Eridu — Lost Knowledge and Anomalies

About Eridu — Lost Knowledge and Anomalies

"When kingship descended from heaven, kingship was at Eridu." The line opens the Sumerian King List preserved on the Weld-Blundell Prism, a four-sided cuneiform pillar of Old Babylonian date now catalogued as Ashmolean AN1923.444 and recovered not at Eridu itself but at Larsa in 1922. The Sumerians who copied that text three thousand years ago already treated Eridu as the place where order arrived from above, and the soil under Tell Abu Shahrain has, ever since the Iraqi excavations of 1946 to 1949, returned a sequence of eighteen superimposed temples on a single platform — Levels XIX through VI of the Ubaid period followed by Levels V through I of the Uruk period, the longest continuously rebuilt temple footprint documented in southern Mesopotamia. The textual claim sits on top of a physical fact. The lost-knowledge layer of Eridu lives in the gap between what those eighteen rebuilds and the texts about them once carried, and what we can recover today.

The temple platform at Eridu, excavated by Fuad Safar and Seton Lloyd between 1946 and 1949, holds a stratigraphic record that no other Mesopotamian sanctuary matches in continuity. Eighteen successive temples were built one on top of the next on the same patch of ground, each new structure carefully aligned with its predecessor, each enlarging the previous footprint without abandoning it. The earliest, Temple XIX, dates to the Ubaid 1 phase around 5400 BCE — a small mud-brick room of perhaps three by four meters with a niche on its short wall. The latest in the Ubaid sequence, Temple VI, was already a multi-room structure on a raised platform with offerings of carbonized fish piled at the altar. The Uruk-period rebuilds, Levels V through I, raised the platform higher still and gave the complex the formal terrace that the later third-millennium ziggurat would inherit. By Level VII the temple had a tripartite plan, a long central hall flanked by smaller rooms on either side, an arrangement that recurs across Ubaid temple architecture from southern Sumer to as far north as Tepe Gawra in northern Iraq. Each rebuild used mud-brick of distinctive proportions, often with the older walls preserved as the core of a new platform rather than demolished, so that the entire eighteen-temple column reads as a single continuous architectural memory rather than as a series of unrelated structures.

It is important to be precise about what this proves and what it does not. Recent reporting in popular media has called Eridu "the oldest temple in history" or "the earliest place of worship on Earth." Both phrasings are wrong. Göbekli Tepe in southeastern Turkey, with monumental T-pillars and circular enclosures dated to the Pre-Pottery Neolithic A and B between roughly 9600 and 8000 BCE, predates Eridu Temple XIX by about four thousand years. There are still older ritual spaces at sites like Çayönü and Jericho. What the Eridu sequence does prove — and what no other Mesopotamian site has yet matched — is the longest documented continuity of cult on a single footprint in southern Mesopotamia. Eighteen rebuilds across roughly sixteen hundred years say something about the persistence of a sanctuary memory, the willingness of generation after generation to honor the lines drawn by the previous one. That is the genuine claim, and it is a serious one without the inflation. Above the eighteen-temple sequence, in the third millennium BCE, the Ur III king Amar-Sin began a substantial ziggurat on the same platform; the structure was never finished, and its unfinished state is documented in the Ur III royal inscriptions and in the brick courses Lloyd and Safar exposed in the upper levels. The ziggurat belongs more properly to the parent Eridu page; what matters here is that the third-millennium builders understood themselves to be continuing the same sanctuary line that began at Temple XIX.

The Sumerian King List exists in around eighteen surviving manuscripts, but the Weld-Blundell Prism is the most complete and the one most often quoted. It is a four-sided clay prism, about twenty centimeters tall, inscribed in cuneiform on every face. Stephen Langdon published it in 1923 as the cornerstone of the king-list reconstruction, and it now sits in the Ashmolean Museum at Oxford under accession number AN1923.444. Two facts are non-negotiable about the object. First, it was found at Larsa, not Eridu — recovered in the 1922 season at Tell as-Senkereh, the ancient mound of Larsa, and only later identified by Langdon as the most extensive surviving copy of the king list. Second, although the text begins with Eridu and lists eight antediluvian kings spread across five cities, the first two ruling at Eridu, the prism itself is a textual source about Eridu, not an artifact from Eridu's own soil.

