The Kurkh Stele: Anunnaku Rank, Ea’s Clever Devices, and the Limits of Shared Agency

Download the complete illustrated PDF report

What this object actually says

The Kurkh stele names Anu as king of the Igigu and Anunnaku. Nearby it calls Enlil the deviser of cosmic designs and Ea the maker of clever devices. That proximity is important for the history of divine authority and knowledge, but it does not give all three predicates to the Anunnaku. The grammar distinguishes a governed divine collective, named title-bearing gods, and a broader group that authorizes royal power.

The core Anunnaku spelling in Column I1 is unbracketed and unflagged in the controlling ORACC edition. The exact historical copy locates its row; an independently licensed photograph establishes the physical object. This is edition-guided, not independently photograph-decoded. We do not present an enlarged reproduction as a fresh decipherment of individual wedges.

A real Akkadian monument, not a generic “Anunnaki tablet”

The selected physical witness is British Museum BM 118884, registration W_1863-0619-2, CDLI P465006. It is the stone Kurkh stele of Shalmaneser III, not the other Kurkh monument of Ashurnasirpal II. CDLI links this object to RIMA 3.0.102.002 and the composite Q004607 / P463785. The language is Akkadian, not Sumerian; the object is a royal display stele rather than a clay tablet. It belongs in this comparative daily corpus because the task includes Akkadian witnesses to Anunnaku traditions.

The photograph’s catalog description places the monument around 852 BCE and notes the annals’ account of Qarqar in 853 BCE. CDLI dates it more broadly to the Neo-Assyrian period. Those are different levels of chronological description, not competing claims about when the stone was photographed. The present analysis is restricted to the opening divine invocation, Column I1–4, rather than an evaluation of every military claim in the annals.

The inspected front has a standing royal relief, a raised hand, divine emblems above, writing across the relief and adjacent surface, visible fractures and areas of abrasion. These observations come from this exact photograph only. They do not identify any of the depicted symbols as vehicles, planets, or machinery.

Identity, publication rights and the old royal ordinal

The CDLI record locks the museum number, P-number, registration, composition and standard edition together. The independent photograph names the same king, Kurkh provenance and museum, and shows the expected monument. Both visual files were compared to current Wikimedia Commons file hashes and dimensions before use. The museum/CDLI photograph has different, noncommercial terms and is not the image republished here.

The open photograph is by Osama Shukir Muhammed Amin FRCP(Glasg), CC BY-SA 4.0. Its derivative here adds only resizing and a banner; that derivative retains the same license. The line copy is Rawlinson’s historical exact-object copy in The Cuneiform Inscriptions of Western Asia, volume III, distributed as public domain. Its heading says “Shalmaneser II”: this is the historical ordinal used for the king now called Shalmaneser III. The original heading is retained, not silently corrected or treated as a second witness.

The original Column I heading and ruled-row sequence are the location controls. The first four rows precede the printed line-5 control. Modern colored bands mark those whole rows. The photo has no invented word boxes, and the copy has no unsupported sign-cluster boxes. Old hatching stays visible rather than being filled with modern text.

Four rows do not mean one shared actor

Column I1 first invokes Ashur, then Anu as king of the Igigu and Anunnaku, then Enlil as father of the gods and determiner of destinies. The opening of I2 continues Enlil’s title: “who devises the designs of heaven (and) netherworld” in ORACC’s published English. Only after that continuation does I2 name Ea, the wise one, king of the apsu, maker of clever devices. The line break is a physical writing boundary; it is not a grammatical reset.

This is the decisive distinction for the craft-and-knowledge proof gap. The design epithet belongs to Enlil across the I1/I2 boundary, whereas the maker epithet belongs to Ea within I2. The Anunnaku are the complement of Anu’s kingship title in I1. They are not the grammatical subject of either the design or maker expression. “In the same opening invocation” is weaker than “the same actor performed the action.”

