Genesis and Ancient Near Eastern Parallels

This entire story, precisely as it is here written, afterwards travelled to Canaan.¹

In the light of the differences, the resemblances fade away almost like the stars before the sun.²

Because the Flood episode in Gen. 6-8 matches the older Babylonian myth so well in plot and, particularly, in details, few doubt that Noah's story is descended from a Mesopotamian account.³

Summary

This article puts the Mesopotamian creation and flood tablets beside Genesis 1-11 and asks two questions.  How close are the parallels, and which way, if any, did the borrowing run?  It does not ask whether the Flood happened;  that belongs to Fossil Order and the Flood and Noah's Ark:  The Logistics.  It does not ask what Genesis 6-9 claims;  that is The Flood Narrative.  The worldwide flood traditions are in Flood Legends Around the World, the long lifespans and the Sumerian King List numbers in The Long Lifespans of Genesis, and the shape of the cosmos in Ancient Cosmology and Genesis.  This article owns the question of literary dependence.

1. The flood parallel is real, specific, and close.  Gilgamesh Tablet XI and Genesis 6-9 share seventeen plot points in the same order:  a divine decision, a warning, a command to build, obedience, entry, a shut door, the storm, universal death, the end of the storm, grounding on a mountain, an opened window, birds sent out, exit, sacrifice, the gods smelling it, and a blessing on the hero.⁴  Both boats are sealed with bitumen, and the Hebrew word for it in Genesis 6:14, kofer (כֹּפֶר), occurs nowhere else in the Bible and is the Akkadian kupru of Gilgamesh XI 65.⁵  Nobody who has read both texts denies the relationship, and this article does not.

2. The creation parallel is thin, and the Assyriologists say so more bluntly than the apologists.  Alexander Heidel's 1951 audit of every proposed point of contact between Enuma Elish and Genesis 1 ended with "no incontrovertible evidence can for the present be produced for either side."²  W. G. Lambert, who spent fifty years editing Enuma Elish, wrote in 1965 that it "is not a norm of Babylonian or Sumerian cosmology" but "a sectarian and aberrant combination of mythological threads," and that on the one close parallel, the splitting of the waters, there is "no evidence of Hebrew borrowing from Babylon."⁶  The famous Tiamat-tehom etymology is dead:  Heidel showed in 1951 that the Hebrew cannot be a loan from the Babylonian, and Kenneth Kitchen agrees.²

3. The Mesopotamian tablets are older as written objects, and by a wide margin.  The main Atrahasis manuscript carries a colophon dated to the twelfth year of Ammi-saduqa of Babylon, about 1635 BC.⁷  The Sumerian flood story is on a Nippur tablet that Miguel Civil dates no earlier than the late Old Babylonian period, about 1600 BC.⁷  Gilgamesh in Akkadian was already in the written tradition in the eighteenth century BC.³  No surviving Hebrew manuscript of Genesis is older than the Dead Sea Scrolls, and no date anyone proposes for the composition of Genesis is earlier than these tablets.

4. But the flood was not in Gilgamesh until late, and the oldest copies of the Gilgamesh flood are younger than the Hebrew monarchy.  Andrew George, editor of the standard critical edition, reports that in every second-millennium version of Gilgamesh "there is still no sign of the incorporation in any version of an account of the great Flood."³  The flood in Tablet XI was lifted from Atrahasis when the Standard Babylonian edition was made, probably between 1200 and 1000 BC, and no copy of Tablet XI is older than about 750 BC.⁶  The one parallel Lambert called decisive, the birds of Genesis 8:6-12, is attested in Mesopotamia only in Tablet XI and in Berossus.  Neither Atrahasis nor the Sumerian story has any birds.⁶

5. The Megiddo fragment does not contain the flood, and should stop being cited as if it did.  The tablet a shepherd picked up at Megiddo in 1954 is a Late Bronze Age local paraphrase of the death of Enkidu, the material of Tablet VII, not Tablet XI.³³  It proves that Gilgamesh circulated in Canaan in the Amarna age.  It does not prove that the flood story did.

6. The differences are as specific as the resemblances.  The Babylonian boat is a cube, a "field" in area and ten rods high, with six decks and nine compartments;  Noah's is 300 by 50 by 30 cubits with three decks.⁸  The Babylonian storm lasts six days and seven nights;  the Genesis flood a year and ten days.⁸  Utnapishtim sends a dove, a swallow and a raven;  Noah a raven and then a dove three times.  Utnapishtim lies to his neighbours on the god's instruction, loads his gold, silver and craftsmen, and is made immortal;  Noah says nothing, takes his family, and gets drunk.⁵  The gods send the flood because human noise keeps Enlil awake and then cower "like dogs" and crowd "like flies" round the sacrifice because they are hungry;  Genesis gives a moral reason and one God who needs nothing.⁸

7. The bird episode, the strongest single link, cuts in no clear direction.  E. A. Speiser, who translated Gilgamesh XI for ANET, wrote in his Genesis commentary that the Babylonian order was "raven-swallow-dove."  It is dove-swallow-raven, and Speiser got it backwards.⁹  Heidel argued the Genesis order (raven first) showed better sense and so priority;  David Freedman argued the same from seafaring practice;  George rejects both and reads the Babylonian order as an aetiology of bird habits that Genesis "obscures."³  Each side reads its own text as the sensible one.  That is not a test.

