Appearance
To link places to Pleiades correctly, match each of your locations to a single Pleiades place resource, store its numeric ID and full URI, and verify every match by hand because ancient toponyms are reused relentlessly. Treat the Pleiades ID as an external authority link, not as a source to copy wholesale, and leave the field empty rather than forcing a doubtful match. Consistency and documentation across the whole collection matter more than the rate of matches.
What is Pleiades and when is it the right authority?
Pleiades is the reference open gazetteer for the ancient world, with deepest coverage of the Greco-Roman Mediterranean, Egypt and the Near East. It is the right target whenever your sources are classical, late antique or early medieval. For places that only exist in modern records, GeoNames or Wikidata fit better; for cross-period work, the World Historical Gazetteer aggregates Pleiades alongside others.
What exactly should you link to?
A Pleiades place is the conceptual entity (e.g. Eboracum). Hanging off it are name resources (attested spellings) and location resources (coordinates with dating). Always link to the place URI:
https://pleiades.stoa.org/places/89178Linking to a name or location sub-resource is a mistake — those are not stable identity anchors and shift as the record is edited.
How do you find the right place ID?
For a handful of places, the web search interface is fastest. For a collection, pull the bulk export and match locally. Pleiades publishes a daily JSON dump:
bash
curl -L -o pleiades.json.gz \
https://atlantides.org/downloads/pleiades/json/pleiades-places-latest.json.gz
gunzip pleiades.json.gzThen match candidates on name and a coordinate tolerance, and review by hand:
python
import json
data = json.load(open("pleiades.json"))
def find(name):
for p in data["@graph"]:
names = [n["attested"] for n in p.get("names", []) if n.get("attested")]
if name in names or name == p.get("title"):
yield p["id"], p["title"], p.get("reprPoint")
for hit in find("Lugdunum"):
print(hit)Note that Lugdunum alone returns Lyon, Lugdunum Convenarum and others — hand verification against your source's region is non-negotiable.
How should the link be stored?
Keep both the bare ID and the resolvable URI, plus a provenance note:
| Field | Example | Why |
|---|---|---|
pleiades_id | 89178 | Compact, joins cleanly |
pleiades_uri | https://pleiades.stoa.org/places/89178 | Resolves directly |
match_confidence | certain | Filterable later |
geometry_source | pleiades-reprPoint | Honest provenance |
What if a place is not in Pleiades?
Leave it unmatched and log the gap; this is the single most common failure mode to avoid. Forcing Vindolanda onto a nearby fort because the name "looked close" silently corrupts every spatial query downstream. A documented absence is a research finding, not a failure.
How do you keep matches consistent across a collection?
Adopt one written rule set before you start: link to places only, require region verification, record confidence, and never reuse a previous editor's match without re-checking the ID. Run a periodic validation pass that resolves every stored URI and flags any that 404, since Pleiades records are occasionally merged.
Key Takeaways
- Always link to the Pleiades place resource, not a name or location sub-resource.
- Store both the numeric ID and the full resolvable URI, plus a confidence value.
- Verify every match against your source's region; ancient names are heavily reused.
- Leave non-matches empty and document them rather than forcing a link.
- Reuse Pleiades coordinates only with attribution and the representative-point caveat.
- Periodically re-resolve stored URIs to catch merged or moved records.
Frequently Asked Questions
What is Pleiades and what does it cover?
Pleiades is an open gazetteer of the ancient world, strongest for the Greek and Roman Mediterranean and the Near East. It gives each place a stable URI and represents locations, names and connections with explicit dating periods.
What is the right identifier to store from Pleiades?
Store the numeric place ID from the canonical URI, for example 89178 from pleiades.stoa.org/places/89178. Keep the full URI too so the link resolves directly.
How do I find the correct Pleiades place programmatically?
Use the Pleiades search interface or download the bulk JSON/CSV dump and match on name plus approximate coordinates. Always verify candidates by hand because ancient names are heavily reused.
Should I link to a Pleiades place, name or location?
Link to the place resource, which is the conceptual entity. Names and locations are sub-resources of a place and are not stable linking targets on their own.
What if my place is not in Pleiades?
Leave the Pleiades field empty and record the gap. Do not force a match to a nearby or similarly named place; an absent link is honest, a wrong one is corrupting.
Can I reuse Pleiades coordinates directly?
Yes, under its CC-BY licence with attribution, but record that the geometry is sourced from Pleiades and carry its representative-point caveat. Pleiades points are deliberately approximate for many sites.