Structural Analysis of the Voynich Manuscript — Elucidating a Knowledge System through Statistical Methods

We present a computational statistical analysis of the Voynich Manuscript (VMS). The text is shown to be neither random nor a simple cipher, but an intentionally designed information-encoding system (Pareto-optimal transition table, Zipf's law R²≥0.94, Poissonian burstiness B≈0). We identify a four-layer descriptive system (character, morpheme, syntax, scribal dialect) and a pharmaceutical grammar including the sh→ch material-transformation process, a triangular processing cycle (edy→eedy→ain), a Dioscoridean five-attribute system (ct-prefix), and an ok→ch universal grammar (53–78% across all sections). The generative model achieves dist=0.0143. Structural-level reading was completed for 90.3% of all words; content-level decipherment remains unachieved.