The opening passage is short. After the line about kingship descending from heaven, the prism gives Alulim as the first king of Eridu, ruling 28,800 years, followed by Alalngar for 36,000 years. The reigns are clearly mythic in length. After two kings the kingship moves to Bad-tibira, then to Larak, Sippar, and Shuruppak, before the flood washes the antediluvian record clean. What the king list preserves, then, is a theological claim — that civilization itself originated at Eridu, that the institution of legitimate rule was first granted there, and that everything afterward was a transmission from that origin. Whether the claim describes anything historical is a separate question. What it documents beyond doubt is how the Sumerians of the early second millennium BCE chose to remember their own beginning.

The temple at Eridu was named E-abzu, which in Sumerian means roughly "house of the deep waters" or "house of the abyss." The Akkadian cognate is Apsû. In Mesopotamian cosmology the abzu was the great underground freshwater ocean that the gods drew on for life, fertility, and ritual purity, and Enki — the Sumerian god of wisdom, water, craft, and incantation — was its lord. Eridu sat at the southern edge of the alluvial plain, near the lagoons and brackish marshes that fed the lower Euphrates, and the choice of location seems to have been deliberately liturgical. The temple precinct contained ritual purification pools, and excavation has confirmed thick layers of fish bones at successive altars across the sequence, consistent with ongoing offerings to the freshwater god.

What is not fully recoverable is the liturgy itself. A fragmentary corpus of Sumerian incantations and Eridu-cult hymns has come down through copies preserved in later libraries — the Eridu lament composed for the city's mythical destruction, several Enki and Ninhursag narratives, and a handful of incantation series invoking the abzu. Substantial portions of the cult-specific liturgy, the ritual instructions that would have governed daily activity inside the E-abzu, are either lost or undeciphered. The Sumerian language itself is reconstructed from external evidence; there are no living speakers and no continuous tradition. Particular vocabulary used inside the priestly orders, especially the more obscure technical terms for purification rites and offering sequences, remains the subject of ongoing philological work. The ritual practice that physically wore the eighteen-temple platform smooth has not, despite a century of scholarship, been fully recovered.

The fish offerings deserve their own note. Among the most striking finds in the 1946–1949 excavations were thick layers of carbonized fish bones at the foot of successive altars across the Ubaid sequence, the bones still in arranged groupings rather than scattered. Lloyd and Safar interpreted them as ritual deposits to Enki rather than ordinary kitchen refuse, both because of their location at the altar base and because of the consistency of species — predominantly freshwater carp and catfish from the surrounding marshes. The fish-and-water complex of Eridu cult, in other words, is materially attested across more than a thousand years of stratigraphy, even though the spoken or sung liturgy that accompanied the offerings has not survived. The physical evidence and the textual evidence corroborate each other on the broad shape of the practice while leaving the specific words missing.

Adapa is the most famous of the apkallu, the seven antediluvian sages who in Mesopotamian tradition brought the arts of civilization to humanity. He is associated by name with Eridu and with Enki's priesthood, and his myth tells of a refusal: offered the bread and water of eternal life by the high god Anu, Adapa declined to eat, on advice from his patron Enki, and lost immortality for himself and his descendants. The story turns on a misread instruction, a moment of caution that becomes a cosmic loss. It is one of the central etiological narratives in the cuneiform corpus.

