How the Voynich Manuscript Was Generated

This deposit accompanies the preprint “How the Voynich Manuscript Was Generated”, which proposes and tests a ledger-driven, adjacency-constrained writing process as a sufficient explanation for the Voynich Manuscript’s local structural properties. The central claim is narrow and mechanistic. The manuscript’s observable structure can be accounted for by a process that records which character-to-character transitions have occurred, reuses permitted transitions heavily, and admits new transitions conservatively. Tokens arise as walks through this evolving adjacency ledger. No assumptions are made about semantics, grammar, ciphering, or meaning. The deposit includes reproducible analyses demonstrating: construction of a directed adjacency ledger seeded from folio F1R, monotonic accumulation of permitted character transitions across pages, rapid early saturation of adjacency legality followed by dominance of reuse, and page-order independence, shown by randomized ingestion of the first 100 pages yielding identical terminal ledger states across independent shuffles. Cross-transcription verification is reported using both the Takahashi (TTLI) and Zandbergen–Landini (EVA) transcriptions, showing that the observed behavior is not an artifact of a single transcription system. This archive is intended as a provenance and reproducibility record. It contains the scripts, logs, CSV artifacts, and cryptographic hashes required to reproduce every table and figure in the associated manuscript from first principles. A full generative implementation based on the same ledger mechanism is under active development and will be reported separately. This work reframes the Voynich Manuscript as the output of a constrained generative process rather than an encoded message, and establishes a falsifiable baseline that any competing generative or linguistic explanation must satisfy.