The Voynich Manuscript

An illustrated vellum codex of roughly 240 surviving pages, written entirely in an unidentified script that no one has been able to read since at least the 1660s, and quite possibly never could. Its parchment has been radiocarbon-dated to 1404-1438; its earliest secure documentary trace is a 1665 or 1666 letter sending it from Prague to Rome, which reports, at second hand and without the letter-writer's own endorsement, that it once belonged to Emperor Rudolf II and was believed to be the work of Roger Bacon, an attribution the radiocarbon date makes impossible on its own terms.

From the Jesuit library in Rome it was hidden from state confiscation in 1870, most likely at Villa Torlonia in Castel Gandolfo rather than the long-assumed Villa Mondragone, sold in 1912 to the antiquarian dealer Wilfrid Voynich, held by his widow and her circle until 1961, bought and unsuccessfully offered for resale by the dealer Hans Kraus, and given to Yale's Beinecke Library in 1969, where it has remained since. A century of claimed decipherments, from William Newbold's 1921 microscopic-shorthand theory to Gerard Cheshire's 2019 proto-Romance claim, has been examined and rejected by specialists one after another.

What remains genuinely unresolved, and is stated here as unresolved rather than settled in either direction, is whether the text encodes a real language, a working but unbroken cipher, or nothing at all. Named scholars hold each position, on record, and the manuscript itself has not yet given either side a conclusive answer.

A book that resists every category built for it

The Voynich manuscript is a vellum codex of roughly 240 surviving pages, written entirely in an unidentified script that has never been read by anyone since at least the seventeenth century, if it was ever read by anyone at all. It is illustrated throughout, and the illustrations are the only part of it anyone can describe with confidence: drawings of plants that match no reliable botanical identification, star charts and zodiac diagrams, small nude female figures standing or sitting in interconnected pools and pipes, jars and vessels that resemble pharmacy equipment, and pages of continuous unillustrated text broken only by star-shaped paragraph markers.

Cataloguers divide the manuscript into six sections by content rather than by any internal division the maker marked: a botanical section of about 113 unidentified plants, each given a full page and a block of text; an astronomical or astrological section of 25 diagrams including zodiac roundels; a section of small nude women in pools and connecting tubes, usually called balneological, or 'biological' by scholars who are unwilling to commit to the bathing interpretation; a cosmological section built around a large multi-page foldout of circular diagrams; a pharmaceutical section of more than a hundred labeled jars and herb fragments; and a final section of continuous text with no illustration at all, popularly called the 'recipes' section because of its short paragraph structure. No date, name, dedication, title or colophon appears anywhere in it.

How it was physically made

The support is calfskin vellum. Modern protein and codicological study, following the 2009 radiocarbon work, found no evidence that the parchment is a palimpsest and estimated that at least fourteen to fifteen whole calfskins were used to produce the volume, of ordinary to below-average quality, with natural flaws such as holes and thin patches left unrepaired in several places. The manuscript today runs to 102 foliated leaves plus one paper leaf, with a complex collation of single, double, triple, quadruple and one sextuple-folio fold-out leaves, some of them foliated inconsistently, which has made a full physical reconstruction of the original gathering structure difficult.

Chemical analysis of the ink and pigments was carried out by McCrone Associates on behalf of the Beinecke Library. The main text and drawings were done in an iron gall ink typical of the period. Folio numbers were added in a separate ink with unusually high iron content; the quire numbers, and a partial Latin alphabet added to folio 1 recto, were written in a third ink, high in carbon and very low in iron, bound with what appears to be a protein medium rather than gum. The blue paint is ground azurite with minor cuprite; the green is a copper resinate mixed with a copper-chloride compound consistent with atacamite; whites use egg white and calcium carbonate. McCrone's report found the materials and technique consistent with other fifteenth-century European manuscripts, and found no chemical difference in the inks used across sections that paleographers assign to different hands, which the report treats as a negative result for the idea that different scribes used physically different ink batches, not as proof that only one scribe worked on the book.

Paleographic study is a separate line of evidence from ink chemistry, and it tells a different story. Prescott Currier, working in the 1970s, was the first to identify two distinct statistical 'languages' in the text, which he called Currier A and Currier B, correlating loosely with two different hands concentrated in different sections, most visibly in the herbal pages. In 2020 the manuscript scholar Lisa Fagin Davis extended Currier's work using digital paleography and proposed as many as five distinct scribal hands across the volume, distinguished by consistent small differences in letterforms between sections rather than within them, which she has argued points to production by a working group over time rather than by one person working alone.