8. Pan-Babylonianism claimed far more than the tablets support, and its specific claims failed.  Friedrich Delitzsch told the Kaiser in 1902 that the flood story travelled to Canaan "precisely as it is here written," that the tablet was "committed to writing 2000 years before Christ," and that the ten antediluvian kings match the ten patriarchs with "agreement ... perfect in all details."¹  The tablet he showed was a seventh-century copy;  the main King List recension has eight kings, not ten;  and the two-versions-of-the-flood argument he leaned on is answered in The Flood Narrative.¹⁰

9. The conservative counter-arguments are weaker than their users think.  Kitchen's argument that Genesis is "a more concise, simpler account, and not an elaboration" assumes that stories only grow;  an epitomiser leaves the same signature.¹¹  Kitchen's argument that the "primeval protohistory" genre was not composed after 1600 BC, so Genesis 1-11 must be patriarchal-age, rests on the Mesopotamian corpus having stopped, which says nothing about a Hebrew author.¹¹  Gerhard Hasel's demonstration that Genesis 1 is an "antimythical polemic" shows that Genesis knew the mythology it rejects;  a polemic is evidence of contact, not of independence.¹²

10. The "shared memory of a real event" reading is coherent, but the Mesopotamian evidence now leans against it as a written tradition.  Y. S. Chen's 2013 Oxford study finds that the Sumerian King List's antediluvian section is absent from its Ur III copy, that the Early Dynastic versions of the Instructions of Shuruppak have no Ziusudra and no flood, and that "Mesopotamian traditions related to the Flood only emerged from the Old Babylonian period onwards."¹³  A memory that is real can still be written down late.  But the claim that the Mesopotamian flood texts preserve a tradition running back unbroken to the third millennium no longer has a text to stand on.

11. Where the parallels are strongest, they are in literary detail, and that is what borrowing predicts.  Two independent memories of a flood would agree that there was a boat and a survivor.  They would not both send out birds at seven-day intervals from a window after grounding on a mountain, and both have the deity smell the sacrifice.  Shared literary detail points to a shared literary tradition somewhere upstream.  Millard's reply, that Genesis cannot have been borrowed "from any text now known to us," is true and is not the same as showing there was no borrowing.¹⁴

12. What would settle the direction, and does not yet exist.  A Mesopotamian text older than any plausible date for Genesis that carries a Genesis-only feature (a raven-then-dove sequence, seven pairs of clean animals, a year-long flood, a 300-cubit hull) would show borrowing into Genesis.  A West Semitic flood text older than the Old Babylonian period would open the reverse.  A shared idiom beyond kupru would show a textual rather than oral channel.  None of these has been found.  Lambert's own reconstruction, oral transmission westward in the Amarna age, came with his caveat that the borrowing "may have been something altogether more involved and complex than we have suggested."⁶

Verdict.  The parallels are real.  The Mesopotamian flood tradition was written down at least a thousand years before any surviving Hebrew text, and the specific details it shares with Genesis 6-9 are too many and too ordered to be coincidence.  The creation parallel with Enuma Elish is slight and the Assyriologists who know the text best say so.  Literary dependence in either direction is undemonstrated:  the Hebrew story shares no wording with any known tablet beyond one loanword, the tablets that carry the closest parallel are younger than the Hebrew monarchy, and the arguments for Hebrew priority are as circular as the arguments against it.  What can be said is that both traditions descend from something older than either, and that the something included a boat, a mountain, a window, birds, and a sacrifice.  Whether that something was an event or a poem is a question the tablets cannot answer.

Every source consulted is catalogued in the literature review and every quotation with its route of verification in the quotes file.

The texts, and when they were written

Every argument about dependence turns on dates, and two dates get confused:  the date of the surviving tablet and the date of the composition it carries.  The table gives both, from the critical editions.

Three things in the table decide most of what follows.

First, the Mesopotamian flood was a written text by about 1635 BC.  That is the colophon date on Ku-Aya's copy of Atrahasis, "Month Nisan, 21st day, the year when Ammi-saduqa, the king" set up a statue of himself, which Lambert and Millard place in that king's twelfth year, about 1635 BC.⁷  Whatever date one gives Genesis, this is earlier.  The traditional Mosaic date is two to four centuries later.  The critical date for the Priestly source is a thousand years later.

Second, the flood is a latecomer to Gilgamesh.  Andrew George's 2003 edition surveys every Old and Middle Babylonian piece of the epic, from eighteenth-century Babylon to thirteenth-century Emar and Hattusa, and concludes:  "There is still no sign of the incorporation in any version of an account of the great Flood."³  Tablet XI as we have it is the work of the Standard Babylonian editor, and it borrowed the flood from Atrahasis so directly that it twice calls its own hero by the other poem's name:  the crowd assembles "at Atra-hasis's gate" (XI 49), and Ea says "I let Atra-hasis see a dream" (XI 197).⁸

Third, the only Mesopotamian flood text with the birds is that late one.  Lambert stated the consequence in 1965 and it has not changed:  "the only surviving testimony to the most telling parallel happens to be later than the Biblical account."⁶  He went on to say he was nonetheless certain of Hebrew dependence on a Mesopotamian tradition.  That distinction, between a tradition and a text, is the whole of this article.

The flood: Gilgamesh XI beside Genesis 6-9

The table below is built from the electronic Babylonian Library's edition of Tablet XI, which is Andrew George's text and translation, and the ESV.  Line numbers follow George.  Gordon Wenham's 1978 list of seventeen shared plot points, with his Genesis source assignments, is given in the last column, because it lets a reader see which parts of the story the source critics assign to J and which to P.⁴

What the table shows

The spine is the same.  Wenham counted seventeen points shared between the combined Genesis narrative and Tablet XI, twelve shared with J alone and ten with P alone, and drew the conclusion that matters for source criticism, which The Flood Narrative takes up:  the whole is closer to Babylon than either of its supposed parts.⁴  For this article the point is simpler.  Seventeen ordered plot points, including a window, birds, a sacrifice and a deity who smells it, are not what two unrelated tellings of a flood would produce.  Speiser called the correspondence "inescapable proof of basic interrelationship," and Kitchen, who denies copying in either direction, concedes "the basic contents are common to both."⁹

The one verbal link is kupru.  Genesis 6:14 uses kofer for pitch, a word found nowhere else in the Hebrew Bible;  the ordinary Hebrew word, used in Genesis 11:3 and 14:10, is hemar (חֵמָר).  Wenham notes that kupru "a common Akkadian term, is used in the Gilgamesh epic (11:65, which is a remarkable parallel with the biblical tradition)."⁵  Speiser saw the same:  "The same substance and the cognate term for it are found in the corresponding cuneiform passage."⁹  A cognate is not a loan;  kupru and kofer could both descend from a common Semitic root, as tiamtu and tehom do.  But a word used once in the Bible, in the one passage that parallels a Babylonian text, in the one line that parallels it most closely, is the kind of evidence a textual channel would leave.  It is also the only such evidence.  The word for the ark itself, tebah (תֵּבָה), is "probably an Egyptian loan word," used elsewhere only of the basket that carried the infant Moses.⁵  A Hebrew author copying a cuneiform text would not have reached for an Egyptian word for the boat.

