The libraries, transcriptions, scholars and community whose work made the substrate possible
This research record consumes a great deal of upstream work: the digitization labour of the
holding libraries, the transcription corpora maintained by the Voynich research community, the
typological and codicological scholarship done by named researchers over decades, the open
datasets and software that make any of it tractable. Attribution belongs on the record. This
page is where it goes; it is meant to be complete and to grow as new sources are drawn on.
If credit is missing or wrong here, the substrate’s author wants to fix it — reach
out via the contact below.
Compiled2025-08-27–2026-05-23
Substrate manuscripts599 catalogued
Named figures210
Citations rowed6,635
§ 01
Holding libraries & digital collections
The institutions whose digitized manuscripts the substrate reads against.
Image rights remain with the holding library; reproduction here is under each library’s
digital-collections terms.
Institution
Mss drawn on
Digital collection
Beinecke Rare Book & Manuscript Library, Yale University, New Haven — holding institution for MS 408 and the bedrock digital surrogate this work rests on
Universitätsbibliothek Freiburg, Landesbibliothek Coburg, Biblioteka Uniwersytecka Wrocław, Johns Hopkins University
1 each
via individual digital collections
MS 408 imagery on this site is reproduced from Beinecke’s public digitized manifest under the
Beinecke’s reuse terms. Comparator imagery is reproduced under each holding library’s
digital-collections terms; where a library requires explicit per-image attribution, that
attribution is carried on the relevant page.
§ 02
Transcription corpora & encoding
The substrate’s textual layer is not its own transcription. Every Voynichese
token used here comes from corpora maintained by the Voynich research community over decades.
IVTFF
Interlinear Voynich Transliteration File Format — the consolidated multi-transcriber corpus maintained by René Zandbergen and Gabriel Landini. The substrate’s folio-level token map (ZL_ivtff_2b.txt) is the IVTFF release used as the canonical transcription. voynich.nu/transcr.html
EVA
European Voynich Alphabet — the dominant ASCII transliteration scheme. Originally developed by René Zandbergen and Gabriel Landini in the mid-1990s.
Takahashi transcription
Takeshi Takahashi’s line-level transcription, one of the foundational transcribers folded into IVTFF.
Currier transcription
Capt. Prescott Currier’s transcription and the “Currier A / B” hand-distinction that underpins many later analyses.
FSG (First Study Group)
The 1944–46 First Study Group transcription, the earliest mechanized rendering of MS 408 into a tabulated alphabet.
Reeds transcription
Jim Reeds’s transcription work, additionally folded into the modern IVTFF corpus.
§ 03
The Voynich research community
Decades of independent and academic work on MS 408 form the substrate’s
load-bearing prior. The list below is incomplete by construction — it names the
researchers, resources and communities the substrate explicitly draws on. The omissions
are work to do, not statements of credit.
Resources & communities
voynich.nu — René Zandbergen’s long-running canonical reference site. voynich.nu
Voynich Mailing List (VML) — the standing community forum since the mid-1990s.
IVTFF / EVA tooling community — the transcription-tooling work that makes corpus analysis possible.
Named researchers (substrate-cited)
The substrate cites and reads against a long list of named researchers. Among those whose work the substrate has explicitly touched and reads from:
René Zandbergen · Gabriel Landini · Mary D’Imperio ·
Prescott Currier · Jim Reeds · Jorge Stolfi · Takeshi Takahashi ·
Lisa Fagin Davis · Stephen Bax · Claire Bowern · Luke Lindemann ·
Keagan Brewer · Michelle Lewis ·
Cary Rapaport · Marco Ponzi · Nick Pelling ·
the Beinecke digitization team ·
and the broader scholarly + community contributors whose individual citations appear on
the live substrate.
“The point was to get people past the graveyard.”
— Ed Honeycutt, on what this work is for. 2026-05-23.
The full citation chain for any single substrate finding lives on its substrate record —
every finding traces to a primary source; the substrate behind it is an internal research surface (invite-only).
§ 04
Contemporary scholarship the substrate reads against
2015–2025 peer-reviewed and scholarly work the substrate engages with
directly — whether by extending a result, citing it as comparator stratum, or testing
it and finding it broken. Grouped by lane.
