The Voynich manuscript
A book in a script nobody can read, on parchment dated 1404 to 1438. People say it is a hoax, a lost language, a cipher, an alien text, or already decoded. This case checks each claim against the records, including the owner’s own.
The question What is the Voynich manuscript, who wrote it, and has anyone really decoded it?
What the records say
The documented story holds up: 10 of 14 timeline steps are Confirmed by the records (score 93).
5 theories checked: 1 disproven, 2 unverified, 2 disputed. A theory is rated by its key claim.
No reading of the text has been independently confirmed
confirmedCredible: it rests on a primary (tier-1) record. See the step
Down the rabbit hole
The ground line is the documented timeline. Each tunnel is a theory that branches off it; every chamber is a step, coloured by how well the records support it. The deepest chamber is the theory’s key claim. Tap one to open it.
Main timeline
Oldest first- 1404 to 1438confirmed
The book’s parchment is made in the early 1400s
In 2009 a University of Arizona lab radiocarbon-dated four leaves; their combined result is 1404 to 1438 at 95.4% probability. The lab’s own caution: “this exercise dated parchment, not when it was written on.”
T1Hodgins radiocarbon report, 17 Feb 2009Conclusion; Fig. 7T2University of Arizona news release, 10 Feb 2011Sources disagreeThe owner’s own collection page still says the book was produced at the “end of the fifteenth or during the sixteenth century”, later than the 2009 radiocarbon result. - Before 1612likely
Emperor Rudolf II is said to have bought it for 600 ducats
The Beinecke Library names Rudolf II (reigned 1576–1612) as the first confirmed owner, who believed it was by the English friar Roger Bacon. The purchase rests on what a 1665 letter says Marci was told.
T1Beinecke Library, collection pageProvenanceinvolved party: Beinecke Library (owner)T2Snopes, 13 Oct 2022T3voynich.nu, history of the MSMarci letterSources disagreeSnopes says Rudolf got it from John Dee, “apparently”; the Beinecke says research has cast serious doubt on Dee ever owning it. - Before 1622likely
An early owner’s name is found on the first page
Under ultraviolet light a 2009 lab saw a signature, “probably that of Jacobj à Tepenece”, on folio 1r; 2024 imaging made it legible. The Beinecke lists Jacobus Horčický de Tepenec (c. 1575–1622) as an owner.
T1McCrone Associates materials report, 1 Apr 2009p. 1T3Davis, Manuscript Road Trip blog, 8 Sep 2024T1Beinecke Library, collection pageProvenanceinvolved party: Beinecke Library (owner) - 1639 to 1665likely
Two Prague scholars own it, and one sends it to Rome
The alchemist Georg Baresch owned it, then Johannes Marcus Marci, who sent it to the Jesuit scholar Athanasius Kircher in Rome with a letter still kept with the book. A 1639 letter from Baresch to Kircher survives.
T1Beinecke Library, collection pageProvenanceinvolved party: Beinecke Library (owner)T2The Art Newspaper, 25 Sep 2024T3voynich.nu, history of the MSBaresch; MarciSources disagreeThe Beinecke gives 1666 for the sending, and Baresch’s life as 1585–1622 while saying he owned the book by 1637. Davis gives 1665 and Baresch’s death in 1662; Zandbergen gives 1665 and a death between 1646 and 1662. - 1912confirmed
The bookseller Wilfrid Voynich buys it from the Jesuits
Voynich, a Polish-American antiquarian, bought it from Jesuit holdings near Rome, and it has carried his name since.
T2University of Arizona news release, 10 Feb 2011T2Snopes, 13 Oct 2022T2Ars Technica, 15 May 2019 (updated 17 May)T1Beinecke Library, collection pageAbout; Provenanceinvolved party: Beinecke Library (owner)Sources disagreeThe University of Arizona says Villa Mondragone; the Beinecke says the Jesuit College at Frascati; Zandbergen says later research shows the books were not stored at Mondragone. - 1969confirmed
The book goes to Yale’s Beinecke Library as MS 408
The New York bookseller H. P. Kraus gave it to the Beinecke Library, where it has stayed, leaving the building once for a 2014–15 exhibition.
