For centuries, people across South and Southeast Asia wrote religious texts, laws, literature and science on palm leaves. Before anyone picked up a stylus, the leaves went through long treatment, and the methods changed from region to region. Those steps left marks in the leaf itself. Today, tracing how a manuscript was made or where it came from is hard.
An international team decided to study the leaf material directly. They ran X-ray measurements at the PETRA III source at DESY in Hamburg and at the European Synchrotron Radiation Facility in Grenoble, France, then fed the data to machine learning software. The work appears in the journal npj Heritage Science.
The team studied 160 samples. Of those, 85 were historical manuscripts and 75 were modern reference materials from collections across the region. The researchers wanted to know if material traits alone could tell two palm genera apart, show how far a leaf had been processed, and point to where a manuscript was made.
The method leaves every object untouched. Nobody has to cut a sample from a fragile manuscript. Three X-ray techniques gave three views. Small-angle scattering shows how the fiber network is built and which way the fibers point. Wide-angle scattering measures how much of the leaf is crystalline cellulose and how much is softer material. X-ray fluorescence maps the chemical elements spread across the surface. Laura Gallardo, first author at Hamburg University of Technology, says combining the methods lets researchers read several layers of a manuscript’s history in a single experiment.
Machine learning weighed all the measurements together instead of leaning on one number. The software sorted the processing stage with high reliability. Telling the two palm genera apart worked too, but the split was sharper in processed leaves than in raw ones. Treatment pushes the two kinds of palm further apart, so the readings reflect botanical type and also how each palm responded to the work done on it.
Place of origin was harder to pin down. Samples from Sri Lanka and southern India looked alike, which points to related processing traditions. The Thai samples stood out with a profile of their own, likely tied to local methods. Still, the data don’t name a manuscript’s origin with certainty. Some regions had few samples, and the team only partly measured how age and storage change the material. Patrick Huber, who directs the materials institute at TU Hamburg, says the findings need to be interpreted alongside historical and codicological evidence.
Manuscript scholars see value in this work. Giovanni Ciotti of the University of Bologna says physical traces show links between objects. Script, form and catalog records miss those links. The traces gain meaning once scholars match them with historical evidence, for example when they ask whether two manuscripts share a making tradition or a route of knowledge transfer.
The approach scales to large collections without harming a single object. Paired with historical records, the results give scholars a new way to study how palm-leaf manuscripts were made, treated and used, and how traditions now extinct changed over time across South and Southeast Asia.