Moses and the Composition of the Pentateuch

Summary

Two positions are usually presented as the only options: Moses wrote the Pentateuch, or four anonymous sources were stitched together centuries later.  Both are worse than the evidence requires, and the interesting work is in between.

• The Pentateuch itself does not claim Moses wrote all of it. It says he wrote specific things — "Moses wrote down all the words of the Lord" at Sinai (Exodus 24:4);¹ "Moses wrote down this law and gave it to the Levitical priests" (Deuteronomy 31:9).² It also narrates his death and burial, notes that "no one knows his exact burial place to this very day," and adds that "no prophet ever again arose in Israel like Moses" (Deuteronomy 34:5–10).³ That last sentence requires a long enough view to survey the prophets who came after. Whatever one concludes, somebody other than Moses wrote the end of Deuteronomy, and this has been acknowledged in Jewish and Christian tradition for two thousand years.
• There are other markers of later hands, and they are not hostile inventions. Genesis 36:31 lists Edomite kings who reigned "before any king ruled over the Israelites" — a note written by someone who knew Israel eventually had kings.⁴ Genesis 12:6 remarks that "at that time the Canaanites were in the land," which implies a writer for whom that was no longer true.⁵ Numbers 12:3 calls Moses "very humble, more so than any man on the face of the earth," which is an odd thing to write about oneself.⁶ Any serious position has to account for these.
• The classic Documentary Hypothesis has fractured, and this is not an apologetic talking point. Julius Wellhausen's 1878 synthesis — J, E, D and P, in that order, with P latest and post-exilic — dominated for a century. It no longer commands assent. European scholarship has largely abandoned the E source and moved to supplementary or fragmentary models (Rendtorff, Blum, Van Seters, Römer); a North American "neo-documentarian" school (Baden, Stackert, Schwartz) defends a revised documentary model. These groups disagree with each other about method, not merely about details. What has not returned is Mosaic authorship, and no one should present the collapse of a consensus as the vindication of its opposite.
• The argument for a late P is about the prophets, not about style, and it makes Ketef Hinnom decisive rather than decorative. The move that fixed Wellhausen's order was Reuss's: "the pre-exilic prophets betray no familiarity with the Mosaic legal system, and … the ritual law in particular, closely related as it is to Ezekiel, could not have originated earlier than the exilic period."⁷ The Ketef Hinnom amulets carry Priestly material. If they are pre-exilic they are a counter-instance to the premise on which the classical dating of P rests — which is why the six-hundred-year palaeographic dispute over them is not a quibble.
• And the divine-names argument, the one most cited in popular accounts, was never load-bearing. Blenkinsopp: the criterion "was, from the beginning, quite limited in application. Initially it was applied only to Genesis and the first two chapters of Exodus," because after Exodus 3 an Elohist had no reason to avoid the name.⁷ He adds that its incidence "is not nearly as consistent as the hypothesis requires" and "is patient of other explanations."⁷ A critical scholar conceding this is worth more than an apologist asserting it — and it leaves the doublets, the flood chronology and Reuss's argument untouched, which is where the weight actually sits.
• The linchpin of the whole scheme is the seventh-century date for Deuteronomy, and it is under attack in mainstream journals. Aaron Koller, writing at Bible Interp, sets out why this one date matters so much: "the book of Deuteronomy is regularly dated to the seventh century BCE. If other sources are judged to be earlier than Deuteronomy, they are perforce earlier than the seventh century; if they are later than Deuteronomy, they are no earlier than the sixth century."⁸ Everything else in the relative sequence is pinned to that one absolute date.
• The Deuteronomy-treaty debate is the sharpest live argument in the field, and it is a genuine draw. Kenneth Kitchen argued long ago that Deuteronomy's structure follows second-millennium Hittite treaty form rather than first-millennium Assyrian form.⁹ Joshua Berman revived this in the Journal of Biblical Literature, arguing that a fifteenth-century Hittite text, CTH 133, parallels Deuteronomy 13 better than Esarhaddon's Succession Treaty of 672 BC does.¹⁰ Bernard Levinson and Jeffrey Stackert replied that the Assyrian parallels are too strong to dismiss, the Hittite ones too weak to keep, and that there is no known route by which Hittite treaty forms could have reached Israel.¹¹
• Koller, who is not a conservative, adjudicates it against Levinson and Stackert on the transmission point and calls the rest a draw. On transmission: "until 2009 it was not known that copies of the VTE were brought to the west, either, and yet scholars were prepared to accept that the Neo-Assyrian treaties had influenced Deuteronomy."⁸ On the substance: "In the case of Deuteronomy 13 in particular, there seem to be no 'smoking guns' that reveal the influence of either the Vassal Treaties of the seventh century, or the Hittite treaties of the thirteenth century, on Deuteronomy."⁸ And on the structure: "the basic structure of Deuteronomy is more similar to Hittite treaties than to the Neo-Assyrian treaties."⁸ His conclusion: "That Deuteronomy, at least in some form, dates back to the second millennium BCE appears to be a more tenable scholarly position than it did earlier."⁸
• Writing was available. This objection is dead and should stop being answered as if it were live. Alphabetic writing in a Semitic language is attested at Serabit el-Khadim and Wadi el-Hol from roughly 1900–1800 BC, centuries before any Exodus date.¹² Nobody serious now argues that Moses could not have written because writing did not exist.
• Two silver amulets carry Pentateuchal text that may predate the exile, and the dating dispute is still open. The Ketef Hinnom amulets contain the Priestly Blessing of Numbers 6:24–26 and language from Deuteronomy 7:9 — material assigned to two different sources — on a single object.¹³ Erik Waaler argues for 725–650 BC and concludes that "disparate Pentateuchal texts existed and were conjoined prior to the reform of Josiah."¹³ But the specialist range is enormous and the dispute did not end with the excavators' 2004 re-edition. Waaler's own survey records Barkay at the second half of the seventh century, Yardeni at the early sixth, and Renz at the early post-exilic period or the two centuries before Christ.¹³ Barkay, Vaughn, Lundberg and Zuckerman published a new edition with advanced imaging in 2004 supporting the late pre-exilic date; Nadav Na'aman then argued in 2011 for an early Second Temple date on orthographic grounds, and Barkay's team published a rejoinder.¹⁴ Popular apologetic use of these amulets almost never mentions any of this.
• Honest weaknesses. There is no external evidence for Moses of any kind — no Egyptian text, no inscription, nothing. The doublets, the alternating divine names, and the differences in vocabulary and theology between Pentateuchal strands are real features of the text that a defender has to explain rather than deny. And the strongest arguments on the conservative side, including the treaty-form argument, establish at most that material in the Pentateuch is old, not that Moses wrote it.