T1Beinecke Library, collection pageProvenanceinvolved party: Beinecke Library (owner)T2Snopes, 13 Oct 2022Sources disagreeThe Beinecke says Kraus bought it in 1961 from Anne Nill, Voynich’s assistant; Snopes says he bought it from Voynich’s widow. - 20 Jul 2004confirmed
Researcher Gordon Rugg says the text could be meaningless, made with a grille
Rugg, of Keele University, said 16th-century techniques such as the Cardan grille make it “surprisingly easy” to produce complex-looking text with no meaning. He named the alchemist Edward Kelley as a possible maker, while saying he could not prove it.
- 14–15 Jan 2009confirmed
A lab finds medieval-type inks and paints
McCrone Associates found the text and drawings in the same iron gall ink, “written with a quill pen”, with azurite blue, red ochre and a copper green, bound in gum. The Arizona dating team called the colours “consistent with the Renaissance palette”.
T1McCrone Associates materials report, 1 Apr 2009pp. 1, 5–6T2University of Arizona news release, 10 Feb 2011Sources disagreeDavis’s 2020 abstract says the “parchment, ink, and pigments date from the early 15th century”. Only the parchment was dated: the Arizona team said radiocarbon-dating the inks was beyond its abilities. - 21 Jun 2013confirmed
A statistics paper finds word patterns like those of real languages
Montemurro and Zanette found long-range word structure “compatible with those found in real language”, linking the herbal and pharmaceutical sections by their words alone. They say this supports “a genuine message inside the book”.
T1Montemurro and Zanette, PLoS ONE, 2013Abstract; Results - May 2019confirmed
A university publicises a “solution”, then takes it down
The University of Bristol publicised Gerard Cheshire’s paper calling the text “proto-Romance”. Within days it removed the story because “concerns have been raised about the validity of this research”.
T2Ars Technica, 15 May 2019 (updated 17 May)update of 17 May 2019T2Boston Globe, 17 May 2019 - Spring 2020confirmed
A handwriting study finds five scribes
The paleographer Lisa Fagin Davis identified five hands, with one scribe writing in Currier’s “Language A” and four in “Language B”. She says any solution must account for this.
T1Davis, Manuscript Studies, 2020pp. 171–179 - Sep 2024likely
Imaging reveals faded alphabets on the first page
Multispectral photos taken at Yale in 2014 and studied in 2024 show three faded columns on folio 1r: two Roman alphabets and a column of Voynich letters. Davis matches the hand to Marci (owner 1662–1665) and found no older writing under the text.
- 26 Nov 2025confirmed
A peer-reviewed cipher turns Latin into Voynich-like text by hand
Michael Greshko’s “Naibbe cipher”, using dice and playing cards available in the 1400s, reproduces many of the text’s statistics. He presents it as proof that a cipher is possible, not as the solution.
- 2026confirmed
No reading of the text has been independently confirmed
Reporting in 2026 describes the book as still undeciphered. A September 2026 report puts it plainly: “No interpretation has been independently established as a translation.”
T2IBTimes UK, 30 Sep 2026T2Archaeology Magazine, 3 Jan 2026T1Davis, Manuscript Studies, 2020abstractT1Beinecke Library, collection pageAboutinvolved party: Beinecke Library (owner)
Branches
Go down oneA branch is only as strong as its weakest step so far. The count is how many sources back each step.
“It has been decoded”
Has anyone actually read the Voynich manuscript?
Davis records that the first formal published solution, in the 1920s, credited Roger Bacon. Hauer and Kondrak note that Newbold and Kent’s 1928 reading was refuted in 1931, and later claims by Feely (1943) and Strong (1945) were also refuted.
- 519,853 YouTube views, Stephen Bax “Voynich - a provisional, partial decoding of the Voynich script” as of 12 Oct 2026 T3YouTube: Stephen Bax, 2014
- 186,179 YouTube views, “Voynich Manuscript Finally Decoded By An AI... And It’s Not Good!” as of 12 Oct 2026 T4YouTube: “Finally Decoded By An AI…”, 2025
- 2,628,831 YouTube views, “Voynich Manuscript Revealed (2018)” (claims “over 30%” translated) as of 12 Oct 2026 T3YouTube: “Voynich Manuscript Revealed (2018)”
A peer-reviewed journal published Cheshire’s solution
Cheshire: the paper “has been blind peer-reviewed and published in a highly reputable journal”.
The premise: confirmedTrue that Romance Studies published it. Bristol withdrew its story within days, and Davis said its publication “is no sign of peer confidence”.
Artificial intelligence found the language is Hebrew
The University of Alberta: the algorithm found “the most likely language was Hebrew”, and the sample sentence “definitely makes sense”.
