The Voynich manuscript is a small illustrated book held at Yale University's Beinecke Rare Book and Manuscript Library as MS 408, written in an unknown script that no one has been able to read. Its vellum has been radiocarbon-dated to between 1404 and 1438, and its roughly 240 surviving pages are filled with drawings of unidentifiable plants, zodiac wheels and bathing figures. More than a century of announced solutions has not produced a single reading that specialists accept.
That combination is what makes the manuscript matter. It is not a legend known only from later reports; it is a physical object anyone can study in the Beinecke's free online scans. Cryptographers, radiocarbon laboratories, paleographers and computational linguists have all examined it, and it still resists them. For a site built around [the idea of forbidden and hidden knowledge](/what-is-forbidden-knowledge-and-why-4biddenknowledge-exists), it is a useful test case: a genuine mystery where the honest answer to “what does it say?” is still “nobody knows,” and where the history of failed decodings is as instructive as the book itself.
What the Voynich manuscript is
The Voynich manuscript is a handwritten codex on vellum, roughly the size of a modern hardcover book, containing continuous text in an unidentified writing system alongside hundreds of illustrations. About 240 pages survive, several of them large fold-outs. The folio numbers written into the book show gaps, and researchers estimate it originally ran to around 272 pages, so some leaves have been lost.
The script, nicknamed Voynichese, does not appear in any other known document. The book carries no title, author's name or date. It is named after Wilfrid Voynich, the antiquarian bookseller who bought it in 1912.
What is inside: the six sections
Researchers divide the manuscript into six sections based on the illustrations rather than the text, because the pictures are the only part anyone can interpret. The names are conventions of convenience, not proven descriptions of what the text says.
- Herbal. The largest section. Most pages show a single plant, often with roots, leaves and flowers, surrounded by text. Few can be matched confidently to real species.
- Astronomical and astrological. Circular diagrams, including a run of zodiac roundels with small human figures, many holding stars. Month names have been added beside them in a later hand in the Latin alphabet, which makes them among the very few readable words in the book.
- Balneological. Nude women standing or sitting in pools and tubs, often connected by tube-like structures.
- Cosmological. Circular and geometric designs, including the large fold-out known as the Rosettes page, with nine linked medallions.
- Pharmaceutical. Drawings of jars and containers resembling apothecary vessels, set beside labelled drawings of roots and leaves.
- Recipes. Pages of short paragraphs, each marked in the margin with a small star, with no other illustrations.
How old is the Voynich manuscript
The vellum dates to the early 15th century. In 2009 a team at the University of Arizona radiocarbon-dated samples from several leaves to between 1404 and 1438. In the same period, McCrone Associates analyzed the inks and paints and found them consistent with materials in use at that time, with nothing that required modern chemistry.
Radiocarbon dating measures when the animals whose skins became the vellum died, so it dates the writing surface, not the writing. The same limitation applies wherever the method is used on manuscripts, including [the Dead Sea Scrolls](/the-dead-sea-scrolls-explained-discovery-contents-and-why-they-matter). Most researchers still think the Voynich text was written fairly soon after the vellum was prepared, since a much later writer would have needed a large supply of matching blank vellum.
The dating ruled out two long-running ideas: that Voynich forged the book himself, and that it was written by Roger Bacon, the 13th-century English friar who died around 1292.
Who owned it: from Prague to Yale
The documented history begins in 17th-century Prague. A Prague alchemist named Georg Baresch owned the manuscript and wrote to the Jesuit scholar Athanasius Kircher in Rome asking for help reading it. After Baresch's death it passed to his friend, the physician Johannes Marcus Marci, who sent it to Kircher with a covering letter dated 1665 or 1666. That letter was later found with the manuscript.
Marci's letter is the source of the two most famous claims about the book: that Emperor Rudolf II, the Habsburg ruler who made Prague a centre for alchemy and the occult sciences, had paid 600 ducats for it, and that it was believed to be the work of Roger Bacon. The Rudolf connection is secondhand and remains a possibility rather than a certainty. The Bacon attribution has been discredited by the dating.
