The Object
From the prologue of a book in progress on the Voynich Manuscript.
The Beinecke Rare Book and Manuscript Library does not let in natural light. Not exactly. The building’s walls are thin panels of translucent Vermont marble, and what passes through them is something other than sunlight—diffused, amber, as if the air itself had been steeped in tea. From outside, on the granite plaza of Yale’s campus, the Beinecke looks like a modernist reliquary, a windowless cube floating above ground on four concrete piers. Inside, the impression shifts. The central glass tower rises six stories through the building’s core, and inside it some 180,000 volumes glow on their shelves in that honeyed, filtered light. The tower is climate-controlled, sealed off from the reading room. You can see the books. You cannot reach them.
I am here for one book in particular, though “book” is not quite the right word. I fill out a call slip at the front desk: MS 408. The form is simple. Name, institutional affiliation, date, the item number. There is no special protocol for this request, no additional paperwork, no declaration of intent. MS 408 is catalogued alongside thousands of other manuscripts in the Beinecke’s collection. The librarian does not raise an eyebrow.
And yet.
I sit at one of the long reading tables and wait. The room is quiet in the particular way that rare book libraries are quiet—not silence so much as a kind of collective restraint, the hush of people trying not to be the loudest thing in the room. A few other researchers sit at their own tables with their own manuscripts. A conservator moves through a far doorway carrying what looks like a map in a polyester sleeve.
The librarian returns with a foam cradle and, resting on it, a small volume bound in limp vellum. It is surprisingly modest. Roughly 230 by 160 millimeters—a little larger than a modern paperback, a little smaller than you expect. The binding is soft, pale, slightly rumpled, the color of old cream. There is no title on the cover. No author’s name. No tooling, no decoration of any kind. It looks, at first glance, like a notebook someone might have carried in a satchel.
She sets it down in front of me and I realize my hands are bare. This is correct. The Beinecke no longer requires gloves for handling most manuscripts. Cotton gloves, it turns out, reduce tactile sensitivity, making it more likely that a reader will tear a page or crack a binding. Bare, clean, dry hands are safer.11 This shift in conservation practice has been widely adopted by major research libraries over the past two decades. The British Library, the Folger Shakespeare Library, and the Bibliothèque nationale de France have all moved away from mandatory glove use for parchment and paper manuscripts. The policy change is recent enough that it still feels transgressive—like being told you can touch a painting.
I open the cover.
The first thing I notice is the handwriting. It moves across the page in a steady, practiced rhythm, left to right, in lines of clear, well-spaced words. The letter-forms are rounded, regular, and confident. Whoever wrote this did not hesitate. There is no sign of copying difficulties, no crossed-out passages, no false starts. The ink is brown—iron gall, the standard writing ink of medieval and early modern Europe—and it sits firmly on the calfskin vellum, six centuries old and still legible. The text fills the page in neat blocks, leaving room for illustrations in the margins or, on some folios, occupying the spaces between larger painted figures.
It is, on first impression, a perfectly ordinary medieval manuscript.
Then I start to read.
I cannot read it. No one can.
The script is not Latin, not Greek, not Hebrew, not Arabic, not any known alphabet. It is composed of perhaps twenty to thirty distinct glyphs—the exact number depends on how you count certain variant forms—assembled into words that have clear boundaries, arranged in lines that move from left to right across the page. Some characters resemble Latin letters: what looks like an o, something close to an a, forms that suggest c or e. Others are wholly unfamiliar—looping, cursive shapes with no obvious parallel in any European or Middle Eastern script. Scholars call this writing system Voynichese, after Wilfrid Voynich, the Polish-Lithuanian book dealer who purchased the manuscript in 1912 and introduced it to the modern world. The name is imperfect—it credits the buyer, not the maker—but it has stuck.
I turn the pages carefully, one at a time, and the strangeness deepens.
The manuscript divides, roughly, into sections. Scholars have assigned them names based on their illustrations, though even these labels carry assumptions we should be cautious about.
The first and largest section is botanical. Page after page of plants, painted in greens and browns and pale yellows, each one filling most of a folio, with blocks of Voynichese text arranged around them. At a glance, they look like the herbal illustrations found in hundreds of medieval manuscripts—a plant shown whole, from flower to root, in the tradition of Dioscorides and his many imitators. Some of the plants seem almost recognizable. A leaf shape suggests a water lily. A flower head could be a sunflower, or a daisy, or neither.