T2University of Alberta Folio, Jan 2018involved party: University of AlbertaThe premise: confirmedTrue that the method ranked Hebrew first. The researchers’ own paper says no output was grammatical, calls the sample “not quite a coherent sentence”, and allows the results may be “artifacts”.
T1Hauer and Kondrak, TACL, 2016section 5Plant and star names can be read: Taurus, juniper, coriander
Bax proposed readings for ten words in 2014.
The premise: unverifiedNot accepted. The 2021 Yale review notes that on folio 15v the first word is not the one his reading needs, and the plant “doesn’t look at all like a juniper tree”.
The script is abbreviated Latin for medical recipes
Nicholas Gibbs, in the Times Literary Supplement, 2017.
The premise: disprovenDavis: “It doesn’t result in Latin that makes sense.” The Yale review: Gibbs does not abbreviate the way medieval Latin writers did, and the Latin is not grammatical.
- confirmed
2014: Stephen Bax proposes readings for ten words
Bax, of the University of Bedfordshire, read plant and star names, including Taurus and juniper, and said the book was not a hoax.
StartStrength so far:confirmed· 2 sources on this step - confirmed
2016: an AI paper ranks Hebrew first but reads no coherent sentence
Hauer and Kondrak say a native speaker found the sample line “not quite a coherent sentence”, and that their results may be “simply artifacts of the combinatorial power of anagramming and language models”.
T1Hauer and Kondrak, TACL, 2016section 5T2New Atlas, 29 Jan 2018= HoldsStrength so far:confirmed· 2 sources on this step - confirmed
2017: specialists reject Nicholas Gibbs’s Latin-abbreviation “solution”
Gibbs published two lines of translation. Davis said the Latin was not grammatical and his lost-index idea had no evidence.
= HoldsStrength so far:confirmed· 2 sources on this step - confirmed
2019: Cheshire’s “proto-Romance” reading is rejected and Bristol withdraws
Davis wrote that “‘proto-Romance language’ is not a thing.” The computational linguist Greg Kondrak said the mapping of symbols disagrees with other proposals.
= HoldsStrength so far:confirmed· 2 sources on this step - unverified
Key claim: someone has really decoded the manuscript
No claimed reading has been independently confirmed. The best-publicised ones were rejected by specialists, and the 2016 Hebrew result by its own authors’ caveats.
T2IBTimes UK, 30 Sep 2026T1Davis, Manuscript Studies, 2020abstractT1Hauer and Kondrak, TACL, 2016section 5▼ WeakerStrength so far:unverified· 3 sources on this step
Bullshit meter
- Key claim: someone has really decoded the manuscript. Unverified.
- Key claim contradicted by tier 1? Partly. The 2016 Hebrew result is undercut by its authors’ own tier-1 paper; the 2017 and 2019 claims were rejected by specialists in tier-2 reporting; no claimed reading has been independently confirmed.
Fluff meter: possibly planted
- Can’t be traced to an original record. Absent: each claim traces to its author’s own paper, article or video.
- First appeared from new or anonymous accounts. Not measured.
- Coordinated identical wording. Not measured.
- Cited source doesn’t exist. Absent: the papers and articles cited exist.
- Later contradicted. Present: Bristol withdrew its 2019 story, and specialists rejected the 2017 and 2019 claims in print.
● present · ○ absent · – not measured. Lists signals only. It never accuses anyone.
Personal interest meter
- University of Bristol, first publicised Cheshire’s paper as research by one of its honorary associates, then removed the story. T2Ars Technica, 15 May 2019 (updated 17 May)
- University of Alberta, publicised its own researchers’ result; its 2018 story says the sentence “definitely makes sense”, while their paper calls it “not quite a coherent sentence”. T2University of Alberta Folio, Jan 2018involved party: University of Alberta
- Nicholas Gibbs, said in his article that he did the research for an unnamed television network. T2Ars Technica, 10 Sep 2017
“Has a documented interest” only, each one sourced. Never an implication of guilt.