The textual transmission deserves precision. The longest and most coherent surviving fragment of the Adapa myth, known in scholarship as Fragment B, did not come from Eridu's soil. It was found at Tell el-Amarna in Egypt in 1887, recovered from the diplomatic archive of Pharaoh Akhenaten, and is now catalogued as EA 356 (Berlin VAT 348) in the Vorderasiatisches Museum, Staatliche Museen zu Berlin. The tablet is fourteenth-century BCE, inscribed in Akkadian using cuneiform, and was likely used as a scribal training text in Egypt rather than as a piece of cult literature. Five smaller fragments, labeled A, C, D, E, and F in subsequent editions, were recovered from the great library of Ashurbanipal at Nineveh in the seventh century BCE. Eridu is the setting of the myth — Adapa breaks the wing of the south wind while fishing on Eridu's coast, and from there is summoned to heaven — but no Adapa manuscript has yet emerged from Eridu's own excavated layers. The myth survives because it was copied and circulated across the literate world of the late Bronze Age, not because Eridu kept its own records intact.

The Sumerian flood story closest in setting to Eridu is the composition that scholarship calls the Eridu Genesis. Its principal manuscript is the tablet catalogued as CBS 10673, recovered in the 1893 Pennsylvania excavations at Nippur and now in the collection of the Penn Museum. Roughly two-thirds of the original text is missing — only the lower portion of one column survives intact, with fragments of the others — and what remains was first read by Poebel in 1912 and formally published in Publications of the Babylonian Section vol. 4 (Historical Texts), 1914. The tablet is dated on paleographic grounds to the late seventeenth or early sixteenth century BCE, the Old Babylonian period.

What the surviving text gives is recognizable. After the gods create humans and animals and bestow kingship on the first cities, including Eridu, a great flood is sent. A pious king named Ziusudra is warned by Enki, builds a vessel, and survives the deluge with his family. After the waters recede the gods grant Ziusudra eternal life and resettle him in Dilmun, "the place where the sun rises." The narrative arc is the same one preserved much more completely in the later Akkadian flood account in Tablet XI of the Epic of Gilgamesh, where the survivor's name becomes Utnapishtim. As with Adapa, Eridu is the setting and the place where the warning originates, but the surviving manuscript came from another city. The text is fragmentary not because Eridu's archives were destroyed but because the single Nippur tablet that preserves it has been broken for at least three thousand years.

The recovery of Eridu by modern archaeology runs through five distinct campaigns across roughly one hundred and seventy years. The first was John George Taylor's 1854 to 1855 sounding, conducted under the British vice-consulate at Basra. Taylor identified Tell Abu Shahrain with ancient Eridu on the basis of inscribed bricks of Nabonidus, but the trenches he opened produced little, and in his own published report he wrote that his work at the site "has been unproductive." The honesty of that admission has tended to be overlooked. Taylor matters because he was the first to put the site on the map, not because his digging recovered the layers that would later prove decisive.

R. Campbell Thompson sounded Eridu in April–May 1918 for the British Museum, publishing "The British Museum Excavations at Abu Shahrain in Mesopotamia in 1918" in Archaeologia 70 (1920). H. R. Hall returned the following season, April–May 1919, as part of the British Museum's Babylonia mission to Ur, al-'Ubaid, and Abu Shahrain (published 1930). The third and most consequential campaign was the joint Iraqi Directorate-General of Antiquities work directed by Fuad Safar with Seton Lloyd as technical advisor, between 1946 and 1949. This is the campaign that uncovered the full eighteen-temple sequence, the Ubaid cemetery (roughly 193 graves excavated, of an estimated thousand or more across the burial field), and the ancillary palace and domestic quarters. Lloyd and Safar published interim reports in Sumer volume III in 1947, volume IV in 1948, and volume VI in 1950; the full final report, by Safar, Mustafa, and Lloyd, did not appear until 1981, more than thirty years after the trenches were closed.

