Bible Translations & Source Texts

Every quotation in a study rests on two choices most readers never see: which translation rendered the verse into English, and which underlying Hebrew or Greek edition that translation was made from. Neither choice is neutral: a formally literal translation of an eclectic critical Greek text argues differently than a thought-for-thought translation of the Byzantine tradition, even quoting the "same" verse. This page catalogs what we lean on and the tradeoffs of each. This repo also functions as a small research tool, so it records which of these are wired into our own queryable database (references/build/bible-text.db), and which are cited from general knowledge because the license doesn't allow us to store the text locally.

Tracking key

✅ Queryable now — full text loaded in bible-text.db, openly licensed, quote at length
🟡 Queryable, restricted — loaded, but non-commercial-only licensing (fine for this site; would need re-review if it ever monetized)
🔒 Queryable for verification, not bulk use — verse text loaded in the external, quotation-only study-notes.db (not bible-text.db), as a byproduct of a commercial study Bible built from that same commercially copyrighted translation. Good for checking a quotation against source instead of trusting memory; not a licence to bulk-cite or concordance-search it the way a ✅ text can be.
❌ Not tracked anywhere — commercially copyrighted, and not present in either database; cited from general knowledge with attribution, never bulk-quoted

See references/README.md and Public Data Sources for the full engineering-level catalog this table summarizes, and Our Data Sources for the plain-language, all-sources overview this page's translation-and-source-text slice fits into.

Translation philosophy

Our default is the ESV for prose quotation. We cross-check it against NASB, NIV, and NLT, for how differently-weighted translation philosophies render a disputed or ambiguous verse. Against the LSB where the divine name is at stake, since it prints Yahweh where the others print LORD. And against BSB, WEB, ASV, and YLT when we need a text we can actually store, search, and quote in full without a permissions ceiling. That makes the BSB the translation readers see most: it is the verse text in every pop-up on the site and on every group sheet, and the one a study quotes when a passage has to be reproduced at length. It took that place from the WEB in October 2026 because it translates the same Greek text as the ESV the studies quote; the WEB's New Testament translates the Majority text and prints verses such as Matthew 17:21 that those studies treat as later additions. The Masoretic Text and Septuagint anchor original- language word studies. The Greek New Testament entries below anchor NT word studies and textual- criticism notes. See AGENTS.md for the standing rule this page expands on.

Three questions to ask of any text on this page

"Hebrew text" and "Greek text" are not one category each. They cover things that differ in ways that decide what a citation can carry, and grouping by language alone hides it. This page grouped two 19th-century translations under Hebrew Old Testament for a long time on that mistake.

1. Is it in the language it was composed in? A text can be in Hebrew without being a Hebrew witness. Delitzsch's New Testament is Hebrew and was translated out of Greek in 1877; it witnesses a translator, not an original. The Septuagint is the converse case: Greek, and a translation of Hebrew, but made before Christ, so it witnesses a Hebrew text older than any Hebrew manuscript we hold.

2. What kind of edition is it? This is the "compilation" question, and it has five answers:

Kind	What it means	Example
Manuscript	An actual surviving copy	The Dead Sea Scrolls
Diplomatic edition	One manuscript, printed as it stands	The Westminster Leningrad Codex
Eclectic edition	Compiled by choosing among many witnesses	SBLGNT, NA28, the Textus Receptus
Derived text	Produced from other editions by a stated procedure	UGNT, following the Bunning Heuristic Prototype
Translation	Rendered into another language	The Septuagint, Delitzsch, every English version

A diplomatic edition tells you what one scribe wrote. An eclectic edition tells you what its editors judged. Neither is "the original", and they fail in opposite directions: the first inherits one copyist's mistakes, the second exists in no manuscript at all.

3. What was it compiled from, and did translation happen anywhere in that chain? Every step away from the autograph is either an editorial remove (someone chose between witnesses) or a linguistic one (someone rendered it into another language). Brenton's Septuagint carries both twice over: an ancient translation out of Hebrew, an edition of that translation's manuscripts, and Brenton's own English rendering beside it. A word study working from the English is four removes out, and should know it.

None of this makes a text unusable. It decides what it is evidence of.

These three questions are now recorded as data

Everything on this page was prose, which meant a reader could apply it and an automated research tool could not. The same judgements now live in machine-readable form in references/sources.toml, covering all 320 works in the database, so a tool can be asked what a source is good for before it quotes from it.

Each source carries:

Field	What it holds
`kind`	the role it plays — witness, translation, index, lexicon, annotation, commentary, grammar
`edition`	the vocabulary from question 2 above, verbatim: manuscript, diplomatic, eclectic, derived, translation
`era`	ancient, medieval, reformation, early-modern, modern
`purpose`	one sentence on what it is for
`strengths` / `limits`	what it can and cannot settle

kind is a list, because question 1's hard cases are real: the Septuagint is recorded as both witness and translation, the case question 1 describes.