Correlation meter
- The 2009 radiocarbon report, main step 1, and the hoax and alien branches. T1Hodgins radiocarbon report, 17 Feb 2009Fig. 7
- The 2009 McCrone materials report, main steps 3 and 8, and the hoax and alien branches. T1McCrone Associates materials report, 1 Apr 2009
- Lisa Fagin Davis’s 2020 handwriting study, main step 11 and the hoax, language, decoded and alien branches. T1Davis, Manuscript Studies, 2020
- Bowern and Lindemann’s Yale review, used in the hoax, language, cipher, decoded and alien branches; its authors work at the owner’s university. T2Bowern and Lindemann, Annual Review of Linguistics, 2021
- The faded alphabets on folio 1r, main step 12, the hoax branch (against a 1912 forgery) and the cipher branch (an early decoding attempt). T2The Art Newspaper, 25 Sep 2024
Links to other cases appear once those cases exist.
Who worked on this
This case was assembled by the site from public records. Contributions are not open yet. Nothing here is invented, so the counts are zero.
This is the first published version of the case.
Sources
- T1involved party: Beinecke Library (owner) “The Beinecke Cipher (Voynich) Manuscript”, collection page, Beinecke Rare Book and Manuscript Library, Yale University (page undated; read 12 Oct 2026). The library owns the manuscript, so its account of its own holding is marked as the owner’s.. beinecke.library.yale.edu
- T1Greg Hodgins (NSF-Arizona AMS Laboratory, University of Arizona), letter-report to Christine McCarthy, Chief Conservator for Special Collections, Yale University Library, “Re: Dating Results for Voynich Manuscript Parchment Leaves 8, 26, 47 and 68”, 17 Feb 2009. Not formally published; released later and hosted on René Zandbergen’s site.. www.voynich.nu
- T2“University of Arizona experts determine age of book ‘nobody can read’”, University of Arizona news release, 10 Feb 2011, as posted on EurekAlert! (AAAS). www.eurekalert.org
- T1Joseph G. Barabe, McCrone Associates, “Materials Analysis of the Voynich Manuscript” (project MA47613), letter-report to Kevin Repp, Beinecke Library, 1 Apr 2009, 6 pp. Commissioned under a purchase order from ProOmnia Film & Video Promotion GmbH (a documentary maker); hosted by Yale Library.. library.yale.edu
- T3René Zandbergen, “Voynich MS - History of the MS”, voynich.nu (independent researcher’s site, read 12 Oct 2026). Secondary.. www.voynich.nu
- T3René Zandbergen, “The Radio-Carbon Dating of the Voynich MS”, voynich.nu (read 12 Oct 2026). Secondary.. www.voynich.nu
- T2Nur Nasreen Ibrahim, “What Does the Mysterious Voynich Manuscript Really Say?”, Snopes, 13 Oct 2022 (corrected 16 Oct 2022). www.snopes.com
- T2“Central Europe: Is 16th-Century Voynich Manuscript A Hoax?”, Radio Free Europe/Radio Liberty, Prague, 20 Jul 2004 (interview with Gordon Rugg, Keele University). www.rferl.org
- T1Marcelo A. Montemurro and Damián H. Zanette, “Keywords and Co-Occurrence Patterns in the Voynich Manuscript: An Information-Theoretic Analysis”, PLoS ONE 8(6): e66344, published 21 Jun 2013. journals.plos.org
- T2Claire L. Bowern and Luke Lindemann (Yale University), “The Linguistics of the Voynich Manuscript”, Annual Review of Linguistics 7 (2021), doi:10.1146/annurev-linguistics-011619-030613. A review article. Read as the authors’ preprint of 8 Sep 2020 on LingBuzz (the journal page returned 403).. doi.org · copy read: ling.auf.net
- T1Lisa Fagin Davis, “How Many Glyphs and How Many Scribes? Digital Paleography and the Voynich Manuscript”, Manuscript Studies 5(1), Spring 2020 (Schoenberg Institute, University of Pennsylvania). repository.upenn.edu · copy read: archive.org
- T3Lisa Fagin Davis, “Multispectral Imaging and the Voynich Manuscript”, Manuscript Road Trip (her blog), 8 Sep 2024. A scholar’s own write-up, not peer-reviewed.. manuscriptroadtrip.wordpress.com
- T2Garry Shaw, “Voynich Manuscript scans reveal early decoding attempt”, The Art Newspaper, 25 Sep 2024 (read through a summarising fetch tool; nothing quoted from it). www.theartnewspaper.com
