---
title: The Voynich manuscript – the book nobody can read
description: >-
A fifteenth-century book in a script no one can read: the object, its history, its seven kinds of page and its writing.
published: 2026-09-25
tags: ["Voynich manuscript", "manuscripts"]
draft: false
series: The Voynich manuscript
card:
image: voynich/f2r
alt: a page of the Voynich manuscript with a single plant, root to flower, and text around it
---
In the Beinecke Rare Book & Manuscript Library at Yale there is a small book,
about the size of a hardback novel, written by hand on calfskin made in the
early fifteenth century. Nearly every page carries drawings: plants that no botanist
has managed to name, circles of stars, naked women bathing in green pools
connected by pipes. Around the drawings, and filling the pages between them,
there is text. It is neat and fluent, laid out in paragraphs, with what look
like words. Nobody can read a line of it.
The library catalogues it as MS 408. Everyone else calls it the Voynich
manuscript, after the bookseller who bought it in 1912 and spent the rest of his
life looking for someone who could tell him what it said.
This is the first of a series of posts about it, and it is only an
introduction to the object. It goes in this order:
1. **The book at a glance**: the basic facts in one table.
2. **How it surfaced**: Wilfrid Voynich and the purchase of 1912.
3. **The book as an object**: parchment, gatherings, missing leaves, ink, paint
and date.
4. **Where it had been**: the owners from Prague in the early 1600s to Yale in
1969.
5. **What is on its pages**: the seven kinds of page, each with an example.
6. **The script**: how the text sits on the page, the letters, how words are
built, the scribes, how the text is transcribed, and the little that is
written in Latin letters.
7. **Why nobody has read it**.
I have kept interpretation out as far as I can. Who has tried to read the book,
and how, is the subject of the [next](/posts/voynich-readers/) [two](/posts/voynich-measured/) posts. After that the series sets out
the measurements any explanation of the text has to reproduce, and then turns
to what such measurements show.
## The book at a glance
| | |
| --- | --- |
| **Holder** | Beinecke Rare Book & Manuscript Library, Yale University, MS 408 |
| **Size** | about 23.5 × 16.2 × 5 cm |
| **Material** | calfskin parchment, from at least 14 or 15 skins; limp goatskin cover |
| **Date of the parchment** | 1404–1438 (radiocarbon, 95 per cent probability) |
| **Extent today** | 102 leaves in 18 quires; about 240 pages if every panel of the foldouts is counted |
| **Extent originally** | at least 116 leaves in 20 quires; 14 leaves are missing |
| **Illustrations** | on almost every page, in brown ink with coloured washes |
| **Text** | about 38,700 words in all, some 35,000 of them in paragraphs; written left to right, in an unknown script |
| **Scribes** | five, in Lisa Fagin Davis's analysis (2020); the number is disputed |
| **First certain mention** | a letter from Prague, April 1639 |
| **In Yale's hands since** | 1969 |
The word count is of groups of letters separated by spaces, in the standard
transcription, counting the spaces its makers marked as uncertain as spaces
too. Calling them words already assumes something, and I come back to that
below.
## How it surfaced
Wilfrid Voynich was born Michał Habdank-Wojnicz in 1865 in Telšiai, then in the
Russian Empire and now in Lithuania, into a family of the Polish minor nobility.
As a young man he was arrested for his part in a Polish revolutionary circle and
sent to Siberia. He escaped, reached London in 1890 and moved among the Russian
émigrés there. Within a decade he had made himself into an antiquarian
bookseller, and a very good one, with shops in London and later New York. In
London he married Ethel Lilian Boole, the youngest daughter of the
mathematician George Boole. She was a novelist in her own right: her *The
Gadfly* sold in the millions in the Soviet Union.
In 1912 (or possibly 1911) the Society of Jesus quietly sold a collection of
about 280 manuscripts and 30 printed books that had come from the old library
of the Collegio Romano. Most of it went to Pope Pius X, who gave it to the
Vatican Library. Voynich bought part of the rest, and this book was among his
share. The sale was conducted in strict confidence, and Voynich kept it. In
public he said he had found the book in a castle, in chests whose owners did
not know what was in them. In 1937 the Jesuit origin came out anyway: slips
found in his London shop named the library the books had come from, and a
census of manuscripts printed it the same year. What remained secret until
Ethel's death in 1960 was the rest of the story, which a sealed letter she had
left supplied: the promise of secrecy, and the go-between, a Father Joseph
Strickland.[^history] For a long time the books were thought to have been kept
at Villa Mondragone, the Jesuit college at Frascati, and most accounts still
put the purchase there. More recent work in the Vatican archives points instead
to the Villa Torlonia at Castel Gandolfo.[^torlonia]
Voynich exhibited the manuscript, most famously at the Art Institute of Chicago
in 1915. By 1921, when he read a paper on it to the College of Physicians of
Philadelphia, he had settled on an author, the thirteenth-century English friar
Roger Bacon. That attribution has not survived, for reasons that come up below.