Look closer. The roots are wrong. They fork and bulge in ways no actual root system does, or they spiral into coiled forms, or they terminate in shapes that resemble tubers from one plant grafted onto the stem of another. The leaves, on second inspection, belong to no single species. A flower sits atop a stem that could not support it. Some of these plants may be stylized depictions of real species—medieval botanical illustration was not always precise, and artistic convention varied widely. But others appear to be composites, or inventions, or something else entirely. Over a century of effort by botanists, historians of science, and amateur enthusiasts has failed to produce a consensus identification for more than a handful of these plants.22 Various researchers have proposed identifications for individual plants. Some, such as the possible Nymphaea (water lily) on folio 2r, have gained modest support. Most remain contested. For a useful survey, see Tucker and Talbert, “A Preliminary Analysis of the Botany, Zoology, and Mineralogy of the Voynich Manuscript” (2013).
The botanical section alone occupies more than half the manuscript. It is followed by what researchers call the astronomical or cosmological section. Here the illustrations shift from organic to geometric. Circular diagrams fill the pages—concentric rings filled with text and small stars, radiating spokes, nested circles within circles. Some contain what appear to be zodiac symbols: a ram, a bull, twins, a crab. These are among the very few elements in the entire manuscript that correspond to any known iconographic tradition, and they have given scholars one of their few firm handholds.
Around the zodiac figures, written in a different hand from the main Voynichese text, are month names. They are not in Voynichese. They are in a Romance language—a Northern French or Occitan dialect, rendered in a slightly clumsy Latin alphabet. Mars, abril, may.33 The month names were first noted by various early researchers and have been analyzed in detail by several scholars. Their dialectal features point toward a Northern French or Occitan origin, which has implications for the manuscript’s provenance but has not settled the question of where, precisely, it was made. These labels are among the only readable text in the entire manuscript, and they raise as many questions as they answer. Were they added by the original scribes or by a later owner? Do they indicate where the manuscript was made, or only where it traveled? The month names are a small, bright window in an otherwise opaque wall, and like all small bright windows, they illuminate mainly their own edges.
After the astronomical pages comes the section scholars call biological, for lack of a better term. It is, by any measure, the strangest part of a strange manuscript.
Dozens of small naked female figures—painted with round bellies and thin limbs, their bodies unshaded and anatomically vague—stand or recline in pools of green and blue liquid. The pools are connected to one another by an elaborate network of tubes or channels that snake across the pages, sometimes branching, sometimes converging, forming a plumbing system of no recognizable kind. Some figures appear to hold objects. Others seem to float. The liquid in the pools is painted in washes of green and blue that bleed into the vellum. There is nothing overtly erotic about the figures; they are schematic, almost diagrammatic, as if the artist were more interested in their positions within the system than in their bodies as such. The overall effect is less like pornography and less like anatomy than like a diagram of some process—bathing, or circulation, or something for which we do not have a name.
The pharmaceutical section follows. Here are drawings of jars and containers—some recognizable as the kind of apothecary vessels depicted in other medieval manuscripts, others less familiar—alongside isolated plant parts: roots, leaves, stems, shown cut and separated from any whole plant. Text accompanies these as well, still in Voynichese, still unreadable.
Throughout all of these sections, the manuscript includes foldout pages. Some fold out once; others twice or more, expanding into large sheets that, when fully opened, are several times the size of a normal page. One foldout, in the astronomical section, extends to roughly six panels and contains an elaborate circular diagram dense with text and small illustrations. These foldouts are original to the manuscript’s construction—the vellum was cut and folded before the illustrations and text were added.
In total, 102 leaves survive out of what was likely an original 116. Some folios are missing from the middle of gatherings, suggesting they were removed deliberately at some unknown point. Counting both sides of those leaves, what remains amounts to roughly 204 pages of calfskin vellum covered in a script no one can read, illustrating subjects no one can confidently name, in a system no one can explain. Foldouts complicate the count further; some descriptions of the manuscript reach higher page totals by treating each panel as a page. This book uses 102 surviving leaves as the physical fact, and says “pages” only when the sense is plainly the surface you turn.
In 2009, researchers at the University of Arizona subjected four samples from the manuscript to radiocarbon dating. The results placed the vellum’s origin between 1404 and 1438, with 95% confidence.44 The radiocarbon analysis was conducted by Greg Hodgins at the University of Arizona’s Accelerator Mass Spectrometry Laboratory. See Hodgins, “Measuring the Age of the Voynich Manuscript” (2011). The dating applies to the vellum itself—the animal skin—and not necessarily to the ink or pigments, which could theoretically have been applied later, though most scholars consider this unlikely given the manuscript’s physical coherence. The vellum was made in the first half of the fifteenth century. This is one of the few things about it that is not in dispute.
