By Emilia Wellesley · Published May 5, 2026 · Updated May 6, 2026
What Is the Voynich Manuscript?
The Voynich Manuscript is a 234-page early fifteenth-century codex written in an undeciphered script, illustrated with unidentifiable plants, naked figures bathing in green pools, and circular astronomical diagrams. Carbon-dated to 1404-1438 and held at Yale’s Beinecke Library as MS 408, it has resisted every serious translation attempt for more than a century. No one has reliably read a single sentence.
The manuscript arrived in modern hands in 1912, when the Polish-Lithuanian rare-book dealer Wilfrid Voynich (1865-1930) bought it from the Jesuit library at the Villa Mondragone outside Frascati. He believed it had once belonged to the thirteenth-century English friar Roger Bacon and spent the rest of his life trying to prove it; he died in 1930 with the script as locked as he had found it. The book is small in the hand — about 23.5 by 16.2 centimetres — and its vellum is soft, almost suede-like, the kind of skin a working scribe would have considered ordinary stock for a private notebook (Beinecke Rare Book and Manuscript Library catalogue entry, MS 408).
What follows is a working historian’s account of what the codex physically is, what its illustrations show, who has tried to read it and why each attempt failed, and what current scholarship can responsibly say. Where the record is silent, the silence is named. As the Encyclopaedia Britannica entry on the Voynich manuscript notes, no consensus reading exists; the manuscript sits inside the broader landscape of historical and archaeological mysteries and remains the most stubborn open question in that field.
The Physical Object: Vellum, Ink, and Carbon Date
The codex is bound in 102 limp vellum folios, several of which fold out into composite pages. The largest — the rosettes folio (f86v) — opens to roughly six page widths and contains nine connected circular diagrams whose function remains unsettled. The 234 surviving pages preserve original foliation that suggests as many as fourteen folios are missing, including a stretch in the herbal section and what was likely a final colophon page.
Carbon Dating and the Forgery Question
In 2009 a team led by Greg Hodgins at the University of Arizona AMS Laboratory radiocarbon-dated four vellum samples to a 95% confidence interval of 1404-1438 [1]. The result ruled out the long-running theory that Voynich or a late nineteenth-century contemporary had forged the book. Producing a 234-page codex on blank period vellum would have been implausible in 1912 and effectively impossible earlier.
The Inks
Separate analysis at McCrone Associates in 2009 examined the inks under polarized-light microscopy and Raman spectroscopy. The text ink is iron-gall — the standard medieval recipe of oak-gall tannins precipitated with iron sulfate. The colored inks include malachite green, vermilion-tinted reds, and an azurite blue used sparingly on the astronomical roundels. The pigments are consistent with the early-fifteenth-century date but not diagnostic of a specific region (McCrone Associates technical report, 2009).
The Five Sections of the Manuscript
Scholars working from the imagery alone have divided the codex into five conventional sections. The names are descriptive, not original; the manuscript itself carries no chapter markers a modern reader can recognize. Each section repeats the same script, in the same hand or hands, and the sections appear to share an internal logic that the imagery alone cannot fully decode.
The Herbal Section
The longest section — about 130 surviving folios — pairs a single drawn plant with several paragraphs of text. The plants do not match any known botanical taxonomy. A few have plausible candidate identifications, but most are visibly composite, with leaves, roots, and flowers borrowed from real species and recombined. The botanist Sergio Toresella catalogued the plates in 1995 and concluded that the section follows the layout conventions of late-medieval alchemical herbals while showing none of their species.
The Astronomical and Zodiac Section
Roughly twenty-five folios show circular diagrams with small naked figures (called nymphs in the literature) standing in barrels or holding stars. Ten roundels carry month names in what appears to be Occitan or northern Italian dialect, written in a different hand and likely added later. The Pisces folio anchors the section in conventional zodiacal symbolism, and each zodiac roundel holds thirty nymphs — suggesting a degree-by-degree astrological scheme rather than a calendar.