The differences

The differences are as concrete as the resemblances, and the conservative commentators have listed them since Heidel.  Kitchen's eight are representative:  the reason (noise versus sin), the secrecy (one god's private tip versus an open command), the boat (cube versus ship), the duration, the passengers (craftsmen versus family), the birds, the hero's initiative in leaving and sacrificing, and the replenishing of the earth.¹¹  Wenham adds the character of the hero.  Utnapishtim "is a very active figure, hurrying hither and thither building his ship, spinning a yarn to his neighbors," loads "a large crew and all his gold and silver," describes the storm and his own tears, and ends among the gods;  Noah "never speaks," is shut in by God, and "the last thing we hear of him is his disgracing himself through drinking too much."⁵

Two of these deserve more weight than the lists give them.

The boat.  Ea's instruction is that "her breadth and length should be the same," and Utnapishtim builds a vessel an "acre" in area and ten rods high, which George's commentary calls "this strange boat," whose "configuration ... was obviously not drawn from practical experience."³  Genesis gives the proportions of a ship.  Noah's Ark:  The Logistics treats this at length and gives the rule that governs its use here:  the Hebrew specification is better engineering, and that is not evidence that it is earlier.  A later writer improving a cube into a hull is at least as likely as an earlier hull degrading into a cube.

The gods.  In Tablet XI the flood is a mistake that its authors regret.  The gods flee to the highest heaven and "curled up like dogs" (116);  Belet-ili wails that she "spoke evil in the assembly of the gods" (120);  they "gathered like flies round the sacrificer" (163) because they have gone without offerings for a week;  Ea tells Enlil "how could you lack counsel and cause the deluge?" and proposes that a lion, a wolf, a famine or a plague would have thinned the people more sensibly (184-195).⁸  The same scene in Atrahasis, five hundred years older, already has the gods "gathered like flies over the offering."⁷  Genesis has one God, a moral cause, a covenant, and no hunger.  Wenham's reading is that "Genesis appears to be self-consciously avoiding" the Babylonian kind of narration, and that the flood story may be "quietly protesting about Mesopotamian kingship ideologies."⁵  That is a reading of intent, and it is plausible.  What it shows, if it is right, is that the Genesis author knew the Mesopotamian shape well enough to invert it.

The bird episode, which everyone calls decisive and nobody can use

Lambert wrote that "the episode of the birds in Gen 8:6-12 is so close to the parallel passage in the XIth tablet of the Babylonian Gilgamesh Epic that no doubt exists" of a connexion.⁶  He is right that the episode is the closest point of contact:  a survivor on a grounded boat, releasing birds at seven-day intervals, the last bird failing to return because the waters have gone down.  He is also the scholar who noted that this episode appears in no Mesopotamian text older than the Standard Babylonian Tablet XI, of which "no copies earlier than 750 B.C. are known."⁶

The sequence is different, and the difference has been argued in both directions.  Gilgamesh sends a dove, then a swallow, then a raven.  Genesis sends a raven, then a dove three times.  Heidel thought Genesis showed better sense, since a raven, a scavenger, would find carrion on the receding waters and not return, while the dove's three flights track the drying of the land;  by this reading the Babylonian order is a corruption of the biblical one.  David Freedman argued from ancient seafaring that Genesis preserves correct practice and that the Babylonian poet "borrowed a Hebrew topos."  George's commentary on Tablet XI works through both and rejects them.  Heidel's inference, he writes, was "based as it was on a theory of bird behaviour extrapolated from the biblical account and thence applied unaltered to the cuneiform tradition," and is "methodologically suspect";  Freedman's evidence that doves were not used to scout for land is answered by "evidence for sailors in different parts of the world releasing doves with exactly the last aim."  George's own reading is that the Babylonian sequence is an aetiology:  doves and swallows nest with people because their ancestors came back to the boat, ravens do not because theirs flew off, and "the order of the birds in Genesis obscures such an explanation."³

Each scholar finds his own text's order to be the intelligible one and the other's to be a corruption of it.  Heidel and Freedman were reading Genesis;  George edits Gilgamesh.  Nothing in the birds themselves decides between them, and this article does not pretend otherwise.  One small fact does bear on how carefully the case has been made.  Speiser, who translated Tablet XI for ANET and knew the sequence, wrote in his Anchor Bible commentary that the order of the test flights was "raven-swallow-dove in Gilg."⁹  It is dove-swallow-raven.  The most cited English commentary on Genesis has the Babylonian birds backwards.

The older flood texts: Atrahasis, the Sumerian story, and the Ark Tablet

Gilgamesh XI is the fullest Mesopotamian flood, but it is the youngest.  The three older texts matter because they show what the tradition looked like before the Standard Babylonian editor got to it, and because two of the most quoted parallels are missing from all three.