Kircher was a natural person to ask. He had published interpretations of Egyptian hieroglyphs that he believed were decipherments, though modern scholars regard them as fanciful; the real breakthrough came in 1822 with Jean-François Champollion, as covered in [the Ancient Egypt topic hub](/topics/ancient-egypt). There is no record that Kircher made progress with the Voynich text.
The manuscript stayed in Jesuit hands, in the library of the Collegio Romano, and later at the Villa Mondragone near Frascati, where Voynich bought it in 1912 among a group of manuscripts the Jesuits were selling. After his death in 1930 it passed to his widow, the novelist Ethel Voynich, and then to his longtime associate Anne Nill. It was sold to the New York rare-book dealer H. P. Kraus, who failed to find a buyer at his asking price and donated it to Yale in 1969. Yale published a full-colour facsimile in 2016.
What the text looks like
Voynichese is written left to right in an alphabet of roughly 25 to 30 distinct glyphs, depending on how ligatures are counted. The writing is fluent, with few corrections, and falls into word-like units separated by spaces.
Statistically, the text is a puzzle in both directions. Word frequencies follow Zipf's law, the pattern in which a few words are very common and most are rare, which is typical of natural languages. But other features are unusual: the same word often repeats two or three times in a row, many words differ from their neighbours by a single glyph, certain glyphs appear only at the beginning or end of words, and the text is far more predictable at the level of individual characters than European languages are. Each camp, real language or generated text, points to the features that suit it.
Why every decoding claim has failed
Every claimed solution so far fails the same test: it cannot turn continuous passages into coherent text using fixed rules that another researcher can apply and get the same result. Genuine decipherments meet that standard. When Michael Ventris deciphered Linear B in 1952, other scholars could apply his values to new tablets and read Greek. No Voynich proposal has come close.
William Newbold and the Roger Bacon theory
In 1921 William Newbold, a University of Pennsylvania professor, announced that the manuscript was Roger Bacon's work, written in a microscopic shorthand hidden in the pen strokes. In 1931 the medievalist John Manly showed that the supposed microscopic marks were cracks in the dried ink, and that Newbold's method was flexible enough to produce almost any reading.
Feely, Levitov and later claims
Joseph Feely proposed abbreviated Latin in the 1940s. Leo Levitov argued in 1987 that it was a text of the Cathars, the dualist Christian movement of southern France. Neither produced consistent readings.
Gordon Rugg and the hoax hypothesis
In 2004 the computer scientist Gordon Rugg showed that a Cardan grille, a card with cut-out windows laid over a table of syllables, could generate text with many of Voynichese's statistical quirks. That showed a hoax was feasible, not that the manuscript is one, and critics noted that Gerolamo Cardano described the grille in 1550, more than a century after the vellum date.
Stephen Bax, Nicholas Gibbs and Gerard Cheshire
In 2014 the linguist Stephen Bax proposed provisional readings for a small number of words, mostly labels beside plants and stars. He called them provisional, and they have not been extended to continuous text. In 2017 Nicholas Gibbs claimed in the Times Literary Supplement that the text was abbreviated Latin forming a women's health manual; specialists, including voices associated with the Beinecke, dismissed it within days because his sample did not yield grammatical Latin. In 2019 Gerard Cheshire claimed the text was a lost proto-Romance language. Specialists rejected it: no such language is attested and his readings were inconsistent.
Hebrew, Turkish, Nahuatl and AI solutions
Claims that the text is Hebrew, Turkish, Nahuatl or dozens of other languages appear regularly; a Mexican origin sits awkwardly with a vellum date before 1492. In 2018 the computer scientists Greg Kondrak and Bradley Hauer ran an algorithm that suggested Hebrew with rearranged letters, then rendered a sample sentence through Google Translate. Hebrew specialists found the output did not hold together, and the result is not accepted.
What serious research has established
The solid findings are more modest than the headlines. William Friedman, one of the most celebrated American codebreakers of the 20th century, organized study groups on the manuscript and came to suspect it was an early attempt at an invented language. Prescott Currier, a US Navy cryptologist, identified two statistically distinct varieties of the text in the 1970s, now called Currier A and Currier B, and linked them to different hands. Mary D'Imperio's 1978 monograph for the National Security Agency, The Voynich Manuscript: An Elegant Enigma, summarized decades of work. In 2020 the paleographer Lisa Fagin Davis identified five distinct scribes. René Zandbergen's website remains the standard reference on the manuscript's history.