A long quiet period followed. The Iran–Iraq War, the Gulf War, the sanctions regime, and the 2003 invasion all kept large international missions away from southern Iraq for most of the late twentieth and early twenty-first centuries. One useful consequence of that quiet, confirmed by satellite analysis after 2003, is that Eridu and the surrounding Eridu Basin escaped the catastrophic looting that gutted other southern Iraqi sites. The mound came through the post-2003 years remarkably intact. The current campaign, the Iraqi-Italian Archaeological Mission at the Seven Mounds of Eridu, abbreviated AMEr, has been directed since 2019 by Franco D'Agostino (Sapienza University of Rome) with co-direction from Philippe Quenet (University of Strasbourg) and epigraphic work by Anne-Caroline Rendu Loisel (Strasbourg); the institutional permit was signed in 2014, and sustained fieldwork resumed in 2018–2019. AMEr is the first sustained excavation at Eridu in seven decades, and its publications — initially in Italian Assyriological journals and now in international conference proceedings — are beginning to fill in the gaps that the 1981 final report left.

One detail of the Adapa story is worth holding for what it says about the lost-knowledge frame. The myth turns on misunderstood ritual instruction. Enki tells Adapa not to eat the food he will be offered in heaven, expecting Anu to offer the bread of death; Anu in fact offers the bread of life, and Adapa, obedient to his patron's earlier warning, refuses and forfeits immortality for himself and for humanity. The Eridu tradition that the myth comes from is therefore self-aware about the fragility of transmitted instruction. The story acknowledges that wisdom can be technically correct in form and catastrophically wrong in application, and that the gap between teacher and student is itself a place where knowledge gets lost. The myth is partly about the conditions of its own preservation.

The latest ancient witness to the Eridu tradition is Berossus, a Babylonian priest of Bel-Marduk who wrote the Babyloniaca, a three-volume history of Babylonia in Greek, around 281 BCE. Berossus claimed direct access to the temple archives at Babylon and presented his work as a faithful summary of Babylonian cosmogony, history, and ritual for a Hellenistic readership. In the surviving excerpts he tells of the apkallu Oannes — a fish-bodied sage who emerged from the Persian Gulf at Eridu in primordial times and taught humans writing, mathematics, agriculture, and the arts. After teaching, Oannes returned to the sea each evening and emerged again the following day. Six similar sages appeared in subsequent ages. The Oannes tradition is, on the cuneiform side, the apkallu tradition, and the connection to Eridu is explicit in both transmissions.

The text of the Babyloniaca itself is lost. What survives are fragments quoted by later writers — primarily Alexander Polyhistor in the first century BCE, whose own work also does not survive but was excerpted in turn by Eusebius of Caesarea around 320 CE and by Flavius Josephus in the late first century CE. The chain of transmission is therefore long and the wording of any particular Berossus passage has to be reconstructed from the citations of citations. For most of the modern period scholars treated Berossus skeptically, suspecting Hellenistic invention. The German Uruk excavations of the 1970s, however, recovered cuneiform sources containing apkallu lists and Oannes-cognate material that confirm Berossus was working from genuine ancient documents rather than fabricating a tradition for Greek readers. The skepticism has softened. What Berossus reports about Oannes and Eridu now has independent cuneiform corroboration.

Two claims that often attach to Eridu in popular treatments deserve a direct answer. The first is the identification of Eridu with the biblical Eden — a thesis associated principally with the Egyptologist David Rohl, who in the 1990s argued that the four rivers of Genesis described a real geography centered on southern Mesopotamia. The argument relies on selective river identifications, contested toponymies, and a chronological framework rejected by mainstream Assyriology. There is no evidence on the ground at Eridu, in the cuneiform record, or in the biblical text itself that supports the equation. The second is the recurring claim that Eridu is the "oldest city in the world." That phrasing depends entirely on the definition of "city," and several other candidates — Jericho, Çatalhöyük, Tell Brak — have older substantial settlement layers. Sumerian tradition's first city is the precise, defensible claim. Eridu is the place the Sumerians themselves named as the first.

Significance

Eridu matters in the lost-knowledge category for a reason that is the inverse of the standard fringe frame. The popular telling treats ancient sites as archives of forbidden technologies waiting for the right researcher to crack them open. Eridu's record runs the other way. The site has been read carefully for more than a hundred and seventy years, the eighteen-temple sequence is one of the most thoroughly published stratigraphic columns in Mesopotamian archaeology, and the surrounding cuneiform tradition has been edited and translated across generations of philological labor. What is missing from Eridu's record is missing for ordinary, recoverable reasons, not because something was hidden.