I3 moves through Shamash and Ishtar to “the great gods” who favor the king. I4 continues that broader group’s royal authorization. We should not silently replace “great gods” with “Anunnaku,” even though the named divine group occurs earlier. The invocation supports a differentiated theological order, not an undifferentiated divine workforce.

Published translation and cautious working translation

The records below separate the published ORACC/RIAo rendering from our cautious working English and identify exactly what the photograph, copy and edition each contribute. Transliteration excerpts retain determinatives and logograms; they are not a complete new edition of the four rows. The published edition is based on Grayson’s RIMA3 and its open ORACC adaptation.

Notation matters. {d} marks a divine determinative, not a spoken syllable to be translated as a separate god. MAN and ZU.AB are logographic writings in an Akkadian text. Square brackets would mark supplied losses; half brackets in nik-⸢la⸣-ti mark a damaged-readable sign. Parentheses in an English translation can supply connective words without claiming a restored cuneiform sign. The ellipsis in the final excerpt below is our abbreviation, not an artifact lacuna.

The Anunnaku core is not entirely restored, parallel-only, or borrowed from another stele. Its exact row is available here. Nevertheless the photograph-visible sign values were not independently verified. “Unbracketed in the edition,” “present in the copied row,” and “independently readable in a photograph” remain different evidential claims.

Image-to-text record 1

Exact image location

Obverse/front, Column I, line 1; first ruled row under Column I; original copy x20–815, y79–95. Gold band.

Edition transliteration

Excerpt: {d}a-nu MAN {d}NUN.GAL.MEŠ u {d}a-nun-na-ki

Published translation

ORACC/RIAo (Grayson/RIMA3): “the god Anu, the king of the Igīgū and Anunnakū gods”.

Cautious working translation

Anu, king of the Igigu and the Anunnaku.

Photograph-visible signs

None independently verified from the photograph. It establishes exact-object identity and preservation, not this reading.

Exact-copy-visible / unbracketed sequence

The exact copy preserves the registered row; ORACC prints the entire Anunnaku spelling and governing kingship phrase unbracketed and without damage marks. The relevant sequence is outside the conspicuous left-side hatched interval; no independent sign-by-sign collation is asserted.

Damaged signs

No damage marks on the selected Anu/Anunnaku phrase in the controlling edition. Other portions of the historical copy’s row have hatching; it is not evidence for missing words here.

Restored, composite or parallel-supplied signs

None in the selected phrase. NUN.GAL.MEŠ is the edition’s logographic writing for Igigu, not the Anunnaku spelling.

Confidence

High for object identity, numbered whole-row location and edition attribution; limited for autonomous sign recognition. Reading is edition-guided.

Limit of visual claim

Exact historical line art locates the complete row. No pixel-level sign recognition or word-sized bounding box is claimed; the modern scholarly edition controls sign values. The old copy’s hatching is preserved, not filled.

Textual evidence class

exact_line_art_or_squeeze_visible

Image-to-text record 2

Exact image location

Obverse/front, Column I, end of line 1 and opening of line 2; teal second-row band, x20–815, y95–111, with the gold row as antecedent.

Edition transliteration

I1: {d}BAD a-bu DINGIR.MEŠ mu-šim NAM.MEŠ; I2 opening: mu-ṣir e-ṣu-rat AN-e KI-tim

Published translation

ORACC/RIAo: Enlil is the one “who devises the designs of heaven (and) netherworld”.

Cautious working translation

Enlil, father of the gods and determiner of destinies, deviser of the designs of heaven and the lower world.

Photograph-visible signs

None independently verified from the photograph. It establishes exact-object identity and preservation, not this reading.

Exact-copy-visible / unbracketed sequence

Registered end-of-row and next-row sequence. The design phrase is unbracketed in ORACC; the printed line boundary does not create a new subject.

Damaged signs

No damage marks on the quoted design phrase in ORACC. The old copy includes hatching within the second row; this reproduction does not independently recover those signs.