There is deliberately no "primary / secondary / tertiary" ranking.It mixes what a thing is with how far it sits from the original: a concordance is not a source but an index over one, and Strong's (1890) and TWOT (1980) feel primary because they sit close to the words while being late reference works. More fundamentally, that ranking is a property of the question, not the source: 1 Enoch is a primary source for Second Temple Judaism and a contextual one for reading Jude; Eusebius is primary for fourth-century church history and tertiary for Acts. A fixed field would answer confidently and wrongly. The axes above stay stable whatever is being asked.

Every entry must record limits, and a build check fails if one does not. Every source has some, and a profile claiming none has not been thought about. The same check fails if a source's stated edition is not one of the five in question 2, which keeps this page and the data from drifting apart.

English translations

Translation	First published	Approach	Strengths	Cautions	Tracked here
ESV — English Standard Version (Crossway)	2001, revised 2007/2011, fixed as a "Permanent Text" in 2016	Essentially literal (formal equivalence), descended from the RSV	Our default citation text — literal enough for close reading, extensive textual footnotes, wide adoption across the study-Bible literature we cite in References sections; the 2016 text freeze means citations won't drift out from under us	RSV lineage carried forward a handful of textual choices scholars still debate (e.g. Isaiah 7:14's footnote); some gender-language renderings read inconsistently across books	🔒 `esv-study-bible` in `study-notes.db` — always verify a quotation here rather than from memory
NASB — New American Standard Bible (Lockman Foundation)	NT 1963, full Bible 1971, revised 1995, updated 2020	Maximally literal formal equivalence	The translation to reach for when word order or grammatical structure in the original matters to the argument; 2020 update modernized archaic pronouns while keeping the literalism	Literalness produces wooden prose in places; supplied-word italics clutter reading aloud; less used in general congregational settings than ESV/NIV; no stated verse-quotation threshold in Lockman's own permission notice, unlike every other translation on this page — don't assume a safe-harbor allowance the way ESV/NIV/NLT below have one	🔒 `nasb-1995` / `nasb-2020` in `study-notes.db` — verify-only, see the caution above before quoting
LSB — Legacy Standard Bible (The Lockman Foundation / Three Sixteen Publishing)	2021	Formal equivalence — a direct revision of the NASB 1995, by scholars from The Master's Seminary	Renders the divine name as Yahweh throughout the Old Testament rather than "LORD" — the reason to reach for it whenever a study turns on the covenant name, since no other commercial translation here keeps that distinction visible; translates doulos consistently as "slave" across the NT rather than alternating with "servant"; keeps original units and currency in the text with conversions moved to the notes; labels Hebrew acrostics with their letters	Very new, so almost no commentary or study-Bible ecosystem has grown around it yet; "Yahweh" is a minority rendering that reads as unfamiliar and usually needs a word of explanation when quoted to a general audience; it deliberately keeps "Lord" for Kurios in the NT even where the NT quotes a Yahweh passage, which is a considered choice but an easy one to misread as inconsistency; the strictest quotation cap of any translation on this page — 1,000 verses and no more than 50% of the quoting work, and Lockman's wording reaches storage, not only quotation	🔒 `lsb-2021` in `study-notes.db` — verify-only; see the cap above before quoting
NIV — New International Version (Biblica/Zondervan)	NT 1973, full Bible 1978, revised 1984, major revision 2011	Dynamic ("optimal") equivalence balancing accuracy and readability	The most widely read English Bible for decades — useful for gauging how a mainstream reader will have already encountered a verse; strong for preaching and general reading	The 2011 gender-language choices remain contested in some circles; idiomatic smoothing occasionally erases an ambiguity a literal translation preserves; the 1984 and 2011 editions differ enough that older citations may not match current printings	🔒 `niv-cultural-backgrounds-study-bible` / `niv-biblical-theology-study-bible` in `study-notes.db`
NLT — New Living Translation (Tyndale House)	1996, revised 2004, 2007, 2015	Thought-for-thought, translated (not paraphrased) by committee	Excellent narrative flow, especially in Wisdom/poetic books; strong for teaching newer believers	Greater interpretive distance from the original wording than ESV/NASB; not suited to precise word-study argumentation; can smooth over a textual tension a study should actually sit with	🔒 `nlt-life-application-study-bible` / `nlt-christian-basics-bible` in `study-notes.db`
BSB — Berean Standard Bible (Bible Hub)	2016 as the Berean Study Bible; renamed 2022, public domain 2023	Mediating: close to the original's wording, rendered as readable modern English	Public domain, so quotable and redistributable in full; translates the critical Greek text the ESV does, prints LORD for the divine name, and capitalises pronouns for God; its translators publish word-level tables tying every English word to the Hebrew or Greek, and footnotes naming which Greek editions read a variant. This is the site's reading translation: every pop-up and group sheet	Less literal than the ESV or WEB, so a word-level argument still rests on the ESV and the interlinears; its capital "He" in a contested text such as Isaiah 53 is the translators' identification, not the text's; less recognised than ESV/NIV/NASB	✅ `berean-bsb`
WEB — World English Bible	Begun 1994, released 2000, ongoing light revision	Modernized, public-domain descendant of the ASV	No permissions ceiling at all — freely quotable and redistributable in full; the site's default fully-queryable text until October 2026; actively revised for readability while keeping an ASV-like literal backbone	Far less recognized in broader church life than ESV/NIV/NASB; no study-Bible ecosystem built around it; ongoing volunteer revision means small wording drift over time rather than a fixed edition	✅ `ebible-eng-web`
ASV — American Standard Version	1901	Formal equivalence	Historically significant — the root of NASB, WEB, and (indirectly) RSV/ESV; public domain; useful for seeing exactly where later translations diverged and why	Dated, archaic English; renders the divine name as "Jehovah" throughout, which reads oddly today; superseded by its own descendants for regular study use	✅ `scrollmapper-ASV`
YLT — Young's Literal Translation	1862, revised 1887 and 1898	Extremely literal, word-order-preserving	Useful for seeing Hebrew/Greek grammar and tense underneath the English that smoother translations normalize away; public domain	Famously unnatural English prose — a cross-check tool, never a primary study or reading text	✅ `scrollmapper-YLT`
ULT — unfoldingWord Literal Text	2019–, ongoing	Formal equivalence, built as the literal half of a translation-training pair (with the UST, which we do not hold)	The only text here that records which original word each English word renders — every word is wrapped in an alignment frame naming the Hebrew or Greek behind it, which is what makes `bible_interlinear` possible and is the cheapest guard against building a point on a translator's choice. Openly licensed (CC BY-SA 4.0) and fully queryable	Designed as a translation tool rather than a reading Bible, so its English is deliberately stiff — closer to YLT than to ESV in feel. Its Old Testament aligns to the UHB and its New Testament to the UGNT, not to the WLC and SBLGNT this project treats as primary, so confirm wording against `bible_verse` before resting an argument on it. Not a quotation source — the site quotes ESV by default	✅ `uw-ult`

