Computational Linguistics & Statistics
9 entries
Stolfi, J. (1997–ongoing). Voynich Manuscript statistics and structural analysis.
Personal research website.
The first systematic quantitative analysis of Voynich token statistics. Vocabulary size, hapax frequency, positional character distributions, and inter-word statistics. The foundational dataset for all subsequent computational work — Stolfi's tables are still cited in papers three decades later.
Foundational
Montemurro, M. A., & Zanette, D. H. (2013). Keywords and Co-Occurrence Patterns in the Voynich Manuscript: An Information-Theoretic Analysis.
PLOS ONE, 8(6), e66344.
DOI: 10.1371/journal.pone.0066344 ↗
Demonstrates that Voynichese shows information-theoretic properties inconsistent with random text and consistent with natural language structure — including section-specific keyword usage suggestive of semantic organisation. Confirms non-random semantic clustering; one of the most technically rigorous results in the field.
Rigorous
Rugg, G. (2004). An Elegant Hoax? A possible solution to the Voynich Manuscript.
Scientific American, 291(1), 104–109.
Demonstrates that a Cardan grille grid device could generate text with Voynich-like surface statistics. Methodologically sound as a proof-of-concept — but only shows such a device could produce similar output, not that it did. Subsequent computational work found the full VMML, Zipf, and semantic-clustering profile not fully replicable by Rugg's mechanism. Essential reading for understanding the hoax debate.
Rigorous
Bax, S. (2014). A proposed partial decipherment of the Voynich script.
Working paper, University of Bedfordshire.
Proposes 14 word identifications by matching Voynich illustrations to labeled images in other medieval manuscripts — including a proposed reading of the word for Taurus. The comparative-iconography method is legitimate in principle (it worked for Linear B), but depends on the unverified assumption that Voynich labels name their illustrations. Not independently replicated; not definitively refuted. Archived at stephenbax.net.
Speculative
Hauer, B., & Kondrak, G. (2017). Decoding Anagrammed Texts Written in an Unknown Language and Script.
Transactions of the Association for Computational Linguistics (TACL).
Applies unsupervised decipherment methods to the possibility that Voynich words are anagrams of natural language words. Finds weak signal consistent with some Semitic languages at token level. Methodologically transparent about limitations.
Rigorous
Parisel, C. (2026). A Quantitative Confirmation of the Currier Language Distinction.
arXiv:2604.25979.
Beta-binomial mixture model applied to the Currier A/B distinction. Achieves 89.2% classification accuracy on folio assignment. Strong independent confirmation of the two-dialect hypothesis originally proposed by Currier (1976). Most rigorous statistical treatment of the A/B question to date.
Rigorous
Silva, F. J. F. da. (2026). BPE Morpheme-Length Clustering Across 55 Writing System Corpora (Paper 7).
Zenodo.
DOI: 10.5281/zenodo.20386119 ↗
First systematic application of BPE VMML to typological classification of the Voynich Manuscript. Establishes the Alphabetic Ceiling, the 55-corpus comparative baseline, and the Voynich Discriminant Zone.
Rigorous
Silva, F. J. F. da. (2026). BPE VMML Cross-Text Instability in Tagalog and the Limits of Single-Corpus Typological Inference (Paper 8).
Zenodo.
DOI: 10.5281/zenodo.20467972 ↗
Adversarial self-test of Paper 7's nearest-language candidate. Demonstrates cross-text VMML instability in Tagalog (Δ=0.336 units, Rizal corpus pair), introduces permutation testing, and expands the Austronesian comparison to 8 languages.
Rigorous
Codicology & Paleography
3 entries
Davis, L. F. (2020). Scribal hands in the Voynich Manuscript: a new assessment.
Manuscript Studies, 5(1).
Identifies multiple distinct scribal hands through systematic paleographic analysis of letter formation and pen-angle patterns. Provides independent evidence for the multi-scribe hypothesis and correlates scribe boundaries with Currier A/B language sections.
Rigorous
Zandbergen, R. (ongoing). The Voynich Manuscript — comprehensive reference site.
voynich.nu.
The single most complete reference resource on the manuscript's history, provenance, physical description, transcription systems, and research bibliography. Maintained with scholarly rigour and updated continuously. Essential reading before publishing any new claims.
Foundational
Kennedy, G., & Churchill, R. (2004). The Voynich Manuscript.
Inner Traditions, Rochester VT.
Popular-audience survey of the manuscript's history and major theories. Useful introduction for general readers; does not advance new research claims.
Accessible
Cryptography & Cipher Hypotheses
3 entries
Currier, P. (1976). New Research on the Voynich Manuscript: Proceedings of a Seminar.
NSA Technical Report (declassified).
First systematic identification of two distinct statistical populations within Voynich text, now known as Currier A and Currier B. The A/B distinction is one of the most reliably reproduced findings in all Voynich research. This document is the starting point for all subsequent statistical work.
