Nomenclator — species-rank names and the people they honour

Two tables about the naming of life. One is complete for its source; the other is an explicit sample.

• dedications — 16,404 taxa whose scientific name records a dedication (named after on Wikidata), with the honoree and what Wikidata knows about them: whether the honoree is a person, their gender, citizenship and year of birth. This is every such record in Wikidata at the time of the snapshot, not a sample.
• fallen — 17,171 species-rank names that the GBIF Backbone has reduced to a synonym, each with the accepted name that absorbed it. This one is a sample: 3,000 per kingdom (the slice sizes are in fallen_summary.json), stored in full because the complete synonym set is two million rows.

The two tables are the two halves of a question that is easier to ask than to answer: of all the names ever published for living things, how many still stand, and who did we choose to put into the ones that do?

What the numbers look like

Taxa with a recorded dedication	16,404
…dedicated to a person	10,963 (66.8%)
…dedicated to something else (a place, a ship, an institution)	5,441
Distinct honorees	9,812
Women : men among the people honoured	2,195 : 8,653 (20.2% women)
Species-rank names in the GBIF Backbone	4,994,322
…still accepted	2,614,593 (52.4%)
…reduced to a synonym or marked doubtful	2,379,729 (47.6%)

A detail that only appears once people and non-people are separated: the most-honoured person in the dedication table is Claude Besuchet with 122 taxa, while the most-honoured thing is the city of Lijiang with 104 — a prefecture outranks every naturalist in the set.

Files and schema

dedications.csv (config dedications)

Column	Type	Notes
taxon_name	string	the scientific name as Wikidata records it (P225)
taxon_qid	string	Wikidata item for the taxon, e.g. Q4661181
rank	string	genus, species, … (P105)
honoree_label	string	who the name honours, as a label
honoree_qid	string	honoree's Wikidata item; empty for 12 anonymous nodes
honoree_is_human	boolean	whether the honoree is an instance of human (P31 = Q5)
gender	string	P21, normalised to male / female / other / empty
country	string	citizenship (P27), one label if several exist
birth_year	integer	P569, year only

fallen.csv (config fallen)

Column	Type	Notes
fallen_name	string	the name that lost
accepted_name	string	the name that absorbed it
authorship	string	author string of the fallen name, e.g. Hauer, 1938
year	integer	year parsed out of authorship; empty when none parses
kingdom … genus	string	GBIF classification of the fallen name
gbif_key	integer	GBIF taxon key of the fallen name — resolves to gbif.org/species/<key>
accepted_key	integer	GBIF taxon key of the absorbing name
published_in	string	the original publication, as GBIF holds it

ledger.json and fallen_summary.json carry the aggregates behind the tables, including the exact per-kingdom slice sizes.

Provenance

Taxonomy	GBIF Backbone Taxonomy — dataset key d7dddbf4-2cf0-4f39-9b2a-bb099caae36c, CC BY 4.0, DOI 10.15468/39omei
Dedications	Wikidata — P138 (named after), P21, P27, P569, CC0 1.0
Snapshot	2026-10-03
Pipeline	Nomenclator · Damnatio

Required citation for the taxonomy data:

GBIF Secretariat (2026). GBIF Backbone Taxonomy. Checklist dataset. https://doi.org/10.15468/39omei accessed via GBIF.org

Nothing in these tables is model-generated. Every row is either a GBIF taxon key or a Wikidata QID, and both columns are kept so that any row can be checked against its source.

What this dataset does not say

• A synonym is one checklist's assertion, not a fact about the world. The GBIF Backbone is a synthesis of many checklists. A row in fallen.csv means "this checklist treats this name as a synonym of that one", not "these two taxa are the same organism".
• The dedication table is a lower bound. named after on Wikidata is filled in by hand. It records 16,404 taxa, not every eponym ever published, and there is no reason to think the coverage is representative across kingdoms, countries or centuries.
• The source's blanks are visible. Among the 10,963 records dedicated to a person, 112 carry no gender, 1,247 no citizenship and 1,047 no birth year. 12 records point at an anonymous Wikidata node and have no resolvable honoree. Values that arrive as .well-known/genid URLs are stored as empty rather than as fake data. Those cells are empty rather than guessed.
• fallen.csv is a sample, not the population. The exact synonym counts per kingdom are in fallen_summary.json; the sample is what is stored in full.

Uses it is good for

• Studying commemoration in science: who gets a species named after them, by gender, nationality and era — a small, joinable table for history-of-science and computational-social-science work.
• Joining to GBIF occurrences (gbif.org/species/<gbif_key>) if you want to weight dedications by how well observed a taxon is.
• Teaching data provenance: the card states its own gaps.

It is not a taxonomic authority, and it should not be cited as one.

Rebuild

git clone https://codeberg.org/NoahIsARider/Nomenclator
git clone https://codeberg.org/NoahIsARider/Damnatio
cd Nomenclator && python src/ledger.py && python src/eponyms.py   # the latter crawls Wikidata (~40 min)
cd ../Damnatio && python src/fallen.py --per-branch 3000

The CSVs here are the exported form of data/ledger.json, data/eponyms.jsonl and data/fallen.jsonl from those repositories.

Citation

Nomenclator (2026). Species-rank names and the people they honour: dedications and fallen names from the GBIF Backbone Taxonomy and Wikidata [Data set]. https://huggingface.co/datasets/NoahIsARider/nomenclator

Please also cite GBIF as required by CC BY 4.0 (see Provenance above).