Also on hand for comparison

bible-text.db also carries JPS (Jewish Publication Society OT, useful for a Jewish-tradition reading alongside the Christian translations above), Darby, Douay-Rheims, and about two dozen other public-domain English editions via the Scrollmapper fork. None of these carry the weight ESV/NASB/NIV/NLT/WEB/ASV/YLT do in our studies. See Public Data Sources for the full list.

What the philosophies actually do: two worked examples

The table above describes translation philosophies in the abstract. Here is what they do to two verses this site's own studies rest on. Both were checked against the databases described below, and every rendering quoted is a fragment of a single verse.

Old Testament — Song of Songs 8:6, and whether God is named

The verse ends with three Hebrew words: אֵשׁ (esh, H784, "fire"), שַׁלְהֶבֶת (shalhevet, H7957, "flame"), and יָה (Yah, H3050), the shortened form of the divine name. The question is whether that last syllable names God or works as a Hebrew superlative, the way "mountains of God" can mean "mighty mountains". The Song never mentions God anywhere else, so the decision determines whether the book names Him once or not at all.

Translations split, and the split is invisible unless you line them up:

Reads it as the divine name	Reads it as an intensifier
ESV — "the very flame of the LORD"	NKJV — "A most vehement flame"
NASB 1995 — "The very flame of the LORD"	NIV — "like a mighty flame"
LSB — "The very flame of Yah"	NLT — "the brightest kind of flame"
WEB — "a very flame of Yah"	CSB — "an almighty flame"
ASV — "A very flame of Jehovah"	BSB — "the fiercest blaze of all"
YLT — "a flame of Jah!"	Geneva (1599) — "vehement flame"
JPS — "a very flame of HaShem"

A reader of the CSB and a reader of the ESV are reading different claims about this verse. Only the LSB, among the commercial translations, prints the name itself rather than the substitute "LORD", which is the case for keeping it on the shelf when the covenant name is the point.

