Thank you

Unread Atlas stands on generations of work by excavators, museum curators, cataloguers, epigraphers and programmers, most of whom we will never meet. Every tablet on this site was dug up, conserved, numbered, photographed, copied or transliterated by someone else first. This page is our thank-you to them, and the place where we keep the licences straight.

Text corpora

The catalogue, find-spots, transliterations (ATF) and human English translations come from these projects. Transliterations are shown only where the source allows republication.

CDLI, the Cuneiform Digital Library Initiative: The backbone of the catalogue: around 427,000 records, find-spots, periods, museum numbers and ATF, from the cdli-gh bulk data. Reused with citation of CDLI, as CDLI asks. We do not copy CDLI images.
ORACC, the Open Richly Annotated Cuneiform Corpus, and its many projects: Lemmatised editions and human translations from RINAP, SAAo, ETCSRI, DCCLT, CAMS and dozens more. CC BY-SA 3.0. Some ORACC JSON files are marked CC0. Adaptations of ORACC material are shared under the same terms.
eBL, the electronic Babylonian Library (LMU Munich): Catalogue metadata and the ATF-to-Unicode sign readings used by our cuneiform converter and font choice. Dataset MIT. eBL transliterations are searched for discovery only and are not republished here.
TLHdig, Thesaurus Linguarum Hethaeorum digitalis: Hittite and Anatolian texts. CC BY 4.0
Linguistically Annotated Achemenet Babylonian Texts (University of Helsinki), from Achemenet: Achaemenid-period Babylonian archives. CC BY 4.0
ETCSL, RINAP, SAA and the print series: We link to these and flag their tablets as already translated. Their publications are the scholarly record, and our machine drafts are not a substitute for them.

Museums & collections

The tablets live in these institutions, which hold the most tablets in our catalogue. Hundreds of smaller collections and private holdings are credited on each tablet's own record. Thank you for keeping them safe, and for every catalogue entry that made a find-spot or museum number possible.

British Museum, London
Penn Museum, Philadelphia
National Museum of Iraq, Baghdad
Istanbul Archaeology Museums (Arkeoloji Müzeleri)
Museum of Anatolian Civilizations, Ankara
Yale Babylonian Collection, New Haven (incl. Nies, Morgan and Newell collections)
Vorderasiatisches Museum, Berlin
Musée du Louvre, Paris
Cornell University, Ithaca
National Museum of Syria (Damascus, Aleppo, Idlib, Deir ez-Zor)
ISAC Museum (formerly Oriental Institute), University of Chicago
Ashmolean Museum, Oxford
Harvard Museum of the Ancient Near East
Schøyen Collection, Oslo
de Liagre Böhl Collection, NINO, Leiden
Horn Archaeological Museum, Andrews University
Princeton Theological Seminary
Free Library of Philadelphia
State Hermitage Museum, St Petersburg
National Museum of Iran, Tehran

Map

Shaded relief: NOAA NCEI ETOPO 2022: The map's terrain and sea floor, rendered by us into our own label-free tiles. U.S. Government work, public domain. NOAA National Centers for Environmental Information (2022), ETOPO 2022 15 Arc-Second Global Relief Model, doi:10.25921/fd45-gt74.
Coastlines, rivers and lakes: Natural Earth: Made with Natural Earth. public domain. Used on the map tiles and the home page coastline.
Find-spots: Coordinates come from the CDLI and ORACC catalogues. Uncertain or region-only find-spots are not pinned.

Fonts

Cuneiform is drawn in a face matched to each tablet's period, following the eBL cuneiform converter.

Assurbanipal (Neo-Assyrian), Esagil (Neo-Babylonian), Santakku and SantakkuM (Old Babylonian cursive and monumental), UllikummiA (Hittite): Created by Sylvie Vanséveren, from the Hethitologie Portal Mainz. free for scholarly use. Files as distributed by eBL.
OB Freie (Old Babylonian literary): Corvin R. Ziegeler, Institut für Altorientalistik, FU Berlin (publication). SIL OFL 1.1
Noto Sans Cuneiform: The Noto project, for other periods and missing signs. SIL OFL 1.1
IBM Plex Sans, IBM Plex Mono and Source Serif 4: The site's text faces, by IBM and Adobe, served by Google Fonts. SIL OFL 1.1

Translation model

montycrypto/atf-to-english: Every English line marked machine draft comes from this model, a fine-tune of Google's T5-small (Apache 2.0) trained on ATF and English pairs drawn from CDLI and ORACC translations. Model MIT. The drafts are unreviewed and are not scholarly translations.

Software

Transformers.js (Hugging Face): Runs the translator in your browser. Apache 2.0
SQLite, Python, GDAL / rasterio, Shapely, NumPy, Pillow: Search index, data pipeline and basemap rendering. SQLite public domain; the rest BSD, MIT or similar permissive licences.

Notes

Unread Atlas is an independent project and is not endorsed by any institution listed here. If we have credited you wrongly, missed you, or you would like a link changed, please tell us on GitHub or @roguesamurai. We will fix it quickly and gladly.