Open the Voynich Manuscript and almost everything about it looks tantalizingly close to comprehension. Lines of neat handwriting run from left to right. Words are separated by spaces. Paragraphs appear to follow rules. Around them are drawings of plants, stars, zodiac-like circles, strange plumbing systems and groups of naked human figures bathing in green pools.
It looks like a book that should be readable.
Yet more than a century after it became famous, nobody can say with scholarly confidence what a single full page means. The script does not match any securely identified writing system, the language has never been established, and repeated claims of decipherment have failed to produce a solution accepted by specialists.
The object itself is real and remarkably accessible. It is catalogued as Beinecke MS 408 at Yale University's Beinecke Rare Book & Manuscript Library, where high-resolution scans of the surviving manuscript are available to researchers. The mystery is not where the book is. The mystery is what its pages say.
A medieval object with an uncertain birthday
The Voynich Manuscript is usually described as a fifteenth-century codex, and there is strong scientific evidence for that broad date. Radiocarbon analysis performed on samples of its parchment placed the animal skins used to make the writing material in the early fifteenth century, commonly reported as roughly 1404–1438.
There is an important distinction hidden inside that famous date. Radiocarbon dating measures the parchment, not the moment when a scribe placed ink on it. In principle, prepared parchment could be stored before use. Chemical analysis by McCrone Associates found inks and pigments compatible with historical manuscript materials and did not reveal a modern substance that would turn the book into an obvious recent forgery, but the ink itself was not directly radiocarbon-dated.
The surviving codex contains roughly 240 pages, including elaborate foldouts, and some leaves are missing. Its pages are made of vellum or parchment, with brownish-black writing and colored illustrations. The script is so distinctive that researchers often call it “Voynichese,” a convenient label that does not imply anyone knows what language — if any — lies underneath.
The illustrations give the illusion of clues. One large section shows plants, many of them difficult or impossible to identify securely. Other pages contain astronomical and astrological diagrams. A particularly strange group depicts small nude figures, mostly women, immersed in pools or connected by tubes. Later pages resemble pharmaceutical illustrations of roots and containers, while another section consists largely of short blocks of text marked by star-like symbols.
These categories — botanical, astronomical, balneological, pharmaceutical and so on — are descriptions based on appearances. They are not chapter titles translated from the manuscript. That difference matters: researchers can see what a drawing resembles without knowing what the author believed it represented.
From an emperor’s library to a bookseller named Voynich
The manuscript's early history contains gaps, but by the seventeenth century it was circulating among scholars in Central Europe. A letter sent in 1666 by Johannes Marcus Marci to the Jesuit polymath Athanasius Kircher survives with the book. Marci reported that the manuscript had once belonged to Holy Roman Emperor Rudolf II and that it was believed to be the work of the medieval English philosopher Roger Bacon.
That attribution is now untenable. Bacon died in the thirteenth century, long before the parchment's radiocarbon range.
The manuscript also passed through the hands of figures including Jacobus Horčický de Tepenec and the Prague alchemist Georg Baresch, who was already frustrated by its unreadability. Baresch sent samples of the mysterious writing to Kircher, hoping that the celebrated scholar might decipher it. He could not.
The book eventually entered Jesuit collections. In 1912, the Polish-born antiquarian bookseller Wilfrid M. Voynich purchased it from a Jesuit collection at Villa Mondragone near Frascati, outside Rome. Voynich enthusiastically promoted the mysterious volume and the possible Bacon connection, bringing it to the attention of scholars and cryptographers. The manuscript has carried his name ever since.
After Voynich's death, it passed through several owners before rare-book dealer H. P. Kraus donated it to Yale in 1969.
Why is the writing so difficult to crack?
Ordinary decipherments usually begin with anchors. The Rosetta Stone famously offered versions of related text in different scripts. Unknown languages can sometimes be attacked through recognizable names, known formulas, related languages or a securely identified writing system. Ciphers may yield when researchers know the underlying language or can infer how the encryption works.
The Voynich Manuscript denies researchers almost all of these advantages at once.
There is no bilingual key. No author is known. No title can be read. No outside document is securely written in the same script. The drawings rarely identify their subjects unambiguously enough to supply a dependable word. Researchers do not even know whether the symbols directly represent sounds, syllables, words, abbreviations, encoded letters or something more artificial.
