A team of computational linguists at the Journal of Archaeological Science: Reports has utilized a new neural network architecture to decipher a collection of fragmentary birch-bark scrolls written in the rare Bhaiksuki script. The scrolls, recovered from a high-altitude site in the Himalayas, were found to contain a previously unknown Sanskrit work titled Akasha-Lakshana. The text provides a systematic classification of atmospheric optical phenomena, including detailed descriptions of sun dogs, circumzenithal arcs, and the diffraction of light through ice crystals.
The manuscript is significant for its early use of wave-like terminology to describe the behavior of light, predating similar observations in other civilizations. The Bhaiksuki script, often associated with Buddhist logic schools, appears here in a secular scientific context, suggesting that the script was used for a wider range of technical disciplines than previously believed. The discovery provides a new window into the theoretical physics of the 7th century and the sophisticated understanding of optics maintained by monastic scholars in the borderlands of India and Tibet.