Verdict: the Pentateuch contains genuinely old material, was edited over a long period, and cannot on any defensible reading have been written entirely by Moses.  The Wellhausen synthesis that dominated the twentieth century has broken up, and its central dating peg is now argued over in mainstream journals rather than assumed.  That is a substantial change and conservatives are entitled to notice it.  What has not happened is a return to Mosaic authorship of five books, and claiming otherwise misrepresents the state of the field.  The position the evidence supports is a Pentateuch with a Mosaic core — legal and covenantal material shaped in a world whose treaty forms are second-millennium — transmitted, supplemented and edited into the text we have.

Part 1 — What the text claims

It is worth being precise, because both sides overstate.

What is claimed for Moses.  Exodus 24:4: "Moses wrote down all the words of the Lord."¹  Deuteronomy 31:9: "Moses wrote down this law and gave it to the Levitical priests."²  Numbers 33:2 has him record the itinerary.  These are claims about specific documents — a covenant text, a law, a travel record — not about the Pentateuch as a five-book whole.

What is claimed against.  Deuteronomy 34 narrates Moses's death, his burial, the thirty days of mourning, and then: "no one knows his exact burial place to this very day.  … No prophet ever again arose in Israel like Moses."³  This is written from a distance.

Jewish tradition handled this openly.  The Talmud (Bava Batra 14b–15a) discusses whether Joshua wrote the last eight verses of the Torah.  So the question of post-Mosaic material in the Pentateuch is not a modern critical invention;  it is an ancient observation.

Three further markers.

Genesis 36:31: the Edomite king list is introduced as covering kings who reigned "before any king ruled over the Israelites."⁴  Written by someone on the far side of the monarchy's beginning.

Genesis 12:6: "At that time the Canaanites were in the land."⁵  A parenthesis explaining an unfamiliar situation to a reader for whom it no longer obtains.

Numbers 12:3: "Now the man Moses was very humble, more so than any man on the face of the earth."⁶  Not impossible for Moses to have written, but it reads as a narrator's assessment.