The earliest secure trace: a letter to Kircher

The earliest point at which the manuscript can be tied to a specific, datable, surviving document is a letter, most likely written in 1665 or 1666, from the Prague physician and rector of Charles-Ferdinand University Johannes Marcus Marci (Jan Marek Marci) to the Jesuit polymath Athanasius Kircher in Rome, sent together with the manuscript itself. The letter survives, bound into the front of the volume, and its text has been transcribed and translated many times since.

In it, Marci reports what he had been told by a third party, Dr. Raphael Mnishovsky, tutor in the Bohemian language to the future Emperor Ferdinand III: 'Dr. Raphael... told me the said book had belonged to the Emperor Rudolph and that he presented to the bearer who brought him the book 600 ducats. He believed the author was Roger Bacon, the Englishman.' Marci's own response to that claim is on record in the same letter: 'On this point I suspend judgment.'

Every later popular account of the manuscript that states flatly that Rudolf II owned it, paid 600 ducats for it, and believed it was the work of Roger Bacon is repeating a claim that Marci himself explicitly declined to endorse, reported to him at second hand, decades after the fact, by a man recalling a transaction from his own earlier life. It is the earliest attribution the manuscript has, and it is also, on its own terms, unverified tradition rather than record. The radiocarbon dating of the vellum to 1404-1438 makes the Roger Bacon authorship chronologically impossible on its own account, since Bacon died in 1292, more than a century before the parchment existed; whatever else the 600-ducats story preserves, it cannot preserve a true attribution to Bacon.

Before the letter: the search for Rudolf II and John Dee

Wilfrid Voynich, who bought the manuscript in 1912, was the first to try to fill in the roughly two centuries between the parchment's radiocarbon date and Marci's letter. He proposed that the English mathematician and astrologer John Dee had brought the manuscript to Prague and sold it to Rudolf II, since Dee and his associate Edward Kelley were resident at Rudolf's court in the 1580s and Dee is known to have owned and sold manuscripts to the emperor. This hypothesis is no longer accepted by specialists: no document ties Dee to this specific book, Dee's own detailed diaries do not mention it, and the theory rests entirely on the coincidence of Dee's presence in Prague at roughly the right period. A related and more colorful theory, that Dee and Kelley forged the manuscript themselves as a confidence trick on Rudolf, fails on the same radiocarbon evidence that undermines the Bacon attribution: the vellum was already a century and a half old by the time Dee and Kelley were active, ruling out the two of them as its makers, whatever their reputations as tricksters.

Archival research since the late 1990s, principally by the German researcher Karin Guzy working from Fugger and Rudolfine court financial records, found a documented purchase in March 1599: the bookseller Karl Widemann sold a consignment of books to the Augsburg financier Zacharias Geizkofler, who passed at least part of it on to Hans Popp, a chamberlain at Rudolf II's court, for 600 Rheinische gulden, a figure roughly equivalent, depending on the exchange rates used, to the 600 ducats Marci later reported. Whether this transaction is the same event Marci was describing, or merely a coincidentally similar number attached to an unrelated purchase, is disputed among specialists; the manuscript itself is not named or described in the surviving Widemann-Geizkofler correspondence, so the identification, however suggestive, is an argument, not a document.

Between the manuscript's creation and this cluster of claims about the 1590s, no owner at all is documented. That gap runs to roughly a century and a half and is recorded here as a gap, not bridged by supposition.

From Prague to the Collegio Romano

An ex libris inscribed on the manuscript's first folio identifies an early, datable owner: Jacobus Horcicky de Tepenec (sometimes called Sinapius), a Prague apothecary and naturalist who was ennobled by Rudolf II in 1608, since his ex libris uses the 'de Tepenec' form of his name that he adopted only after that date. How Tepenec obtained the manuscript, and whether it came to him directly from the imperial court, is not documented. Some time after Tepenec, and before 1637, the manuscript belonged to Georg Baresch, a little-known Prague alchemist who is independently documented writing to Kircher in 1639 about a mysterious book in his possession that he could not read, sending a partial transcription and asking for help; scholars now generally accept that this was the Voynich manuscript, on the strength of matching detail rather than an explicit statement in the letter.