Atrahasis is the source.  Lambert and Millard's 1969 edition rests on Ku-Aya's three-tablet copy, dated by colophon to the reign of Ammi-saduqa, and they summarise the story it tells:  the junior gods strike against their labour, man is created from clay and the blood of a slain god to do the work, mankind multiplies and its noise keeps Enlil awake, plague and drought and famine are sent and fail, and at last the flood, of which Atrahasis is warned by Ea through the wall of his reed hut.⁷  This is the "Creation-Rebellion-Flood" frame that Millard, in the lecture that first set the newly reconstructed epic beside Genesis, said changed the comparison:  "it is no longer legitimate to describe the Hebrew Flood story as 'borrowed' from a Babylonian 'original' without including its complementary Creation account."¹⁴  Wenham draws the same conclusion in his commentary:  "the Atrahasis epic from the early second millennium shows that the basic plot of Gen 1-11 was already known then."⁵  What Atrahasis does not have, in any surviving piece, is the birds;  the relevant part of Tablet III is broken, and Lambert's 1965 statement that "so far there is nothing about birds" still stands.⁶

The Sumerian flood story is one broken tablet.  Miguel Civil's edition in the same volume describes CBS 10673 as "the bottom third, approximately, of the complete tablet, apparently from Nippur," with no duplicates and no colophon, dated on internal grounds "not earlier than Late Old Babylonian."⁷  What survives runs:  the gods make "the black-headed people" and the animals;  kingship comes down from heaven and five cities are founded, Eridu first;  the gods swear an oath to send a flood;  Ziusudra, "the king, the gudug priest," hears the decision through a wall;  "all the windstorms and gales arose together, and the flood swept over," and "waves and windstorms had rocked the huge boat for seven days and seven nights";  Utu the sun comes out, Ziusudra opens the boat, sacrifices "oxen and offered innumerable sheep," and is given "life like a god" in Dilmun, "where the sun rises."²⁰  Thorkild Jacobsen renamed it the "Eridu Genesis" in 1981 because, in his reading, it "prefigures ... the biblical Genesis in its structure":  creation, then a list of rulers with their reigns, then the flood, "combined simply by arranging them along a line in time."²¹  He drew the conclusion that the chronological interest in both is the mark of "a mytho-historical genre," and that the biblical Priestly source, which he dated about 500 BC, was "influenced by Mesopotamian writings" but used them "as models rather than having been directly borrowed from."²¹  A reader should notice that the middle member of Jacobsen's tripartite structure, the king list, is not in the surviving tablet at all;  he restored it from the King List and from the later bilingual fragment.

The Ark Tablet is sixty lines on a palm-sized Old Babylonian tablet, "probably about 1750 BC," that Irving Finkel published in 2014.¹⁵  Its boat is round.  Enki tells Atrahasis to "draw out the boat that you will make on a circular plan," and the surviving lines are almost all about its construction:  ropes and palm-fibre, thirty ribs, 3,600 stanchions, cabins above and below, and bitumen, again kupru, "one finger" thick inside and out, with the kilns loaded with 28,800 measures of it.¹⁵  Finkel's argument that Utnapishtim's "acre" of floor with equal length and breadth was a coracle description misread by the later poet as a cube is attractive and cannot be checked;  it is discussed with the NOVA reconstruction in Noah's Ark:  The Logistics.  For this article the tablet's importance is that a third boat shape is now in the Mesopotamian record, which makes the Genesis hull one of three, not the odd one out from a single Babylonian original.

Set the three older texts beside Tablet XI and the pattern is plain.  The plot elements that Genesis shares with all of them are the divine decision, the private warning to one man, the boat, the seven-day unit, the sacrifice and the divine response to it.  The elements that Genesis shares with Tablet XI alone are the window and the birds.  Those two are the closest parallels, and they are the ones with the shortest Mesopotamian pedigree.

The creation texts: Enuma Elish beside Genesis 1

Alexander Heidel's The Babylonian Genesis was written to test exactly the claim its title suggests.  Its third chapter takes every point of comparison that had been proposed between Enuma Elish and Genesis 1:1-2:3 since George Smith, and works through them one at a time.  The table below is Heidel's list, with his section titles, his verdict on each, and the primary lines in Lambert's 2013 translation.²

What Heidel found, and the three explanations he weighed

Heidel's chapter ends by laying out three ways to account for what the table shows.  The first is that the Babylonian story derives from the Hebrew.  He rejects it, with a caution that both sides of the later debate should keep:  "the Babylonian version antedates the Hebrew account," but "since priority of publication does not imply priority of existence, this argument must be used with a certain amount of caution.  The Hebrew story may have been current in some form or other many centuries before it assumed its present form."²  The second is that Genesis borrowed from Babylon, which he thought possible in a limited sense, and about which he wrote the sentence that has been quoted against him ever since:  "I personally fail to see why it should be incompatible with the doctrine of inspiration to assume that Gen. 1:1-2:3 might in a measure be dependent on Enûma elish.  But I reject the idea that the biblical account gradually evolved out of the Babylonian;  for that the differences are far too great and the similarities far too insignificant.  In the light of the differences, the resemblances fade away almost like the stars before the sun."²  The third is a common source.  His conclusion:  "no incontrovertible evidence can for the present be produced for either side;  I believe that the whole question must still be left open."²

Heidel was a Lutheran and worked at the Oriental Institute;  a skeptic may discount him.  The two scholars who have edited Enuma Elish since are harder to discount, and they are harsher on the parallel than he was.

Lambert: Enuma Elish is not Babylonian cosmology

W. G. Lambert's 1965 Oxford lecture, "A New Look at the Babylonian Background of Genesis," is the paper that ended the old consensus among Assyriologists, and his 2013 Babylonian Creation Myths is the critical edition he spent the rest of his life on.  The 1965 paper opens by recalling that "by the turn of the century the idea of dependence on a Babylonian original in the two cases of creation and flood was an accepted opinion in critical circles," and that the creation case was propped up by the flood case:  "even though the case for the creation narrative is dubious, the better case of the flood can be used to prove it, a very debatable procedure."⁶

His first conclusion is about the Babylonian side.  "The first major conclusion is that the Epic of Creation is not a norm of Babylonian or Sumerian cosmology.  It is a sectarian and aberrant combination of mythological threads woven into an unparalleled compositum."⁶  In 2013 he set out why:  the poem exists to show that Marduk, "utterly unimportant in Sumerian times," had replaced Enlil as head of the pantheon, and the date of its composition therefore cannot be earlier than that elevation, which he places under Nebuchadnezzar I.¹⁶  "Competent scholars have expressed opinions ranging from 1700 to 750 B.C.," he writes, against "our date of c. 1100 B.C."¹⁶  Speiser's Old Babylonian date in ANET is the older view Lambert was correcting.¹⁷