The eighteen-temple sequence is the structural example. Eighteen rebuilds across sixteen hundred years on a single footprint say something serious about cult memory, but that fact has often been inflated into the false claim that Eridu is the earliest temple in human history. Göbekli Tepe is older by roughly four thousand years; so are several smaller Pre-Pottery Neolithic sanctuaries. The honest claim is that Eridu's continuity of cult on a single footprint is the longest documented in southern Mesopotamia, and that claim is impressive on its own terms.

The Apsû liturgy is the second example. The E-abzu was named "house of the deep waters" and sat near the marshes Enki was thought to govern; the precinct contained purification pools and absorbed centuries of fish offerings. Substantial portions of the priestly liturgy are lost or undeciphered. The Sumerian language is reconstructed from external evidence, and the technical vocabulary of the abzu cult is among the harder strata of the lexicon to recover. This is not an esoteric secret; it is a corpus problem.

The Adapa myth, the Eridu Genesis, and the Berossus transmission are the third example. Each preserves a tradition Eridu generated but did so through a manuscript found somewhere else — Adapa's Fragment B from Amarna in 1887, the Eridu Genesis from Nippur in 1893, the Berossus excerpts via Polyhistor and then Eusebius and Josephus more than a thousand years later. The texts are fragmentary because tablets break and intermediaries paraphrase. Where scholars once suspected Berossus of Hellenistic fabrication, the 1970s German Uruk excavations recovered cuneiform sources that confirmed his sources were real.

The lesson Eridu offers a careful seeker is the lesson of any well-published site: the genuine mysteries are smaller, more specific, and more philological than the fringe versions claim, and no less interesting for being smaller. The textual claim that kingship descended from heaven sits next to a physical platform rebuilt eighteen times. The two records corroborate each other on how Sumerian civilization remembered its beginning. The remaining work is to read what has not yet been read.

Connections

Eridu's lost-knowledge threads sit inside a wider Mesopotamian and ancient-world conversation. The cross-links worth following are listed here.

The parent page is the full Eridu entry, which gives the chronology of occupation from Ubaid 1 around 5400 BCE through the Neo-Babylonian revival under Nabonidus, the geography of the alluvial plain at Tell Abu Shahrain, and the relationship of Eridu to the surrounding Sumerian city network. The two sister B-pages on Eridu treat material that this page deliberately avoids. Eridu Astronomical Alignments covers the orientation of the temple platform, what is and is not recoverable about the alignment of the eighteen rebuilds, and the broader question of how Mesopotamian temples encoded astronomical events. Eridu — Comparisons to Other Sites sets Eridu alongside Uruk, Ur, Lagash, and the other great Sumerian cities, treating relative scale, founding date, and ritual function. Reading the three Eridu pages together gives a fuller picture than any one alone.

The Berossus thread on this page connects to the broader category of antediluvian sage traditions in the ancient Near East. The Sumerian flood narrative preserved on CBS 10673 names Dilmun as the place where Ziusudra was resettled after the deluge, "the place where the sun rises." Dilmun, identified with the Bahrain archipelago and the eastern Arabian coast, was Eridu's southern counterpart in the Sumerian cosmography — a paradisiacal land that the gods chose for the immortalized survivor of the flood. The relationship between Eridu as the place where kingship arrived from heaven and Dilmun as the place where one mortal human was given an existence outside time is one of the central polarities of Sumerian sacred geography.

The eighteen-temple-sequence thread connects to the wider Bronze Age conversation about civilizational continuity in the urban revolution. Mohenjo-daro, the great Harappan city in the Indus Valley, was rebuilt across an extended sequence on its own elevated mound in roughly the same period that the Uruk-phase rebuilds at Eridu raised their later platforms. The two sites are independent in their archaeological development, but both demonstrate the pattern of long-duration sanctuary persistence that distinguishes mature urban centers from their earlier village predecessors.

