Three hands who edited the Old Testament after it was written, told in four pictures on one line

The Old Testament was written, and then it was edited, by people who left parentheses: 'unto this day', 'which is Hebron', 'before there reigned any king'. Sorted by what each parenthesis presupposes, the editors come out as three hands. This page shows them one picture at a time. The full count, every verse and every check is on the Edit points page; this one is the story.

3hands that edited the text after it was written

268parentheses they left

3sign-offs, one on each of Butler's three breaks

0times an earlier hand mentions a later one

Three words. Strands are who wrote the text. Hands are who reached back into it later. Parentheses are the evidence: the asides a hand leaves because its readers no longer know a name, a custom, a measure, or a kingdom. The line is Butler's: seventeen marks, one per historical book, in book order. It never changes between the four pictures.

1. The quiet picture: three hands on one line

Butler's seventeen marks run left to right in book order. Under them, one bar per hand: it starts at the earliest story the hand touched and ends at the dot, where the hand signs off. Hand 1 works from the Patriarchs to the Division of the kingdom. Hand 2 reaches from Babel to the Babylonian exile. Hand 3 runs from the Kingdom to the Return. Each dot sits on one of Butler's three breaks. That is the whole claim.

2. The parentheses: the evidence, placed on the hands

Every parenthesis is a dot, on the bar of the hand that left it, under the story it sits in. Hand 1's bar is crowded: the place renamings in Genesis, the 'unto this day' monuments of Joshua and Judges, the source citations of Kings. Hand 2's bar is almost bare: a hand that closes Kings and reaches back for a word, 'Ur of the Chaldees'. Hand 3's dots are Chronicles, Ezra and Nehemiah. The weight of the editing is early on the line and early among the hands.

3. The pointing: each hand treats the one before as scripture

Arcs from a later hand to an earlier layer. Green, the later hand copies the earlier text parenthesis included, so Chronicles keeps 'there it is unto this day' about an ark that was gone. Blue, dashed, it cites the earlier layer as a book, 'as it is written in the book of the law of Moses'. Red, it accuses, Jeremiah 8:8 on the scribes' pen. Gold flags are colophons, a hand marking the edge of what it received. Every arc climbs; no hand ever mentions a later one. The lanes hold the Torah layer the hands received and the three hands, oldest at the top.

4. The strands: who wrote the five books, under the same line

The Documentary Hypothesis sorts the five books by author: J, E, P, D and the Redactor R who stitched them. One thin column per chapter, split by strand. The strands are who wrote; the hands are who reached back in later. J and E are the layer the hands received. D is closest to Hand 1. P edits by framing, not by asides, so it is thick in the columns and thin in the dots. The barcode stops at Deuteronomy because the hypothesis does; the hands run on to the Return.

What to take away

Hand 1: the kingdom-era compiler

writes from the divided monarchy; settles claims to places ('unto this day'), renames, updates seer to prophet, cites the royal annals, signs off at the Division

Hand 2: the exile hand

closes Kings at 2 Kings 17 and 25:27-30, reaches back only for a word: Ur of the Chaldees, Leviticus 26's exile as past; never shares a chapter with Hand 1

Hand 3: the Persian hand

Chronicles, Ezra, Nehemiah: Persian coin in David's offering, Zerubbabel's line carried on, the formula re-used for the province's own present

The one rule the pictures share. Every arc climbs and every bar ends at a break. A later hand copies, cites or accuses an earlier one; no earlier hand knows a later one exists. That is why the hands can be ordered at all, and why the count can say "three" without saying a year.

Living notes: other ways to read the same pictures

The four pictures above stay as drawn. Below them, ideas about the same evidence are collected as they come up; none of them is a decision yet. One rule for every entry: the ruler is order, not the calendar. The text keeps its own time by genealogies and by what each layer knows, copies and cites, so every reading says how it orders the hands from the text itself, and what would test it.

What a sign-off dot means: a floor, not a stopping point explains the pictures added 2026-10-12

A hand's 'today' shows in its asides: 'unto this day' is the editor saying where it stands. The dot is meant to mark that standpoint. Two things set it, and both are floors: the hand stands no earlier than this.