His second is about the points of contact.  Gunkel's 1895 argument had rested on the Old Testament poetry of God's battle with the sea, Rahab and Leviathan, as the missing link between Marduk's war with Tiamat and the calm of Genesis 1.  Ugarit removed it:  the Leviathan allusions "are derived from Canaanite Baal myths, and these show no signs of dependence on Mesopotamian sources.  Accordingly, one of the main supports for assuming the dependence of Genesis on Babylonian myths has gone."⁶  On the one parallel he thought real, the division of the waters on the second day, he found that the Babylonian battle-then-split sequence "may be only the result of conflation of no particular antiquity," attested first in Enuma Elish itself, and summed up:  "there is one close parallel between Genesis and Enūma Eliš, but no evidence of Hebrew borrowing from Babylon."⁶

Lambert did think two features of Genesis 1-11 might have Mesopotamian roots:  the seven-day week ending in rest, which he traced with the Babylonian šapattu to a common Amorite source rather than to Babylon, and the run of long-lived figures before the flood.⁶⁶  On the second, Hasel's 1978 audit in Andrews University Seminary Studies should be read beside him:  Lambert had written of "ten long-lived patriarchs," but "the major recension of the Sumerian King List (WB 444) contains only eight and not ten kings," one text has seven, the bilingual fragment nine, and only Berossus and one tablet, WB 62, give ten.¹⁰  The numbers themselves are in The Long Lifespans of Genesis.

Kitchen, and the one parallel he keeps

Kenneth Kitchen is an Egyptologist, a conservative, and a hostile witness to the older critical consensus.  On Enuma Elish he writes that it "was completed by circa 1000 from older sources" and that "despite the reiterated claims of an older generation of biblical scholars, Enuma Elish and Gen. 1-2 in fact share no direct relationship."  Tehom "is common to both Hebrew and Ugaritic (north Syria) and means nothing more than 'deep, abyss.'  It is not a deity, like Ti'amat."  Creation "is the massively central concern of Gen. 1-2, but it is a mere tailpiece in Enuma Elish."  The shared order is "the inevitable banalities:  creation of earth and sky before the plants are put on the earth, and of plants before animals (that need to eat them) and humans;  it could hardly have been otherwise!"  And then the concession:  "The creation of light before the luminaries is the only peculiarity that might indicate any link between the Hebrew and Enuma Elish narrative;  but where did it earlier come from?  Not known, as yet."¹¹

That is the honest state of the creation comparison.  Water at the start, a divided sky, light before the sun, a rest at the end.  The first is universal, the second is real and single, the third is real and unexplained, and the fourth is two different things with one name.

Hasel's polemic, and what it proves

Gerhard Hasel's 1974 paper in the Evangelical Quarterly, "The Polemic Nature of the Genesis Cosmology," is the standard conservative treatment and is more careful than its title.  Hasel goes through tehom, the separation of the waters, the luminaries and the creation of man, and argues at each point that Genesis 1 does not merely differ from the Babylonian and Canaanite texts but takes aim at them.  On the luminaries:  "Gen. 1 avoids the names 'sun' and 'moon' undoubtedly because these Common Semitic terms are at the same time names for deities," and the phrase "and the stars" in verse 16 "appears to be a parenthetical addition, whose purpose it is, in view of the prevalent star worship in Mesopotamia, to emphasize that the stars themselves are created things and nothing more."¹²  His conclusion is that Genesis 1 "represents not only a 'complete break' with the ancient Near Eastern mythological cosmologies but represents a parting of the spiritual ways brought about by a conscious and deliberate antimythical polemic which meant an undermining of the prevailing mythological cosmologies."¹²

Hasel's case is good, and it is also the reason the polemic argument cannot do the work apologists ask of it.  A polemic requires knowledge of its target.  If Genesis 1 unnames the sun and the moon because their Semitic names are the names of gods, then its author knew the gods.  If the flood story "quietly protests" the Mesopotamian kingship ideology, as Wenham suggests, then its author knew the ideology.  Wenham says this outright:  "Gen 1-11 as we read it is a commentary, often highly critical, on ideas current in the ancient world about the natural and supernatural world."⁵  A commentary is dependent on what it comments on.  The polemic reading and the dependence reading are the same reading with different adjectives.  What the polemic reading defeats is not dependence but the claim that Genesis is a "pale reflection" of Babylon, which is Delitzsch's claim and nobody's now.

The smaller parallels

Four other comparisons appear in every survey.  Each is real, each is slight, and each is stated here at its actual size.

The rib.  In Enki and Ninhursag the god Enki eats eight plants and falls ill in eight parts of his body;  Ninhursag bears a healing deity for each, named by a pun on the ailing part:  "'My brother, what part of you hurts you?'  'My ribs (ti) hurt me.'  She gave birth to Ninti out of it."¹⁸  Sumerian ti means both "rib" and "life," so Ninti is "lady of the rib" and "lady of life."  Eve is made from Adam's rib and named "mother of all living" (Gen 2:21-22, 3:20).  The pun does not work in Hebrew, where tsela (צֵלָע) "rib" and chawwah (חַוָּה) "Eve" share nothing.  Kramer proposed the connexion, Lambert reported it, and it remains a curiosity:  the only route by which a Sumerian pun could reach a Hebrew author who could not hear it is an intermediary who explained the joke.⁶

Dilmun.  The same poem opens on a land where "the lion did not slay, the wolf was not carrying off lambs," where "no eye-diseases said there:  'I am the eye disease,'" and where no one says "I am an old woman" or "I am an old man," and fresh water is made to rise from the ground at Enki's word.¹⁸  It is the land to which Ziusudra is taken after the flood.  The likeness to Eden is a likeness of type, a place without predation, sickness or age, watered from below (Gen 2:6).  Dilmun is Bahrain, on the sea;  Eden is at the head of four rivers, two of them the Tigris and Euphrates.  Nothing in the poem is a garden, a tree, a command or a fall.

