Skip to content
clickidy
🎯 ALL GAMES

READING ¡ 19 MIN

THE BOOK
NO ONE CAN READ.

Roughly 240 pages of parchment, filled with flowing handwriting in an alphabet that matches no known language, and illustrated with plants almost nobody can name. Codebreakers, linguists and the NSA's own top cryptologist have all tried. Nobody has read a single sentence.

A real 15th-century book, not a Victorian hoax ◢

The manuscript is named for Wilfrid Voynich, a Polish-American antiquarian book dealer who acquired it in 1912, reportedly among a collection at the Jesuit Villa Mondragone near Rome. It's been held since 1969 at Yale University's Beinecke Rare Book & Manuscript Library, which has digitised the entire book and made every page freely viewable online. A University of Arizona radiocarbon team dated the parchment itself to 1404-1438, a result announced in February 2011 - so whatever it is, it's a genuine 15th-century object, not a Victorian or modern hoax dressed up as one. What that test dates is the animal skin, though, not the moment somebody wrote on it.

A script that follows rules, but breaks all of ours ◢

The text - cryptographers call it “Voynichese” - uses somewhere between 20 and 30 distinct glyph shapes, written left to right in a flowing, confident hand with no corrections or hesitation marks, as if the scribe already knew exactly what they were writing. Statistically it behaves like real language: word lengths, letter-frequency patterns and repetition all resemble natural text far more than random noise, the kind of internal structure you'd expect from something built on a simple, consistent rule - not unlike working out an unfamiliar alphabet one symbol at a time. But it matches no known language, living or dead, and no proposed key has ever produced a translation that other researchers can independently reproduce.

Herbs nobody can name, and figures in green pools ◢

The illustrations are almost stranger than the text. A long “herbal” section draws plants with roots, leaves and flowers that mostly match no species anyone can name. A handful are the exception, and they are the reason the book's date is still argued about: the Beinecke's own description of the manuscript reads several of them as New World specimens brought back by Columbus, which would put the writing after 1493. That identification is contested, and it does not fit the parchment. An astronomical section shows circular zodiac-style diagrams. A “balneological” section - the strangest of all - shows dozens of small nude female figures standing or sitting in green pools, connected by a network of pipes and tubes. A final short section is text-only, its paragraphs marked with little star-like bullets, as if it were a list or set of recipes. Beyond those few plants, none of the imagery has been matched to a known 15th-century scientific or herbal tradition.

The world's best codebreaker couldn't crack it either ◢

William Friedman, the cryptologist who led the American team that broke Japan's PURPLE cipher machine in World War II, spent decades of his career on the Voynich manuscript on and off, running a small study group on it into the 1960s. He eventually leaned toward it being an attempt at a constructed, universal philosophical language rather than an encrypted natural one - a theory, not a solution. In 2004, researcher Gordon Rugg proposed the opposite extreme: that a 16th-century hoaxer could have generated Voynichese-looking gibberish mechanically, using a table-and-grille method, to sell a fake “ancient” book to a wealthy buyer with no real content inside at all. Neither the language theory nor the hoax theory has been proven, and several claimed “decipherments” since - including a widely reported 2019 claim that it was written in an early form of proto-Romance - have been rejected by the wider academic community for not holding up when checked line by line.

Play the ciphers in this story

SCRIPTSHieroglyphics Translator

Write your name the Ancient Egyptian way.

𓄿-𓊃 A TO ZCODESCaesar Cipher Decoder

Shift the alphabet to encode - or brute-force all 25 to crack it.

SHIFT 1-25 ¡ ROT13CODESCodes & Ciphers

Morse, binary, Caesar, more - encode anything.

10 CIPHERS

More Codes & Ciphers history →

WHAT THE RECORDS SHOW ◢

The parchment of the Voynich Manuscript was made from animals that died between 1404 and 1438. That is a date for the skin, and only for the skin. Nothing has ever dated the writing on it.

That distinction matters more than it sounds, because three published documents date this book three different ways, and one of them will not sit with the others. You can check all three yourself.

Document Date it gives What it is based on
Yale's catalogue record for Beinecke MS 408 "[ca. 1401-1599?]" No basis stated in the record
The Beinecke's codicological description "s. XV ex-XVI", and "could not have been written before 1493" Plants in the drawings identified as New World specimens
The University of Arizona radiocarbon result 1404 to 1438 Carbon-14 in four samples of the parchment

Two of those three sit together without any trouble. The catalogue record's span is wide enough to contain the radiocarbon result, and to contain the description's century as well.

The collision is between the other two. Parchment that was ready to write on by 1438 cannot carry writing that was put there no earlier than 1493. That single 55-year gap is the whole argument, and it is the reason the manuscript's date is still open.