Cipher & operator-grammar lane
Davis, Lisa Fagin (2020). “How Many Glyphs and How Many Scribes? Digital Paleography and the Voynich Manuscript.” Manuscript Studies: A Journal of the Schoenberg Institute for Manuscript Studies 5(1):164–180. — digital-paleographic attribution concluding that MS 408 was written by five distinct scribal hands; the substrate records its own reframing as a content-lane signature, F-DAVIS-5HANDS (Observed). Davis’s 2024 multispectral-image margin work, January 2025 Voynich Codicology piece, 2025 talk “The Materiality of the Voynich Manuscript,” and 2026 LSA paper with Layfield (below) continue the line. Engaged with critically by Nick Pelling at Cipher Mysteries (2021 “What mistakes did Scribe 2 never make?”; 2024 “f1r: Five Scribes, five vibes?”); both readings remain open.
Layfield, Colin & Davis, Lisa Fagin (2026). “The Application of Latent Semantic Analysis to the Voynich Manuscript.” Digital Humanities Quarterly 20(1). DOI 10.63744/2ezxpskcezq4. — LSA-based test of textual coherence across MS 408 sections; control comparator is the Latin Speculum Humanae Salvationis. Finds anticipated similarity-score drops at section / page breaks and an overall coherence pattern — methodologically distinct from the Timm & Schinner self-citing-table hypothesis and structurally consonant with the substrate’s operator-grammar / section-grammar finding (F-CORE-1). Engages Timm & Schinner (2023), Reddy & Knight (2011), and Sterneker, Polish & Bowern (2021) directly.
Zattera, Massimiliano (2022). “A New Slot-Grammar-Based Computer Analysis of the Voynich Manuscript.” CEUR-WS Vol-3313, paper 10 (Malta Voynich conference). — automated F1-optimising slot-grammar inducer (open-source v4j Java); 12-slot positional model with 86.6% of tokens (51.3% of types) fully decomposable; section-specific grammars classify pages at 92% accuracy. SLOT_MACHINE CFG F1 = 0.270 vs Stolfi F1 ≈ 0. Directly parallel to the substrate’s operator-grammar finding (F-CORE-1).
Greshko, Michael Anthony (2025). “The Naibbe Cipher: A Substitution Cipher That Encrypts Latin and Italian as Voynich Manuscript–Like Ciphertext.” Cryptologia. DOI 10.1080/01611194.2025.2566408; open-access supplement 10.5281/zenodo.17219445. — verbose-homophonic dice-and-cards substitution executable with 15th-century materials; reproduces Voynichese B’s slot grammar, glyph frequencies, word lengths, and qok-clustering simultaneously. The substrate treats the natural-language target as contradicted at Tier 1 while logging the cipher-mechanism as a parallel slot-grammar application.
Lindemann, Luke & Bowern, Claire (2020). “Character Entropy in Modern and Historical Texts: Comparison Metrics for an Undeciphered Manuscript.” arXiv:2010.14697. — entropy analysis of MS 408 in comparative context. The substrate carries an explicit caveat against over-reading the result — EVA re-fusion reduces but does not eliminate the conditional-entropy anomaly (h2 ≈ 2 for Voynichese vs 3–4 for natural languages).
Timm, Torsten & Schinner, Andreas (2020). “A Possible Generating Algorithm of the Voynich Manuscript.” Cryptologia 44(1):1–19. — self-citing-table / algorithmic-generation hypothesis. The substrate cites it as the textbook hoax-end of the literature the framework needs to engage.
Zandbergen, René (2022). “Towards a Better Transliteration Alphabet for the Voynich Manuscript.” CEUR-WS Vol-3313 (Malta keynote). — Super Transliteration Alphabet (STA) defined by EVA’s co-creator, fusing pedestalled gallows at family level; explicit admission that EVA is “sub-optimal for most statistical analyses.”
Burgundian-corridor & apothecary lane
Bolt, Nanno — six-paper peer-reviewed corpus plus two monographs on Burgundian-state apothecaries; the substrate’s primary scholarly anchor on the apothecary layer of F-CORRIDOR-1. Database covers ~1,000 named apothecaries across 77+ towns of the Burgundian state.
Bolt, N. (2015a). “Ages at First Mention as Active Master-Apothecary in the Burgundian State.” American Journal of Medicine and Medical Sciences 5(3):113–120.