Restored, composite or parallel-supplied signs

None in the quoted phrase in ORACC. Parenthesized “and” in the English is a translation aid, not a restored Akkadian sign.

Confidence

High for object identity, numbered whole-row location and edition attribution; limited for autonomous sign recognition. Reading is edition-guided.

Limit of visual claim

Exact historical line art locates the complete row. No pixel-level sign recognition or word-sized bounding box is claimed; the modern scholarly edition controls sign values. The old copy’s hatching is preserved, not filled.

Textual evidence class

exact_line_art_or_squeeze_visible

Image-to-text record 3

Exact image location

Obverse/front, Column I, line 2 after the Enlil design phrase; complete second ruled row x20–815, y95–111, teal band.

Edition transliteration

{d}é-a er-šu MAN ZU.AB ba-nu-ú nik-⸢la⸣-ti

Published translation

ORACC/RIAo: “the god Ea, the wise one, the king of the apsû, the creator of clever devices”.

Cautious working translation

Ea, the wise one, king of the apsu, maker of ingenious things / clever devices.

Photograph-visible signs

None independently verified from the photograph. It establishes exact-object identity and preservation, not this reading.

Exact-copy-visible / unbracketed sequence

The row and Ea clause are registered on the exact copy. The text names Ea, not the Anunnaku, as the maker. No sign values are independently read from the photograph.

Damaged signs

The la sign in nik-⸢la⸣-ti is damaged-readable in ORACC. Half brackets mark damage, not a completely supplied word.

Restored, composite or parallel-supplied signs

No square-bracket restoration in the selected Ea clause. The English “devices” is a lexical rendering, not identification of machinery.

Confidence

High for object identity, numbered whole-row location and edition attribution; limited for autonomous sign recognition. Reading is edition-guided.

Limit of visual claim

Exact historical line art locates the complete row. No pixel-level sign recognition or word-sized bounding box is claimed; the modern scholarly edition controls sign values. The old copy’s hatching is preserved, not filled.

Textual evidence class

exact_line_art_or_squeeze_visible

Image-to-text record 4

Exact image location

Obverse/front, Column I, lines 3–4; third and fourth ruled rows x20–815, y111–143, purple bands.

Edition transliteration

I3 end: DINGIR.MEŠ GAL.MEŠ ÁGA-ut MAN-ti-ia; I4: šá EN-ti kiš-šu-ti u šá-pi-ru-ti ú-šar-bu … ma-aʾ-diš iš-ku-nu-in-ni

Published translation

ORACC/RIAo describes “the great gods, who love my kingship” and says they made his dominion, power and leadership great.

Cautious working translation

The great gods who favor my kingship, who enlarged my rule, might and leadership, and established my honored name and exalted command.

Photograph-visible signs

None independently verified from the photograph. It establishes exact-object identity and preservation, not this reading.

Exact-copy-visible / unbracketed sequence

Complete registered rows. The ellipsis in this short citation is an editorial omission by this report, not a break in the artifact.

Damaged signs

No damage marks in the quoted royal-grant sequence in ORACC; old-copy loss/shading remains separate from modern edition readings.

Restored, composite or parallel-supplied signs

None in the selected edition sequence. “The great gods” is broader wording and must not be silently replaced with Anunnaku.

Confidence

High for object identity, numbered whole-row location and edition attribution; limited for autonomous sign recognition. Reading is edition-guided.

Limit of visual claim

Exact historical line art locates the complete row. No pixel-level sign recognition or word-sized bounding box is claimed; the modern scholarly edition controls sign values. The old copy’s hatching is preserved, not filled.

Textual evidence class

exact_line_art_or_squeeze_visible

What “clever devices” can and cannot establish

The edition’s phrase “creator of clever devices” translates an Ea epithet, not a description of a recovered machine. Our alternative “maker of ingenious things” deliberately leaves the category broad. A technical identification would need an object, a procedure, or independent material evidence, none of which is supplied by this short epithet. It would be a mistake either to erase the knowledge-and-ingenuity motif or to upgrade it to evidence of advanced machinery.