The reasonable conclusion, held by conservative scholars for centuries, is that the Pentateuch contains Mosaic material within a framework edited later.  The argument is about how much of each.

Part 2 — The Documentary Hypothesis and what happened to it

The classic form

Julius Wellhausen's Prolegomena zur Geschichte Israels (1878) synthesised a century of source criticism into a scheme that dominated until the 1970s:

• J, the Yahwist, using the divine name YHWH from the start, dated tenth or ninth century.
• E, the Elohist, using Elohim until Exodus 3, dated ninth or eighth.
• D, Deuteronomy, identified with the law book found in Josiah's temple repair in 622 BC.
• P, the Priestly source, with the genealogies, the cult, and Genesis 1, dated post-exilic.

The observations behind it are real and a defender must answer them: the alternation of divine names, the doublets (two creation accounts, two flood chronologies, wife-sister episodes told three times), differences in vocabulary and theology between strands, and the apparent development of Israelite religion from a simple to an elaborate cult.

The argument behind the late date for P, which is usually left out

This article listed P's post-exilic dating without giving the reason for it, which is unfair to a real argument.  Joseph Blenkinsopp's standard introduction supplies it.  The decisive move toward Wellhausen's order was made by Edouard Reuss:

Reuss pointed out that the pre-exilic prophets betray no familiarity with the Mosaic legal system, and that the ritual law in particular, closely related as it is to Ezekiel, could not have originated earlier than the exilic period.⁷

That is a checkable claim about the prophets, and it is the load-bearing argument for the whole JEDP sequence, since P's lateness is what forced the chronological order.  Anyone defending an early P has to answer it, and pointing at the collapse of the Elohist does not.

And it makes the Ketef Hinnom amulets far more important than their usual apologetic use suggests.  Reuss's premise is that Priestly material was unknown before the exile.  The amulets carry the Priestly Blessing of Numbers 6:24–26.  If they are pre-exilic, they are not merely a nice early attestation — they are a direct counter-instance to the argument on which the classical dating of P rests.  That is exactly why the palaeographic dispute over them, set out below, matters so much and why Na'aman's 2011 argument for an early Second Temple date is not a minor quibble.

The divine-names criterion, and its limits

The alternation between YHWH and Elohim is the argument most often given in popular accounts, and Blenkinsopp is careful about what it can carry:

It is important to bear in mind that the criterion of divine names was, from the beginning, quite limited in application.  Initially it was applied only to Genesis and the first two chapters of Exodus—in other words, down to the point in the narrative at which the divine name YHWH was revealed to Moses (Ex 3:13–15, with a parallel version in Ex 6:2–3).⁷

After Exodus 3 there is no reason for an Elohist to avoid the name, so the criterion stops working — which means it never applied to most of the Pentateuch.

He then reports the criticism, from a critical scholar rather than an apologist:

it has since been pointed out by several critics of the theory that the incidence of divine names is not nearly as consistent as the hypothesis requires, that in particular the Garden of Eden story combines both in a single designation (YHWH Elohim), and that the alteration of names is patient of other explanations.⁷

Conservatives should note what this does and does not give them.  It weakens the single most-cited popular argument for source division, and Blenkinsopp — who accepts a composite Pentateuch — says so plainly.  It does nothing about the doublets, the flood chronology, or Reuss's argument about the prophets, which are where the real weight sits.

Why the identification of D with Josiah's law book was so powerful

Because it supplied the one absolute date.  Koller states the structure plainly:

Scholars have often argued that the Torah is comprised of originally disparate sources.  They have often disagreed about sequencing those sources on a timeline, however.  What they have usually agreed on is the linchpin: the book of Deuteronomy is regularly dated to the seventh century BCE.⁸

He also notes that the Josiah connection is not a nineteenth-century idea: Jerome, the commentator printed as Rashi on Chronicles, and Hobbes all held some version of it, and W. M. L. de Wette in 1805 "was apparently only reflecting what he took to be the conventional view."⁸

The fracture

Three things happened.

E dissolved.  The Elohist has been progressively abandoned as a continuous source;  most European scholars no longer work with it.

The model changed shape.  Rolf Rendtorff and Erhard Blum argued for a supplementary model — a growing text repeatedly expanded — rather than four parallel documents spliced together.  John Van Seters and Thomas Römer developed further variants.  The disagreement is about the basic mechanism of composition, not about details.