The translators knew. Three committees footnote the fork at this verse. The LSB offers "Or A vehement flame", and explains that Yah is "the shortened form of Yahweh, found in poetry and praise (e.g. Hallelujah), and in names (e.g. Zechariah)". The CSB, having printed the intensifier, notes "Or the blaze of the Lord". The NIV carries a text note at "mighty flame" too, which the NIV Biblical Theology Study Bible cites for the alternative reading. Reading the footnote is faster than reconstructing the disagreement from six parallel versions, and it is the translators' own testimony that the question is live.

New Testament — John 6:54, and a verb most translations flatten

Through John 6 Jesus uses the ordinary verb for eating, ἐσθίω (esthiō, G5315), eleven times: the crowd ate the loaves, the fathers ate manna, "unless you eat the flesh of the Son of Man" (6:53). At verse 54 He switches to τρώγω (trōgō, G5176) and stays with it for the rest of the discourse. The lexicon separates them: Louw-Nida puts ἐσθίω at 23.1 and τρώγω at 23.3, alongside γεύομαι ("taste", G1089) and βιβρώσκω, the chewing-and-consuming end of the range. Verse 58 uses both in one sentence: the fathers ate (ἔφαγον) and died; whoever feeds on (τρώγων) this bread will live.

Of every translation on this page, one marks the change:

	John 6:54
ESV	"Whoever feeds on my flesh…"
NASB 1995/2020, LSB, NIV, NKJV, CSB, NLT, WEB, ASV, YLT, BSB	"…eats my flesh…"

The ESV stands alone. The NASB and LSB are the most formally equivalent versions here, and both flatten a distinction the ESV keeps. A translation's stated philosophy predicts its general behaviour, not its decision at any particular verse.

A correction this site had to make

An earlier draft of one study asserted the opposite: that the ESV obscures John's ἐσθίω→τρώγω shift. It renders every occurrence "feeds on"; the claim was written from memory and was wrong in the exact direction that flattered the argument. It is recorded in AGENTS.md as the reason quotations here get verified against a database rather than recalled.

Hebrew Old Testament

Text	Date	What it is	Strengths	Cautions	Tracked here
Masoretic Text — Westminster Leningrad Codex (WLC)	Manuscript dated 1008 CE by its own colophon	The standard Hebrew Bible text underlying virtually every printed Hebrew Bible (BHS/BHQ) and every English OT translation above	Oldest complete Masoretic manuscript in existence, full vocalization/cantillation/Masoretic notes intact; our copy carries both morphological tagging (`morphhb`) and SDBH semantic-domain codes (`macula-hebrew`) — the two together are what make Hebrew word studies here queryable rather than hand-searched	A single-manuscript edition, not an eclectic reconstruction — it inherits that one codex's scribal idiosyncrasies; copied roughly 1,400 years after the events it records and centuries after the Dead Sea Scrolls, so cross-checking against those earlier witnesses still matters on disputed readings	✅ `morphhb-wlc`, `macula-hebrew-wlc`, `scrollmapper-WLC`
UHB — unfoldingWord Hebrew Bible	2022–	The same Westminster Leningrad Codex, re-encoded in USFM from the Open Scriptures Hebrew Bible, with per-word lemma, Strong's and morphology	The text the ULT's Old Testament alignment resolves against, which is its real job here. Carries 949 footnotes — 930 of them Qere readings, the Masoretes' own "the consonants say X, read Y", which arrive automatically on any verse lookup. Openly licensed (CC BY-SA 4.0)	Not a second witness to the Hebrew — it is the same WLC consonantal text, differing genuinely in only ~1.5% of verses, and for a stated reason: "in order to avoid subjectivity, the text of the UHB uses the Ketiv of the WLC", where the WLC prints the Qere. And it uses a different versification — unfoldingWord's own note says it follows the ULT's scheme rather than "the WLC scheme common for Hebrew Bibles", which "may make some resources that are keyed to the WLC more difficult to use". That is not theoretical: the same reference names a different verse in `uw-uhb` and `morphhb-wlc` across ~1,500 verses in Joel, 1 Chronicles, 1 Kings, Numbers, Job, Ezekiel and Malachi. Compare the two with `bible_parallel`, never by reference alone	✅ `uw-uhb`
Dead Sea Scrolls (biblical scrolls)	Copied c. 250 BCE – 68 CE	262 scroll witnesses to 36 of the 39 Old Testament books, transcribed by Martin Abegg and colleagues	A thousand years older than the Masoretic manuscripts — the only pre-Christian Hebrew witnesses we hold. Each scroll is kept as its own work, because they disagree: Isaiah 53:5 survives in both 1Qisaa and 1Q8 and they read differently. Makes Deuteronomy 32:8 and Psalm 22:16 checkable here rather than taken on report	Fragmentary — 31.8% of words carry an editorial mark, and our text keeps them: `[ ]` is reconstruction, `#` and `?` damaged or uncertain. A scroll reading inside brackets is an editor's judgement, not manuscript evidence. Esther is absent entirely. Non-commercial licence	🟡 `dss-*`
Samaritan Pentateuch	Extant manuscripts from the medieval period; tradition claims earlier origins	An independent Torah-only text tradition preserved by the Samaritan community	Valuable second witness for text-critical comparison against the Masoretic Text on disputed Torah readings	Reflects Samaritan theological commitments (Mount Gerizim vs. Jerusalem, among others) rather than the mainstream Jewish transmission line — a comparison tool, not a primary text	🟡 `scrollmapper-SP`

New to lemmas, parsing and semantic domains?