Bolt, N. (2015b). “Apothecary Clothing in Burgundian Colors.” American Journal of Medicine and Medical Sciences 5(6):273–278.
Bolt, N. (2016). “Apothecary Clientele in the State of Burgundy.” Revue d’Histoire de la Pharmacie 64(390):227–240. PMID 29485780; Persée.
Bolt, N. (2017). “Apothecaries in the Low Countries Registered in the Burgundian-Ducal Archives.” Revue d’Histoire de la Pharmacie 104(395):389–396. Persée.
Bolt, N. (2018). Middeleeuwse apothekers in het Bourgondisch Rijk. Boekscout. — the database book.
Bolt, N. (2021). “Guild Regulation of Burgundian Apothecaries.” Pharmaceutical Historian (BSHP) 51(1):1–5.
Bolt, N. (2024). Apothecaries in Medieval Burgundy (1200–1600). Eburon Academic. — current consolidated monograph.
Baveye-Kouidrat, Laurie (2015). Burgundy Apothecary Archival Research (doctoral thesis). — the primary scholarship the substrate’s Burgundian dossier extracts from; read directly off Vol. I (306 pp, ~12k lines).
DALME database — Documentary Archaeology of Late Medieval Europe; uncited in any Voynich publication located by the substrate’s scholarship survey; treated as live comparator-data surface.
Doutrepont (1906); De Winter (1985); Wijsman (2010); Falmagne & Van den Abeele (2016); Vaughan — the Burgundian-court manuscript-culture scholarship the substrate reads against on the Bavaria-Straubing ↔ Burgundian-Wittelsbach kinship branch. All return zero Voynich citations in the surveyed corpus; the substrate carries them as comparator-stratum primary anchors.
Padua-scholastic, computus & Alfonsine lane
Battisti, Eugenio & Battisti, Giuseppa Saccaro (1984). Le Macchine Cifrate di Giovanni Fontana. Arcadia Edizioni. — the standard reference on Fontana’s enciphered-manuscript work; the closest Quattrocento parallel to MS 408 in the literature.
Chabás, José (2007). “From Toledo to Venice: The Alfonsine Tables of Prosdocimo de’ Beldomandi of Padua (1424).” Journal for the History of Astronomy 38:269–281. — Padua–Alfonsine astronomical-tables anchor.
Chabás, José & Goldstein, Bernard R. (2003). The Alfonsine Tables of Toledo. Springer. — foundational astronomical-tables scholarship.
Chabás, José & Goldstein, Bernard R. (2012). A Survey of European Astronomical Tables in the Late Middle Ages. Brill. — astronomical-tables scaffold the substrate’s zodiac / cosmological reading reads against.
Schuba, Ludwig (1981; 1992). Palatine catalogues (medical, 1981; quadrivium, 1992). — Pal.lat. shelfmark scaffold; absent from every located Voynich bibliography but the index the substrate uses to read against Pal.lat. 1115 / 1262 / 1373 / 1368 / Barb.lat. 186.
Crisciani, Chiara; Agrimi & Crisciani; Pesenti, Tiziana; Nicoud, Marilyn; Park, Katharine — Padua / Italian-scholastic medical-tradition scholarship the substrate reads against for Cermisone, Montagnana, Hugo Senensis, Prosdocimo, the Santasofia hand.
Sniezynska-Stolot, Ewa (2003). Astrological Iconography in the Middle Ages: The Decanal Planets. Jagiellonian University Press. — natural framework for reading VMS zodiac’s 30 nymphs as decanal / paranatellonta. Does not itself mention the VMS.
Blume, Dieter (2016). Personal correspondence with René Zandbergen, August 2016. — reaffirms “the probable relevance of the manuscripts of Alfonso el Sabio” (Castilian Libros del saber) as iconographic anchor for the VMS zodiac. The strongest external-expert anchor available; never elaborated in any peer-reviewed venue.
Wallis, Faith (1999). Bede: The Reckoning of Time. Liverpool University Press. — computus framework; never applied to VMS in peer-reviewed work; the substrate carries it as the standard computus reference.
