Research session · Voynich Manuscript

Session 6: Phase 2: registered tests, brute force, five surveys; book v1.1

A pre-registration framework with surrogates, a 103-language reference, planted-key controls and a held-out split; nineteen hypotheses tested, including substitution and homophonic searches over 187 language samples; five methods surveys (cryptanalysis, linguistics, bioinformatics and signal, imagery, emerging methods). Book v1.1 added Part V.

Session 622 Oct 2026Test0

What it left open

{'fate': 'Built in session 7 (fragmenting-cipher solver).', 'text': 'The joint segmentation solver (the one search not made)'}
{'fate': 'Done in session 7.', 'text': 'Acquire v101/GC2a for the transliteration check'}
{'fate': 'Done in session 7.', 'text': 'List-type reference texts'}

Questions raised 2

Is it a cipher of a known language?

Q-012

QST

Is it a list, index or shorthand?

Q-015

QST

Answers proposed 20

A known language under a homophonic cipher

H-CIPHER-HOMOPHONE

ANS

A natural language spelled letter by letter

H-LANG-ANY

ANS

Common words fall into distributional classes

H-STRUCT-WORDCLASS

ANS

Each word encodes one letter (Naibbe-type)

H-CIPHER-WORDLETTER

ANS

Each word is a set of letters written out in one fixed canonical order (an anagram cipher…

H-CIPHER-ANAGRAM

ANS

Formulaic repeated phrases, as in recipes

H-STRUCT-PHRASES

ANS

Generated by copying and modifying nearby words

H-MEANINGLESS-AUTOCOPY

ANS

One of 103 languages under simple substitution

H-LANG-SUBST

ANS

One or two glyphs are nulls

H-CIPHER-NULLS

ANS

Paint was applied in campaigns organised by section/quire rather than by scribe.

H-IMAGE-PAINT

ANS

Predictive information reaches beyond the word: a model that sees more of the preceding…

H-STRUCT-CONTEXT

ANS

Repeated multi-word phrases occur as in natural-language prose (test T4d).

H-STRUCT-REPEAT

ANS

Text written in rings round diagrams, and radial text, is written in the same register as…

H-STRUCT-RING

ANS

The glyph stream has been transposed with a fixed period (columnar, route or decimation),…

H-CIPHER-TRANSPOSITION

ANS

The line is a unit of the system

H-STRUCT-LINE

ANS

The present binding order preserves the original sequence of sheets.

H-ORDER-BINDING

ANS

The sheet is the unit of composition

H-ORDER-SHEET

ANS

The substitution changes with position in the line (a line-keyed or position-keyed…

H-CIPHER-POLYALPHABETIC

ANS

Whatever the symbols are, the word sequence carries information the way a language's…

H-LANG-WORDLEVEL

ANS

Zodiac labels are unique names of degrees or stars.

H-IMAGE-ZODIAC-LABELS

ANS

Evidence found 11

A three-state HMM recovers onset, nucleus and coda glyph classes.

F-019

EVD

Homophonic substitution of five languages fails

F-021

EVD

Nearest languages: CV-syllable and agglutinative

F-020

EVD

No one-to-one substitution of a known language fits

F-013

EVD

No repeated four-word phrases

F-016

EVD

Paint campaigns do not follow quires, scribes or languages (herbal green coverage).

F-023

EVD

Ring text has the word-initial profile of labels, not paragraphs.

F-018

EVD

The sheet, not the page, is the unit of composition

F-014

EVD

Word order carries almost no information

F-017

EVD

Word-as-letter, periodic transposition, position-keyed alphabets, single nulls and sorted…

F-022

EVD

Words spelled alike are used alike

F-015

EVD

Runs 23

Canonical-order (sorted anagram) test: share of words monotone under the best single ranking of units

RUN-013

RUN

Entropy law and word-order gain: bits per word saved by word order (random word ids)

RUN-028

RUN

Entropy law and word-order gain: residual from the H1 vs entropy-rate law (word level), sd units

RUN-027

RUN

Entropy law and word-order gain: residual from the H1 vs entropy-rate law (word level), sd units

RUN-017

RUN

Form-function coupling (spelling vs context similarity): Spearman(form sim, context sim) top-300 words

RUN-008

RUN

Form-function coupling (spelling vs context similarity): share of 5 nearest context-neighbours within edit distance 1

RUN-009

RUN

Held-out context scaling (Kneser-Ney orders 1-8): held-out bits/glyph at order 1,2,4,6,8

RUN-018

RUN

Homophonic solver with planted-key controls and permutation null: held-out gap (own minus decoded), nats/char, best of 19 configurations

RUN-019

RUN

Homophonic solver with planted-key controls and permutation null: held-out gap, 300 words as units

RUN-020

RUN

Homophonic solver with planted-key controls and permutation null: held-out gap, six configurations re-run with the stronger frequency…

RUN-054

RUN

Homophonic solver with planted-key controls and permutation null: homophonic solver recovers planted 80-unit Latin key, no spaces

RUN-010

RUN

Null-glyph deletion: max h2 after deleting one glyph unit

RUN-012

RUN

RUN-005

RUN

RUN-004

RUN

RUN-003

RUN

Periodicity, transposition and isomorph battery: coincidence between consecutive lines at same position

RUN-015

RUN

Periodicity, transposition and isomorph battery: glyph-disjoint isomorph pairs at same line position / anywhere

RUN-016

RUN

Periodicity, transposition and isomorph battery: period-bigram peak (period, count) in text vs within-word shuffle

RUN-014

RUN

Profile screen against 103 languages: h2 vs 103 languages (min)