He died in 1930 without learning anything about what the text said.
## The book as an object
### Parchment, size and binding
The book measures about 23.5 by 16.2 centimetres and is a little over five
centimetres thick.[^physical] It is written on calfskin parchment, and at least
fourteen or fifteen whole skins went into it. The parchment is of ordinary
quality, with the holes, tears and stitched repairs common in any medieval
codex, but it was carefully prepared: the hair side and the flesh side are hard
to tell apart. The present cover is limp goatskin, put on in the eighteenth or
nineteenth century. Inside it the leaves are sewn on old fibre thongs,
possibly the original fifteenth-century ones, reinforced with leather in the
1960s.
### Quires, missing leaves and foldouts
A medieval book was made from quires: a few sheets of parchment folded in half,
nested inside each other and sewn through the fold. Each folded sheet is a
*bifolium*, two leaves joined at the spine. A standard quire in this book has
four bifolia, so eight leaves and sixteen pages. Someone numbered the quires at
the foot of the last page of each, with two exceptions, at a date nobody knows
but earlier than the leaf numbers; someone else later numbered the leaves in
the top right corner.
Those numbers show how much has gone. The quire numbers run to 20 and the leaf
numbers to 116, but only 18 quires survive, and fourteen leaves are missing:
leaf 12, leaves 59 to 64, leaf 74, the whole of quires 16 and 18 (leaves 91–92
and 97–98), and leaves 109–110. All of that was already gone when Voynich
bought the book. Some of the pages that remain are folded into it. Several open
out to two or three times the width of a page, and the largest, at the heart of
the book, opens out to about six pages' worth of parchment. Pages on foldouts
are numbered by panel, so leaf 67 has pages called 67r1, 67r2 and so on. That
is why the number of pages is always given as "about 240". The standard
transcription, which the counts in this post follow, divides the book into 227
page units: most foldout panels get a unit of their own, but the whole side of
the largest foldout is one.
### Ink, paint and date
The text and the outlines of the drawings are in iron gall ink, the standard
brown-black writing ink of medieval Europe, and a materials analysis judged
them in all probability the same ink. The colours were painted over the
outlines: blue from ground azurite, greens from copper compounds, a red-brown
ochre. The text ink and the green paint are bound in gum; the Latin alphabet
in the margin of folio 1r is in a different, carbon-based ink. That analysis, made in
2009 for a documentary film and published by the library, identified materials
a European workshop of the period would have used.[^mccrone] Multispectral
photographs of ten pages, taken in 2014, found no earlier writing scraped off
underneath, and nothing in the forensic work suggests that the book is a
palimpsest.
In 2009 four samples of the parchment, taken from different leaves, were
radiocarbon dated at the University of Arizona. They agree with each other, and
together they place the death of the animals between 1404 and 1438, with 95 per
cent probability.[^carbon]
That date belongs to the parchment, not the writing. Parchment could be stored
before use, and the dating cannot tell for how long. But writing on fresh
parchment was the normal thing, and the date fits what the drawings and the
handwriting suggest on their own. It rules out a thirteenth-century author such
as Roger Bacon. It makes the most famous sixteenth-century candidates, John Dee
and Edward Kelley, very unlikely, since they would have had to write on
parchment already a century and a half old. The old suspicion that Voynich
forged the book himself is disposed of by documents, not by the date: by 1912
the book had a documented history going back nearly three hundred years.
## Where it had been
### Prague
The history begins in Prague. The first page of the book carries a faded
signature in its lower margin, visible today only under ultraviolet light. It
names Jacobus Horčický de Tepenec, chemist to the Emperor Rudolf II and keeper
of his botanical garden, who used the title *de Tepenec* from his ennoblement in
1608. How the book came to him is not known. A similar inscription, with a
number beside it, appears in other books he owned. One doubt has been raised: it
does not match a signature of his on a document found in 2003.
