Subtle Signs of Scribal Intent in the Voynich Manuscript

TL;DR

This study investigates whether the Voynich Manuscript’s Voynichese encodes meaningful content or is a visually convincing artifact. It analyzes token-length distributions and token-position propensities relative to intrinsic script structure (lines/paragraphs) and extrinsic features (drawing intrusions) using the Zandbergen-Landini transliteration on a carefully curated Herbal corpus from Scribe 1. The results reveal significant differences in token lengths and usage patterns near drawings and at line boundaries, and identify tokens with strong positional propensities, suggesting scribal constraints linked to extrinsic features. These findings open new avenues for decipherment by considering how layout and drawings may influence token choices, while acknowledging that interpretations remain speculative and require further validation. The work also provides a catalog of positionally influenced tokens and a framework for assessing scribal intent through statistical analysis with $p$-value thresholds of $0.01$ and $\log(B) \ge 5$.

Abstract

This study explores the cryptic Voynich Manuscript, by looking for subtle signs of scribal intent hidden in overlooked features of the "Voynichese" script. The findings indicate that distributions of tokens within paragraphs vary significantly based on positions defined not only by elements intrinsic to the script such as paragraph and line boundaries but also by extrinsic elements, namely the hand-drawn illustrations of plants.

Figures (6)

• Figure 1: Voynichese Script on Folio f36r. The script exhibits a noticeable conformity to the outline shapes of the drawn illustrations.
• Figure 2: Folio Page Count by Scribe and Illustration Type
• Figure 3: Venn Diagram of Cohorts. Note that the reference and subject cohorts (shown with solid borders) are all mutually exclusive.
• Figure 4: Schematic of Cohort Tokens. Each diagram shows a hypothetical folio page with two paragraphs of tokens, the second paragraph conforming around a drawing. The shaded tokens indicate cohort members.
• Figure 5: $\chi^2$ Statistical Significance Matrix. Lighter cells indicate that the pair of cohorts that are most probably governed by different causal phenomena resulting in observed differences between their distributions of token lengths.