The Voynich Manuscript's Positional Entropy Collapse: Comparative Evidence from Languages and Ciphers

This is a preprint of a manuscript currently under peer review at Cryptologia. The paper presents a leakage-free, character-level structural analysis of the Voynich Manuscript, comparing positional entropy, n-gram cross-entropy, and local edit-distance overlap against natural languages and a constructed verbose substitution cipher. The analysis employs strict recto–verso train–test separation, transcription-layer isolation, and identical preprocessing across corpora. The version deposited here includes author identification. A blinded version was submitted for peer review. Figures associated with the manuscript are provided as supplementary files. v1.01 corrects a normalization inconsistency in the bits-per-character (BPC) computation used for the reported BPC2/BPC3 values and the derived ΔBPC (BPC2 − BPC3). Specifically, the implementation included the log-probability contribution of the end-of-word symbol in the cross-entropy sum, but did not include that symbol in the character count used for normalization. This results in a uniform, mechanically predictable inflation of the reported BPC-based values that depends on average token length. The issue affects only the absolute scale of BPC and ΔBPC. It does not affect relative comparisons between corpora, legality rejection rates, positional entropy patterns, train/test partitioning, or the qualitative conclusions of the manuscript.