Key Takeaways

  • The radiocarbon test measured four samples of parchment, each about 1 by 6 millimetres, cut from four pages Yale conservators had picked out as never rebound or repaired.
  • The date belongs to the animal skin. The scientist who took the samples said sampling ink cleanly off the parchment is "currently beyond our abilities".
  • The Beinecke's own codicological description argues from the plant drawings that the book postdates 1493, which is 55 years after the latest date the parchment can be.
  • The 2009 materials report identified iron gall ink, azurite, red ochre and a copper green, and never dated anything. It is an inventory, not a verdict.
  • The text ink and the drawing ink cannot be told apart, but the page numbers, the quire numbers and a Latin alphabet written on folio 1 recto are three further inks, all different from one another.

What did the Voynich Manuscript radiocarbon test actually date?

Four pieces of skin, and nothing else on the book.

Greg Hodgins, of the National Science Foundation Arizona Accelerator Mass Spectrometry Laboratory, travelled to Yale to take them. "I sat down with the Voynich manuscript on a desk in front of me, and delicately dissected a piece of parchment from the edge of a page with a scalpel," he told the University of Arizona's news office. He cut four samples from four pages, each about 1 by 6 millimetres.

The pages were not picked at random. Yale conservators had already "identified pages that had not been rebound or repaired". A repair introduces recent carbon, and the measurement is sensitive enough for that to matter. Hodgins named the two contaminants he expected on a page margin: finger oils picked up over centuries of handling, and adhesive, in case a spot that now sits on the edge had once sat on the spine.

Each sample was then burned, stripping everything but its carbon and leaving "a small dusting of graphite at the bottom of the vial". That is what the accelerator weighs. The result, announced in February 2011, put the parchment between 1404 and 1438.

The writing was never in reach. "It would be great if we could directly radiocarbon date the inks, but it is actually really difficult to do," Hodgins said. He gave three reasons: ink sits on a surface in trace amounts, its carbon content is very low, and "sampling ink free of carbon from the parchment on which it sits is currently beyond our abilities". Some inks, he added, are ground from minerals and contain no carbon at all.

So the careful statement of the finding is narrower than the headline. The animals whose skins became these pages died in the first four decades of the fifteenth century. Somebody wrote on those pages afterwards. The size of the gap between those two events has never been measured.

Why does the Beinecke's own description argue for a later date?

Because it was written from the evidence inside the book, and that evidence points somewhere the parchment cannot follow.

The Beinecke's codicological description of MS 408, by Barbara A. Shailor, dates the manuscript to "Central Europe [?], s. XV ex-XVI [?]", meaning the end of the fifteenth century or during the sixteenth. Then it gives its reasoning, and the reasoning is a positive claim rather than a shrug: "The identification of several of the plants as New World specimens brought back to Europe by Columbus indicates that the manuscript could not have been written before 1493."

That is an argued earliest possible date. It rests on a handful of identifications inside a botanical section of 113 species that are otherwise unidentified, which is a thin base for a heavy conclusion, and it lands 55 years after the latest date the parchment can be.

The library has not withdrawn it. The catalogue record still carries the date span "[ca. 1401-1599?]" and the place "Central Europe?", and the Beinecke's current page for the manuscript still opens by saying it "was produced in Central Europe at the end of the fifteenth or during the sixteenth century".

Three possibilities survive. Either the plant identifications are wrong. Or the parchment sat unwritten for at least half a century before somebody used it. Or the radiocarbon result is wrong. Choosing between those three is exactly the work nobody has finished, and a reader who wants to know why this manuscript is still argued about will get further here than in any decipherment claim.

What did the ink and pigment analysis find, and what did it never claim?

On 14 and 15 January 2009 the manuscript was examined and sampled in the Sterling Library conservation laboratory. The work was done by McCrone Associates of Westmont, Illinois, and the report went to Kevin Repp, Curator of Modern European Books and Manuscripts at the Beinecke, on 1 April 2009. It is signed by Joseph G. Barabe, Senior Research Microscopist.

A few details in it are worth pausing on, because the study is easy to read as a library's own investigation. The report says the work "was performed under authority of the purchase order dated 11/28/2008 from ProOmnia Film & Video Promotion GmbH". It says the sampling locations were recommended by Dr Alfred Vendl of the Universität fßr angewandte Kunst in Vienna, "and in consultation with ProOmnia representatives". It says the samples themselves were forwarded to Andreas Sulzer on 27 February 2009. Yale received the report and later published it. Yale is not who the report says raised the purchase order.

