Source

Tell el-Amarna Project (Cambridge/EES) — Object Database

Tell el-Amarna (ancient Akhetaten), Middle Egypt · Eighteenth Dynasty (Akhenaten-Tutankhamun) primary; some Late Antique strata

7,386

Objects

1,132

Inscriptions

Browse all 7,386 objects → Visit source →

No imaged objects to show. Try the full browse view.

Source details

Type: excavation_archive
Access: bulk_dump
Stack: wordpress-excel
License: Project-published academic dataset (CC-BY-NC presumed; verify with Cambridge/EES)
Modality: sign_name_vocab
ID system: EES registration number (YY/NN format)
ID tier: tier-3
Inscription richness: 6/10
Cleanliness: 7/10

Form breakdown

other: 2,917
jewelry: 1,216
architecture: 778
vessel: 755
statue: 609
tool: 293
textile: 191
cosmetic_object: 187
ostracon: 158
document: 84
seal: 68
stela: 53
figurine: 36
scarab: 16
amulet: 12
relief_fragment: 8
funerary_equipment: 4
papyrus: 1

Notes

## Discovered schema (Tell el-Amarna Object Database, surveyed 2026-05-24) ### HEADLINE: Excavation archive is BULK-ONLY (Excel), but RICHLY STRUCTURED The mini pre-flight tagged Tell el-Amarna `open` because `/wp-json/wp/v2/` returned real JSON. That JSON was 10 blog posts + ~79 WordPress pages — the project's news/PR site. The real excavation data is a **2.7 MB Excel file** at `/wp-content/uploads/Amarna-Object-Database.xls` plus a PDF schema key. This is a fifth stack class beyond MET-proprietary / Axiell+CIIM / KE EMu+ Solr / EDAN+Solr+api.data.gov: **5. WordPress site + downloadable Excel bulk data (Tell el-Amarna)** It's also the **first excavation-archive shape** we've surveyed — and genuinely different from museum-collection shapes, justifying a parallel comparison column or table in future surveys. ### Dataset overview - **7,599 records** with registration numbers (35 structured columns each) - Year-of-excavation range: 1921-1936 (EES Petrie-era seasons 1921-1936 dominate) - Materials top 5: {'faience': 1796, 'limestone': 988, 'pottery': 751, 'bronze': 746, 'clay': 354} - Areas (Amarna sub-sites): {'C.C.': 2436, 'N.S.': 2010, 'M.C.': 1903, 'W.V.': 363, 'N.C.': 333, 'Maru Aten': 269, 'N.P.': 138, 'River Temple': 28, '?M.C.': 5, '?N.C.': 4} - License: not explicit on download; CC-BY-NC presumed per Cambridge norms ### Test 1 — Shape model (the brief's headline question) **Excavation hierarchy is DENORMALISED into column groups on one row per object** rather than separate entity tables: - Site = always Tell el-Amarna (implicit, not a column) - Season = `Year of excavation` (numeric) - Sub-area = `Provenance > Area/Suburb` (controlled-vocab: C.C., N.S., M.C., W.V., N.C., Maru Aten, N.P., River Temple) - Trench/cell = `Provenance > Subdivision/building label` (Petrie grid: e.g., 'O49.22' — letter+number locates a 200m grid square, post-dot locates building within it) - Context = `Provenance > Further details` + `Stratigraphic details` (prose; e.g., "pit (?well) just to west of west enclosure wall of O.49.7&8") - Object = the row itself Implication for our canonical schema: when we ingest excavation-archive sources we DON'T need a separate `season` / `trench` / `context` table. We just need denormalised stratigraphic columns on the object row: `find_spot_site`, `find_spot_subarea`, `find_spot_grid`, `find_spot_further`, `find_spot_strat_details`, `excavation_year`. ### Test 2 — Stratigraphic metadata (STRONGLY STRUCTURED) | field | populated | type | |---|---|---| | Year of excavation | 7589/7,599 | numeric (season) | | Area/Suburb | 7507/7,599 | controlled vocab | | Subdivision/building | 7374/7,599 | Petrie grid ref | | Further details | 2976/7,599 | prose | | Stratigraphic details | 169/7,599 | prose (locus / feature) | Sample sample: 70/70 with Area, 69/70 with grid ref, 16/70 with stratigraphic details. This is **the cleanest stratigraphic metadata of any source surveyed so far.