From that work, four main possibilities remain open. The text could be an unknown or poorly recorded natural language written in an invented script. It could be a cipher over a known language, though simple substitution ciphers of the kind used in the 15th century do not fit its statistics. It could be a constructed or glossolalic text, produced by someone writing in a devotional or altered state, a question that touches on themes explored in [the Consciousness topic hub](/topics/consciousness). Or it could be an elaborate hoax, made to impress or sell to a wealthy collector. None of these has been proven.
Billy Carson's reading
Billy Carson's interest in the Voynich manuscript is as an example of knowledge that resists decoding. Carson has not claimed to have deciphered it, and 4biddenknowledge does not present a translation of its own. What draws his attention is what the book represents: a real, dated, physical record whose meaning has been lost, which is a different thing from a legend.
[The Book of Thoth](/the-book-of-thoth-explained-the-egyptian-text-the-legend-and-the-modern-books) is known mainly through Egyptian story and later esoteric tradition, a book of hidden knowledge whose physical existence cannot be shown. The Voynich manuscript is the reverse: the book survives, but its contents are closed. And [the Nag Hammadi texts](/the-gnostic-gospels-and-nag-hammadi-explained-the-gospel-of-thomas-and-why-they-were-buried) were hidden and recovered, then read and translated, because their language was known. In Carson's framing, the Voynich manuscript sits at the far end of that spectrum, and it demands the same discipline as any ancient text: separate what the object shows from what people want it to say.
Watch, read and go deeper
Billy Carson discusses ancient texts, lost knowledge and unsolved mysteries on [the 4biddenknowledge Podcast on 4BK TV](/watch/billy-carson-4biddenknowledge-podcast). For the wider theme of what counts as hidden knowledge and why it matters, start with [why 4biddenknowledge exists](/what-is-forbidden-knowledge-and-why-4biddenknowledge-exists). The [Consciousness topic hub](/topics/consciousness) collects related articles on the mind, altered states and the texts that try to describe them. For the Egyptian thread, see [the Ancient Egypt hub](/topics/ancient-egypt).
Frequently asked questions
Has the Voynich manuscript been decoded?
No. Many people have announced solutions since 1921, but none has produced a consistent, repeatable translation of continuous text that specialists accept.
How old is the Voynich manuscript?
Its vellum was radiocarbon-dated by the University of Arizona in 2009 to between 1404 and 1438. The dating applies to the writing surface, but most researchers think the text was written not long afterwards.
Where is the Voynich manuscript kept, and can I see it?
It is held at Yale University's Beinecke Rare Book and Manuscript Library as MS 408. The Beinecke has digitized the entire manuscript, the scans are free to view online, and Yale published a printed facsimile in 2016.
Is the Voynich manuscript a hoax?
It might be, but that has not been proven. Gordon Rugg showed in 2004 that a hoax was technically feasible, and some statistical oddities fit a generated text. Other features, including Zipf-like word frequencies and the work of several scribes, lead many researchers to think it encodes something real.
Who wrote the Voynich manuscript?
Nobody knows. The old attribution to Roger Bacon is ruled out by the dating, and Lisa Fagin Davis's 2020 paleographic study identified five scribes, which suggests a group effort rather than a single author.
Sources and further reading
- The Voynich Manuscript, edited by Raymond Clemens (Yale University Press, 2016), the full facsimile.
- Mary E. D'Imperio, The Voynich Manuscript: An Elegant Enigma (National Security Agency, 1978).
- Lisa Fagin Davis, “How Many Glyphs and How Many Scribes? Digital Paleography and the Voynich Manuscript,” Manuscript Studies (2020).
- Gordon Rugg, “An Elegant Hoax? A Possible Solution to the Voynich Manuscript,” Cryptologia (2004).
- John M. Manly, “Roger Bacon and the Voynich MS,” Speculum (1931).
- René Zandbergen, The Voynich MS (reference website on the manuscript's history).
- Beinecke Rare Book and Manuscript Library, MS 408, complete digital facsimile.
