve words, word divisions, corrections, uncertain letters, gaps, nomina sacra, omissions, and lacunae as distinct states.
Compare every Vaticanus token with CNTR using exact and declared reversible normalization. INTF remains governing where they disagree.
Place tokens between monotonic source anchors. Existing shared rows are exhausted before a new row is introduced. Another tradition may guide placement but cannot supply Vaticanus text.
Admit English only when lexical identity is supported by the declared source chain. Site-generated results are orange; manuscript conditions are red.
Leave English blank when evidence cannot support one result. A blank is a controlled scholarly outcome, not permission to guess.
Run invariant, source-concordance, generated-English, and production-build checks. Publish the decision hash and retain all disagreements.
49,975
tokens corroborated exactly by the CNTR shadow
13,338
corroborated after declared reversible normalization
198
INTF-governed disagreements retained rather than harmonized
Both the live four-Gospel source audit and the generated-English audit completed with zero invariant failures. The English decision ledger is fixed by SHA-256 61cf5809fb7921635a072cd87e3802b3b3284901d912aed2ad414f365b0bf1c9.
Reproducibility is necessary, but it is not the last scholarly step. External certification would add qualified manuscript specialists, published review criteria, independent reruns from the pinned sources, adjudication of the retained disagreements, image-level spot checks where open scans permit them, versioned corrections, and signed reviewer findings.