A North American school defended a revised documentary model.  Joel Baden, Jeffrey Stackert and Baruch Schwartz argue for a documentary approach on narrative-continuity grounds, without Wellhausen's evolutionary theory of Israelite religion.

What this does and does not mean.  It means there is no consensus model of Pentateuchal composition, and that anyone who tells you "scholars agree the Pentateuch is JEDP" is describing 1950.  It does not mean the underlying textual observations went away, and it emphatically does not mean the field has moved toward Mosaic authorship.  It has not.

The critical side describes the situation the same way, and it is worth having in their words rather than in a conservative summary of them.  Finkelstein and Silberman:

In the last few decades scholarly opinions about the dates and authorship of these individual sources have varied wildly.  While some scholars argue that the texts were composed and edited during the existence of the united monarchy and the kingdoms of Judah and Israel (c. 1000–586 BCE), others insist that they were late compositions, collected and edited by priests and scribes during the Babylonian exile and the restoration … or even as late as the Hellenistic period.  Yet all agree that the Pentateuch is not a single, seamless composition but a patchwork of different sources, each written under different historical circumstances to express different religious or political viewpoints.¹⁵

Two things follow, one for each side.  For conservatives: the disagreement about dates is enormous — a six-hundred-year spread among people who all work critically — and that is a genuine weakness in the field's confidence, worth saying.  For everyone: "all agree" that it is composite.  The fracture of the Wellhausen synthesis did not produce a single-author Pentateuch and nobody in the field thinks it did.

Part 3 — The treaty-form argument

This is the most substantive live argument bearing on the date of Deuteronomy, and it deserves to be laid out properly because both sides misreport it.

The claim

Ancient Near Eastern treaties have a recognisable structure.  Second-millennium Hittite vassal treaties typically include a preamble, a historical prologue rehearsing the suzerain's past kindness, stipulations, provisions for deposit and periodic public reading, a witness list, and blessings and curses.  First-millennium Neo-Assyrian treaties, including Esarhaddon's Succession Treaty of 672 BC, generally lack the historical prologue and are curse-heavy.

Deuteronomy has the full second-millennium shape: a historical prologue (chapters 1–3), stipulations, a deposit-and-reading provision (31:9–13), witnesses, and blessings and curses (chapter 28).

Kenneth Kitchen made this argument from 1966 onward and set it out at length in On the Reliability of the Old Testament.⁹  It was largely dismissed for decades.

His version is more specific than the summary above, and the specificity is the point.  Kitchen sorts the whole ancient Near Eastern corpus of treaty, law and covenant documents into six chronological phases, and compares the Sinai covenant and its two renewals against them element by element.  His result:

the glaring contrast between Hittite phase V plus Sinai on the one hand and both variants of the later phase VI on the other.  In VI there is no historical prologue, no blessings to match with curses, no deposition, no reading arrangements.  But all these are common to phase V and Sinai, and … not to any other phase at all!  Sinai and its two renewals — especially the version in Deuteronomy — belong squarely within phase V.⁹

Phase V is the Hittite corpus of roughly 1400–1200 BC.  Phase VI is the first-millennium material, Sefire in the west and the Assyrian treaties in the east, roughly 900–650.  So the claim is not a loose resemblance;  it is that four specific structural features cluster in the second-millennium group and in the biblical texts, and are absent from the first-millennium group.

He also anticipates the obvious objection.  The Exodus–Leviticus version splits its stipulations either side of the tabernacle material, which looks untidy.  Kitchen answers from the Hittite corpus itself: the treaty between Tudkhalia IV and Ulmi-Tesup "intercalates the deposition (4a) between two lots of stipulations (3), and then inserts further stipulations after curses and blessings … and then adds more witnesses at the end."  In treaties with peoples rather than kingdoms — Israel's case — the Hittites varied the order in the same way.⁹

And then he overclaims, which should be reported too.

The impartial and very extensive evidence (thirty Hittite-inspired documents and versions!) sets this matter beyond any further dispute.  It is not my creation, it is inherent in the mass of original documents themselves, and so cannot be gainsaid, if the brute facts are to be respected.⁹

It was disputed, in print, by Levinson and Stackert, and the dispute is unresolved.  "Beyond any further dispute" is the language of someone who has stopped expecting an answer, and it is the same register as Steven Collins's "absolutely incontrovertible" about Sodom.  The argument is strong;  the sentence is not evidence.