First, what the aside presupposes. 'So Israel rebelled against the house of David unto this day' (1 Kings 12:19) needs a writer who knows two kingdoms, so after the Division. The dots use this rule only.

Second, where the aside sits in the story. 'The rest of the acts of Jehoiakim, are they not written in the book of the chronicles of the kings of Judah?' (2 Kings 24:5) speaks of a finished reign, so its writer stands after Jehoiakim. The dots do not yet use this rule.

That is why Hand 1's bar runs past its dot in picture 1: 28 of its markers sit under Assyria and 17 under Babylon, most of them that source-citation formula. Two ways it could be settled, neither chosen: move Hand 1's dot to Babylon, where it meets Hand 2 and the two may be one hand; or split the source citations into a layer of their own, a compiler who copied the royal records up to the exile, leaving Hand 1 as the renamer and 'unto this day' writer. Only Hand 2 (2 Kings 25:27-30, the close of Kings) and Hand 3 (2 Chronicles 36:22-23, Cyrus's decree, repeated as Ezra 1:1-2) end on a passage that is a real ending.

Orders by: what each aside presupposes, and which events it narrates as already past

What would test it: re-place each dot at the later of the two floors and see whether the three hands stay three

Sources: diagram_data/Edit-Points-Of-The-Old-Testament.json, plotted.hits by stratum and mark (2026-10-12)

All three hands after the story was finished open idea added 2026-10-12

Chad's deduction, 2026-10-12: the asides only ever say 'no earlier than'. Nothing in them stops all three hands from working after the whole story was complete, each reaching into a finished text in turn. On that reading the bars keep their left ends, the dots slide right, and the three hands could even sit together after the Return.

What survives the move is the order. A later hand copies, cites or accuses an earlier one, and never the reverse: Chronicles keeps Kings' 'there it is unto this day' about an ark that was gone (2 Chronicles 5:9), so the Kings text already existed when the Chronicles hand worked. What is lost is the spacing: the pictures could no longer say how far apart the hands stood, only who came after whom.

Some asides were first written early even if the editing was late. A line that went false at a known turn in the story ('rebelled against the house of David unto this day' only reads true while the north stands) was written before that turn; a later hand that carries it unchanged is copying, not composing. That separates two questions: when an aside was first written, and when the text was last worked over.

Orders by: copy, cite and accuse arcs between the hands; stale 'unto this day' lines a hand carries without correcting

What would test it: for each hand, count the 'unto this day' lines already false by the Return: a hand carrying many stale lines is copying older material. Where two surviving copies of a passage differ, the shorter one stands before the edit, which orders the edit without a date. The genealogy numbers of Genesis 5 and 11 differ between the Hebrew, Samaritan and Greek texts: each tradition's reworking of the chronology is a hand of its own to place in order.

Sources: conversation with Chad, 2026-10-12; the scholarly models below, corpus folder AcademicRedactionHistory (46 documents, gathered 2026-10-12)

Three editors, all after the fall of Jerusalem: the Göttingen layers scholarly model added 2026-10-12

The closest scholarly match to the idea above. Smend found a law-minded addition in Joshua 1:7-9 that turns a battle speech into a call to obey the law of Moses, and named its writer DtrN, the nomist. Dietrich added DtrP, a prophetic editor who set the prophet stories into Kings. With the first historian, DtrH, that makes three layers on Joshua to Kings, all working after Jerusalem fell: historian, then prophetic editor, then law editor.

Pakkala pushes further: every editor of Kings stands after the fall, and only the excerpts from the royal annals are older. His ordering tool is dependence: a passage that borrows another's wording and cannot be understood without it comes after it. His monotheistic passages are 'dependent on nomistic texts', so they come after the nomists, who are themselves additions to the historian's text.

Römer keeps three stages too but starts the first earlier, under the kings, partly because 'unto this day' fits a writer who still has a king and a temple. Others (Otto, Knauf, Westermann) deny that one Deuteronomistic history ever existed.