RUN-007

RUN

Profile screen against 103 languages: statistics outside attested range

RUN-001

RUN

RUN-011

RUN

Sequence structure: MI decay, compressibility, BPE curve, repeated phrases: repeated 4-grams real/shuffled

RUN-002

RUN

Sequence structure: MI decay, compressibility, BPE curve, repeated phrases: repeated word 4-grams

RUN-006

RUN

Methods used 15

Canonical-order (sorted anagram) test

M-T10

MTH

Entropy law and word-order gain

M-T12

MTH

Form-function coupling (spelling vs context similarity)

M-T6

MTH

Held-out context scaling (Kneser-Ney orders 1-8)

M-T13

MTH

Hidden Markov model of glyph classes

M-T7

MTH

Homophonic solver with planted-key controls and permutation null

M-T3

MTH

Null-glyph deletion

M-T9

MTH

One-to-one substitution over 187 samples

M-T2

MTH

M-T5

MTH

Periodicity, transposition and isomorph battery

M-T11

MTH

Pigment map from the page images

M-I1

MTH

Profile screen against 103 languages

M-T1

MTH

M-T8

MTH

Sequence structure: MI decay, compressibility, BPE curve, repeated phrases

M-T4

MTH

Transliteration dependence (GC v101)

M-T18

MTH

Datasets used 1

Zandbergen-Landini transliteration ZL3b (EVA, IVTFF 2.0, 13 May 2025)

DAT

Sources consulted 275

A Bayesian framework for word segmentation: Exploring the effects of context

SRC

A Computational Evaluation of Syllabic Hypotheses for Rongorongo: Evidence from N-gram Analysis

SRC

A Generative Grammar Underlying the Voynich Manuscript, the Pastiche Hypothesis: Evidence from Large Language Models

SRC

A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space

SRC

A Massive Machine-Learning Approach For Classical Cipher Type Detection Using Feature Engineering

SRC

A Methodology for the Cryptanalysis of Classical Ciphers with Search Metaheuristics

SRC

A Mutual Information-based Approach to Quantifying Logography in Japanese and Sumerian

SRC

A Note on Zipf's Law, Natural Languages, and Noncoding DNA regions

SRC

A Preliminary Analysis of the Botany, Zoology, and Mineralogy of the Voynich Manuscript

SRC

A Statistical Comparison of Written Language and Nonlinguistic Symbol Systems

SRC

A Statistical Model for Lost Language Decipherment

SRC

A Zipf-preserving, long-range correlated surrogate for written language and other symbolic sequences

SRC

A massively parallel corpus: the Bible in 100 languages

SRC

A miscellany of nine rosette links

SRC

A new Transliteration Alphabet brings new Evidence of Word Structure and Multiple "languages" in the Voynich Manuscript

SRC

A possible generating algorithm of the Voynich manuscript

SRC

A prefix-midfix-suffix decomposition of Voynichese words; the core–mantle–crust word paradigm

SRC

A proposed partial decoding of the Voynich script

SRC

A sticky HDP-HMM with application to speaker diarization

SRC

A superset of transliteration alphabets

SRC

A tailored Handwritten-Text-Recognition System for Medieval Latin

SRC

A tool for empowering Symbol Detection through Technological Integration in Library Science. A case study on the Voynich manuscript

SRC

Adaptor Grammars: A Framework for Specifying Compositional Nonparametric Bayesian Models

SRC

Allographic Variations and Statistical Analysis of the Rongorongo Script

SRC

An Analysis of the Relationship between Words within the Voynich Manuscript

SRC

An elegant hoax? A possible solution to the Voynich manuscript

SRC

Ancient Symbols, Computational Linguistics, and the Reviewing Practices of the General Science Journals

SRC

Anti Dependency Distance Minimization in Short Sequences. A Graph Theoretic Approach

SRC

Applications of natural language processing to archaeological decipherment: a survey of Proto-Elamite

SRC

Attention-Augmented LSTMs for Automatic Homophonic Ciphertext Decipherment

SRC

Bayesian Inference for Zodiac and Other Homophonic Ciphers

SRC

Bayesian Online Changepoint Detection

SRC

Beam Search for Solving Substitution Ciphers

SRC

Benchmarking Large Language Models for Cryptanalysis and Side-Channel Vulnerabilities

SRC

Beyond Linguistic Decipherment: A Structural Analysis of the Voynich Manuscript as a Diagrammatic Information System (DIS)

SRC

Beyond Word Frequency: Bursts, Lulls, and Scaling in the Temporal Distributions of Words

SRC

Botanical observations on the Voynich MS

SRC

Breaking Homophonic Substitution Ciphers with CrypTool 2 (workshop handout)

SRC

CD2a-n.txt: Currier-D'Imperio transliteration (IVTFF 2.0)

SRC

Can Sequence-to-Sequence Models Crack Substitution Ciphers?

SRC

and 235 more

What it produced 6

Methods survey: bioinformatics and signal processing

SRC

Methods survey: cryptanalysis

SRC

Methods survey: emerging mathematics and machine learning

SRC

Methods survey: imagery and symbology

SRC

Methods survey: linguistics and datasets

SRC

Reading Without a Key

SRC

Cite it

Mysteries Department. (2026). Session 6: Phase 2: registered tests, brute force, five surveys; book v1.1 [Research session SES0D8C46P04BVW]. https://mysteriesdept.com/r/SES0D8C46P04BVW

@misc{mysteriesdept_ses0d8c46p04bvw,
  title     = {Session 6: Phase 2: registered tests, brute force, five surveys; book v1.1},
  author    = {{Mysteries Department}},
  year      = {2026},
  note      = {Research session, catalog key SES0D8C46P04BVW},
  url       = {https://mysteriesdept.com/r/SES0D8C46P04BVW}
}