Baresch died sometime between 1662 and 1665. Marci's letter states that he inherited the book from Baresch's estate before sending it on to Kircher in Rome. From that point, custody of the manuscript is documented as passing into the library of the Society of Jesus: Kircher, who lived in Rome at the Jesuit Collegio Romano until his death in 1680, is understood to have kept the manuscript there, and it remained part of the college's library holdings for roughly two centuries.

Hidden from confiscation: Villa Mondragone or Villa Torlonia

In 1870, the unification of Italy and the accompanying anticlerical laws threatened the Society of Jesus with the confiscation of Church property in Rome, including the Collegio Romano's library. To keep the most valuable holdings out of state hands, the Jesuits moved a substantial portion of the library's private collection out of the city under private ownership, principally that of the Jesuit Superior General Pieter Jan Beckx, and hid it from the confiscation. For over a century, and repeated in essentially every published account of the manuscript including earlier editions of this record's own reference material, that hiding place was said to be the Jesuit villa at Frascati known as Villa Mondragone, the location where Voynich was said to have found the manuscript in 1912.

That location has since been challenged by archival research. Voynich himself, in his own published and private accounts, never named Villa Mondragone; he referred only to 'a castle in Austria' in 1915 and 'an ancient castle in Southern Europe' in 1921, citing a confidentiality agreement. The Villa Mondragone identification appears to originate with Hans P. Kraus, who stated it publicly roughly thirty years after Voynich's death, apparently conflating the villa with the presence there of Voynich's Jesuit intermediary, Father Joseph Strickland. Research published in 2022 by Francesca Potenza, working from newly located 1912 inventory catalogues headed 'Catalogue of manuscripts of the archive of the house at Castel Gandolfo' and from letters between the Jesuit Father Augusto Spinetti and the Vatican Library's prefect Franz Ehrle describing an inspection of manuscripts at 'the Jesuit house in Castel Gandolfo' on 24 June 1912, argues that the collection, and with it the Voynich manuscript, was in fact kept at Villa Torlonia in Castel Gandolfo, not at Villa Mondragone in Frascati, and that no evidence places Voynich at either villa in person. This record follows Potenza's identification as the better-documented one and flags the long-standing Villa Mondragone story as an error traceable to Kraus rather than to any contemporary record.

Wilfrid Voynich's purchase, and what he actually said about it

The Jesuits sold a large part of their hidden library, some 30 or more manuscripts including this one, around 1911-1912 for financial reasons, in a transaction that both the Vatican and Voynich needed to keep quiet because Italian law still treated Church property acquired before 1870 as potentially subject to state claim. Voynich's own public account, repeated in a 1921 presentation in Philadelphia, was that he had personally discovered the collection, unknown to its owners, in chests in an old European castle. That account is now understood to be an embellishment: Voynich was invited to inspect and purchase part of the collection by its Jesuit owners, under a promise of absolute secrecy about the source, and did not discover anything by accident. The clearest evidence against Voynich's own story comes from his own household: his wife Ethel Voynich wrote down what she had been told about the transaction, including the Frascati association and Father Strickland's role, in a sealed letter that was opened roughly thirty years after it was written and confirmed a rather more mundane arrangement than the discovery story Voynich told the public.

The price Voynich paid the Jesuits for the manuscript specifically has not been published in any source consulted for this record; contemporary accounts of the sale of the broader collection mention a Vatican-related transaction in the tens of thousands of lire, but no figure has been reliably tied to this one item. What can be stated as documented is that Voynich acquired the manuscript in 1912, that it stayed in his possession until his death in 1930, and that its physical binding today still carries a typewritten label reading 'Ex Bibliotheca privata P. Petri Beckx,' recording its presence in Beckx's private library during the hidden decades.

Ethel Voynich, Anne Nill, and the sale to Hans Kraus

After Wilfrid Voynich's death in 1930, the manuscript passed to his widow, Ethel Lilian Voynich, herself a novelist. On her death in 1960 it passed briefly to Anne Nill, her longtime friend and Wilfrid Voynich's former secretary and business associate, who sold it in 1961 to the New York rare-book dealer Hans P. Kraus for 24,500 dollars.

Kraus, one of the most prominent antiquarian booksellers of the twentieth century, tried for years afterward to place the manuscript with a buyer, listing it at an asking price of 160,000 dollars, a figure that reflected the growing cryptological and popular fame the book had accumulated since Voynich first publicized it, rather than any settled scholarly value. No buyer came forward at that price. In 1969, after roughly eight years of failed private sale attempts, Kraus donated the manuscript to the Beinecke Rare Book and Manuscript Library at Yale University, where it has remained since, catalogued as MS 408.