The earliest certain mention of the manuscript is a letter of April 1639 from
Georg Baresch, an alchemist in Prague, to Athanasius Kircher, the Jesuit
polymath in Rome. Kircher had announced in a book of 1636 that he could read
Egyptian hieroglyphs, which he could not, and Baresch hoped he could do the
same for this book. He had sent Kircher a copy of some of its pages a year and
a half earlier. He would not part with the book itself. A historian of
Kircher's correspondence had recorded the letter, but its connection with the
manuscript went unnoticed until René Zandbergen, following that lead, obtained
a copy in 1999.[^history]
When Baresch died, the book passed to his friend Johannes Marcus Marci,
physician and rector of the university in Prague. In 1665 Marci sent it to
Kircher with a covering letter in Latin. The letter was still tucked into the
book when Voynich bought it, and it supplies most of what we know of the book's
early story:
> This book, bequeathed to me by an intimate friend, I destined for you, my very
> dear Athanasius, as soon as it came into my possession, for I was convinced it
> could be read by no one except yourself. […] Dr. Raphael, tutor in the
> Bohemian language to Ferdinand III, then King of Bohemia, told me the said
> book had belonged to the Emperor Rudolph and that he presented the bearer who
> brought him the book 600 ducats. He believed the author was Roger Bacon, the
> Englishman. On this point I suspend judgment.
This is where both Rudolf II and Roger Bacon enter the story. Marci passes both
on as hearsay, and says so. Rudolf certainly collected books of this kind, and
the imperial accounts record a payment of 600 florins in 1599 to the collector
Carl Widemann for books, delivered, in a clerk's words, in "a small barrel with
a couple of remarkable books".[^guzy] (Ducats and florins were different
coins; the hearsay and the accounts may still describe the same sale.) Whether
that was this book, nobody can say. The Bacon attribution Marci declined to endorse, and the radiocarbon date
has since disposed of it.
Someone in Prague also seems to have tried a key on the book itself. The
right-hand margin of the first page has a faint column of Latin letters.
Multispectral photographs show that it is really three columns: a Latin
alphabet, a column of the manuscript's own letters beside it, and a second
Latin alphabet shifted by one place. Lisa Fagin Davis, comparing the letter
forms, attributes it to Marci.[^fagin]
### Rome
After Marci's letter the book disappears for two centuries into Kircher's
library at the Collegio Romano, the Jesuits' university in Rome. Italian troops
took Rome in 1870, and in 1873 the new state's law confiscating Jesuit property
reached the city. Before it took effect, the Jesuits moved part of the library
into the private collections of members of the order, which were exempt. One
of them was that of Petrus Beckx, the head of the order: the books from it
carried slips naming his private library, which Voynich later removed to keep
his promise of secrecy. From there the book went, eventually, to the villa
where Voynich bought it.
### From Voynich to Yale
After Voynich's death the manuscript passed to Ethel and to their friend and
assistant Anne Nill. Ethel died in 1960. The next year Nill sold the book to the
New York dealer Hans P. Kraus for 24,500 dollars. Kraus asked 160,000 for it and
found no buyer, and in 1969 he gave it to Yale.[^history] The library put
near-complete colour scans online in 2004, and the whole book can now be paged
through in its digital collections.[^beinecke]
The whole chain, with the uncertain links marked:
| Owner, and when | How we know |
| --- | --- |
| Emperor Rudolf II, perhaps bought from Carl Widemann in 1599 | hearsay in Marci's letter; a matching payment in the imperial accounts |
| Jacobus Horčický de Tepenec, after 1608 | erased inscription on the first page, like those in his other books; one doubt raised |
| Georg Baresch, Prague, by 1637 until his death, between 1646 and 1662 | his letter to Kircher of 1639 |
| Johannes Marcus Marci, Prague, by 1662 until 1665 | his letter to Kircher of 1665 |
| Athanasius Kircher and the Collegio Romano, Rome, 1665–1873 | Marci's letter; then no records for two centuries |
| The private library of Petrus Beckx, from before 1873 | slips naming his library, removed by Voynich |
| Wilfrid Voynich, 1912 (or 1911) until 1930 | the sale, kept secret by Voynich; the Jesuit origin known from 1937 |
| Ethel Voynich and Anne Nill, 1930–1961 | inheritance |
| Hans P. Kraus, 1961–1969 | purchase |
| Yale University, MS 408, since 1969 | gift |
## What is on its pages
Nobody can read the text, so the book has been divided up by its pictures. The
section names are modern and descriptive: they say what the drawings look like,
not what the text is about. Here they are in the order of the book as it is
bound now, with the number of pages and words each takes up in the standard
transcription. Each is described below, with the quires it occupies and an
example page.
