Which ancient remedies
did science never test?
Rhizotomos audits therapeutic claims from Dioscorides' De Materia Medica, the pharmacology reference used for 1,500 years. Each claim is linked to a modern taxon through GBIF, checked against live PubMed evidence and tested against its own negative controls.
Two thousand years of documented use. Hundreds of lab papers. Not one human trial. That gap is where the next clinical study should start.
Four steps from a Greek plant name to a research priority
Every number on this page comes from an open API call that you can rerun. Every formula is in score.py.
Read the claim
Greek name, paraphrased use and two doubt flags: is the plant identification secure, and is the reading of the claim secure?
Resolve the taxon
The GBIF Backbone gives the accepted name, family, match confidence and occurrence records worldwide and in Greece.
Weigh the evidence
PubMed counts records, clinical trials and systematic reviews, then divides by the plant's total literature to cancel popularity bias.
Score the gap
Preclinical signal × lack of trials × certainty. Each plant is compared with its own unrelated use controls.
Every claim, its evidence and its doubts in one view
The bottom right of the map holds claims with plenty of laboratory work and no clinical trial. Click a point or a row to open the full record.
Historical claim vs modern evidence
| ID | Greek | Taxon | Use | Records | Trials | Reviews | Lift | Certainty | Gap | Status |
|---|
The signal holds up against controls, doubt and regulators
A famous plant shows up in many papers whatever you search for. So each plant is compared with itself: the same taxon paired with two unrelated uses drawn with a fixed seed.
What this atlas does not claim
- Literature co-occurrence measures research attention, not efficacy or safety. Trial counts say nothing about trial results.
- GBIF matched every name with 99 to 100% confidence, yet several historical identifications are disputed. Modern taxonomy cannot see that problem, so we record it in the data.
- Some evidence migrated into isolated compounds such as morphine and scopolamine. Plant name queries miss it, so those claims are labelled as translated via compound and are not counted as gaps.
- The seed table paraphrases a public domain text and needs verification against the Wellmann edition. Mandragora, Hyoscyamus, Veratrum and Colchicum are highly toxic. Nothing here is medical advice.