What the 2009 radiocarbon dating actually established

In 2009, a team at the University of Arizona's Accelerator Mass Spectrometry Laboratory, led by Greg Hodgins, radiocarbon-dated four vellum samples taken from folios 8, 26, 47 and 68 of the manuscript, using samples between one and six millimeters in size cut from areas that Beinecke conservators had identified as unrebound and unrepaired original parchment. The measured carbon-14 fractions, calibrated against the IntCal curve, produced a combined date range for the parchment's creation of 1404 to 1438 at 95 percent probability, with all four samples statistically consistent with one another. The results were first shared publicly at a Voynich centenary event at Yale in May 2012.

The dating establishes when the animal whose skin became the parchment was alive, or very shortly after, since vellum is normally used within a few years of preparation. It does not, and does not claim to, establish when the text and illustrations were added to that parchment, and Hodgins himself has been explicit that direct radiocarbon dating of the ink was not attempted, because the trace quantities of carbon in ink applied to a surface make it extremely difficult to date independently of the substrate. The chemical consistency of the ink and pigments with other fifteenth-century materials, established separately by the McCrone analysis, supports but does not prove that the writing followed reasonably soon after the parchment's manufacture. Every claim that flatly states the manuscript 'is from 1420' or similarly collapses the parchment date and the writing date into one fact is going beyond what either study established.

A century of claimed solutions

The manuscript has attracted more announced decipherments than perhaps any other undeciphered document, and each has failed to persuade the specialist community. The first came from William Romaine Newbold, a University of Pennsylvania philosophy professor who worked on the text at Voynich's own request beginning in 1919 and announced in 1921 that he had found a system of microscopic shorthand embedded within the visible letters, supporting an authorship by Roger Bacon that fit Voynich's own hopes for the manuscript's importance. Newbold's full theory was published posthumously in 1928 as 'The Cipher of Roger Bacon.' It was dismantled in 1931 by John Matthews Manly, a University of Chicago medievalist and former wartime codebreaker, in the journal Speculum; Manly, despite having been a friend of Newbold's, showed that what Newbold had read as deliberate microscopic shorthand marks were simply natural cracks in the aged ink, and that Newbold's decoding method was flexible enough to extract almost any desired message from the same text.

Other announced solutions followed across the twentieth century without gaining acceptance: Joseph Martin Feely proposed in 1943 that the text was a private Latin shorthand, a claim scholars rejected outright; Leonell C. Strong and later Robert S. Brumbaugh each proposed different cipher systems yielding medical or herbal readings that other cryptanalysts could not reproduce; the Ukrainian-born engineer John Stojko proposed in 1978 that the text was Ukrainian with the vowels removed, a claim never accepted by Slavicists. William F. Friedman, the U.S. Army's leading twentieth-century cryptanalyst, spent decades of his career and organized two informal after-hours research groups of fellow military cryptanalysts, the First and Second Voynich Manuscript Study Groups of the 1940s, applying the same statistical and combinatorial methods used against wartime ciphers; Friedman never published a solution and by the end of his life concluded only that the text might encode an artificial or constructed language rather than a natural one in cipher, a guess rather than a decipherment.

Claimed solutions have continued into the twenty-first century. In 2014 the applied linguist Stephen Bax proposed a partial, bottom-up decoding identifying ten proper names, mostly of plants and one constellation, built from fourteen recurring symbol clusters matched against botanical and astronomical vocabulary in medieval Arabic, Hebrew and other sources; other Voynich researchers have found his character-to-name matches statistically indistinguishable from chance, and no further extension of his method was completed before Bax's death in 2017. In May 2019, the University of Bristol publicized a claimed full decipherment by its research associate Gerard Cheshire, who argued the text was a lost proto-Romance language and that the manuscript was produced by Dominican nuns for Maria of Castile, Queen of Aragon, publishing the claim in the journal Romance Studies. The claim drew immediate and pointed rejection from specialists in medieval Romance philology, who found that Cheshire assigned phonetic values to symbols without consistent method and then selected readings from multiple Romance languages until a plausible-sounding result emerged; the medievalist Lisa Fagin Davis, executive director of the Medieval Academy of America, called the work 'just more aspirational, circular, self-fulfilling nonsense,' and the University of Bristol subsequently withdrew its promotional coverage of the paper. As of this record, no proposed reading of the Voynich manuscript has achieved acceptance from the specialist community, and the Beinecke Library's own catalogue description states plainly that cryptographic approaches have failed to decipher the text, if it is indeed encoded at all.