Adapa.  The sage of Eridu breaks the wing of the south wind and is summoned to heaven.  Ea, his god, warns him:  "When they offer thee bread of death, thou shalt not eat it.  When they offer thee water of death, thou shalt not drink it."  Anu offers him the bread and water of life;  Adapa refuses;  Anu laughs and sends him back to earth, mortal.¹⁷  The parallel with Genesis 3 is the lost immortality and a food that could have given it.  Everything else is inverted:  Adapa obeys his god and loses life by obedience, Adam disobeys his and loses it by disobedience.  Speiser, who translated the text, listed it with Gilgamesh as one of "two suggestive analogues" for the tree of life and left it there.⁹

Creation by word.  The Memphite Theology on the Shabaka Stone has Ptah conceive the gods in his "heart" and bring them into being by his "tongue," which John Wilson in ANET called "an approach to the Logos Doctrine" and dated to an original "more than two thousand years older" than the stone.¹⁷  Miriam Lichtheim printed it among the Old Kingdom texts of her first volume in 1973, as "a work of the Old Kingdom, but its precise date is not known," and in her third volume in 1980 withdrew it:  Friedrich Junge's 1973 article "makes it appear certain that it is a work of the Twenty-fifth Dynasty," about 710 BC, whose "strongly archaizing language" and King Shabaka's own claim to have copied a worm-eaten original "misled generations of Egyptologists."  Readers, she wrote, "should in their minds transfer this text" into the pseudepigrapha.¹⁹  The nearest Egyptian parallel to "And God said" is therefore a text of about the same age as the Hebrew monarchy, which claims to be old.  Whatever a reader makes of Genesis 1, the Egyptian side of this comparison is not an ancient original.

The direction of dependence

What Delitzsch claimed, and what survived

Friedrich Delitzsch, professor of Assyriology at Berlin, gave the first of his Babel und Bibel lectures before the Deutsche Orient-Gesellschaft, with Kaiser Wilhelm II in the audience, in 1902.  The English edition of 1903 preserves what he said about the flood.  The tablet from "the library of Sardanapalus at Nineveh" was, he told his audience, "committed to writing 2000 years before Christ."  The Babylonian Noah, Xisuthros, builds a ship "of certain dimensions," coats it "thoroughly with pitch," sends out a dove, a swallow and a raven, and offers a sacrifice whose "sweet odor was scented by the gods, etc., etc."  Then:  "This entire story, precisely as it is here written, afterwards travelled to Canaan," where "we accordingly find in the Bible two distinct versions of the Flood, which are not only absolutely impossible from the point of view of natural science, but are also at diametrical variance with each other," and "the ten Babylonian kings who reigned before the Flood have also been accepted in the Bible as the ten antediluvian patriarchs, and the agreement is perfect in all details."¹

Every checkable claim in that paragraph has since failed.  The tablet Delitzsch held up was a seventh-century copy, and the version of the story it carries was, on Lambert's dating, put in shape between 1200 and 1000 BC, not 2000.⁶  The two-versions argument is examined in The Flood Narrative, where Wenham's finding that the combined Genesis text is closer to Gilgamesh than either supposed source turns it around.  The ten kings are eight in WB 444.¹⁰  "Precisely as it is here written" was never true of a story whose boat, duration, birds, passengers, motive and theology all differ.  Hasel's 1974 summary of where the field had gone is fair:  the pan-Babylonian school "attempted to show that there was nothing in the Old Testament that was not but a pale reflection of Babylonian ideas," and "today the situation has changed radically."¹²

The change was not a retreat from dependence.  It was a retreat from Delitzsch's kind of dependence.  Speiser, writing the Anchor Bible Genesis in 1964, still held that on creation "P's opening account goes back to Babylonian prototypes," that "the Babylonians did not take over these views from the Hebrews, since the cuneiform accounts ... antedate in substance the biblical statements," and, most sweepingly, that "nor is there the slightest basis in fact for assuming some unidentified ultimate source from which both the Mesopotamians and the Hebrews could have derived their views about creation."⁹  He also wrote that the two accounts are "both genetically related and yet poles apart."⁹  On the flood he held that "Hebrew tradition must have received its material from some intermediate, and evidently northwesterly, source."⁹  That is the mainstream position today, restated by George in 2003 and Finkel in 2014.  Finkel's version:  "Comparison of the Hebrew text with Gilgamesh XI highlights such a close and multi-point relationship between the accounts that the dependence of one upon the other is unavoidable," and his mechanism is Judaean exiles learning cuneiform in Babylon.¹⁵

What tablet dates prove, and what they do not

The dates in the first table settle one thing:  a Mesopotamian flood story with a warned hero, a bitumen-sealed boat, a seven-day storm and a sacrifice the gods hurry to was in writing by about 1635 BC.  No one dates the writing of Genesis earlier.  If the direction of dependence were decided by the age of the oldest surviving manuscript, the question would be closed.

It is not decided that way, and both sides have said so.  Heidel's "priority of publication does not imply priority of existence" is the conservative form.²  Lambert's is sharper and comes from the other side:  the parallel he called decisive survives only in a text younger than the Hebrew one, "but nevertheless I hold that there is certain dependence of the Hebrew writers on a Mesopotamian tradition."⁶  Lambert was arguing from a tradition he could not see to a text he could.  So is anyone who argues the reverse.

