A systematic methodology to assess the identity of plants in historical texts: A case study based on the Byzantine pharmacy text John the Physician’s Therapeutics

Ethnopharmacological relevance
In recent decades, the study of historical texts has attracted research interest, particularly in ethnopharmacology. All studies of the materia medica cited in ancient and medieval texts share a concern, however, as to the reliability of modern identifications of these substances. Previous studies of European or Mediterranean texts relied mostly on authoritative dictionaries or glossaries providing botanical identities for the historical plant names in question. Several identities they suggest, however, are questionable and real possibility of error exists.

Aim of the study
This study aims to develop and document a novel and interdisciplinary methodology providing more objective assessment of the identity of the plants (and minerals) described in these resources.

Materials and methods
We developed an iterative experimental approach, using the 13th century Byzantine recipe text John the Physician's Therapeutics in its Commentary version (JC) as a case study. The methodology has six stages and relies on comparative analyses including statistical evaluation of botanical descriptions and information about medicinal uses drawn from both historical and modern sources. Stages 1–4 create the dataset, stage 5 derives the primary outcomes to be reviewed by experts in stage 6.

Results
Using Disocorides’ De Materia Medica (DMM) (1st century CE) as the culturally related reference text for the botanical descriptions of the plants cited in JC, allowed us to link the 194 plants used medicinally in JC with 252 plants cited in DMM. Our test sample for subsequent analyses consisted of the 50 JC plant names (corresponding to 61 DMM plants) for which DMM holds rich morphological information, and the 130 candidate species which have been suggested in the literature as potential botanical identities of those 50 JC plant names. Statistical evaluation of the comparative analyses revealed that in the majority of the cases, our method detected the candidate species having a higher likelihood of being the correct attribution from among the pool of suggested candidates. Final assessment and revision provided a list of the challenges associated with applying our methodology more widely and recommendations on how to address these issues.

Conclusions
We offer this multidisciplinary approach to more evidence-based assessment of the identity of plants in historical texts providing a measure of confidence for each suggested identity. Despite the experimental nature of our methodology and its limitations, its application allowed us to draw conclusions about the validity of suggested candidate plants as well as to distinguish between alternative candidates of the same historical plant name. Fully documenting the methodology facilitates its application to historical texts of any kind of cultural or linguistic background.