The revival and the reply

Joshua Berman argued in the Journal of Biblical Literature (2011) that a fifteenth-century Hittite text, CTH 133, is a closer parallel to Deuteronomy 13 than Esarhaddon's Succession Treaty is, and in a second article (2013) that Deuteronomy 1–3 belongs to the Hittite treaty prologue tradition.¹⁰

Bernard Levinson and Jeffrey Stackert replied in the Journal of Ancient Judaism (2012) with three arguments: the Deuteronomy–Esarhaddon parallels are too strong to dismiss, the Deuteronomy–Hittite parallels are too weak to keep, and there is no known historical route by which Hittite treaty forms could have reached Israelite scribes.¹¹

Berman replied;  Levinson and Stackert replied again.  The exchange ran through 2013.

How it stands

Aaron Koller of Yeshiva University, who is not writing as a conservative and does not endorse Berman's conclusions, assesses it as follows.

On the transmission objection — the argument that Hittite forms could not have reached Israel — he points out that the same objection would have blocked the Assyrian thesis until very recently: "until 2009 it was not known that copies of the VTE were brought to the west, either, and yet scholars were prepared to accept that the Neo-Assyrian treaties had influenced Deuteronomy."⁸  He grants that Assyrian influence on the eighth-century Levant is massively attested and Hittite influence is not, so the two cases are not symmetrical.  But he concludes that transmission should be a question rather than a precondition: "If they are [compelling], then the 'how' of the transmission can be left as a question, rather than a necessary pre-condition for analysis."⁸

On the parallels themselves he finds neither side decisive: "there seem to be no 'smoking guns' that reveal the influence of either the Vassal Treaties of the seventh century, or the Hittite treaties of the thirteenth century, on Deuteronomy."⁸

On the larger structure he sides with the second-millennium reading: "the basic structure of Deuteronomy is more similar to Hittite treaties than to the Neo-Assyrian treaties."⁸  He notes the qualification that one Neo-Assyrian treaty, with the king of Qedar, apparently had a historical prologue, which weakens the argument — and then observes that this raises its own problem, since it means the Hittite form somehow reached Neo-Assyrian times by a route nobody can trace either.⁸

His verdict:

That Deuteronomy, at least in some form, dates back to the second millennium BCE appears to be a more tenable scholarly position than it did earlier.⁸

How much weight to put on this.  Not too much, and it should be quoted with its hedges intact.  "More tenable than it was" is not "established."  Koller is one scholar summarising an unresolved exchange.  But the exchange itself is real, it ran in the field's leading journals, and it is about the single date on which the absolute chronology of the whole Pentateuch has hung for two centuries.  That is worth knowing.

Part 4 — Could Moses have written?

Writing

The objection that alphabetic writing did not exist in Moses's time is dead.  Proto-Sinaitic inscriptions at Serabit el-Khadim in the Sinai, and the Wadi el-Hol inscriptions in Egypt found by John and Deborah Darnell in 1999, are alphabetic writing of a Semitic language, dated to roughly 1900–1800 BC.¹²  That is centuries before the earliest proposed Exodus and in exactly the right cultural setting: Semitic speakers working in an Egyptian context.

This does not show that Moses wrote anything.  It removes an objection that used to be made and should not be made now.

Egyptian colouring

The Pentateuch, and the Joseph and Exodus narratives especially, contain Egyptian personal names, loanwords, and administrative details.  "Moses" itself is plausibly the Egyptian element msy, "born of," found in names like Thutmose and Ahmose.

How much this proves.  Less than it is usually made to.  Egyptian material was available to Israelite writers throughout the first millennium — Judah had continuous contact with Egypt, and there were Judean communities in Egypt from at least the sixth century.  Egyptian colouring is consistent with an early origin and does not establish it.  This is the diagnostic-versus-consistent test applied to our own side's argument.

Ketef Hinnom

In 1979 Gabriel Barkay excavated two rolled silver amulets from a burial chamber at Ketef Hinnom in Jerusalem.  Unrolled, they carry inscribed Hebrew text including the Priestly Blessing of Numbers 6:24–26 and language paralleling Deuteronomy 7:9.¹³

Why it matters.  Numbers 6 is assigned to P and Deuteronomy 7 to the frame of D — on the classic scheme, two different sources, one of them supposedly post-exilic.  Erik Waaler's argument is that the inscriber "is likely to have used a single source for these two quotations, a source that probably included more Pentateuchal material," and therefore that "disparate Pentateuchal texts existed and were conjoined prior to the reform of Josiah."¹³

The dating problem, which apologetic uses routinely omit.  Palaeographers have not agreed, and the disagreement has a history worth setting out in order.