Yet the text is not obviously random.
Words and glyphs occur in highly patterned ways. Certain characters prefer particular positions within words. Some terms cluster in particular sections. Statistical studies have found properties resembling those seen in natural language, and recent computational work continues to examine its vocabulary distributions, word frequencies and scribal variation.
That does not prove the manuscript is an ordinary natural language written in a strange alphabet. Clever encoding systems and structured generative methods can also create regularities. The central problem is that statistics can reveal structure without telling researchers what that structure means.
One author? Paleography suggests a team
Modern research has overturned another popular image of the Voynich Manuscript: that of one eccentric genius spending years filling an entire book with secret writing.
Paleographer Lisa Fagin Davis identified five distinct scribal hands in the Voynich text. Yale linguist Claire Bowern has highlighted this finding in discussing the manuscript, and computational analyses have found differences in word use associated with the proposed scribes.
If that interpretation is correct, the manuscript was a collaborative production. Several people knew how to write its mysterious system consistently enough to work on different portions of the same book.
That observation complicates the simplest hoax theories. It does not make a hoax impossible, but a meaningless prank now has to explain not only the book's scale and structured text but also coordinated participation by multiple trained writers on expensive parchment.
Other possibilities remain open. The writing might encode a known language through an elaborate cipher or abbreviation system. It could represent an otherwise unattested language or constructed system. It could combine meaningful and formulaic material. Scholars continue to debate what kinds of mechanisms could generate its unusual statistical properties.
Has artificial intelligence finally solved it?
No.
Claims that computers or artificial intelligence have “decoded the Voynich Manuscript” appear regularly in headlines, but there is still no accepted translation. As recently as 2025, Yale described the manuscript as resistant to translation and its script as unique and unidentified.
Computational methods are genuinely useful. Algorithms can compare symbol frequencies, test language models, group pages by vocabulary, detect scribal patterns and ask whether Voynichese resembles known languages or particular types of cipher. Machine learning can expose regularities that would be exhausting to find by hand.
But AI faces the same fundamental obstacle as a human cryptanalyst: there is no verified ground truth. A model can produce a plausible-looking interpretation without demonstrating that the interpretation is historically or linguistically correct. With enough assumptions about how characters should be mapped, which language to expect and which transformations to permit, it becomes dangerously easy to manufacture apparent words.
A convincing decipherment would need to do far more than turn a few selected strings into meaningful phrases. It should explain the writing system consistently across the manuscript, account for grammar and vocabulary, work on unseen passages, make sense of the illustrations and historical context, and allow independent researchers to reproduce the reading without continually changing the rules.
No proposed solution has yet cleared that bar.
The mystery may be solvable without a dramatic “code-breaking” moment
The popular image of Voynich research is a cryptographer suddenly discovering the key and reading the manuscript overnight. The real path may be slower.
Paleography can identify who wrote which pages without translating them. Material science can constrain when and how the book was made. Linguistics can describe word patterns without knowing the words. Imaging can recover faded marks. Historical research can reconstruct the manuscript's ownership and compare its illustrations with medieval traditions. Computational analysis can test whether proposed explanations actually fit the whole corpus.
Each method reduces the space of possibilities.
Even a definitive demonstration that the text is not decipherable in the conventional sense — for example, because it was generated through a particular nonsemantic process — would count as solving an important part of the mystery. Conversely, evidence that it encodes a genuine natural language would narrow the problem dramatically without immediately revealing which language it is.
For now, the Voynich Manuscript occupies a rare position between knowledge and ignorance. We can measure its parchment, analyze its ink, distinguish scribal hands, map its vocabulary and inspect every surviving page at enormous magnification. We know far more about the physical book than Wilfrid Voynich did in 1912.
And still we cannot read it.
That is what makes MS 408 so enduring. Its mystery does not depend on a lost treasure or a supernatural legend. The evidence is sitting in a climate-controlled library at Yale, digitized for anyone to inspect. The unknown is visible in plain sight: line after line of careful handwriting left by people who apparently understood exactly what they were doing — and who, six centuries later, have not yet told us what it means.