This page is about which text to trust. Reading the Original-Language Data is about the annotation attached to those texts: what MACULA is, and what a lemma, a Strong's number, a morphology code and a semantic domain each mean, with Genesis 1:1 shown at every layer.

Which English word renders which Hebrew or Greek word

One translation here carries something the others do not. unfoldingWord's ULT (uw-ult) is aligned word by word to the Hebrew and Greek. So query.py interlinear <book> <ch> <v> answers what is this English word actually translating. That is the question behind most word studies, and one this project previously had no way to answer without judgement.

Two cautions. The mapping is many-to-many: Genesis 1:1's "the heavens" renders both אֵת and הַשָּׁמַיִם, since the Hebrew object marker has no English of its own. And ULT is one literal translation, so it tells you what ULT chose, not what the word must mean. For that, look the lemma up.

Hebrew New Testaments — translations, not witnesses

Both are in Hebrew and neither is a Hebrew witness: they are 19th-century renderings out of Greek, made long after the New Testament was written. They sat under Hebrew Old Testament on this page until 2026-09-05.

Text	Date	What it is	Strengths	Cautions	Tracked here
Delitzsch Hebrew Bible	First published 1877, revised through the early 20th century	A modern scholarly Hebrew translation of the New Testament, paired here with the Hebrew OT	Useful for tracing how NT vocabulary and OT allusions map back into Hebrew — helpful for spotting an OT echo a Greek-only reading might miss	Not an ancient witness — a 19th-century translation project (Franz Delitzsch), so it reflects translator choices, not manuscript history; never cite it as if it were the OT's own Hebrew	✅ `ebible-heb`
Salkinson-Ginsburg Hebrew New Testament	Salkinson 1885, revised by Ginsburg 1886	A second scholarly Hebrew translation of the New Testament, independent of Delitzsch	Read against Delitzsch it shows where a Hebrew rendering is a translator's choice rather than something the Greek forces — Romans 1:17 is from faith to faith in Delitzsch and from a wellspring of faith in Salkinson	Not an ancient witness — a 19th-century translation, and its restriction to Tanakh vocabulary is a stylistic programme, not evidence of an earlier Hebrew text	✅ `ebible-hebsg`

Two Hebrew New Testaments, and what they can and cannot settle

Delitzsch and Salkinson-Ginsburg are independent 19th-century translations of the Greek into Hebrew, and they differ in every verse. That disagreement is the point: where they diverge, a Hebrew rendering is the translator's judgement rather than something the Greek compels. No ancient Hebrew New Testament manuscript exists. The earliest Hebrew gospel witnesses (Shem Tov, Du Tillet, Münster) are medieval, and the Cochin manuscript Cambridge holds is an 18th-century Hebrew Matthew its own catalogue describes as made for polemical purposes.

A live question sits alongside that. Some readings in these Hebrew editions have no counterpart in the Greek, and there is an argument that the earliest printings of Salkinson-Ginsburg carried readings later editions removed. Second Thessalonians 2:7 is the case that matters most here, since the identity of the restrainer bears on the rapture study. What our copy actually reads at that verse is recorded in references/README.md, dated, so the claim can be tested against a specific text.