Let me draw the line clearly, because this book depends on it.
Here is what we know.
The vellum dates to the early fifteenth century. The manuscript is written primarily in an undeciphered script using twenty to thirty distinct glyphs, depending on the classification system.55 The glyph count varies among scholars because certain forms may be variants of the same character or distinct characters in their own right. The most widely used transcription alphabet, developed by the European Voynich Alphabet (EVA) project, identifies approximately twenty to twenty-five basic characters. The text is written left to right with clear word boundaries. In 1976, a Navy cryptographer named Prescott Currier demonstrated that the text could be divided into two distinct “languages”—now called Currier A and Currier B—which differ in their statistical properties and glyph frequencies, though both use the same character set.66 Prescott Currier presented this analysis at a seminar at the Naval Academy in 1976. His paper, “Some Important New Statistical Findings,” remains a foundational document in Voynich studies. The text obeys Zipf’s law, which describes the frequency distribution of words in natural languages: the most common word appears roughly twice as often as the second most common, three times as often as the third, and so on.77 Zipf’s law compliance was demonstrated by several researchers independently. It is a necessary but not sufficient condition for the text to be a natural language—some artificially generated texts can also satisfy Zipf’s law. The glyphs follow strict positional rules—certain characters appear only at the beginnings of words, others only at the ends, still others only in the middle. Lisa Fagin Davis, a paleographer, has argued that five distinct scribal hands can be distinguished in the manuscript, which would mean multiple people contributed to its production.88 Lisa Fagin Davis, “How Many Glyphs and How Many Scribes? Digital Paleography and the Voynich Manuscript,” Manuscript Studies 5, no. 1 (2020). The five-scribe reading is influential and remains contested; a 2025 computational critique argued that continuous variation in a single hand may be more parsimonious. This book treats the finding as a serious hypothesis, not as consensus.
These measurements have been tested and, in the main, replicated. They tell us that the manuscript is a genuine medieval artifact, that its text has internal structure, and that its production was careful, consistent, and almost certainly collaborative.
Here is what we do not know.
We do not know what the text says. We do not know what language, if any, lies beneath the script. We do not know whether it is enciphered natural language, a constructed language, an elaborate code, or something else. We do not know who wrote it, or where, or for whom. We do not know what the plants are, or whether they are meant to depict real plants at all. We do not know what the circular diagrams represent—whether they are astronomical, astrological, cosmological, or none of these. We do not know what the naked figures are doing in their connected pools, or what the pools and tubes signify. We do not know why some pages were removed, or by whom, or when. We do not know how the manuscript traveled from its place of origin to the court of Rudolf II in Prague, where it first surfaces in the historical record, or how it passed through the hands of the various owners who held it before Wilfrid Voynich found it in a Jesuit college in 1912.
We do not know, in short, almost anything that matters.
This asymmetry—between the solidity of what can be measured and the vastness of what cannot be interpreted—is the condition that has governed the manuscript’s modern life. For more than a century, it has attracted cryptographers, linguists, historians, botanists, computer scientists, artists, conspiracy theorists, and enthusiasts of every description. They have proposed solutions in dozens of languages. They have identified the author as Roger Bacon, as John Dee, as a Renaissance con artist, as a community of medieval women, as an anonymous Aztec scribe. They have found in its pages a lost herbal, a guide to women’s medicine, a work of alchemical philosophy, a meaningless hoax. Each solution has been presented with confidence. None has been accepted.
The manuscript endures. It sits on its foam cradle under the amber light of the Beinecke, 102 leaves of calfskin covered in a handwriting that is beautiful, consistent, and completely opaque. I turn another page. A plant with no name fills the folio, its leaves splayed in careful detail, its roots dissolving into spirals. Beside it, four lines of Voynichese run to the margin and stop.
I close the cover.
What happens when we encounter something that refuses to be known? Not something that is difficult, or obscure, or encrypted in a code we have not yet broken—but something that may be, in some fundamental sense, resistant to the very tools we bring to bear on it? What happens to the people who try? What do their attempts reveal—about the object, yes, but also about the methods, the assumptions, the desires they carry with them into the reading room?
I leave the manuscript with the librarian and step outside, into the flat Connecticut light. The plaza is bright after the Beinecke’s interior. Students cross it without looking up. The building behind me holds, among its hundreds of thousands of volumes, one small book that has outlasted every claim made upon it.
This is a book about that persistence. Not a book about solving the Voynich Manuscript. A book about what it has meant to try.
I cannot read it. No one can.