Orders by: dependence: who borrows whose wording, and whose additions only make sense on top of another's

What would test it: for each of the three hands here, check whether its asides borrow the wording of another hand's asides; borrowing gives order with no date

Sources: ARH-roemer-2013-deuteronomistic-history-ebr; ARH-roemer-2015-current-discussion-so-called-dtrh; ARH-person-2009-conversation-with-roemer-so-called-dtrh; ARH-peterson-2014-authors-dtrh-ch1-since-noth; ARH-pakkala-2007-monotheism-dtrh; ARH-pakkala-2010-why-cult-reforms-probably-did-not-happen; ARH-wikipedia-2026-deuteronomist

'Unto this day' may be many hands, not one scholarly model added 2026-10-12

Childs (1963) read every 'until this day' in the Old Testament. His conclusion: the formula 'reflects the age of many different redactors'. It is a writer's personal testimony added to a tradition he received, and most of it sits inside older source material (the Acts of Solomon, the Chronicles of the Kings of Judah, the prophet stories), not in the frame around it; the frame uses it only rarely (2 Kings 17:23, 34).

Geoghegan (2003, 2006, 2007) reads the same formula the other way: one editor writing while the temple and the kings still stood, which would make it a single early layer.

This bears straight on Hand 1, which is mostly 'unto this day'. If Childs is right, Hand 1 is several hands folded together, and the split suggested in the first card (the source citations as their own layer) is the start of pulling them apart.

Orders by: whether an aside sits inside a quoted source or in the frame that quotes it: the frame is later than what it frames

What would test it: tag each of the 143 first-point markers as inside source material or in the editor's frame, and see whether the two groups behave like different hands

Sources: ARH-childs-1963-formula-until-this-day; ARH-geoghegan-2003-until-this-day-preexilic-redaction; ARH-geoghegan-2007-levites-literature-late-seventh-century

Edits you can see: two copies of the same passage scholarly model added 2026-10-12

Some edits are not inferred at all; two surviving copies show the text before and after. A Dead Sea Scroll of Judges (4QJudg-a) runs from 6:6 straight to 6:11, without the unnamed prophet's speech of 6:7-10 that critics had already marked as an insertion into a finished story. The Greek Jeremiah, with Hebrew fragments from Qumran, keeps a short first edition; the longer Hebrew Jeremiah adds headings, fuller names and stock phrases 'during the final stage of the growth of the book'. 1 Samuel 17-18 lays a second David and Goliath story over a first.

Berman's survey of such documented cases gives a ceiling: they rarely show more than two or three stages of major revision. Three hands sits at the upper edge of what the documented cases support.

The limit: a shorter copy can be the result of a cut, not the text before an addition, and the study of 4QJudg-a weighs both.

Orders by: two copies of one passage: the edition the other presupposes comes first

What would test it: list which of the 268 markers sit in passages where the Greek or a scroll differs from the Hebrew; any marker missing from an older edition is caught being added

Sources: ARH-ausloos-2014-judges-6-4qjudga-literary-textual-criticism; ARH-tov-1985-literary-history-jeremiah-textual-history; ARH-tov-1985-composition-1-samuel-16-18-septuagint; ARH-tov-2003-large-scale-differences-lxx-mt; ARH-tov-2008-septuagint-and-deuteronomists; ARH-tov-2009-textual-harmonizations-deuteronomy; ARH-tov-2015-source-of-source-criticism; ARH-muller-2014-evidence-of-editing-introduction; ARH-dyck-2014-review-evidence-of-editing; ARH-berman-2016-empirical-models-textual-growth; ARH-sigrist-2017-review-ulrich-developmental-composition

The genealogies as an edited chronology scholarly model added 2026-10-12

The text keeps its own time by genealogies, and the genealogies were edited too. The ages in Genesis 5 come in three editions: Hebrew, Samaritan and Greek. Hendel reads them as three rival fixes for one problem the first text created, patriarchs who outlive the Flood; all three come before the Greek translation and before Jubilees, and the Samaritan sectarian touches come last. That is an order of editions built from the numbers alone.

The 480 years of 1 Kings 6:1 is twelve generations of forty: a schematic count, read by Hughes as a late adjustment to the priestly scheme and by Laato as older than the Deuteronomistic editors. Wilson and Johnson show genealogies changing with use, so a genealogy is itself a layer with hands of its own.