- T3“Hidden Secrets of the Voynich Manuscript Revealed Through Cutting-Edge Scans”, Medievalists.net, 9 Sep 2024 (no named author). www.medievalists.net
- T3“Voynich Manuscript partially decoded, text is not a hoax, scholar finds”, Medievalists.net, 20 Feb 2014 (no named author; quotes Stephen Bax). A record of Bax’s claim.. www.medievalists.net
- T3“Voynich - a provisional, partial decoding of the Voynich script”, Stephen Bax’s own channel, YouTube, published 2 Feb 2014. A record of his claim.. www.youtube.com
- T1Bradley Hauer and Grzegorz Kondrak (University of Alberta), “Decoding Anagrammed Texts Written in an Unknown Language and Script”, Transactions of the Association for Computational Linguistics 4 (2016): 75–86, doi:10.1162/tacl_a_00084. aclanthology.org · copy read: aclanthology.org
- T2involved party: University of Alberta “Using AI to uncover the mystery of an ancient manuscript”, Folio, University of Alberta news, January 2018. The university publicising its own researchers’ work.. www.ualberta.ca
- T2Rich Haridy, “Has AI finally cracked the mysterious Voynich manuscript?”, New Atlas, 29 Jan 2018. newatlas.com
- T2Annalee Newitz, “So much for that Voynich manuscript ‘solution’”, Ars Technica, 10 Sep 2017. arstechnica.com
- T2Jennifer Ouellette, “No, someone hasn’t cracked the code of the mysterious Voynich manuscript [Updated]”, Ars Technica, 15 May 2019, updated 17 May 2019. arstechnica.com
- T2“The Voynich manuscript, a famously mysterious medieval text, has been decoded. Or has it?”, The Boston Globe, 17 May 2019. www.bostonglobe.com
- T3Gerard Cheshire, “The Language and Writing System of MS408 (Voynich) Explained”, Romance Studies 37(1): 30–67, published 29 Apr 2019, doi:10.1080/02639904.2019.1599566; abstract read on the University of Bristol research portal. A record of what the author claims.. doi.org · copy read: research-information.bris.ac.uk
- T2Sergio Prostak, “Historically Plausible Cipher Recreates Statistical Signature of Voynich Manuscript”, Sci.News, 7 Jan 2026 (on Michael A. Greshko, “The Naibbe cipher”, Cryptologia, published online 26 Nov 2025, doi:10.1080/01611194.2025.2566408, which returned 403). www.sci.news
- T2Dario Radley, “A new study suggests the mysterious Voynich Manuscript may be a medieval cipher”, Archaeology News Online Magazine, 3 Jan 2026. archaeologymag.com
- T3René Zandbergen, “The Cardan grille approach to the Voynich MS taken to the next level”, arXiv:2104.12548 (2021), abstract read. A preprint, not peer-reviewed.. arxiv.org
- T2Seneca Cabrera, “Voynich Manuscript ‘Decoded’ by WitchTok? Viral Theory Claims Women Used Period Blood to Grow Mysterious Plants”, International Business Times UK, 30 Sep 2026. www.ibtimes.co.uk
- T3“This Famous Medieval Book May Be a Hoax”, SciShow channel, YouTube, published 18 Jun 2025. Its description says the book “might be a medieval fraud, created by an algorithm executed with nothing more than a pen”.. www.youtube.com
- T3“Voynich Manuscript Revealed (2018)”, Voynich Manuscript Research channel (tagged “Ahmet Ardic” and “Ata Team Alberta”), YouTube, published 22 Feb 2018. A record of the channel’s claim to have translated “over 30%” as old Turkic.. www.youtube.com
- T4“Voynich Manuscript Finally Decoded By An AI... And It’s Not Good!”, The Ultimate Discovery channel, YouTube, published 29 Jun 2025. www.youtube.com
- T4“Did Aliens Write the Voynich Manuscript? | COLOSSAL MYSTERIES”, Colossal Cranium channel (a Peacock Kids show), YouTube, published 15 Sep 2019. www.youtube.com
- T4“AI Just Read the Voynich Manuscript. It’s Terrifying”, Odd Whys channel, YouTube, published 29 Jan 2026. Its description asks “What if the most mysterious books in history weren’t written for humans?”. www.youtube.com
- T3“The UnXplained: Medieval Book Contains SACRED Info (Season 6)”, HISTORY channel, YouTube, published 12 Jan 2024 (a clip from the TV series hosted by William Shatner). www.youtube.com
- T3“No, TikTok Witches Did Not Solve the Voynich Manuscript”, Anthony Gramuglia channel, YouTube, published 29 Sep 2026. A critic’s video about the 2026 TikTok claim.. www.youtube.com