The samples were lifted with "an extremely fine-pointed tungsten needle and a microscalpel" and examined by polarised light microscopy, infrared spectroscopy, X-ray diffraction and elemental analysis in a scanning electron microscope. The findings, in the report's own words:

  • The ink of the text and the ink of the drawings are both iron gall inks.
  • The blue paint is ground azurite, with minor cuprite.
  • The green is copper-stained organic material consistent with copper resinate, plus copper-chloride compounds consistent with atacamite, with no resins detected.
  • The red-brown is a red ochre, its crystal phases identified as hematite and iron sulfide.
  • A clear white sample was proteinaceous with a large amount of calcium carbonate, and "a mixture of glair (eggwhite) and calcium carbonate is likely".
  • The binding medium for the inks and the green paint is gum, though "the binder for Sample 20 is undetermined".

Not one of those is a modern synthetic, and the report itself points to a medieval recipe for the green, citing Theophilus on "Salt Green" in Book I, Chapter 35 of On Divers Arts.

But read what the conclusions do not say. Not one of them dates anything. The report identifies substances; it never states the manuscript's age and never uses the word forgery. Hodgins drew the same line from the other side: "It was found that the colors are consistent with the Renaissance palette", he said, "but it doesn't really tell us one way or the other, there is nothing suspicious there."

The report is also candid about its loose ends, which is a good reason to trust the rest of it. Trace copper and zinc in the inks are called "a little unusual", possibly from a brass inkwell, "the actual source is unknown". Two samples held small amounts of mercury with no red pigment to explain it, "so its origin remains unknown". X-ray diffraction on one sample found lead-bearing crystals while the elemental spectrum "shows no lead at all". Infrared spectra of the ink binder held sharp peaks not expected for a gum, suggesting other constituents "which remain unidentified as of this date".

What the inks say about the order the book was made

This is the part of the report that a reader interested in the manuscript should take away, and it is not about chemistry so much as sequence.

The analysts "found no significant differences between the writing inks and the drawing inks used throughout the document" and "tentatively conclude that the text and drawings were most likely created contemporaneously". One page was found where paint clearly lies over the writing. So on the material evidence, the illustrations and the script are not two separate projects separated by centuries.

Three other inks on the book behave completely differently. The report's conclusion is that "in all probability" all three "are all different from the text/drawing ink and from one another":

  • The page number sampled from folio 26 recto is an iron gall ink of a different formulation, carrying more iron than any text or drawing ink in the book.
  • The quire number is an iron gall ink of unusually low iron content.
  • The Latin alphabet written on folio 1 recto, sampled at the letter "a", is a third distinct ink. Here the report checks itself: so little material was available that it "lowers the level of confidence in any conclusions we might wish to draw".

That is physical evidence that the foliation and the quire marks were written in ink unlike the book's own, which is what you would expect if they were added later. It bears on a long-running provenance argument, because the Beinecke's codicological description attributes that foliation to the English astrologer John Dee, on the strength of a comparison of his Arabic numerals with two manuscripts in Oxford's Bodleian Library. The library's current page has moved away from that reading, saying research "has also cast serious doubt on suggested ownership by the English alchemist John Dee". The ink tells you the numerals are not the book's own. It does not tell you whose hand held the pen.

What is missing from the Voynich Manuscript

The book is not whole, and the codicological description says so precisely.

MS 408 is 102 leaves of parchment plus one of paper, measuring 225 by 160 mm, in a vellum case binding of the eighteenth or nineteenth century. It is not a stack of ordinary pages. It includes five double-folio, three triple-folio, one quadruple-folio and one sextuple-folio folding leaves, which is why a codex of 102 leaves is fairly described as running to about 240 pages. McCrone's examiners measured the whole codex at about 5 cm deep and recorded its condition as fair.

Fourteen leaves are named as gone: folio 12, folios 59 through 64 from the centre of one quire, folio 74, folios 91, 92, 97 and 98, and the central bifolium at folios 109 and 110. Whatever the manuscript says, a reader has never had all of it.

Folio 1 recto has been damaged in a more deliberate way. Under ultraviolet light it shows an erased ownership signature, read by the codicological description as "Jacobi de Tepenecz" and independently picked up in 2009, when ultraviolet examination "revealed the presence of a signature, probably that of Jacobj Ă  Tepenece". The same examination found the page blotchy in a pattern consistent with a chemical stain which, the report notes carefully, "it was said" had been applied to enhance the writing. Then it adds the sting: the stain "may also have had the long term result of washing away some of the original writing". Somebody trying to make this page more legible may have destroyed part of what was on it.

How far back does the ownership trail really go?

To a letter, and the letter is more cautious than the legend built on it.