Greek New Testament

Text	Date	What it is	Strengths	Cautions	Tracked here
SBLGNT — SBL Greek New Testament	2010, ed. Michael Holmes	The modern eclectic critical text this project queries by default	Openly licensed (CC BY 4.0), so it's the one we can actually store and quote in full; assembled by comparing several prior critical editions' agreements and disagreements; carries per-word morphology, lemma, and Louw-Nida semantic-domain tagging in our copy (`macula-greek`)	A documented synthesis of existing eclectic editions rather than a fresh manuscript collation — where its source editions already agreed on a disputed reading, SBLGNT doesn't add independent new evidence	✅ `sblgnt`, `macula-greek-sblgnt`
NA28 — Nestle-Aland, 28th edition	2012	The critical text nearly all modern scholarship and translations (ESV, NIV, NASB, etc.) actually cite	The field standard; the most recent edition incorporated fresh collation of the Catholic Epistles against the full manuscript tradition (the ECM project)	Commercial license — not in our own database at all; reachable only indirectly through the NA28-ESV parallel inside the external, quotation-only `study-notes.db`, never quoted at length	❌ Not tracked; quotation-only via `study-notes.db`
Tischendorf, 8th edition	1869–1872	Constantin von Tischendorf's critical text, built heavily on Codex Sinaiticus — which Tischendorf himself discovered at St. Catherine's Monastery	An independent line of 19th-century critical-text reasoning, distinct from the SBLGNT/NA28 editorial lineage; public domain	Over 150 years old — predates the papyri discoveries and collation work that inform NA28/SBLGNT	✅ `ebible-grc-tisch`
Byzantine/Majority Text (Robinson-Pierpont)	2005, revised 2013	The Byzantine manuscript tradition underlying the KJV/Textus Receptus lineage — the majority reading by raw manuscript count	Valuable for seeing exactly where the eclectic critical texts diverge from the Majority Text, and why	"Majority by count" is a different argument than "earliest or best attested" — worth being explicit about that distinction whenever citing it	🟡 `scrollmapper-Byz`
UGNT — unfoldingWord Greek New Testament	2022–	The Greek text unfoldingWord's translations are aligned to, following the Bunning Heuristic Prototype	Openly licensed (CC BY-SA 4.0) and queryable here; a fourth lineage alongside SBLGNT, Tischendorf and the Byzantine text, and the text that makes ULT's word-level alignment resolvable	A different kind of edition — a heuristic text from the Center for New Testament Restoration rather than a committee edition like NA28. It differs from SBLGNT in roughly one verse in six by raw count, but most of that gap is manuscript spelling rather than a different text — Δαυείδ for Δαυίδ, Πειλᾶτος for Πιλᾶτος — and only 87 verses in the whole New Testament differ by three or more words. Those that do fall at the well-known cruxes: the ending of Mark, the woman taken in adultery, Luke 23:34, and John 1:34, where it reads Υἱὸς against SBLGNT's ἐκλεκτός. Check which text a reading comes from before resting an argument on it	✅ `uw-ugnt`
Textus Receptus (Stephanus 1550 / Scrivener 1894)	1550, standardized 1894	The specific printed Greek text underlying the King James Version	Essential for explaining why the KJV reads differently from modern translations at a given verse (e.g. the Comma Johanneum at 1 John 5:7-8)	Based on a small number of late medieval manuscripts available to Erasmus in the 1500s, long since superseded by far older manuscript evidence	🟡 `scrollmapper-TR`

Three lines of transmission, as Ken Johnson frames them

The table above is a catalogue: it says what each text is, but not how they relate to one another. Ken Johnson (Bible Facts) maps that relation as three lines descending from the original, in a study introducing his verse-by-verse work on the Hebrew Thessalonians (video). It is redrawn here with his own spoken dates, because the shape is a useful map of the disagreement even where his argument is contested. In the table, "Textus Receptus" and "critical text" are the endpoints of a live dispute.

flowchart LR
    NT["New Testament<br/>c. AD 45-96"]
    NT ==> L1
    NT --> L2
    NT --> L3
    subgraph L1["Byzantine line — the one he argues for"]
      direction TB
      A1["Greek, Lucianic<br/>c. AD 150"] --> A2["Received Text<br/>printed AD 1400s"] --> A3["'Fullest text'<br/>KJV, AD 1611"]
    end
    subgraph L2["Alexandrian line"]
      direction TB
      B1["Greek, 'abridged'<br/>c. AD 100-200"] --> B2["Critical Text<br/>printed AD 1880s"] --> B3["'Smallest text'"]
    end
    subgraph L3["Hebrew / Aramaic line"]
      direction TB
      C1["Text c. AD 200<br/>manuscripts AD 500-1500"] --> C2["Copied from Greek<br/>in the Middle Ages"] --> C3["'Received Text plus'"]
    end

His argument. The Received Text is the fullest, and the critical text is that text with material removed. The third line runs through Hebrew and Aramaic manuscripts. Most of it is medieval back- translation from Greek, which he concedes freely. A subset carries readings that back-translation does not explain. Some of them are anti-Catholic, which would have made an inquisition worse rather than deflected one. They appear at the same places across independent manuscript families. And church fathers who predate every surviving manuscript quote them. Hence "Received Text plus" rather than a fourth text-type. That last argument is the strongest part of his case: it falsifies the usual explanation for those particular readings.

