wariq
Viking-age silver hoards and Islamic dirhams, machine-readable — with the catalogues’ own uncertainty kept intact.
Viking-age silver hoards and Islamic dirhams, machine-readable — with the catalogues’ own uncertainty kept intact.
All files CSV, UTF-8. These are facts extracted from published, openly-available catalogues — the sources were already public; what was missing was any form in which they could be queried, joined or counted.
| File | Rows | Contents |
|---|---|---|
| hoards.csv | 548 | Hoard, TPQ, findspot, discovery, container, deposit class, soil, silver weight, catalogue reliability scores |
| contents.csv | 548 | Per-hoard coin counts by origin (Oriental, Byzantine, German, English, other European, undetermined), whole and fragmentary |
| cns_coins.csv | 242 | Islamic-tradition coins: striking date (AH and AD), mint, dynasty, ruler, weight, diameter, die axis |
| cns_coins_european.csv | 152 | English, Danish and German coins from the same hoards |
| mint_alignment.csv | 77 | Mint names to Nomisma URIs, with match evidence |
Catalogues record how much of a legend the cataloguer could actually read. Square brackets mark characters restored from type — and they fall mid-word, as in al-Muhammad[iyya] or Madinat Nays[abur]. In the CNS volumes, 29.0% of mint attributions and 32.2% of Hijri dates carry such a mark.
Almost every digitisation strips it, including this project’s own first attempt. Here the printed form is kept in mint_raw alongside the normalised mint_canonical, with a mint_restored flag. Nothing is cleaned in place.
The dataset is deposited and citable:
Reid, C. (2026). wariq: Viking-age Baltic silver hoards and Islamic dirhams, machine-readable [Data set]. Zenodo. https://doi.org/10.5281/zenodo.21756407
That DOI covers the structured form and the code that produced it, not the evidence. For the coins and hoards themselves, cite the catalogues: Gruszczynski (2019) for the hoards, Corpus Nummorum Saeculorum IX–XI for the coins. Both are named in every row’s provenance and on the silver road page.
ans_mint_reconciliation.csv is ODbL, not CC-BY. Everything in the table above is extracted from the Gruszczynski and CNS catalogues and carries our own licence. This file does not: it is derived from the American Numismatic Society’s collection data, which is published under the Open Database License. ODbL is share-alike, so a derivative of it must also be ODbL. Keep it separate from the tables above, or the obligation travels.
It holds counts only — 50 mints, with how many ANS dirhams each struck across all periods and within AH 180–400, and the AH span. Every row carries its own source and license column so the provenance survives being merged into something else. The ANS coin records themselves are not redistributed here; reproduce them with scripts/15_ans_dirhams.py --fetch, which puts the query in code rather than shipping someone else’s database.
Applying our registry to their dirhams resolved 4,003 of 8,193 mint-bearing coins (48.9%) across 50 mints with no new aliases. What that comparison shows is on the mints page.
The hoards are also published as RDF for Nomisma.org-compatible harvesting, with a stable URI per hoard.
| File | What it is |
|---|---|
| wariq-hoards.rdf | 548 hoards: 509 with a closing date, 539 with a findspot resolved to a Wikidata gazetteer URI, 506 with composition |
| void.rdf | VoID dataset descriptor — the file a harvester reads first |
The nmo: and crm: triples are conformant Nomisma. Composition and legibility ride in a local wariq: namespace, because the Nomisma ontology has no property for either — there is no way to say “482 Oriental coins, 12 German”, which is the primary datum of Viking-age hoard scholarship. A harvester that does not know our namespace can ignore it safely.
Coins are not offered for harvesting. Contributed coin records require nmo:hasTypeSeriesItem, a URI in a published type series, and no Islamic type series exists in the ecosystem — which is why Islamic coinage is absent from it.
Licensing. The hoard data is extracted from Gruszczynski (2019), deposited open-access in the Oxford Research Archive. The coin data is extracted from the CNS volumes published freely by the Stockholm Numismatic Institute. Neither source PDF is redistributed here; the pipeline re-fetches them. Cite the catalogues, not this site.
A separate corpus — a structured parse of the unpublished Noonan–Kovalev dirham hoard catalogue — underlies some of this project’s analysis but is not published here and nothing derived from it appears on this site, pending its author’s permission.