The Beinecke keeps the original with the manuscript, and its Latin text is transcribed in full in the codicological description. Johannes Marcus Marci of Cronland sends the book to the Jesuit scholar Athanasius Kircher, flattering him that "sphinxes of this kind obey none but their own Kircher". He passes on what he has been told about its history: a Dr Raphael, Bohemian-language tutor to Ferdinand III, had said the book belonged to Emperor Rudolph, who paid its bearer 600 ducats, and thought the author was Roger Bacon the Englishman.

Then Marci does the thing almost nobody quotes. Having relayed the attribution, he refuses to endorse it: ego judicium meum hic suspendo, "here I suspend my own judgement."

So the Roger Bacon attribution, which gave this manuscript its early fame and its early nickname, enters the record as a second-hand report that the man reporting it declined to believe. The Beinecke's current verdict is blunter: "The association with Roger Bacon is pure speculation." It is also arithmetically hard. The Library of Congress name authority heading gives his dates as "1214?-1294", which puts his death more than a century before the earliest date the parchment can carry.

Even the letter's date is not firm. The description transcribes its closing as "Pragae 19 Augusti. AD 1666", and then adds a bracket of its own: "[or 1665?]".

So how old is the Voynich Manuscript?

The defensible answer has three parts, and only the first is a number.

The parchment dates to 1404 to 1438. The writing and the drawings went onto it at an unmeasured time after that, in inks and paints whose ingredients were all available in medieval and Renaissance Europe and none of which is anachronistic. The earliest hard documentary sighting of the book as an object is a letter of 1665 or 1666, and everything claimed about the two centuries in between rests on inference.

On the text itself, the correct finding is a null one, and the owning library states it plainly: "Cryptographic approaches have failed to decipher the text, if it is indeed encoded." That last clause is doing real work. It is not settled that there is a message to recover, which puts the Voynich in a different position from a cipher whose method is already known, where only the key is missing.

The Beinecke has stopped trying to keep up with people who think otherwise, noting that it "is unable to respond to individual theories about the manuscript due to the volume of proposals received". The manuscript has left the building exactly once since 1969, for an exhibition at the Folger Shakespeare Library in 2014 and 2015. It is completely digitised, and anyone can read the catalogue record, the codicological description and the 2009 materials report without asking permission from anybody.

Frequently asked questions

How old is the Voynich Manuscript?

Its parchment was made between 1404 and 1438, measured by accelerator mass spectrometry at the University of Arizona on four samples cut from four pages, and announced in February 2011. That figure dates the animal skin rather than the writing, and the two are not the same thing. The open question is a 55-year gap: the Beinecke's own codicological description argues from plants identified as New World specimens that the book "could not have been written before 1493", which the parchment date cannot accommodate. Yale's catalogue record for Beinecke MS 408 gives a much wider bracket of about 1401 to 1599, wide enough to contain both.

Does the radiocarbon date prove the manuscript is not a forgery?

It establishes that the parchment is genuinely fifteenth-century, and the 2009 materials analysis found nothing anachronistic in the inks or paints written onto it, which is as far as the physical evidence reaches. Neither test dates the moment the writing went onto the skin, because ink cannot currently be sampled free of the parchment's own carbon. The evidence therefore narrows the possibilities sharply without formally closing the question, and the scientist who took the samples summarised the pigment results as showing "nothing suspicious there".

Who commissioned the ink and pigment analysis?

The report by McCrone Associates, dated 1 April 2009, was addressed to a Beinecke curator, but it records the work as performed under a purchase order dated 28 November 2008 from ProOmnia Film & Video Promotion GmbH, and says the samples were forwarded to Andreas Sulzer in February 2009. The examination and sampling took place at Yale, in the Sterling Library conservation laboratory, on 14 and 15 January 2009. The Beinecke published the resulting report, which is why anybody can read it.

Has anyone deciphered the Voynich Manuscript?

No decipherment has been accepted. The library that owns the manuscript summarises the position as "Cryptographic approaches have failed to decipher the text, if it is indeed encoded", and the hedge in that sentence is deliberate, because it has never been established that the writing encodes a language at all. Claimed solutions arrive often enough that the Beinecke states it cannot respond to individual theories about the manuscript given the volume of proposals it receives.

Can you see the Voynich Manuscript in person?

Access to the original is restricted because of the manuscript's fragility and the number of requests, and scholars must submit a written proposal explaining why the original is necessary. The manuscript has physically left the Beinecke only once since it arrived in 1969, for the Folger Shakespeare Library exhibition of 2014 and 2015. Every leaf has been digitised at high resolution and is free to view, and facsimile editions published in 2016 and 2017 can be consulted in the reading room.

If undeciphered books are your thing, the Voynich is not the only one on the shelf. The Rohonc Codex is another illustrated manuscript nobody has read, the Beale Ciphers are a shorter and stranger problem, and Kryptos is the modern entry on the same list.

Sources