Genesis is the hardest book on this site to check, and it is worth saying so at the outset. It describes a period for which there is almost nothing to check it against, and its closest documented relatives are not Israelite.
What the book is
Fifty chapters in two very unequal halves. Chapters 1 to 11 are what scholars call the primeval history: creation, the garden, the first murder, a genealogy of long-lived ancestors, the flood, and the dispersal of peoples at Babel. Chapters 12 to 50 narrow abruptly to a single family across four generations, ending with it settled in Egypt. The change of scale at chapter 12 is the most important structural fact in the book: before it the subject is humanity and the timescales are enormous; after it the narration slows to quarrels over wells and negotiations about burial plots.
How it is put together
A repeated formula — “these are the generations of” — introduces each new section and passes the narrative from one figure to the next. It is a deliberate editorial spine, and it does not always fall where a modern reader would put a chapter break. The opening is the clearest place to see the book’s composite character. Genesis 1:1 to 2:3 gives an ordered week: light, sky, sea and land, luminaries, creatures, and finally humanity, male and female together, as the last act before rest. From 2:4 a second account begins on different terms. The man is formed first, from the ground, before the garden is planted; the animals are made afterwards as candidate companions and found wanting; the woman comes last. The two differ in order, in vocabulary, in the name used for God, and in tone.
The flood narrative shows the same doubling from the inside. Within one continuous story the animals are taken aboard in pairs and also in sevens of the clean kinds, the waters last for figures that do not reconcile neatly, and the divine name alternates. The traditional reading takes these as complementary detail within a single account; the critical reading takes them as two versions interwoven. The evidence is the same either way, and anyone can see it by reading chapters 6 to 9 slowly with a pencil.
The documentary hypothesis, in outline
The standard critical account of the Pentateuch, developed in nineteenth-century European scholarship and refined since, proposes that Genesis and the four books after it were assembled from distinct sources, conventionally labelled J, E, D and P, each with its own vocabulary, theology and preferred divine name. On this model the priestly material supplies the creation week, the genealogies and the covenant formulae; older narrative strands supply the garden, the wells and the family quarrels.
Two things should be said fairly. This is a position, not a finding: no manuscript of J or P has ever been recovered, and the sources are reconstructions from the text itself. And it is not held today in the confident four-document form it once had — much recent work argues for fewer sources, more supplementation, or a longer and messier editorial process. What survives across almost all critical approaches is the modest claim that Genesis is composite. The traditional position, Mosaic authorship of the whole, is a real alternative and is defended in print, but it is a minority view in university scholarship.
The flood and its Mesopotamian relatives
Genesis 6 to 9 is not the only ancient flood story from the region, and not the earliest we have in writing. Two Akkadian works are usually set alongside it: Atrahasis, which tells of the creation of humanity and its near-destruction by flood, and the flood episode inside the epic of Gilgamesh, where the survivor recounts the story to the hero. A Sumerian version of the story also survives.
The overlaps are not thematic generalities. A deity resolves to destroy humanity; one man is warned in advance; he builds a sealed vessel to specified dimensions and takes aboard family and living creatures; the vessel grounds on a mountain; birds are released to test whether the waters have gone down; a sacrifice is offered on landing and the gods respond to its smell; the survivor is singled out for blessing afterwards.
The differences matter as much. In the Mesopotamian versions the flood is sent because humanity has become too numerous and too noisy for the gods to sleep, and the gods themselves are frightened by what they have unleashed. Genesis makes the cause moral — the earth is corrupt and full of violence — and its God is neither surprised nor overmatched. That the texts are related is agreed on all sides; the direction of dependence is not.
Where the external record touches it
Here the honest answer is uncomfortable. No Abraham, Isaac, Jacob or Joseph appears in any external document. No inscription refers to the events of chapters 12 to 50. The coalition of kings in chapter 14, the only episode with international politics in it, has never been matched to a known campaign. The patriarchal narratives are, in the strict sense this site uses, unattested — not contradicted, but unsupported from outside.
What the record supplies is friction over details. The narratives give Abraham camels as ordinary pack animals, but excavated animal remains from the southern Levant point to camels coming into regular domestic use considerably later than the second-millennium setting the stories are usually given. Abimelech is called a king of the Philistines, though Philistines appear in the archaeological and Egyptian record only from around 1200 BCE. Abraham’s home is “Ur of the Chaldeans”, but Chaldeans are a first-millennium presence in southern Mesopotamia. And chapter 36 dates a list of Edomite kings to a time “before any king reigned over the Israelites”.
None of these is decisive alone, and each has a defence. Camel remains are patchy and argue mainly about frequency; “Philistine” and “Chaldean” may be later editorial updating of older place-names, a practice visible elsewhere in the Bible. Taken together they are the main reason most critical scholars read the patriarchal stories as later compositions set in a remembered past. An older argument in the opposite direction — that patriarchal customs matched second-millennium legal texts from Mesopotamian archives — was widely made a few generations ago and is now much less confidently held, because the same customs turn out to be attested across a far longer span.
Where to start
Read Genesis 1:1 to 2:3, then 2:4 to 2:25, without a break, writing down the sequence of creation in each. Then read chapters 6 to 9 and mark every place a detail is stated twice. Under an hour, and the whole compositional argument in miniature.
For the reasoning behind the dating claims above, see how we date a text. For why “no external attestation” is a description rather than a verdict, see what external attestation means. Where we have got something wrong, it is listed at corrections.