1989–1992.  Barkay published the amulets and dated them to the late pre-exilic period — on Waaler's summary, "the second half or last quarter of the seventh century."¹³

1995.  Johannes Renz, in his Handbuch der Althebräischen Epigraphik, dated them far later.  Waaler states the range this produced: "Most scholars have dated the two amulets to the late seventh or early sixth century BC, but Renz dates them to the early post-exilic period or to one of the two centuries leading up to the Christian era."¹³  Ada Yardeni dated them to the early sixth century.¹³

2003–2004.  Barkay, with Andrew Vaughn, Marilyn Lundberg and Bruce Zuckerman, re-read the amulets using advanced imaging and published a new edition and evaluation in BASOR 334, supporting the late pre-exilic date.¹⁴

2011.  Nadav Na'aman published "A New Appraisal of the Silver Amulets from Ketef Hinnom" in the Israel Exploration Journal, arguing on orthographic grounds for an early Second Temple date, no earlier than the second half of the sixth century BC.  Barkay's team published a rejoinder.¹⁴

So the spread across specialists runs from about 725 BC to the first century BC, and the argument did not stop when the excavators re-published.  This project has read none of the 2003, 2004 or 2011 publications, and reports their existence and positions rather than adjudicating between them.

What can be said.  If the amulets are pre-exilic, they are a strong datum against a wholly post-exilic P, and two "different sources" on one object before Josiah is exactly the kind of evidence that constrains composition models.  If Na'aman or Renz is right, they constrain nothing about the pre-exilic period.  The honest statement is that this is a good argument whose foundation is contested by named specialists in print, and that anyone citing it should cite the contest.

Common objections

"Moses wrote the Pentateuch."  Not the account of his own death and burial, not the note that no prophet like him arose afterwards, and not the Edomite king list dated relative to Israel's monarchy.  This has been acknowledged in Jewish tradition since the Talmud.

"The Documentary Hypothesis has been disproved."  Its classic Wellhausen form no longer commands consensus and the E source has largely been abandoned.  But the field replaced it with other composite models, not with Mosaic authorship, and the textual observations that generated it are still there.  Overstating this is the most common error in conservative popular writing on the subject.

"Scholars all agree on JEDP."  Also false, in the other direction.  European and North American scholarship disagree about the basic compositional mechanism.

"Writing hadn't been invented yet."  Alphabetic Semitic writing is attested around 1900–1800 BC.¹²

"The doublets prove multiple sources."  They are evidence for it and are not conclusive.  Ancient Near Eastern narrative uses repetition and variation deliberately, and some proposed doublets are better read as parallel episodes.  But the flood chronology and the divine-name alternation are harder to dismiss than popular conservative writing admits, and a serious defence has to engage them individually rather than by appeal to literary convention in general.

"The treaty form proves a second-millennium Deuteronomy."  It makes it more tenable than it was, in the words of a non-conservative assessor of the debate.⁸  It does not prove it, the exchange is unresolved, and the strongest single counter — that Assyrian influence on the Levant is massively attested while Hittite influence is not — has not been answered.

"None of this matters theologically."  It matters to some claims and not others.  A Pentateuch edited over centuries is compatible with inspiration on most accounts of inspiration;  it is not compatible with a claim that Moses personally wrote every word.  Which of those a reader needs is a theological question, not a historical one.

What would settle it

• A pre-exilic manuscript of any Pentateuchal passage of substantial length. Ketef Hinnom is currently the whole of the evidence in this category and its date is disputed. A dated ostracon or inscription quoting narrative Pentateuchal material would be decisive.
• Publication and general acceptance of a redating of the Ketef Hinnom amulets. The palaeographic range of six hundred years is doing enormous work and could in principle be narrowed.
• A Hittite treaty text found in the southern Levant. This would remove Levinson and Stackert's transmission objection at a stroke. Its absence is currently their strongest card.
• Statistical stylometry with agreed methods. Attempts exist and have not converged; the field has not settled on a method that both camps accept, which is itself informative.
• Any external attestation of Moses. None exists, and the honest expectation is that none will, since a Semitic leader in the second millennium would not normally leave Egyptian records.

Sources and further reading

The literature reviewed for this article, including sources considered and not used, is in literature-review.md.