| Section | Pages | Words |
| --- | ---: | ---: |
| Herbal | 129 | 11,500 |
| Astronomical | 8 | 900 |
| Cosmological | 11 | 2,200 |
| Zodiac | 12 | 1,300 |
| Balneological | 19 | 6,400 |
| Pharmaceutical | 16 | 2,500 |
| Recipes | 25 | 11,600 |
Seven further pages, with about 2,300 words, have text and no drawing, among
them the first page of the book. The counts follow the transcription's 227 page
units, so the whole side of the largest foldout counts as one page, and they
include the labels beside the drawings.
### Herbal
*Quires 1–8, and again in quires 15 and 17. 129 pages, about 11,500 words.*
The first and by far the longest section in pages is the herbal; in words the
recipes at the end of the book are as long. Page after page shows
a single plant, drawn large, usually complete from the root to the flower, with
two or three blocks of text fitted around it.
```figure
{ "image": "voynich/f2r", "alt": "Folio 2r: a single plant with a red root, long narrow green leaves and two spiky flower heads, with blocks of text written above and below the leaves", "caption": "Folio 2r, from the herbal section. One plant, root to flower, with the text fitted around it. Beinecke Rare Book & Manuscript Library, Yale University." }
```
The plants look like the ones in the medieval herbals that were copied all over
Europe, and that resemblance is why the section is called a herbal. But only a
few have been identified with any confidence, and botanists disagree even about
those. Many look like composites, with the root of one plant, the leaves of
another and the flowers of a third. The drawing generally came first and the
text was written around it: lines stop short where they meet a leaf or a stem
and carry on on the other side.
### Astronomical
*Quire 9. 8 pages, about 900 words.*
The astronomical pages are circular diagrams, divided into sectors and rings,
with suns, moons and stars, short labels in the sectors and lines of text
running round the rings. Several are on foldouts.
```figure
{ "image": "voynich/f67r", "alt": "Folio 67r opened out: on the left a sun with a face and sixteen blue and red rays inside rings of stars and text; on the right a ring of twelve moons around a central star, with labels in red ink", "caption": "Folio 67r, opened out to its two panels, 67r1 and 67r2: a foldout from the astronomical section. On the right, twelve moons in twelve sectors, and some of the labels in red. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
### Cosmological
*Scattered through quires 8–10 and 14. 11 pages, about 2,200 words.*
The cosmological pages are a looser group: large circular or many-part
diagrams that are not obviously astronomical. Folio 57v is one of the most
curious pages in the book. Its four rings hold three kinds of writing:
running text, short separated words, and, in the second ring from the outside,
a sequence of single letters, each standing on its own.
```figure
{ "image": "voynich/f57v", "alt": "Folio 57v: four concentric rings of writing around four human figures and a flower-like shape at the centre; one ring holds single separated letters", "caption": "Folio 57v. Four figures at the centre, and four rings of writing. The second ring from the outside holds single letters set apart from each other. Beinecke Rare Book & Manuscript Library, Yale University." }
```
The largest foldout in the book belongs here too. It is usually called the
Rosettes. Opened out, it shows nine circles in a three-by-three grid, joined by
causeways, with towers, something like a castle and what may be a volcano, and
text in the rings. It is the largest single drawing in the manuscript.
```figure
{ "image": "voynich/f85v-86r", "alt": "The Rosettes foldout opened out: nine large circular rosettes arranged in a three by three grid, connected by bands, with small buildings and dense patterning", "caption": "The Rosettes foldout, folios 85v and 86r, opened to its full size. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
### Zodiac
*Quires 10–12. 12 pages, about 1,300 words.*
The zodiac pages are the one place where the pictures can be read with some
confidence. Each has a familiar sign at the centre: two fish for Pisces, a ram,
a bull, and so on. Around it are concentric rings of small naked female
figures, most standing in tubs or barrels and each holding a star on a short
line. There are about thirty figures to a sign. The cycle starts with Pisces,
not with Aries as it usually does, and the last two signs, Capricorn and
Aquarius, are missing with the leaf that once held them. Aries and Taurus each
take two pages, with fifteen figures on each.