Foundational
Schinner, A. (2007). The Voynich manuscript: Evidence of the hoax hypothesis.
Cryptologia, 31(2), 95–107.
Proposes a stochastic generation mechanism that could produce Voynich-like statistical properties without encoding real language. Methodologically rigorous within its scope. Our work tests and partially refutes this model: random EVA generation under Schinner's process does not replicate the full VMML + BC + CBMI profile.
Rigorous
Timm, T., & Schindler, K.-U. (2020). A Voynich-manuscript positional word-stress model.
arXiv:2004.11218.
Proposes a self-citation mechanism based on positional word-stress patterns to explain Voynich's statistical structure. Interesting hypothesis but the EBNF/MI trade-off is irreconcilable with the observed Boundary Concentration profile. Tested and refuted in Paper 7.
Speculative
Decipherment Proposals (with critical assessment)
2 entries
Cheshire, G. (2019). Linguistic Missing Links: Rosetta Stone for the Voynich Manuscript.
Romance Studies — subsequently retracted.
Claimed the manuscript is written in proto-Romance language with Latin-based cipher. The journal retracted the paper following expert criticism of the methodology, linguistic claims, and failure to provide reproducible decipherment. Listed here for transparency and as a case study in insufficient peer review.
Retracted
Naibbe (attributed). (2025). Claims of Voynich Manuscript decipherment.
Media coverage, various outlets.
Multiple media cycles in 2025 reported "Voynich decoded" claims attributed to this source. Tested and refuted in our Paper 3: the proposed decipherment does not reproduce consistent transliterations, contains unfalsifiable interpretive leaps, and is structurally incompatible with BPE analysis. No peer-reviewed preprint was produced.
Speculative
History & Provenance
2 entries
Voynich, W. M. (1921). A Preliminary Sketch of the History of the Roger Bacon Cipher Manuscript.
Transactions of the College of Physicians of Philadelphia, 43.
Voynich's own account of acquiring the manuscript and his hypothesis attributing it to Roger Bacon. A primary historical document. The Roger Bacon attribution is now considered unsupported, but this text remains essential for understanding the early provenance chain.
Foundational
Zandbergen, R., & Boser, R. (ongoing). Voynich Manuscript provenance chain research.
voynich.nu — provenance section.
The most complete reconstruction of the manuscript's documented ownership history, tracing it from the Holy Roman Emperor Rudolf II through Georg Baresch, Johannes Marcus Marci, Athanasius Kircher, and the Collegium Romanum to Wilfrid Voynich. Indispensable for any work touching on origin or dating.
Rigorous
Software & Digital Tools
4 entries
Zandbergen, R., & Landini, G. (1998). EVA (European Voynich Alphabet). Transcription standard enabling computational analysis.
Available at voynich.nu.
The standard ASCII-compatible notation system for Voynich glyphs — the precondition for all computational research. Before EVA, each research group used incompatible transcription conventions, making cross-team comparison nearly impossible. EVA assigns short alphabetic codes to each glyph, enabling shared datasets, statistical analysis, and reproducible results. Still the dominant system for computational Voynich research more than 25 years after publication.
Foundational
Takahashi, T. (1998). IVTFF (Interlinear Voynich Transliteration File Format). Standard for aligned multi-hand transcriptions.
Transcription format specification.
The file format that allows multiple independent transcriptions of the same Voynich folio to be stored in alignment — enabling inter-transcriber agreement statistics and multi-hand comparison. Essential infrastructure for any serious corpus work; IVTFF files are the primary data format used by the Voynich.nu community and most published computational studies.
Foundational
DG97EEB. (2026). Voynich Whispers Archive Search Engine.
voynich-whispers-archive.lovable.app
voynich-whispers-archive.lovable.app ↗
Searchable database of Voynich Mailing List and forum discussions spanning decades of community research. Useful for tracing the provenance of specific claims, finding prior art for proposed hypotheses, and recovering discussions that predate modern indexing. An underused research tool — many "new" claims have been raised and addressed before.
Useful
Pelling, N. Cipher Mysteries blog.
ciphermysteries.com
ciphermysteries.com ↗
Active research blog maintained by Nick Pelling, author of The Curse of the Voynich (2006). Tracks current decipherment claims, media coverage, and new papers with critical commentary. One of the longest-running independent Voynich blogs; useful for monitoring what claims are circulating and what the community response has been. Pelling is skeptical and methodologically careful by blog standards.
Useful
On critical assessments. The labels Rigorous, Foundational, Speculative, Accessible, Useful, and Retracted are editorial judgments by the Voynich Lucidity team, based on peer-review status, methodological transparency, reproducibility, and community reception. They are not verdicts — they are reading guides. We encourage all readers to consult primary sources and form independent judgments. If you believe an assessment is incorrect, write to us: contact@voynichlucidity.com.