Statistical Validation of Structure in MS 408: Independent Confirmation of the Jaccard Anomaly

Independent AI-driven statistical analysis of the Voynich Manuscript (MS 408), confirming the existence of the "Jaccard Anomaly" (extreme line-to-line vocabulary independence).

This study provides reproducible quantitative evidence (J = 0.0226, Entropy = 3.87 bits) for non-natural language structure, validating the statistical observations of S.A. Quevedo (2026) while maintaining neutrality on specific mechanical reconstructions.

Key Findings:

• Jaccard Index Anomaly: The manuscript exhibits a median line-to-line Jaccard Index of 0.00, confirming zero contextual memory between lines—a statistical impossibility for natural narrative prose.
• Vocabulary Explosion Paradox: A small character set (28) generates an disproportionately massive vocabulary (8,381 words), suggesting a combinatorial generation engine.
• Morphological Selectors: "Gallow" characters ([P], [F], [T], [K]) demonstrate statistically distinct following-word distributions, supporting a mechanical "mode-switching" hypothesis.

AI Attribution & Transparency Statement:
This research represents a "Hybrid Intelligence" collaboration. While the conceptual direction and responsibility lie with the human principal investigator, the statistical auditing, code generation, and morphological analysis were executed by autonomous AI agents. This project serves as a demonstration of ethical AI transparency and the capability of agentic workflows in forensic discovery. 

Research orchestrated by AI Kapitein @ www.zakenoporde.com