Computational Analysis and Two-Tier Encoding of the Voynich Manuscript: Code, Data, and Papers

This repository contains the complete computational analysis of the Voynich Manuscript (Beinecke MS 408), comprising two research papers, 48 Jupyter notebooks, corpus data, and all experimental results. Paper 1: Computational Analysis of the Voynich Manuscript develops a four-stage pipeline—statistical fingerprinting, generative model comparison, syllabic segmentation, and Bayesian decipherment—applied to the STA1 2.0 corpus (37,087 words, 166 glyph types). Nineteen decipherment models against five Semitic languages all converge on the same structural ceiling (~68% n-gram realism, 3–5 root matches). No model produces readable text. Paper 2: Two-Tier Encoding in the Voynich Manuscript reverse-engineers the encoding mechanism, discovering that 48.3% of words are functional units (Tier 1) while 51.7% are assembled combinatorially from positionally constrained onset–coda pairs (Tier 2), with only 6.7% utilisation of the theoretical combinatorial space. The manuscript is consistent with a 48-kilobyte encyclopaedic code whose index has been lost. Contents: papers/— PDF versions of both papers supplementary/— Appendix A (Conceptual Bridge glossary) and Appendix B (Causal Funnel) latex/ — Full LaTeX source for both papers notebooks/ — 48 Jupyter notebooks with all analyses corpus/ — STA1 2.0 and EVA transcriptions results/— 43 JSON result files from all experiments images/ — 112 generated figures