For the broader Egyptian context of long-continuous-use sacred precincts, the entry on Karnak Temple is the natural companion. Karnak's expansion across Middle Kingdom, New Kingdom, Late Period, and Ptolemaic phases produced a layered architectural record that, like Eridu's eighteen-temple sequence, documents the persistence of cult memory across regime change and language shift. The lost-knowledge layer at Karnak — the publication backlog of the Cachette material — sits alongside the lost-knowledge layer at Eridu — the partly undeciphered Apsû liturgy — as parallel cases of recovery rather than concealment.

Further Reading

• Safar, F., Mustafa, M. A. & Lloyd, S. (1981). *Eridu*. Baghdad: State Organization of Antiquities and Heritage (Ministry of Culture and Information). 360 pp., 160 figs., 24 colour pls. The full final report on the 1946–1949 Iraqi excavations, including the eighteen-temple sequence, the Ubaid cemetery, and the surrounding domestic quarters.
• Lloyd, S. & Safar, F. (1947). "Eridu: A preliminary report on the first season's excavations." *Sumer* 3: 100–111. The first interim report from the 1946 season.
• Lloyd, S. & Safar, F. (1948). "Eridu: A preliminary report on the second season's excavation." *Sumer* 4: 115–127. Interim report on the second season.
• Safar, F. (1950). "Eridu: A preliminary communication on the third season's excavations." *Sumer* 6. Third-season interim report covering 1948–1949 work.
• Langdon, S. (1923). *The Weld-Blundell Collection, Volume II: Historical Inscriptions, containing principally the chronological prism, W-B 444*. Oxford Editions of Cuneiform Texts 2. Oxford: Clarendon Press. The editio princeps of the Sumerian King List as preserved on the Weld-Blundell Prism, with the Ashmolean object record at AN1923.444.
• Cuneiform Digital Library Initiative wiki entry for the Sumerian King List and the Weld-Blundell Prism. https://cdli.mpiwg-berlin.mpg.de/wiki/. Open-access transliteration and bibliography.
• Izre'el, S. (2001). *Adapa and the South Wind: Language has the Power of Life and Death*. Mesopotamian Civilizations 10. Winona Lake: Eisenbrauns. The standard modern edition of the Adapa fragments, including Fragment B (EA 356) from Amarna and the smaller Nineveh manuscripts.
• Izre'el, S. (1997). *The Amarna Scholarly Tablets*. Cuneiform Monographs 9. Groningen: Styx (Brill). The volume containing the chapter on EA 356 and the Adapa fragment in the context of the Amarna scribal archive.
• Penn Museum object record CBS 10673. https://www.penn.museum/collections/. Catalogue record for the Eridu Genesis tablet recovered at Nippur in 1893; first read by Poebel in 1912; published in PBS 4, 1914.
• Encyclopaedia Iranica, "Berossus." https://iranicaonline.org/. Survey of Berossus and the *Babyloniaca*, including the transmission via Polyhistor, Josephus, and Eusebius.
• Verbrugghe, G. P. & Wickersham, J. M. (1996). *Berossos and Manetho, Introduced and Translated: Native Traditions in Ancient Mesopotamia and Egypt*. Ann Arbor: University of Michigan Press. Annotated translation of all the surviving Berossus fragments.
• Ramazzotti, M., D'Agostino, F., et al. (2021). "The Iraqi-Italian Archaeological Mission at the Seven Mounds of Eridu (AMEr): preliminary report on the 2019 and 2020 seasons." Sapienza University of Rome and the State Board of Antiquities and Heritage, Iraq. The first AMEr campaign reports under D'Agostino's direction (Sapienza) with co-direction from Philippe Quenet (Strasbourg) and epigraphic work by Anne-Caroline Rendu Loisel (Strasbourg). Permit signed 2014; sustained fieldwork resumed 2018–2019.