The Biological or Balneological Section
Twenty folios depict groups of naked female figures bathing in green pools connected by ribbed tubes, anatomical chambers, and plumbing-like networks. Reading the section as balneological — concerning balneotherapy, the medical use of mineral waters — is the most common interpretation, though some scholars read the imagery as a coded gynaecological treatise. The art historian Jennifer M. Rampling has noted the visual debt this section owes to fifteenth-century alchemical “bath of the philosophers” imagery, where green water indicates the nigredo or dissolution stage [2].
The Cosmological Section and the Rosettes
The cosmological section culminates in the six-panel foldout known as f86v, the rosettes folio. Nine connected roundels depict cities, castle-like structures, suns, and waves. One roundel contains a structure crowned with swallow-tail merlons — a battlement style associated with Ghibelline fortifications in northern Italy in the fourteenth and fifteenth centuries. The detail anchors a substantial scholarly argument for an Italian region of production.
The Pharmaceutical and Recipe Sections
The final folios shift format. Small drawings of root and leaf fragments stand beside what look like apothecary jars; the closing pages drop the imagery entirely and present dense, star-marked paragraphs in the layout of a recipe book. The marginal star-glyphs recall medieval recipe collections, where a single mark introduced each new instruction.
The Script: What “Voynichese” Is and Is Not
The language of the manuscript — Voynichese, by convention — uses an alphabet of roughly 25-30 distinct glyphs depending on how one counts ligatures. The text reads left to right. Words are separated by spaces and grouped into paragraph blocks. The handwriting is fluid, consistent, and shows no signs of hesitation or correction; whoever wrote it wrote with confidence, and the same person or small workshop wrote nearly all of it (paleographic analysis published in Cryptologia by Stephen Bax, 2014).
Statistical Properties
Statistical analysis has produced two conflicting findings that shape the entire field. The text behaves like a natural language: word frequency follows Zipf’s law, characters cluster into stable position-classes within words, and vocabulary differs measurably between sections. At the same time, the text resists every standard cipher technique — no detectable polyalphabetic substitution, no plausible nomenclator key, no phonetic mapping to any known European or Asian language. The cryptanalyst Mary D’Imperio summarized the result in her 1978 NSA monograph: a text that is “either a very strange language or a very strange code” [3].
Hands and Scribes
In a 2020 Manuscript Studies article, the computational paleographer Lisa Fagin Davis identified at least five distinct scribal hands, with hand-1 contributing the bulk of the herbal section and hand-2 dominating the biological folios. Earlier analysts read the entire codex as the work of one person; close inspection of letterform variation now suggests a small workshop or a single scribe working across several years.
Theories That Have Been Tested and Rejected
For more than a hundred years the Voynich Manuscript has attracted decoders. The pattern of failure is itself informative; each failed reading rules out a class of explanation and narrows what is left.
The Roger Bacon Theory
Voynich’s own claim that the codex was written by Roger Bacon (c. 1219-1292) was disproved by the 2009 carbon date; Bacon died more than a century before the vellum was produced. Early Bacon-attribution decoders, including William Romaine Newbold (1865-1926), produced “translations” by reading microscopic shorthand they alone could see; subsequent inspection showed the marks were ink-shrinkage cracks.
The Hoax Theory
Computer scientist Gordon Rugg argued in 2004 that the manuscript could have been generated by a sixteenth-century hoaxer using a Cardan grille — a paper template with cut-out windows shifted across a syllable table to produce language-like nonsense. The grille method reproduces some statistical features of Voynichese but not the sectional vocabulary differences, and 234 pages of grille output would have demanded implausible labor for an undocumented market. The hoax thesis remains a minority position.
Recent “Solutions” That Did Not Hold
Several decipherment claims have appeared in the last decade. The University of Bristol scholar Gerard Cheshire announced in 2019 that the script was a self-reconstructed proto-Romance vernacular; the announcement was withdrawn after medieval linguists noted that the “translations” produced no internally consistent grammar. AI-based attempts using Hebrew or Turkic hypotheses have produced suggestive partial matches but no readable continuous sentence. Each headline candidate has failed the same test: produce one paragraph, in one section, that a second reader can verify.