** No museum source has provided this level of find-spot resolution. ### Test 3 — Inscription content (THIRD CONTENT MODALITY) - **3,061 / 7,599 (40%) records have populated decoration text** - **1,022 / 7,599 (13.5%) records mention hieroglyph / inscription content** - Pattern: `Decoration > Obverse/Reverse` uses **`category: hieroglyph_name`** vocabulary — e.g., "cartouche: Akhenaten", "hieroglyph: nefer", "cartouche: Tutankhamun" This is a THIRD inscription-content modality across surveyed sources: - **Fitzwilliam**: MdC transliteration of full text ("Htp di nsw in wsir...") - **ISAC**: English description prose ("for the benefit of Yartiuerow") - **Tell el-Amarna**: structured hieroglyph identification with controlled sign-name vocabulary (NOT a transcription; an enumeration of named hieroglyphic signs/cartouches present on the artifact) For our Phase 2 inscription table, this adds a fourth populatable column alongside `transcription_mdc`, `description_en`, and now `decoration_signs` (structured list of hieroglyph-category : sign-name pairs). Sample inscribed records: - 21/65: obverse='royal: part of figure offering to the Aten', reverse=None - 21/527: obverse='non-royal: Shed and Isis', reverse=None - 22/271A-M: obverse='royal: Aten worship', reverse=None - 22/273: obverse='royal: Aten worship', reverse='royal: Aten worship' - 22/274: obverse='inscribed', reverse=None ### Test 4 — Cross-refs (NO TM, NO PLEIADES; LOCAL PUBLICATION REFS INSTEAD) The database has NO Trismegistos, Pleiades, Wikidata, JdE, or Porter-Moss fields. Cross-referencing is **internal to the EES publication tradition**: - COA I (City of Akhenaten I, Peet & Woolley 1923): 800 records (10%) - COA II (Frankfort & Pendlebury 1933): 1,175 records (15%) - COA III (Pendlebury 1951): 2,246 records (29%) - Union (at least one COA ref): ~4,219 records (~55%) - EES negative number (photo archive): 2,899 records (38%) For Phase 2: when we cross-link Amarna small finds to text-database sources (TLA / TM), the bridge MUST be built via fuzzy match on (registration number, year, area, identification keywords, hieroglyph signs). There's no shortcut via Wikidata or other shared IDs. The COA publication refs ARE the bridge to print-OCR work — same shape as ISAC's `BibUrl → JSTOR` pattern, except COA volumes are not generally on JSTOR. Most are in HathiTrust / Internet Archive / EES print archive. ### Test 5 — Publication linkage (STRUCTURED, see Test 4) Three dedicated columns for the City of Akhenaten EES Excavation Memoir volumes. Each cite is in the form `"p. 29, pl. XIII.5"` — page + plate.figure. The EES Excavation Memoirs series itself is digitally available through: - HathiTrust (mostly): https://catalog.hathitrust.org/Search/Home?lookfor=City+of+Akhenaten - Internet Archive: https://archive.org/search?query=city+of+akhenaten - JSTOR (some EES volumes): https://www.jstor.org The EES Negative Numbers column links to the **EES Photographic Archive** (housed at the Society in London, partially digitised on the EES website). ### Sample (n=70) by form category - {'stela': 14, 'papyrus': 1, 'statue': 14, 'jewelry': 14, 'vessel': 14, 'architecture': 14} Inscription content found in our sample: 15/70 records. Publication refs: 45/70. EES negative: 23/70. ### Amarna Letters note (the brief flagged this) The cuneiform "Amarna Letters" (discovered 1887, ~382 tablets) are NOT in this database. Their primary holdings are at: - Vorderasiatisches Museum Berlin (most of them) - British Museum (most other significant cluster — but BM is Cloudflare-gated) - Cairo / Egyptian Museum This database is the 20th-and-21st-century EES + McDonald Institute SMALL FINDS register from Akhenaten's city, not the 19th-century cuneiform correspondence archive. ### Nine-way museum/archive comparison (museum pattern N=9; first excavation archive) | metric | MET | BM | BKM | Fitz | Leiden | ISAC | Manch | Smith | **Amarna** | |---|---|---|---|---|---|---|---|---|---| | total Egyptian (sampled / available) | 77 of 27,972 | 13 (WD) | 4 (WD) | 80 of 16,801 | 0 | 166 of 38,912 | 0 | 84 of ~2,007 NMNHANTHRO | **70 of 7,599 small finds** | | has inscription field | 0 | 0 | 0 | 6 (MdC) | gated | yes (English) | gated | 0 | **15/70 (sign-name vocab)** | | inscription content modality | — | — | — | MdC transliteration | gated | English prose | gated | prose mentions | **hieroglyph sign-name enum** ⭐ | | multi-inscription/object | n/a | n/a | n/a | yes | gated | yes | gated | no | partial (obverse+reverse) | | with cross-source IDs | 0 | 2 (WD-mediated) | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | with image (in this sample) | yes | (WD) | (WD) | yes | gated | yes | (Eur) | yes | indirect (EES neg→archive) | | with structured find spot | prose | no | no | yes+coords | gated | yes+locus | gated | hierarchical L1/L2/L3 | **5-field stratigraphic** ⭐ | | with structured period | partial | prose | prose | yes (dynasty) | gated | yes (dynasty) | gated | decade-binned only | year-of-excavation (mostly C18 dynasty implicit) | | federation modeling | n/a | n/a | n/a | n/a | n/a | dataset code | n/a | unitCode | n/a (single site) | | direct API access | yes | gated | gated | key+JWT | gated | yes (Solr) | gated | yes (api key) | **no (Excel bulk dump)** | | stack | proprietary | proprietary | proprietary | Axiell+CIIM | WP+CF | KE EMu+Solr | WP+React+CF | EDAN+Solr | **WordPress + Excel bulk** ⭐ | | publication / OCR linkage | no | no | no | publications field | gated | RA.BibUrl→JSTOR | gated | no | **COA I/II/III refs + EES negs** ⭐ | | coverage on direct path | 100% (27,972) | <0.1% | <0.01% | 100% (16,801) | 0% | 100% (38,912) | 0% | 100% (~2,007) | **100% (7,599)** | ### Flag interpretation for Tell el-Amarna - Flag 1 (shape model): NEW MODALITY. Excavation-archive denormalisation, not museum-style per-object record. The canonical schema should support BOTH shapes via the same `object` table with conditional stratigraphic columns. - Flag 2 (stratigraphic metadata): 5-field structured, the cleanest seen. - Flag 3 (inscription): third content modality (sign-name vocabulary). - Flag 4 (cross-refs): no shared IDs; bridges via fuzzy match required. - Flag 5 (publication linkage): YES, structured to EES City of Akhenaten volumes. Bridge to print-OCR work like ISAC's RA→JSTOR but to EES print memoirs instead. ### Operational gotchas - The /wp-json/wp/v2/ JSON is the project's BLOG, not the data. The data is a bulk Excel download. Pre-flight lesson reinforced: confirm the JSON actually contains collection data before tagging `open`. - Excel multi-row header (top group + sub-column) must be parsed with `header=[0,1]` and the resulting MultiIndex columns. Some columns have no sub-header (year, registration number, date alt) and read as "Unnamed: N" pandas defaults — filter those. - Year-of-excavation has at least one typo (9132 instead of 1932). Filter to plausible range before computing date spans. - The Petrie grid reference format `O49.22` is the Egyptian-archaeology standard from Petrie's 1894 work; not a generic format you'll find in museum catalogs. Useful identifier for cross-source spatial joins IF other sources retain Petrie grid refs. ### Open questions - Are there sister-database Excel files for other parts of the Amarna Project (textiles, glass, bioarchaeology)? Worth a side investigation of /wp-content/uploads/. - Does the EES website host machine-readable versions of City of Akhenaten I/II/III? Worth probing https://www.ees.ac.uk for a digital memoirs collection. - Egypt Artefacts at Griffith Institute (egyptartefacts.griffith.ox.ac.uk) was linked from the Amarna small-finds-database page — worth surveying as a sibling resource later (per-object detail beyond Amarna). - The "Distribution" columns (Registration card / Publication / Confirmed) hint at which museums received the objects post-excavation. Mining those for accession-number bridges to MET/Fitzwilliam etc. is a cross-ref opportunity for the future fuzzy-match pipeline.