Apsu here is Ea’s divine domain. Its occurrence is relevant to the Enki/Ea and Abzu strand of the corpus, but this clause does not describe a laboratory, an underwater industrial site, or an engineering installation. Likewise “designs” expresses divine ordering in this invocation; it does not give dimensions, a fabrication sequence, or an operational blueprint.

The conventional English Anunnaki covers traditions whose scholarly forms include Akkadian Anunnaku and Sumerian Anuna/Anunna. This Akkadian witness is not itself a bilingual equation and does not prove that every use of those labels designates an identical membership or office. Its local contribution is the title relationship expressed by this particular invocation.

Corpus relevance: support, limits and a falsifiable next step

Support: an exact monumental witness places the Anunnaku under Anu’s kingship in an organized divine order. In the same tightly bounded opening, Enlil owns cosmic-design language, Ea owns apsu and ingenious-making language, and the broader great-gods group sanctions kingship. These are positive data for differentiated divine authority, knowledge theology and royal legitimacy.

Limit and contradiction: the clause structure contradicts a reading that makes the Anunnaku the makers of Ea’s devices merely because the names are close together. The text does not prove extraterrestrials, Nibiru, spacecraft, genetic engineering, mining programs or advanced technology. Nor does royal ideological language by itself demonstrate a historical nonhuman organization operating the Assyrian state. A text can be strong evidence for an ancient belief while remaining weak evidence for a modern literal interpretation of that belief.

The new contribution beyond the previously studied Black Obelisk is a distinct exact Kurkh witness and its cross-line Enlil/Ea craft-and-design distinction. This is not a renamed republication of the other monument. The next high-value test is a nonduplicate, rights-cleared witness in which an un-restored Anuna/Anunnaku subject directly governs a manufacturing, measuring or knowledge-transfer predicate. Until that grammar appears, adjacent divine titles must remain separate.

Daily decoding improvement: cross-line epithet ownership

This run adds a Cross-Line Epithet Ownership Register: physical row, continuing grammatical head, start of a new divine name, predicate, collective’s role, damage/restoration, old-versus-modern identifier, and visual-claim ceiling. Applied here, it prevents the start of I2 from being misassigned to Ea and prevents Ea’s maker predicate from being transferred to the Anunnaku.

The accompanying identity improvement is equally concrete: preserve historical royal ordinals in primary-copy captions and reconcile them through museum and composition identifiers. A mismatch of old and modern numbering is not automatically a mismatched object, but it must be explained before image-to-text use. The photographic and copy evidence remain separately classified and credited.

Sources, attribution and reproducibility

Primary controls were checked for this 15 October 2026 Türkiye run. The local ORACC certificate chain failed once; a narrowly scoped public-source-only transport fallback retrieved the expected complete Q004607 edition. Artifact identity, textual content, file hashes and publication rights were checked separately from transport. No private credentials or publisher connection used that fallback.

Quoted and adapted ORACC scholarly material is attributed to the RIAo/RIMA edition and its CC BY-SA 3.0 license. The photograph and its derivative have their separate CC BY-SA 4.0 credit; the historical line art is public domain. Modern overlays, cautious interpretation and the evidence register do not replace the primary edition. The PDF and article share the same report content and exact annotated panels.

• CDLI P465006: exact BM 118884 witness, museum registration and Q004607/P463785 cross-reference
• ORACC/RIAo Q004607: Shalmaneser III 002; A. Kirk Grayson, RIMA3 (1996), adapted by Jamie Novotny and Nathan Morello
• Exact Kurkh photograph and photographer/license metadata: Osama Shukir Muhammed Amin FRCP(Glasg)
• Rawlinson exact Kurkh copy, Cuneiform Inscriptions of Western Asia III, public-domain source
• Photograph and annotated photographic derivative: Creative Commons Attribution–ShareAlike 4.0
• ORACC annotated-edition license: Creative Commons Attribution–ShareAlike 3.0