What Current Scholarship Cautiously Holds
A small consensus has formed among working specialists. The manuscript is genuinely early-fifteenth-century. It was probably produced in northern Italy, on the basis of the swallow-tail merlons in the rosettes folio, the month-name dialect, and the herbal-imagery conventions. It encodes some kind of structured information, given its statistical regularities and sectional vocabulary, but whether that information is a constructed artificial language, a heavily abbreviated shorthand, or a sophisticated cipher is genuinely undetermined. The Beinecke catalogue entry for MS 408 documents the 1969 acquisition from the bookseller Hans P. Kraus, and the library treats the codex as an open research object [4].
What Would Count as a Solution
A genuine decipherment will need to read continuous text, predict the content of an unread section before it is examined, and produce a system other scholars can apply independently. Partial matches have appeared regularly; none has met the predictive bar. The manuscript may yet yield to that test. It may also turn out, as the historian of science Anthony Grafton has cautioned, that some objects from the past keep their content because they were designed by their makers to do so [5].
Why the Manuscript Still Matters
The Voynich is unsolved, and its survival exposes how thin the historical record actually is. A scribe in 1420 produced a substantial book — months of labor, the cost of a small herd’s worth of vellum — and the entire context of its production has vanished. No commission, no patron, no library catalogue entry survives from before its 1665 sale to Athanasius Kircher. The codex is a useful corrective to confident histories of the late medieval world.
It is also an instructive limit case. The Voynich rewards the slow disciplines — paleography, codicology, pigment analysis — and resists the fast ones. The manuscript sits within a wider tradition of Hermetic and alchemical books that hide their argument in image and code; readers can place it alongside the Magnum Opus and other alchemical great-works, where similar visual conventions structure encrypted material. Since 2014 the Beinecke’s open-access facsimile has brought amateur cryptographers and computational humanists into the field. The volume of attention has not produced a solution; it has produced a steadily improving understanding of what the manuscript is not, which is how progress in historical philology is usually measured.
Frequently Asked Questions
Where is the Voynich Manuscript kept?
The codex is held at Yale University’s Beinecke Rare Book and Manuscript Library in New Haven, Connecticut, catalogued as MS 408. High-resolution scans have been freely available since 2014.
How old is the Voynich Manuscript?
Radiocarbon dating at the University of Arizona in 2009 placed the vellum within a 95% confidence range of 1404 to 1438, ruling out both medieval Roger Bacon authorship and modern forgery.
Has the Voynich Manuscript been deciphered?
No verified decipherment exists. Several decoders have announced solutions, but none has produced continuous text that other scholars can independently verify.
Who was Wilfrid Voynich?
Wilfrid Voynich (1865-1930) was a Polish-Lithuanian rare-book dealer who acquired the manuscript in 1912 from the Jesuit library at the Villa Mondragone near Frascati, Italy.
What language is the Voynich Manuscript written in?
The script, conventionally called Voynichese, does not match any known European or Asian language. It uses an alphabet of about 25 to 30 distinct glyphs and shares statistical properties with natural languages, though no decoded mapping has held up under review.
Is the Voynich Manuscript a hoax?
Most specialists no longer treat the hoax thesis as the most likely explanation. The vellum is genuinely early fifteenth century, and the text shows linguistic structure hard to reconcile with random nonsense. A constructed language or a sophisticated cipher are the leading candidates.
Where was the Voynich Manuscript probably made?
Visual evidence — the swallow-tail merlons in the rosettes folio and the dialect of the month-name labels — points to northern Italy. The case is circumstantial but consistent across multiple independent details.
Why is the Voynich Manuscript so difficult to decode?
The script combines natural-language statistical features with resistance to standard cipher methods. Whatever system underlies the text was either deliberately constructed to resist substitution analysis or represents a language unrelated to known linguistic families.
Who owned the manuscript before Voynich?
Documented owners include the Jesuit scholar Athanasius Kircher (1602-1680), who received it in 1665. It appears to have circulated earlier in the courts of Holy Roman Emperor Rudolf II, possibly purchased from John Dee or Edward Kelley.
Can amateur researchers contribute to Voynich studies?
Yes. Since 2014, mailing lists, GitHub repositories, and preprint servers have brought amateur cryptographers, linguists, and computational humanists into the field. Most progress on the script’s statistical properties has come from this distributed community.