Hartlieb & women’s-secrets lane
Brewer, Keagan & Lewis, Michelle (2024). “The Voynich Manuscript, Dr Johannes Hartlieb and the Encipherment of Women’s Secrets.” Social History of Medicine 37(3):559–582. — women’s-secrets reading of MS 408 anchored on the Bavaria-Munich Hartlieb tradition. The substrate’s primary-thesis test fails (F-RETIRED-BREWER-LEWIS) — the Bavaria-Munich anchor is the wrong Wittelsbach branch and slightly post-C14. The methodological precedent — court-physician encipherment of medical / women’s secrets — carries forward.
Schneider; Bosselmann-Cyran — Hartlieb-tradition catalogue scholarship; the GNM / BSB catalogue traditions; the scaffold the substrate’s Hartlieb-corridor dissection wave reads against.
EVA transcription-critique lane
Cham, Julian J. & Jackson, Patrick (2014). “The Curve-Line System and the Voynich Manuscript.” — independent statistical confirmation of glyph affinity; cited downstream by Zattera, Lindemann-Bowern, Tucker-Janick, Feaster.
Ponzi, Marco (Medium + ViridisGreen blog); Ponzi, M. & Smith, K. (2019). “Statistical Properties of the Voynich Text.” Cryptologia 43(6). — comparative transcription work showing how transcription choice alters Sukhotin/Hulden vowel detection.
Feaster, Patrick (2020). “Voynichese: A New Transliteration Approach.” Griffonage-Dot-Com (blog post). — addresses the arc-class undercounting issue head-on.
Claston, Glen / Petersen, Theodore C. (c. 2009). v101 transcription (~200 distinct characters). — foundational alternative to EVA.
Provenance & transmission lane
Zandbergen, René & Prinke, Rafał T. (2016); Hunt; Rampling, Jennifer; Harkness, Deborah E. — the standard published transmission-history work (Tepenec / Baresch / Marci / Kircher post-1608).
Guzy, Stefan (2022). “The Augsburg Hoffinanz Transaction of 1599.” CEUR-WS Vol-3313, paper 16 (Malta). — March 1599 Hoffinanz transaction Rudolf II → Carl Widemann of Augsburg for “barrel of remarkable books.” The substrate carries this as the late-corridor Augsburg anchor.
Pelling, Nicholas (2006). The Curse of the Voynich. Compelling Press; ongoing work at Cipher Mysteries. — long-running Northern Italian / Milan-Sforza production framework with the swallowtail-merlon and “4o”-cipher arguments. The substrate reads against this as the most well-developed competing framework.
D’Imperio, Mary E. (1978). The Voynich Manuscript: An Elegant Enigma. National Security Agency. — with Currier, Prescott H. (1976 NSA seminar); Tiltman, John H. (1967, repr. 2002 NSA Technical Journal); Reeds, James; Stolfi, Jorge (2000) — the foundational priors the substrate’s operator-grammar lineage descends from.
Source-list audit method: cross-referenced against the 2026-04-26 AC-relay scholarship
survey of Cryptologia, Manuscript Studies, Social History of Medicine,
Annual Review of Linguistics, Early Science and Medicine, Journal for
the History of Astronomy, Pharmaceutical Historian, Revue d’Histoire
de la Pharmacie, the Yale 2016 Clemens facsimile, Voynich-2022 Malta CEUR-WS Vol-3313,
arXiv, academia.edu, ResearchGate, and principal monographs (Pelling 2006; Kennedy &
Churchill 2004; D’Imperio 1978; Brumbaugh 1978; Janick & Tucker 2018; Edwards
2024). The published Voynich field has not engaged most of these scholarly anchors against
MS 408; the substrate carries them as primary-source comparator material.
§ 05
Research instruments & the working stack
What the lab actually uses to produce, store, and read against the
substrate — the analytical equipment, the pipelines, and the open-source standards
the work runs on.
The substrate database — SQLite with FTS5 full-text search. Persists 326 findings, 599 catalogued comparator manuscripts, 210 named figures, 21 working axes, 7,079 cross-relationships, 6,652 citations. The lab’s memory of its own work.
The Coviewer — the lab’s multi-pane image-reading workbench. Places Beinecke MS 408 folios side-by-side with comparator-archive folios; reads IIIF manifests and direct-image URLs from an allowlist of established holding libraries. Public iteration is in development; see Coviewer.