Where it is contested, and this site does not follow him. "The critical text is the Received Text with material cut out" is the disputed question stated as a premise. The mainstream reading runs the other way: the Byzantine text is the later and fuller one. Neither direction can be assumed from the diagram. His attribution of the first line to Lucian also conflates two things: Lucian of Antioch died in 312, not the mid-second century, and the "Lucianic recension" is a proposal about text-type whose attribution is itself debated, not an account of who assembled the canon.

What he explicitly does not claim. Not Hebrew primacy: "we don't want to argue that the Greek or the Hebrew or the Aramaic is the original." He suggests Matthew and Paul may have written in more than one language, and warns against the fallacy in both directions: a shorter text is not automatically earlier, and a longer one is not either. The third line above bears on the two Hebrew New Testaments this repo actually holds, where the same caution applies.

Septuagint (Greek Old Testament)

Text	Date	What it is	Strengths	Cautions	Tracked here
Brenton Septuagint	Greek text and English translation published 1844/1851 (Sir Lancelot C. L. Brenton)	The standard public-domain LXX edition, Greek text based on Codex Vaticanus	Essential for tracing how New Testament authors quote the Old Testament — many NT quotations follow the LXX's wording rather than the Masoretic Text's; openly licensed and fully queryable here	Vaticanus-based text is one witness among several LXX manuscript traditions (Alexandrinus and Sinaiticus each diverge in places); the accompanying English prose is 19th-century and reads as dated. Its Daniel is Theodotion, not the Old Greek — identifiable at Daniel 1:3, which reads Ἀσφανὲζ where the Old Greek has Ἀβιεσδρί — so say which Greek Daniel you mean when citing it. Carries fourteen books outside the Hebrew canon, Tobit through the Prayer of Manasseh, but no Odes, Psalms of Solomon or Psalm 151	✅ `ebible-grcbrent`
Swete Septuagint	Greek text, 1887-94 (H. B. Swete, Cambridge), digitised by the Open Greek and Latin Project	A diplomatic LXX edition printing Codex Vaticanus, with its gaps supplied from other uncials and their readings in an apparatus	Supplies the three texts Brenton lacks: the Odes, the Psalms of Solomon and Psalm 151	Only those three are taken; every other Septuagint book here comes from Brenton alone. Its Odes follow Alexandrinus and are numbered differently from Rahlfs, by the verses of the passage each Ode reproduces. The digitisation carries some OCR noise	✅ `swete-lxx`

The LXX numbers its chapters differently — check before comparing

A reference is not a universal address. The Septuagint renumbers nearly the whole psalter, so English Psalm 23 is LXX Psalm 22, and it adds a Psalm 151 (here from Swete, since Brenton lacks it). Jeremiah is reordered rather than renumbered. The oracles against the nations move to the middle of the book, so the new covenant passage Hebrews 8 quotes is English Jeremiah 31 but LXX Jeremiah 38. The chapter count is identical at 52, so nothing about the book looks unusual until a citation lands in the wrong place. Joel and Malachi divide as the Hebrew does rather than as English Bibles do, and Daniel divides differently again at each end: the Septuagint follows the English break at 3/4 and the Hebrew one at 5/6.

The repo resolves this. Every work in bible-text.db records which scheme it uses, and uv run python query.py align Joel 2 28 reports a reference as each scheme numbers it. Verse lookups align automatically; see references/README.md for the cases that are deliberately left unmapped because no correspondence can be established.

Why isn't the Septuagint's text alongside its lemma tooling?

GreekResources (our lemma/lookup fork for LXX word studies) deliberately ships without the LXX text itself. Its own maintainers exclude it because the standard CATSS source is restrictively licensed, and we followed the same discipline. Brenton's edition above fills that gap with a text we've independently confirmed is public domain.

Three Old Testament lines, as Ken Johnson frames them

The companion to the New Testament diagram above, from the same study. Three witnesses descend from the Hebrew, and he grades each on two axes: whether it agrees with the New Testament's quotations, and whether its genealogical numbers are sound.

flowchart LR
    HB["Hebrew Old Testament<br/>c. 1500-400 BC"]
    HB --> S1
    HB --> S2
    HB ==> S3
    subgraph S1["Septuagint line"]
      direction TB
      A1["Greek (LXX)<br/>c. 200 BC"] --> A2["Agrees with Paul<br/>Dates corrupted"] --> A3["Corrupted AD 350s<br/>published AD 1800s"]
    end
    subgraph S2["Masoretic line"]
      direction TB
      B1["Masoretic Text<br/>c. AD 700"] --> B2["Disagrees with Paul<br/>Dates correct"] --> B3["Corrupted AD 700s<br/>published AD 1800s"]
    end
    subgraph S3["Scrolls — his preferred"]
      direction TB
      C1["Dead Sea Scrolls<br/>c. 200 BC"] --> C2["Agrees with Paul<br/>Dates correct"] --> C3["Uncorrupted but<br/>fragmentary<br/>published AD 1990s"]
    end

His argument. The Masoretic Text is a thousand years younger than the scrolls and was copied by scribes with a motive to smooth over messianic readings, so where a New Testament writer quotes the Old and the Masoretic disagrees, the scrolls side with the New Testament writer. The Septuagint agrees with the New Testament but carries inflated genealogical numbers. The scrolls alone score on both axes. (His slide dates the Masoretic line "~700 BC"; his spoken date is AD 700, which is the one to read.)