Orders by: which version's numbers solve a problem another version created: the fix comes after the problem

What would test it: lay the three versions of Genesis 5 and 11 side by side, mark which edit fixes which, and place those chronology hands on the same line as the asides

Sources: ARH-hendel-2012-hasmonean-edition-mt-genesis-chronology; ARH-klein-1974-archaic-chronologies-textual-history; ARH-holloway-1991-review-hughes-secrets-of-the-times; ARH-laato-2016-mysterious-number-480-1-kings-6-1; ARH-ray-2016-role-functions-biblical-genealogies; ARH-wilson-1975-old-testament-genealogies-recent-research; ARH-tenhoopen-2017-genesis-5-formation-primeval-history; AKP-northcote2007 (already in the corpus)

The genealogy test, run: Genesis 5 and 11 in three editions measured added 2026-10-12

Run on the Hebrew (KJV), the Greek (Brenton's English, with Codex Alexandrinus from his notes) and the Samaritan, each counted in its own years from creation. Most of the genealogy is one shared text: every lifespan from Adam to Mahalalel, and Enoch's, Noah's and Shem's, is the same in all three.

The differences cluster where the numbers collide with the story. Jared, Methuselah and Lamech, the three lives that reach the Flood, are the only places all three editions disagree. The Hebrew has Methuselah die in the Flood year. The Samaritan shortens all three lives so they die in the Flood year. Brenton's Greek reads 167 for Methuselah and leaves him alive 14 years after the Flood, not on the ark; Alexandrinus and most Greek copies read 187, the Hebrew number, and he dies before it. Which Greek reading came first is disputed: Cassuto holds 187 original and 167 a slip, and Brenton's base for this part of Genesis is a 15th-century supplement to Codex Vaticanus (Smith 2018).

Each edition has a hand you can recognise by its method. The Samaritan moves deaths onto story events: the three in the Flood year, and Terah (145 years) dying the year Abram leaves Haran, where the Hebrew and Greek keep him alive 60 years longer. The Greek adds a hundred years at a stroke and inserts a second Cainan. The Hebrew has its own schematic numbers (Lamech 777, Eber 464, low Genesis 11 ages) and leaves Noah alive into Abram's life.

The order this gives, with no calendar: a shared text with the Flood problem in it; then the split; then three separate fixes. Every fix presupposes the finished story (Noah 600 at the Flood, Genesis 7:6; Abram 75 leaving Haran, 12:4). For this layer at least, the editing came after the story was complete.

Orders by: a fix comes after the problem it fixes; a number two editions share comes before the branches that share it

What would test it: open: Genesis 11 reads two ways (the Hebrew lowered the ages at birth, or the Samaritan and Greek raised them); the Samaritan is the hinge, sharing ages with the Greek and lifespans with the Hebrew. A competing order: Smith 2018 argues the Greek keeps the original numbers, the Hebrew was cut down after AD 70, and the Samaritan Genesis 5 was cut to match the Book of Jubilees, which would put the Samaritan hand after Jubilees

Sources: genealogy_editions.py; research/genesis-5-11-editions/genesis-5-11-editions.json; ProjectNotes/Genealogy-Test-Genesis-5-and-11; ARH-hendel-2012-hasmonean-edition-mt-genesis-chronology; ARH-smith-2018-case-for-septuagint-chronology-gen-5-11; AKP-northcote2007; ARH-tov-2009-textual-harmonizations-deuteronomy

Jubilees as a fourth witness: checking Smith's claim measured added 2026-10-12

Jubilees dates every birth on its own clock of jubilees, weeks of years and years, and frames the whole story as fifty jubilees from Adam to the crossing of the Jordan (50:4). Turned into years from creation, its Flood falls in year 1308; the Samaritan's falls in 1307, the Hebrew's in 1656, the Greek's in 2242.

Smith's first claim holds: the Samaritan pre-Flood chronology matches Jubilees, with the same low ages for Jared, Methuselah and Lamech. It matches in shape and total, not number for number; three ages differ by a year or two.

His second claim, that the Samaritan was cut down to match Jubilees, the numbers lean against. Where the Genesis editions agree, Jubilees copies them exactly (Adam 130, Seth 105, Enosh 90, Kenan 70, Enoch 65), so here it follows a Genesis text. After the Flood it invents freely: Peleg fathers Reu at 12, and 568 years separate the Flood from Abram, a figure no edition has. A Samaritan editor copying Jubilees would have copied only half of it. The simplest order: a Genesis text with Samaritan-type pre-Flood numbers came first, and Jubilees followed it.