AI as a research instrument — the lab uses large language models (currently Anthropic’s Claude family) as instruments for transcription parsing, comparator dissection, hostile-referee review of candidate findings, and substrate write-gating. Every model output passes through a tier-disclosure step before any finding lands on record. The AI is a tool, not an oracle, and never closes a finding by itself.
HTR & image pipeline — Tesseract for OCR / HTR passes on PDF-backed scholarly sources; pdftotext for body-text extraction from monograph PDFs; Pillow and ImageMagick for plate transcoding, cropping, and IIIF caching.
IIIF (International Image Interoperability Framework) — the standard the holding libraries publish manifests in. Every comparator-manuscript image the substrate reads against is fetched against the holding library’s published IIIF manifest (BSB · BAV · Bodleian · Beinecke · Wellcome · BnF · e-codices · HAB · SBB · ONB · GNM · SLUB · UB Heidelberg · Morgan · Cambridge · manuscripta.at).
Python ecosystem — lxml, beautifulsoup, requests, sqlite3, FastAPI, and the broader PyPI toolchain the transcription-extraction, comparator-dissection, and analysis pipelines are built on.
Working environment — the lab is local-first by design. The substrate, the Coviewer, the HTR + image pipelines, and the orchestrator that ties them together all run on the lab’s own machine; nothing about the research depends on a third-party SaaS. Hosting of the public-facing record is on AWS (S3 / CloudFront / Route 53 / ACM).
§ 5b
Lab deposits on the record
Honeycutt AI Labs preprints and datasets directly relevant to the
substrate, deposited on Zenodo with live DOIs. Cite these where they support a claim
below.
Foundation deposits (substrate + methodology, 2026 Q1)
Voynich Manuscript Paduan Medical Reference (dataset, 2026-02-19) —
characterises MS 408 as a mechanical, modular, analog-computational system in the Paduan
medical–philosophical tradition. Introduces the VMS_OS framing the substrate’s
operator-grammar finding (F-CORE-1) rests on.
10.5281/zenodo.18687530.
Epistemic-Boundary Misclassification in Large-Language Models
(working paper, 2026-02-19) — characterises a reproducible failure mode in LLMs
during long-horizon analytical work. Cited where the substrate’s AI-assisted
readings flag boundary risk.
10.5281/zenodo.18690241.
SlopFilter: A Portable Epistemic Hygiene Protocol for AI-Generated Content
(preprint v3, 2026-04-11) — introduces SlopFilter (public-facing name; Narrative
Pressure Index is the institutional framing) and the Epistemic Constraint Profile
(ECP-1). Adjacent to the substrate’s tier-disclosure and adversarial-review
discipline; not VMS-specific.
10.5281/zenodo.19503170.
Voynich substrate deposits (2026-05-24)
Twelve substrate-anchored preprints deposited 2026-05-24, each carrying its own falsification matrix and external-citation register. Comparator-lens, structural, and corridor-provenance work — no decode, plaintext, or authorship-resolution claim.
The Voynich Manuscript as an Operational Page-Architecture: Structural Provenance in the Paduan Medical-Computational Tradition (preprint, 2026-05-24) — the breviarium frame for MS 408: compressed operational reference inside the Paduan medical corridor on parchment 1404–1438.
10.5281/zenodo.20369007.
An Operator Grammar for Beinecke MS 408: Full-Manuscript Validation, Adversarial Testing, and Contemporary Manuscript Comparison (preprint, 2026-05-24) — 31 operator classes across 226 folios and 114,507 transitions; Currier A/B classifier confound disclosed; engages Layfield & Davis 2026 (DHQ) as orthogonal text-channel comparator. F-CORE-1 through F-CORE-8 Confirmed.
10.5281/zenodo.20369025.
The Ghost Pattern Codebook: A Public-Surface Taxonomy of LLM Output Patterns (preprint, 2026-05-24) — seven-pattern taxonomy (Adversarial Loop, Rigorous Wrapper, Formal Dress, Decorative Formalism, Closed Loop, Authority Bleed, Narrative Pressure) plus interaction chains. Public-surface companion to the SlopFilter spec; scoring formula stays internal.
10.5281/zenodo.20369032.