What the dates do exclude is narrower and worth stating.  They exclude any claim that the Hebrew text, as a text, was the source of the Old Babylonian Atrahasis.  A Hebrew literary tradition is not attested before the first millennium, and Lambert's observation stands that "Genesis shows no knowledge of Mesopotamian matters prior to 1500 B.C.":  Babylon, "utterly unimportant" before Hammurabi, is the city of the Tower;  Nimrod's kingdom is a late map;  "the very existence of the Sumerians is nowhere hinted at."⁶  A Genesis that knew Mesopotamia in its third-millennium form would look different.  Kitchen turns this into an argument for a patriarchal-age origin:  the fourfold frame of creation, crisis, flood and recovery was composed in Mesopotamia only in "about 1900-1600," "no further literary works of this kind seem to have been composed in Mesopotamia after circa 1600," and so "it is logical to suggest that the framework and basic content of Gen. 1-11 goes back to the patriarchal period, and came as a tradition with the patriarchs westward from Mesopotamia."¹¹  The premise is about Mesopotamian scribes.  The conclusion is about a Hebrew author.  Nothing in the premise prevents a Hebrew writer in 1200 or 600 BC from using a frame that Mesopotamian scribes were still copying, as Kitchen himself notes they were "down to the seventh century."¹¹

The "simple to elaborate" argument, audited

Kitchen's second argument is from length.  "In terms of length and elaboration, Gen. 6:9-8:22 might be equal in amount to about 120 lines in Sumerian or Akkadian.  Contrast the lengths of at least 370 lines in Atrahasis II-III, some 200 lines in Gilgamesh tablet XI, and the roughly 150/200 lines in the Sumerian account.  Genesis thus offers a more concise, simpler account, and not an elaboration of a Mesopotamian composition."¹¹  Millard made the same point in 1967, and Noah's Ark:  The Logistics has already flagged its weakness:  an epitome is shorter than its source.  The argument would carry weight if there were a known law that stories only grow.  There is not.  The Standard Babylonian editor shortened Atrahasis when he fitted it into Tablet XI, and the Sumerian story is shorter than Atrahasis while being, on every editor's dating, no older.  Length is not a clock.

The same applies to the better-boat argument.  Kitchen's third difference, "Noah's has shiplike proportions, but the Mesopotamian one was a cube!", is true and is not evidence of priority, because a later writer can improve a boat as easily as an earlier one can spoil it.¹¹  And the two arguments pull against each other.  If Genesis is primitive because it is short, it is not primitive because it is better engineered;  if the Babylonian cube is a corruption of a sensible original, then the tradition degraded as it grew, and its greater length is a mark of lateness, which is what the conservative side wants.  The arguments cannot both be used, and neither is decisive alone.

The "shared memory" reading, and what would distinguish it from borrowing

The reading most conservative commentators now hold is neither Delitzsch's nor Kitchen's.  It is Millard's:  "The two accounts undoubtedly describe the same Flood, the two schemes relate the same sequence of events," and "granted that the Flood took place, knowledge of it must have survived to form the available accounts;  while the Babylonians could only conceive of the event in their own polytheistic language, the Hebrews, or their ancestors, understood the action of God in it."  Abraham came from Mesopotamia, so "it could be said that the stories were borrowed from there, but not that they were borrowed from any text now known to us."¹⁴  Wenham's form is that the Genesis writer need never have read Gilgamesh, because "these traditions were part of the intellectual furniture of that time in the Near East, just as most people today have some idea of Darwin's Origin of Species though they have never read it."⁵

This is a coherent position, and Lambert could not refute it.  He set out the two ways of avoiding a Hebrew borrowing from Babylon, an Amorite tradition ancestral to both or a real event remembered twice, and wrote of the first, "I can think of no refutation of such a view."⁶  It is also a position that predicts something, and the prediction can be tested against the tables above.

Two independent memories of one event should agree on the event and disagree on the telling.  They should both have a flood, a survivor, a boat, and a landing.  They should not both have the survivor warned through a wall, the boat sealed with the same word for bitumen, the storm counted in sevens, the hero opening a window after grounding, birds sent at seven-day intervals with the last one not returning, and the deity smelling the sacrifice.  Those are not facts about a flood.  They are choices a storyteller makes.  Where Genesis and Gilgamesh agree, they agree on storytelling choices, and that is what a shared literary tradition predicts and a shared memory does not.

The point can be pressed from the Mesopotamian side, and it is the finding in this article that a conservative reader will like least.  Y. S. Chen's 2013 monograph traces the flood motif through every Sumerian and Akkadian source.  The Sumerian King List's Ur III copy has no antediluvian section;  it was "only secondarily added."  The Early Dynastic III versions of the Instructions of Shuruppak, from about 2500 BC, name the city of Shuruppak but have no Ziusudra;  "the name Ziusudra ... is nowhere to be found in the Early Dynastic III versions."  Every chronographic reference to the flood is "dated to the Old Babylonian or post-Old Babylonian periods."  His conclusion is that "Mesopotamian traditions related to the Flood only emerged from the Old Babylonian period onwards," about 2000 to 1600 BC, and that they are "largely intellectual and cultural constructs" of that period's politics.¹³  Chen may be wrong that the motif was invented rather than first written down;  absence from a few school tablets is thin evidence of absence from a culture.  But the shared-memory reading has often been stated as if the Mesopotamian flood tradition were demonstrably third-millennium, an independent line reaching back toward the event.  It is not demonstrably that.  It is a second-millennium literary tradition, like the Hebrew one only older.

Where does that leave the shared-memory reading?  Alive, and unproven, in the same way the borrowing reading is.  What it explains that borrowing does not is the disagreement in the concrete details, the boat, the duration, the birds, the hero's fate, on which a copyist would have no reason to diverge.  What borrowing explains that it does not is the agreement in narrative devices.  A third reading, that both descend from a common literary source older than Atrahasis which was itself a telling of a remembered event, explains both and is the one Heidel, Lambert (as a possibility), Millard and Wenham all leave open.  It is also the one that no surviving tablet tests.

What would show it

The direction of dependence is undemonstrated because the evidence that would demonstrate it has not been found.  It is worth saying exactly what that evidence would be.

A Mesopotamian text older than any plausible date for Genesis that carries a feature found only in Genesis would show borrowing into Genesis.  The candidates are a raven-then-dove sequence, seven pairs of clean animals, a flood lasting a year, a vessel of 300 by 50 by 30, a rainbow, or a covenant.  None exists.  Every Mesopotamian feature that Genesis shares is also shared among the Mesopotamian texts themselves, so it shows only that Genesis stands downstream of the tradition, not of any tablet.

