Tested K&A at scale, found something, need your help

[CorwinFr]

My earlier phonetic decryption contained a methodological bias (Perseus validation too permissive, ingredient correlation was Zipf noise at p=0.92). A corrected V2 is on Zenodo.

I've since taken a completely different approach, bottom-up structural analysis without assuming any encoding. The main finding: herbal plant codes are systematically embedded inside pharmaceutical compound words, linking the two sections structurally.