Jubilees also carries the extra Cainan, as the Greek does. It is a chronology hand of its own: it retells the finished story and lays it onto a fifty-jubilee frame.

Orders by: where a later work copies its source exactly and where it departs from every source

What would test it: not on disk and not checked: Smith's evidence from Jerome's Samaritan manuscripts (Methuselah 187/782/969); if it holds, single Samaritan numbers kept changing after an older short total

Sources: Pseudepigrapha/PSE-Jubilees-Charles-1913; genealogy_editions.py (jubilees_witness); ProjectNotes/Genealogy-Test-Genesis-5-and-11; ARH-smith-2018-case-for-septuagint-chronology-gen-5-11

Who copied whom: Kings and Chronicles from a shared text scholarly model added 2026-10-12

Picture 3 draws Chronicles copying Kings. Auld reverses that: Kings and Chronicles both draw on an older shared story, and the material only Kings has (the Elijah and Elisha stories among it) was built later out of that shared text. Nihan and Römer argue against him.

Either way the ark line survives as evidence of order: 'there it is unto this day' stands in both 1 Kings 8:8 and 2 Chronicles 5:9, so on Auld's reading it belongs to the shared text and is older than both books that carry it.

Orders by: what both books share against what only one has

What would test it: split the Kings and Chronicles markers into shared and one-book-only; if the one-book-only markers in Kings behave like a later hand, Auld's order fits the count

Sources: ARH-auld-2017-continuing-life-in-kings; ARH-nihan-roemer-1999-source-commune-rois-chroniques-auld

Written late as a whole, and the reply scholarly model added 2026-10-12

A stronger form of the all-late idea: not only the editing but most of the writing happened late, in the Persian or Greek period (Lemche, Thompson and Davies, the Copenhagen school; Gmirkin and Wajdenbaum for a Greek-period Torah). Van Seters puts the Yahwist late and calls him an author, not a collector of sources.

The reply (Dever): 'no one denies late editing', but a text written in the Persian period should show Persian details, and he challenges anyone to name them. In this count the two Persian markers scanned for, drams and 'king of Persia', occur in the KJV only in Chronicles, Ezra, Nehemiah and Daniel, never in Genesis to Kings; David's drams are in Chronicles' retelling, not in Samuel.

Ordering by the Hebrew itself is contested: Hornkohl finds the Masoretic Torah keeps older forms that the Qumran rewritings modernise; Young, Rezetko and Ehrensvärd answer that late scribes could write classical Hebrew, so old-sounding language does not prove an early passage.

Orders by: anachronisms (what a passage could not know if written early) and the stage of the language

What would test it: grow the Persian and Greek marker list beyond drams and 'king of Persia' and see whether any land in Genesis to Kings

Sources: ARH-wikipedia-2026-biblical-minimalism; ARH-lemche-1993-hellenistic-book; ARH-lemche-2000-pre-hellenistic-history; ARH-thompson-2009-history-and-fiction; ARH-davies-2002-minimalism-ancient-israel; ARH-davies-2005-origin-of-biblical-israel; ARH-dever-2003-contra-davies; ARH-wajdenbaum-2016-plato-to-moses; ARH-anthonioz-2018-review-gmirkin-plato; ARH-gmirkin-2018-response-to-anthonioz; ARH-schmid-2020-textual-authority-israel-judah; ARH-vanseters-2007-author-or-redactor; ARH-vanseters-2015-dating-the-yahwist; ARH-hornkohl-2017-hebrew-diachrony-reworked-pentateuch; ARH-young-2010-linguistic-dating-biblical-texts; ARH-young-2019-review-hendel-joosten-how-old

Built from diagram_data/Edit-Points-Of-The-Old-Testament.json by gen_three_hands.py; the count, the strata and the checks live on that page.
Hands are groups of edit points that never share a chapter; they are relative, not dated. A hand's bar says where it reached and where it signed off, not when.
Butler's marks encode book order, not years. The spacing on every panel is his.