A West Semitic or Hebrew flood text older than the Old Babylonian period would open the reverse direction.  None exists, and nothing in Hebrew is attested in writing before the first millennium.

A shared idiom would show a textual rather than an oral channel.  There is one, kupru and kofer, and one word is what an oral channel would also leave, since the substance and its name travelled together with the trade.  A second and a third, in phrases rather than nouns, would change the picture.  A hundred and fifty years of looking have not produced them.

A Mesopotamian flood text with birds older than the Standard Babylonian Gilgamesh would close the gap that Lambert identified.  The broken columns of Atrahasis III are the place to look, and they have been looked at.

Until one of these turns up, the honest summary is the one Heidel gave in 1951 for the creation texts and which now fits the flood texts too:  the question is open, and it is open in both directions.

Common objections

Objection 1: "Genesis copied Gilgamesh. The tablets are a thousand years older, and the story is the same."

The tablets are older, and the story is the same in outline.  But the tablets that are a thousand years older do not contain the closest parallels.  The Old Babylonian Gilgamesh has no flood.  Atrahasis and the Sumerian story have no birds and no window.  The text that carries the birds is Tablet XI, of which "no copies earlier than 750 B.C. are known."⁶  "Copied" also needs a mechanism, and every mechanism proposed, Amarna-age oral transmission (Lambert), a northwesterly intermediary (Speiser), exiles in a Babylonian school (Finkel), is a reconstruction its author could not check.  The safe statement is that Genesis stands in the same tradition, not that it copied any tablet.

Objection 2: "Then Gilgamesh copied Genesis. Noah's story is the sensible one, with a real ship and a real reason."

This is Delitzsch in reverse, and it fails for the same reason.  "Sensible" is a judgment made from inside one text.  George reads the Babylonian bird order as an aetiology that Genesis "obscures";  Heidel read the Genesis order as the one that fits bird behaviour.³  A better boat can be a later improvement.  And there is no Hebrew text, or trace of one, old enough to have been copied by an Old Babylonian scribe.  The reverse claim needs a lost Hebrew or Amorite original, which is the common-source reading under another name.

Objection 3: "The tehom of Genesis 1:2 is Tiamat. Genesis 1 is a demythologised Babylonian creation myth."

The etymology was disproved in 1951 and has not been revived by anyone who edits Akkadian.  Heidel:  "to derive tehom from Tiamat is grammatically impossible, because the former has a masculine, the latter a feminine, ending," and both words "go back to a common Semitic form."²  Ugaritic thm has the same meaning with no Babylonian goddess behind it.  Kitchen:  tehom "means nothing more than 'deep, abyss.'  It is not a deity, like Ti'amat."¹¹  Lambert found "no evidence of Hebrew borrowing from Babylon" even on the one real parallel.⁶  The "demythologised myth" reading survives in introductory textbooks, not in the critical editions.

Objection 4: "The seven days come from the seven tablets of Enuma Elish."

Heidel answered this in one paragraph:  "In the Genesis account acts of creation were performed on all of the first six days, and on the seventh day God rested;  while in Enûma elish Tablets II, III, and most of I and IV do not deal with any part of creation, and the story of Marduk's rest begins as early as the first half of Tablet VI and then extends over virtually all of Tablet VII."  The idea is "lacking all evidence."²  Lambert thought the week might have an Amorite root shared with the Babylonian šapattu, which is a different claim and a weaker one.⁶

Objection 5: "Every culture has a flood story, so the Mesopotamian parallel proves nothing special."

Every culture does not, and the ones that do mostly do not share the Mesopotamian details;  the census is in Flood Legends Around the World.  The Mesopotamian parallel is special because it shares narrative devices, not because it shares a flood.  A Hawaiian or Chinese story with a boat is a flood story.  A story with a warned hero, a sealed boat, a mountain, a window, birds and a sacrifice is this story.

Objection 6: "If Genesis knew the Babylonian myths well enough to argue against them, then it is dependent on them, and the polemic reading concedes the critics' case."

It concedes contact, which no one who has read the texts denies.  It does not concede that the Genesis account is derived from the myths, which is what Delitzsch and Gunkel claimed.  Wenham's formulation covers both:  Genesis "is a commentary, often highly critical, on ideas current in the ancient world," and also "Genesis and the ancient Near East probably have more in common with each other than either has with modern secular thought."⁵  A commentary that disagrees with its subject is not a copy of it.  The polemic reading is a claim about intent, and it is compatible with borrowing, shared memory, and a common source alike.

Objection 7: "Chen has shown the Mesopotamian flood tradition was invented in the Old Babylonian period. So the flood is a second-millennium literary fiction, and Genesis inherited it."

Chen has shown that the tradition is not attested in writing before the Old Babylonian period, which is not the same.  The Early Dynastic evidence is a handful of school copies of one didactic text, and absence from them is weak evidence of absence from Sumerian culture.  What Chen's work does defeat is the claim that the Mesopotamian texts carry a demonstrable third-millennium memory.  Whether a flood happened is a question for Fossil Order and the Flood, and Mesopotamian archaeology's own flood layers, at Ur, Kish and Shuruppak, are river floods spread over a thousand years, as The Flood Narrative and Flood Legends both record.²²

Objection 8: "The Megiddo tablet proves the Gilgamesh flood was known in Canaan before Israel existed."

It proves Gilgamesh was known in Canaan.  The Megiddo fragment, "found by a shepherd in 1954 near the dump of the Chicago expedition," carries "fragments of text similar to parts of SB Tablet VII," the death of Enkidu.³  No second-millennium Gilgamesh anywhere, at Megiddo, Ugarit, Emar or Hattusa, includes the flood.³  The Ugarit and Nippur Atrahasis fragments of the Middle Babylonian period do include the flood, and they are the right evidence for a westward spread of the flood story in the Amarna age;  the Ugarit piece "dealt only with the flood, omitting all the earlier episodes."⁷  That is Lambert's Amarna-age channel, and it is a channel for Atrahasis, which has no surviving birds.

Sources