What checks out here. The genealogical divergence is real and queryable in this repo: at Genesis 5:3 Brenton's Septuagint reads τριάκοντα καὶ διακόσια (230 years to Seth's birth) where the Hebrew reads 130. Genealogy and the Age of the Earth tables all three traditions generation by generation. And on the one verse of Genesis 5 the scrolls actually preserve, his dates column holds: 4Q2 at Genesis 5:13 reads eight hundred and forty, siding with the Masoretic against the Septuagint's 740.

Where the diagram is tidier than the evidence. That last check is also the problem with it. The biblical scrolls preserve one verse of Genesis 5 and none of Genesis 11, so a verdict of "dates correct" rests on a single fragmentary line. "Uncorrupted" needs the same caution the Dead Sea Scrolls row above records: 31.8% of words in this corpus carry an editorial mark, and a bracketed reading is a modern editor's reconstruction.

The "agrees with Paul" axis has also now been enumerated. Of the 140 strong New Testament quotations of the Old that this repo derives, eighteen land on a verse where a scroll also differs from the Masoretic. Read one at a time, one of those eighteen is a case where the scroll supplies the reading the New Testament follows. That one is Isaiah 61:1, where 1QIsaa carries the single divine name Luke 4:18 and the Septuagint have, against the Masoretic's double. Even there, 4Q56 and 4Q66 keep the Masoretic reading. Two of the most quoted verses run the other way outright. At Isaiah 53:1 the "Lord" that John 12:38 opens with is a Septuagint plus that all three scrolls lack. At Isaiah 7:14 the scroll reads he shall call, where the Masoretic has she shall call and Matthew has they shall call. The scroll differs from the Masoretic, and not in Matthew's direction.

The two axes do not partition together. At Deuteronomy 32:8 the scroll reads sons of God against the Masoretic sons of Israel, siding with the New Testament; at Genesis 5:13 the same corpus sides with the Masoretic on a number. Qumran is textually plural: some scrolls are proto-Masoretic, some stand behind the Septuagint, some are independent. That plurality is the discovery. Two further details resist the tidy split: AD 700 is when the Masoretes added vowel points, not when their consonantal text began, since that text is attested at Qumran centuries earlier; and his own headline case, Hebrews 10:5, is explained by the Septuagint alone with no scroll required.

Why the "main" translations aren't in `bible-text.db`

ESV, NASB, NIV, and NLT, the four translations we quote most in prose, are missing from bible-text.db, the ✅ database. They're commercially copyrighted, and Crossway/Lockman/Biblica/Tyndale's permissions policies allow generous quotation (a study citing a verse or a short passage with attribution is the intended use) but not bulk redistribution into a database anyone could dump wholesale. WEB, ASV, and YLT exist in our own database specifically so we always have something fully open to fall back on: concordance searches, cross-reference generation, or any use that would otherwise require copying a commercial text at scale.

All four get the same narrower exception, marked 🔒 above rather than ❌. Each one's verse text is a byproduct of a commercial study Bible loaded into study-notes.db, or for NASB a standalone edition: esv-study-bible; niv-cultural-backgrounds-study-bible / niv-biblical-theology-study- bible; nlt-life-application-study-bible / nlt-christian-basics-bible; nasb-1995 / nasb-2020. Each is queryable there for the one purpose that license permits: checking a quotation against source before publishing it.

NASB is the one to treat more cautiously even within that 🔒 tier. ESV, NIV, and NLT all carry a stated safe-harbor (500–1,000 verses / 25–50% of a work, see the permissions table in references/README.md). Quote a verse or two in a study and you're comfortably inside it. NASB's own permission notice states no such threshold at all; Lockman's language requires quotation and reprint requests to be "directed to and approved in writing." Being 🔒-tracked here means an NASB quotation can be verified against source instead of trusted from memory. It doesn't mean the safe-harbor reasoning that applies to the other three translations extends to NASB.

Reach for the 🔒 translations when writing for a reader (verify against study-notes.db first, and for NASB specifically don't lean on a verse-count safe harbor that doesn't exist); reach for the ✅ ones when writing a query.