```figure
{ "image": "voynich/f70v2", "alt": "Folio 70v2: two fish at the centre of a circular diagram, ringed by small female figures each holding a star, with a short label beside each figure", "caption": "Folio 70v2, the Pisces page. The word under the fish is written in Latin letters, in a different hand, and reads mars, March. Beinecke Rare Book & Manuscript Library, Yale University." }
```
Each figure has a label beside it, a single word in the manuscript's script,
and a line of text runs around each ring. If the labels are names of stars or
of days, nobody has managed to show it.
```figure
{ "image": "voynich/f70v2-labels", "alt": "Close-up of the outer rings of folio 70v2: figures in striped tubs holding stars, each with a short handwritten label beside it, and a line of text running round the ring above them", "caption": "The outer rings of the Pisces page, close up. One label to a figure, and a line of text round each ring. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
### Balneological
*Quire 13. 19 pages, about 6,400 words.*
The balneological section, the whole of quire 13, gets its name from its
bathing scenes. Naked women stand and sit in pools of green or blue water,
linked by tubes and channels that look sometimes like plumbing and sometimes
like something anatomical. Similar bathing scenes appear in medieval
manuscripts about the medicinal baths near Naples, and there too the water is
usually painted green. The text on these pages is dense, and it runs right up
to the drawings and continues on the other side of them.
```figure
{ "image": "voynich/f75r", "alt": "Folio 75r: naked women in a long green channel and a pool, with dense lines of text on both sides and below", "caption": "Folio 75r, from the balneological section. Beinecke Rare Book & Manuscript Library, Yale University." }
```
### Pharmaceutical
*Quires 15 and 19. 16 pages, about 2,500 words.*
The pharmaceutical section returns to plants, but in pieces. Roots and leaves
are drawn in rows, each with a short label, and down the margin stand tall,
elaborately painted containers that look like apothecaries' jars. Many of the
plant parts look like reduced versions of the drawings in the herbal.
```figure
{ "image": "voynich/f88r", "alt": "Folio 88r: three tall painted jars down the left margin, and rows of roots and leaves across the page, each with a short label, between paragraphs of text", "caption": "Folio 88r, from the pharmaceutical section: jars, labelled plant parts, and three paragraphs. Beinecke Rare Book & Manuscript Library, Yale University." }
```
### Recipes
*Quire 20, and both sides of folio 58 in quire 8. 25 pages, about 11,600 words.*
The book ends with the recipes: page after page of short paragraphs, more than
three hundred of them in all, with no drawings except a small star in the margin
at the start of each paragraph. The section's name comes from the look of the
pages, which resemble the collections of short medical recipes common in the
period. The stars are sometimes painted red, sometimes left plain, and some
have a tail.
```figure
{ "image": "voynich/f103r", "alt": "Folio 103r: a page filled with short paragraphs of text, each marked with a small star in the left margin", "caption": "Folio 103r, from the recipes section. Each paragraph starts beside a star. Beinecke Rare Book & Manuscript Library, Yale University." }
```
The order of the sections is the order of the book as it is bound now, and
there are good reasons to think it is not the order in which it was written.
Leaves may have been shuffled when it was rebound. That will come up again in
this series.
## The script
### How the text sits on the page
The first page of the book is a fair place to start.
```figure
{ "image": "voynich/f1r", "alt": "Folio 1r: four paragraphs of text with a few faint green leaf shapes behind them and two large red ornamental initials in the left margin", "caption": "Folio 1r, the first page. Four paragraphs, two red initials in the left margin, and faint green shapes behind the text. Horčický's signature, in the lower margin, cannot be seen in ordinary light. Beinecke Rare Book & Manuscript Library, Yale University." }
```
The writing runs from left to right in horizontal lines, and is grouped into
paragraphs. A paragraph usually opens with one of the tall letters described
below, often standing in the margin or rising high above the line. Its last
line is often short and pushed to the right, a habit you can see on folio 1r.
There is no punctuation of any kind and nothing that looks like numerals. The
lines are broken by spaces into groups of letters. They look like words, and
everyone calls them words, though nobody has shown that they are.
Running text in paragraphs is the commonest arrangement, but not the only one.
There are labels: single words written beside a drawing or part of one, like
the figures on the zodiac pages or the roots on the pharmaceutical pages. There
is text written in rings round circular diagrams, and text running outwards
along the spokes. And there are a few places, such as the second ring of folio
57v, where single letters stand on their own.
The hand is even and practised. It flows, and there are hardly any of the
crossings-out, corrections and second thoughts you would expect from someone
copying something unfamiliar.
### The letters
Close up, the script looks like no known alphabet, and yet many of its letters
look almost familiar. Here are sixteen words cut from two pages of the herbal,
each labelled with the transcription explained further down. Between them they
contain nearly every common letter.
