light 225° · worn field: uncertain · lost surface: unreadable at any angle · missing clay: nothing to light
S25 · World Memory pilot
K.3375 · known as the Flood Tablet
A version of Tablet XI of the Standard Babylonian Epic of Gilgameš
Neo-Assyrian, 7th century BC · from Nineveh · Kuyunjik collection, British Museum
Before it is a story about a flood, it is a piece of hard clay roughly the size of a hand, written on both faces in six columns, with one corner gone. It is a copy — made by a scribe who was himself working from older tablets, some of them already damaged.
Everything anyone can say about Tablet XI passes through objects like this: buried, dug up, shipped, numbered, shelved, read under lamplight, compared with other broken copies, restored where a duplicate allows it, and left open where nothing does. The gaps you saw above are not a failure of the page. They are the condition of the record.
Responsible reconstruction stops where the record stops.
How to read this Station
From the British Museum object record or from published Assyriological scholarship, cited at the foot of this page.
A restoration made by scholars from duplicate manuscripts or from phrasing repeated elsewhere in the poem. Reasoned, not witnessed.
Nothing survives and no duplicate supplies it. Left open here, as editions leave it open.
Made for this Station: the clay surface above, its wedges, its damage, and the sample lines below. It is not a facsimile of K.3375, and the marks are not readable Akkadian. Any real reading belongs to the editions.
Cuneiform is not drawn. The scribe presses the corner of a cut reed into damp clay and lifts it away; the clay keeps a wedge-shaped hollow with one bright wall and one dark one. A sign is a small arrangement of these hollows. A column of literary text is a few hundred of them, placed at speed, in order, by someone trained for years to do it.
This is why light matters so much later on. A worn wedge is not ink that has faded — it is a shallow dent, and whether it exists depends on where the lamp is. Epigraphers still turn the tablet in raking light to read it, and now also photograph it from many angles at once so the surface can be re-lit afterwards.
Scribes also recorded what they could not read. In Neo-Assyrian copies, a copyist who met a break in the tablet in front of him could write ḫepi — “broken” — in the line, sometimes ḫepi eššu, “newly broken.” The oldest gesture in this whole chain of transmission is not invention. It is the honest marking of an absence.
No single tablet carries a whole poem intact. Editors work by collation: they lay one manuscript beside another, line by line, and see whether a second copy happens to preserve what the first has lost. Where two witnesses overlap, a broken line can be read again. Where they break in the same place, or where no second copy exists at all, the line simply stops being available.
Below is a diagram of sixteen consecutive lines — not the real collation of Tablet XI, which involves far more manuscripts than these. Add a witness and watch which lines close. Then watch which ones do not.
Witness β is a second Nineveh manuscript; witness γ is a later Babylonian copy. Both are modeled here for the demonstration.
An Assyriological edition tells you, mark by mark, how much of what you are reading was actually on the clay. Below, the flood tablet’s bird episode is set out with those conventions — in English, so the marks themselves stay legible; in a real edition they sit over transliterated Akkadian. The damage pattern shown here is modeled, not the damage of K.3375. Tap any mark to see what authorises it.
When the seventh day arrived, brought out a dove and let it go.
The dove went off; it found no place to alight, it turned back to me.
I brought out a swallow and let it go. The swallow went off; , and it turned back to me.
I brought out a raven and let it go. The raven went off, saw the waters receding, , and it did not turn back.
the mountain
Select any mark above to see what authorises it.
Read as a smooth translation, the same passage runs: “When the seventh day arrived I brought out a dove and let it go. It found no perch and came back to me. Then a swallow, then a raven — and the raven did not come back.” Every mark disappears. That is what readability costs, and why the edition, not the paraphrase, is the thing a scholar cites.
Excavation at the mound of Kuyunjik, Nineveh, sends tens of thousands of tablets and fragments to the British Museum, where they are catalogued as the Kuyunjik collection. K.3375 is one of them. British Museum collection record, K.3375; excavation associated with Hormuzd Rassam.
The material arrives as fragments in crates. Sorting, numbering and keeping come first: without the museum’s custody and its inventory numbers, there is no way to find the piece again, and no way for anyone to say that two fragments belong together. Institutional practice of the Kuyunjik collection.
George Smith, working through Kuyunjik fragments in the Museum, reads a paper on a Babylonian account of the deluge to the Society of Biblical Archaeology; it is published the following year. He is one documented reader in a long chain, not its author. G. Smith, “The Chaldean Account of the Deluge”, Transactions of the Society of Biblical Archaeology 2 (1873).
The slow work: fragments joined physically, hand-copied in ink as autograph drawings, collated against one another, re-read as more Kuyunjik pieces are identified. The epic is assembled as a text from many manuscripts — none of which is the epic. Standard Assyriological method; see George 2003, vol. 1, on the manuscripts of Tablet XI.
A. R. George publishes the critical edition of the Babylonian Gilgameš epic: every manuscript given a siglum, every restoration bracketed, every gap left visibly a gap. This is the Tablet XI that scholars now argue with. A. R. George, The Babylonian Gilgamesh Epic: Introduction, Critical Edition and Cuneiform Texts, 2 vols (Oxford, 2003).
New fragments continue to be identified and joined; a substantial new source for Tablet V, published in 2014, changed a passage everyone thought was settled. Digital corpora now let scholars search unplaced fragments against known lines. The text is not finished, and no responsible edition pretends it is. F. N. H. Al-Rawi & A. R. George, Journal of Cuneiform Studies 66 (2014); Electronic Babylonian Library project, LMU Munich.
K.3375: one broken Neo-Assyrian tablet from Nineveh, six columns, roughly hand-sized. It carries a version of Tablet XI. It is a copy, not an original, and it is not the Epic of Gilgameš.
A flood narrative was told in Mesopotamia long before this tablet was pressed and beyond it: in the Old Babylonian epic of Atra-ḫasīs, in a Sumerian flood story, in other Babylonian versions. Tablet XI’s account is closely related to Atra-ḫasīs. The tablet joins a stream; it does not begin one.
The Tablet XI you can read today in translation is a composite, built by editors from many manuscripts. Its brackets are load-bearing. Remove them and you get a fluent poem that quietly claims more than the clay can support.
Sources