Voynich Manuscript

The case in 60 seconds

The safest statement about the Voynich Manuscript is this: the book is real, medieval, and its text remains unreadable. But we also lack sufficient grounds to call it “entirely meaningless scribbling”.

The claim

“What world's language does the Voynich Manuscript speak—a book containing strange plants and astronomical maps unlike anything on Earth, which no cryptologist or linguist has been able to decipher for centuries?”

Within this viral statement, four different questions are conflated:

“It has not been deciphered” is true. That does not establish that it is “non-human”, “the language of a lost civilisation” or “from another world”.

What we know

The Voynich Manuscript is an illustrated codex written on parchment, held at Yale University's Beinecke Rare Book & Manuscript Library as MS 408 . Yale's catalogue describes the text as an “unidentified writing system”; the illustrations are broadly divided into herbal, astronomical/astrological, balneological, cosmological, pharmaceutical and recipe-like sections.

Its historical trail — a short version

• 1404–1438: The University of Arizona's radiocarbon analyses identify this interval as the most likely period in which the parchment was produced.
• Fifteenth century: Analyses of the inks and pigments show that the materials are consistent with medieval manuscripts.
• Seventeenth century: Johannes Marcus Marci's letter to Athanasius Kircher is an important link in the book's known early provenance; it recounts a second-hand story of a purchase involving Rudolf II.
• 1912: Bookseller Wilfrid M. Voynich purchases the manuscript from a Jesuit collection in Italy; it becomes known in the modern world by his name.
• 1969: H. P. Kraus donates it to Yale.
• 2009: Material analyses and radiocarbon results substantially weaken the possibility of a modern forgery.
• 2020: Lisa Fagin Davis argues for multiple scribal hands on the basis of palaeographic analysis; the most widely cited model proposes five distinct hands.
• 2026: Layfield and Davis publish two studies examining textual coherence and binding structure; the second argues that the page order in which we read the book today may not be its original arrangement.

An important nuance: radiocarbon analysis measures the age of the parchment. It does not, by itself, establish the precise year in which the writing was added. However, ink/pigment analyses and historical style are broadly consistent with production in the early fifteenth century.

The evidence

A) Physical evidence: a genuine medieval object

Yale's conservation work and the McCrone Associates material analysis show that the book's parchment, inks and pigments are consistent with medieval manuscript production. In 2009, radiocarbon curves from four different parchment samples proved mutually consistent, and the combined interval was reported as 1404–1438.

B) The text does not appear random

Two separate studies published in PLOS ONE in 2013 reached an important point of agreement. Amancio and colleagues reported that the Voynich text largely resembles natural languages on various network and word-statistical measures; Montemurro and Zanette reported topic-like clustering and information-bearing word networks in the long-range distribution of words.

These studies do not claim to have translated the text. Their finding is narrower but important: the distribution of symbol sequences across the pages is more regular than would be expected from simple random scribbling.

C) The text also differs strangely from natural languages

In 2022, Matlach, Janečková and Dostál found strong autocorrelations in the repetition behaviour of symbols that differed from the 21 natural languages and some encrypted examples they tested. “There is language-like structure” therefore does not mean “it behaves like one of the natural languages we know”.

This peculiarity leaves room for both a distinctive encoding system and rule-based artificial generation as hypotheses.

D) Sections and word use are related

There are measurable relationships between changes in word distribution and the visual sections conventionally labelled “herbal”, “astronomical” and so on. This reduces the likelihood that the drawings are merely decorations entirely independent of the text.

E) 2026: we may be reading the pages in the wrong order

This study does not decipher the text. But it changes something important: some “meaningless transitions” might be caused not by the language, but by a later rearrangement of the pages.

Weight of evidence — at a glance

Five major questions

1) Is the text really a language?

The safest answer: it has statistical structure resembling natural language, but it has not been established that it must be a natural language written directly.

In 2021, Yale linguists Claire Bowern and Luke Lindemann reviewed the linguistic research and argued that the natural-language hypothesis should be taken seriously. Other researchers, however, show that Voynich-like structures can also be obtained through meaningless but rule-governed generation.

2) Is it a cipher?

It could be. But simple substitution ciphers of the familiar “letter A equals this symbol” kind have long appeared inadequate. Word positions within pages and lines, symbols occurring only in particular places, and word families may require a more complex mechanism: an abbreviation system, syllable-based code, multilayered cipher, steganographic structure, or a combination of these.

In 2022, Bowern and Gaskell showed that some word-level measures can remain consistent with certain forms of encryption. The inference “its statistics resemble language, so it cannot be a cipher” is therefore incorrect.

3) Could it be meaningless fabricated text?

Yes; that has not been entirely ruled out. In 2004, Gordon Rugg showed that Voynich-like words could be produced quickly using tables and methods resembling the Cardan grille known in the Renaissance. This refutes the claim that “such a hoax was technically impossible in the fifteenth century”.

The difficulty, however, is that an explanation must account not only for the appearance of the words, but also for long-range distributions, differences between sections, repetition patterns and possible behaviour of multiple scribes. A hoax is possible, but it has not been “easily proven”.

4) Do the plant drawings depict real species?

According to the cautious descriptions in Yale and PLOS sources, most plants cannot be unambiguously matched to a single real species. Several ordinary explanations are possible: stylisation within the medieval herbal tradition, combinations of parts from different plants, distortion through successive copying, or symbolic/functional drawing.

“Plants unlike anything on Earth” is a visually appealing statement, but scientifically weak. Failure to identify a plant does not mean it does not exist in nature.

5) What do the astronomical/astrological diagrams show?

Some motifs, such as zodiac signs and star shapes, can clearly be related to the medieval astrological tradition. Nevertheless, it is unknown whether the diagrams specifically represent a calendar, astrological calculations, medical astrology, cosmology, or some other system.

The problem here is the same: we recognise the type of image, but cannot read its caption. Until the text is deciphered, we must settle for strong but uncertain interpretations of the diagrams' purpose.

Theories

There is no “single grand theory” of Voynich. A better approach is to examine which data several competing models explain well or poorly.

Social media / internet claims

What remains unknown

The real mystery: This is not a book about which we know nothing. On the contrary, there is a surprising amount of data about its physical history and textual structure. What is missing is the key that brings all these data together into one testable, readable explanation.

My assessment

In my view, the most important message of the evidence is this: presenting Voynich as an “alien language” inflates the mystery while diminishing the scientific question. The real problem is much more interesting: why does the writing system of a book physically consistent with fifteenth-century Europe resist modern linguistics and cryptology so persistently?

My minimum test for taking a proposed Voynich solution seriously would be: the same rules must yield consistent text on previously unused pages, not merely on a few selected words; other researchers must reach the same result using the same key. So far, no popular “I've solved it” claim has passed this threshold in a way that has gained broad acceptance.

Status