Historical use of grasses in South Indian palm-leaf manuscript production reconstructed from phytolith analysis

We collected thirty fallen off fragments of palm-leaf manuscripts (PLM), from four manuscript collections. Aiming to identify grasses used in the 18-19 centuries for PLM preparation, we analysed these samples for grass silica short cell phytoliths (GSSCP). Due to the complex, diverse phytolith sources and limited literature, we first examined reference grass material from herbaria collected in South India and SE Asia with comparable climates. Material from thirty grass species across six subfamilies was studied. In total, 36 GSSCP phytolith morphotypes were described; subfamily-specific assemblages were identified. All individual morphotypes were analysed for positive and negative co-occurrence patterns, which were then used in the interpretation of the results of phytolith analysis of the PLM samples. We revealed 28 GSSCPs in PLM samples. Preservation of the phytolith material varied from sufficient to low. Many cells demonstrated abnormal morphology, and some GSSCPs were forced into the surface of the palm leaves, mechanically broken, half-brushed off and eroded, providing evidence of active mechanical forces applied to the surface of PLMs during their production. In PLM samples provenancing from Kerala, Arundo donax , Phragmites sp., Bambuseae, and Chloridoideae (most likely Desmostachya bipinnata ) were revealed. In PLM samples from Tamil Nadu, Phragmites sp., domestic rice (possibly with admixture of Arundinoideae and/or Chloridoideae), and Panicoideae material were evidenced. Evidence of Aristida sp. is most likely attributed to historical contamination. The main difference in the PLM samples reflects the amount of preserved material historically accumulated on the folio surfaces as well as the original provenance of the PLMs. • Palm-leaf manuscript samples show regional and subregional difference in phytoliths. • Differences in grass phytolith diversity depends on the geographical provenance. • Samples of manuscripts with closer geographical proximity demonstrate more similarity. • Abnormal phytolith morphologies and mechanically broken phytoliths are regularly observed in the manuscript samples. • Mechanically broken, driven into the surface grass phytoliths evidence brushing. • Diversity of the phytoliths is driven by the available amount of preserved material. • Practices of manuscript brushing should be revised to conserve historical material.