The Twelve-Component Computational Apparatus of BSB Cgm 7958 (Hartlieb, Bavaria 1456) as Comparator-Lens for the Voynich Manuscript (preprint, 2026-05-24) — nine folio-pinned plus three umbrella apparatus components in Cgm 7958, 1456 sanctorale date-lock, cross-witnessed at Cpg 3 / Cpg 5 / Cpg 65 / Chart.A.558. Comparator-lens framing, not identity claim.
10.5281/zenodo.20369033.
Pattern Pressure, Accuracy Drift, and False User-State Attribution (preprint, 2026-05-24) — specimen paper anchored on a single 2026-04-08 archived assistant emotion-misread record; four-axis decomposition (intent / fact / provenance / pattern). Sibling to the Epistemic-Boundary work.
10.5281/zenodo.20369046.
Emulation Diagnostics for Model Self-Reports: A Bridge Between Interpretability and Welfare Assessment (preprint, 2026-05-24) — propagation-stripping diagnostic for LLM self-reports; sibling to Epistemic-Boundary and Pattern Pressure.
10.5281/zenodo.20369052.
The Hartlieb Kräuterbuch 9-Witness Corpus: A Substrate-Complete Census (preprint, 2026-05-24) — nine-witness Hartlieb-herbal census (HSC Werk 2317), five stand-alone witnesses, three Megenberg-embedded, one fragmentary. Structural comparator vs MS 408 herbal section — not identity claim.
10.5281/zenodo.20369053.
The Namenmantik Tradition and the Voynich Alphabetic Apparatus (preprint, 2026-05-24) — Candidate-tier comparator: Hartlieb’s Onomatomancia multi-alphabet apparatus parallels the VMS f1r right-margin Latin alphabet and the V-X-M-I apparatus-exclusive glyph distribution. Greshko 2025 Naibbe-cipher engaged as discriminating falsifier.
10.5281/zenodo.20369054.
Apothecary Readership and Print-Culture Reception of the Regimen-Sanitatis Workflow (1477–1501) (preprint, 2026-05-24) — six-section court-physician workflow mapped to MS 408 section architecture; Benedetto da Norcia 1477 print as consolidation hinge; Bolt corpus (1029 apothecaries, 85 cities, 14c–16c Burgundy) cited as substrate-anchor previously uncited in Voynich literature.
10.5281/zenodo.20369057.
Multilayer Cipher Architecture in Fontana’s Secretum and Bellicorum: A Twenty-Lens Template Return, Pending Negative-Control Validation (preprint, 2026-05-24) — twenty-lens analytical instrument applied to three Fontana witnesses; template-return outcome, not falsifier-passed; negative-control corpus listed outstanding. Tentative-tier framing throughout.
10.5281/zenodo.20369060.
The Despars Avicenna Canon Corpus: A Seven-Witness Substrate-Anchored Census (1431–1498) (preprint, 2026-05-24) — seven-witness corpus, one HTR-walked imaged (Lyon 1498), six catalogue-anchored. HTR cross-witness method observed; Despars commentary tradition as parallel French branch to the Padua-scholastic line.
10.5281/zenodo.20369061.
All deposited CC BY-NC 4.0, with author affiliation Honeycutt Ai Labs (Princeton, TX) and an AI-tools-in-research-composition disclosure section per modern publisher requirements.
The 2026-05-24 deposits each carry an external-citation register; the bibliographies in
§ 04 are the substrate-side index of those citations. This list is updated as additional preprints land public DOIs.
§ 06
Disclosure, license & how to be cited
Where the research lives, how it’s licensed, and how a third party should cite it.
Maintainer
This research record and the substrate database behind it are maintained by Honeycutt Ai Labs, Princeton, Texas. The substrate is a personal research workspace; no institutional funding is disclosed because none is involved.
License
Substrate output (findings, schemas, derived tables, code, this site’s text) — CC BY-NC 4.0. Underlying manuscript images — per each holding library’s digital-collections terms; reproduction here is under those terms.
How to cite
Honeycutt Ai Labs, thevoynich.org — a research record of Beinecke MS 408, working draft, compiled 2025-08-27–2026-05-23.
Corrections welcome
If credit is missing, wrong, or the scope of attribution is off, the substrate’s author wants to know. Reach out at hello@honeycuttailabs.com and the row will be updated. Source-completeness is part of the work, not a finishing flourish.