Hoax, cipher, or unknown language: the live argument

Beneath the parade of failed solutions sits a harder and still unresolved question: whether the text encodes any message at all. In 2004 the psychologist Gordon Rugg, working with Gavin Taylor, proposed that the text could have been generated as meaningless gibberish using a simple mechanical method available in the fifteenth century, a table of syllable fragments combined with a moving grille or stencil, sometimes called a Cardan grille. Rugg demonstrated experimentally that this method could reproduce several of the text's more unusual statistical properties, including its close-to-binomial word-length distribution, and that a single person working with period materials could plausibly produce a manuscript of the Voynich's length, fully illustrated, in the range of 250 to 500 hours. Rugg's argument does not claim to prove the manuscript is a hoax; it claims only that the statistical regularities long cited as evidence of a real underlying language can also arise as an unintended side effect of a simple meaningless production method, which removes those regularities as proof of meaning.

Scholars arguing for a genuine underlying language point to a different body of statistical evidence. In 2013 the physicists Marcelo Montemurro and Damian Zanette published an information-theoretic analysis in PLOS ONE arguing that the long-range distribution of words across the manuscript's sections forms coherent semantic word-networks comparable to those found in real-language texts, structure that Montemurro has argued would be an implausible thing for a hoaxer working in the fifteenth century to have known how to fake, since the relevant concepts in information theory did not exist yet. Other statistical studies, including a 2014 network-based comparison by Diego Amancio and colleagues, have likewise found the text's internal structure closer to that of known natural languages than to random text in a majority of the tests applied. Neither side of this argument has persuaded the other, and this record states both positions, and the names attached to them, without resolving the underlying dispute, because it has not been resolved.

What is missing

The manuscript survives incomplete, and the gaps are old. Numbering on the quires and folios, added while more of the book was still intact, shows that at least six bifolios are missing from within existing quires, and that two entire quires, numbered 16 and 18 corresponding to gaps in the foliation of two leaves each, are missing altogether. Reconstructing the original page count from the surviving numbering gaps suggests the manuscript once ran to as many as 272 pages, of which roughly 240, or about 88 percent, survive today across 18 of an original 20 quires. Because the quire and folio numbers were added before the losses occurred, some of this material was already missing by the time Wilfrid Voynich acquired the manuscript in 1912, and nothing is known about when, how, or by whom the missing bifolios and quires were removed, or whether any of them survive unrecognized elsewhere.

One further identification once cited as physical evidence for the manuscript's date has not held up under scrutiny: an early claim that one of the botanical folios (f93v) depicted a New World sunflower, which would have required a composition date after Columbus's voyages, was reexamined by the computer scientist Jorge Stolfi, who found the resemblance to a sunflower superficial, based mainly on the plant's large central disc, a feature also found in Old World composite flowers already known in Europe before 1492. The identification is not now considered reliable evidence for the manuscript's date, and the radiocarbon dating of the vellum to 1404-1438 in any case places the parchment's manufacture before Columbus's voyages, whatever conclusion is eventually reached about the text and illustrations added to it.

The manuscript today

Yale digitized the entire manuscript and made high-resolution images freely available through its digital library, with the complete set online since 2020, following an earlier partial digitization in 2004; anyone can examine every surviving page, including the fold-outs, at full resolution without visiting New Haven. The Beinecke and Yale University Press published a facsimile edition, with scholarly commentary, in 2016, and a separate full-color facsimile was issued by the Spanish publisher Siloe in 2017. The manuscript has left the Beinecke building only once since arriving there in 1969, for the exhibition 'Decoding the Renaissance: 500 Years of Codes and Ciphers' at the Folger Shakespeare Library in Washington, D.C., in 2014-2015.

Research continues on the manuscript's codicology rather than on new decipherment claims from the library's own side: Lisa Fagin Davis's paleographic study of the multiple scribal hands, presented at Beinecke public programs, is part of an active project examining the manuscript's physical production as a codex, independent of any claim about what the text says. The Beinecke Library's own public description of the manuscript states its content, script, and origin as unresolved, records the standard provenance chain from Rudolf II onward as historically reported rather than uniformly proven, and does not endorse any decipherment. That caution, maintained by the institution that has now held the book for more than half a century, is itself part of the accurate record of what is and is not known about it.