Research materials

The measured characterization of the text: entropy, section specificity, slot structure, what is excluded and what remains open.

Why a baseline is needed, how the Champollion-style cross-constraint works, the validation of the tool.

Reading the book as the knowledge package of an early-fifteenth-century physician-apothecary.

A catalogue of the Latin abbreviations from which the signs graphically derive.

Statistical comparison with a baseline between the Voynich and the oldest Bibles (Greek Codex Sinaiticus, Latin Vulgate): no correspondence beyond the baseline; the Voynich's conditional entropy is too low for a simple cipher.

The full transcription, about 37,000 word-tokens, on which the validator and the analyses run.

The manuscript's alphabet with frequencies.

The (Voynich word, hypothesized meaning) pairs preloaded in the validator.

The labels beside the signs, compared with star names and lunar mansions.

A facing-text edition, for line-by-line consultation.

The volume gathers into a single PDF the archive materials: monograph, state-of-the-art manual, register, library, translated seventeenth-century letters, appendices and source code. The individual parts are available here in the archive; the full volume (130 MB) is available on request.

The main dossier: the Cappelli cipher as a QUBO, the QAOA pipeline, the Born machine, entanglement entropy. Why quantum does not decipher, and why that is a result.

Entanglement entropy of the text (Schmidt spectrum) across EVA, Currier and Cappelli, with Latin and baseline. The Voynich's long-range structure decays within ~12 symbols.

The full manuscript (226 folios): the gradient of 'language-likeness' across sections, prose vs labels, and the constructive Naibbe and Alberti proofs.

What our scripts compute: the text's entanglement entropy (Schmidt spectrum); the Cappelli cipher as a QUBO, solved with QAOA on Qiskit and with a Born machine; the constructive Naibbe and Alberti-disk proofs; the Kondrak-Hauer test with the dictionary curve; the reconstruction of the Llullian combinatorial wheels; the data-driven segmentation of the sections. Every measure is compared with chance (baseline).

The reference lexicon grew from 5,518 to 158,200 lemmas: the real Cappelli (10,500 abbreviations → Latin expansions, Ad Fontes UZH, public domain) and Whitaker's Words (31,707 lemmas). Third-party resources catalogued, not redistributed: download them from the original projects.

mappa_totale_v2.py, estrai_agganci_cappelli.py, validatore_cappelli_formale.py, sistema_segnale.py. Honest outcome: morphological signals B (core|suffix entropy), C (8.4% Latin vs 3.0% baseline) and F (11.1% documented-medical vs 3.3%) confirmed; signals A/D/E/G (section specificity, anchored labels, opener TTR, grammar) absent. No decipherment: a morphological signature, not meaning.

Source: the Carnic German vocabulary of Timau 'Bartarpuach' (a third-party lexicographic work), used as a touchstone for the Alpine-languages hypothesis. Catalogued as a reference, not redistributed.

Greek texts used as a baseline for the Byzantine-Greek hypothesis (medical and alchemical roots). Third-party / public-domain source: catalogued, not redistributed.

Source: plant names from the 'Gart der Gesundheit' (Mainz ~1485, public domain), for comparison with the botanical section. Catalogued as a reference.