```figure
{ "gallery": [
{ "image": "voynich/word-daiin", "label": "daiin" },
{ "image": "voynich/word-chokaiin", "label": "chokaiin" },
{ "image": "voynich/word-shodaiin", "label": "shodaiin" },
{ "image": "voynich/word-qotaiin", "label": "qotaiin" },
{ "image": "voynich/word-qokey", "label": "qokey" },
{ "image": "voynich/word-otchal", "label": "otchal" },
{ "image": "voynich/word-kydainy", "label": "kydainy" },
{ "image": "voynich/word-ypchol", "label": "ypchol" },
{ "image": "voynich/word-fodan", "label": "fodan" },
{ "image": "voynich/word-cthy", "label": "cthy" },
{ "image": "voynich/word-ckhy", "label": "ckhy" },
{ "image": "voynich/word-cphy", "label": "cphy" },
{ "image": "voynich/word-cfhodar", "label": "cfhodar" },
{ "image": "voynich/word-dals", "label": "dals" },
{ "image": "voynich/word-sor", "label": "sor" },
{ "image": "voynich/word-dam", "label": "dam" }
], "caption": "Sixteen words, cut from folios 2r and 8r at the resolution of the library's scan and shown at about three times the size they have on the page. Some cuts take in a stroke of a neighbouring word or a petal of the drawing. Beinecke Rare Book & Manuscript Library, Yale University." }
```
The letters fall into a few families. The transcription letters given for each are
explained in the section after next. For now they are only names.
- **Round letters: `o`, `a`, `y`, `d`.** A small circle; an *a*; a *9* with a
tail, rather like a *y*; and an *8*. They turn up anywhere in a word, and `y`
very often at its end.
- **Benches: `ch`, `sh`.** Two *c*-strokes joined by a bar along the top. In
`sh` a plume rises above the bar. They come at the start and in the middle of
words.
- **Gallows: `k`, `t`, `p`, `f`.** Tall letters with two upright stems and
loops, rising well above the line. `p` and `f` have a long stroke out to the
left, and they stand mostly in the first line of a paragraph.
- **Bench gallows: `ckh`, `cth`, `cph`, `cfh`.** A gallows standing on a bench,
as if on a pedestal. They come inside words and at their start.
- **The *e*-stroke: `e`.** A short stroke like a *c*, often doubled or
tripled. It sits inside words, after the benches and the gallows.
- **Minims: `i`, `n`.** Short downstrokes, like the strokes of a written *m*;
`n` ends in a curl. They come in runs before a closing letter: `in`, `iin`,
`iiin`.
- **Closing letters: `l`, `r`, `m`.** A looped *l*; a *2* with a tail; and
something like an *n* with a tail curling up. Mostly at the end of words.
- **Others: `q`, `s`, `g`, `x`.** `q` looks like a *4* and almost always
stands first in its word, followed by `o`. `s` looks like an *r* or a *2*
with a hook at the top. `g` and `x` are uncommon.
The gallows are the most striking. They are what makes a page of the manuscript
recognisable from across a room, and they are the letters that open a
paragraph. Some shapes resemble the abbreviation signs used in medieval Latin
manuscripts, and the `q` followed by `o`, written in one movement, looks very
much like a *4o*.
How many letters there are depends on how they are counted. Is `ch` one letter
or two? Is `iin` three, or one? Counted one way there are about twenty-five
distinct signs, counted another about forty. Nobody yet agrees which count is
right, and the answer matters a great deal for anything you might try to
measure about the text.
### How words are built
Even without reading a word, you can see that words are not put together at
random. They seem to be built from a few parts in a fixed order. A `q` stands at
the start and is followed by `o`. Gallows and benches come early or in the
middle. A run of `e`-strokes follows them. The minims and the closing letters,
`aiin`, `ain`, `ar`, `al`, `dy`, `y`, come at the end. Put together that way, the
parts give words like `qokeedy`, `chedy`, `shedy`, `daiin`, `okaiin` and `chol`,
and these come up again and again on page after page.
Words also repeat in a way that is unusual for written language. Here is part
of a line from folio 75r, after the drawing that interrupts it:
```figure
{ "image": "voynich/f75r-line13", "alt": "A single line of writing in which the same word appears three times in a row, followed by two similar words", "caption": "Folio 75r, line 13, the part to the right of the drawing: qokedy qokedy qokedy qokain olshedy. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
The same word three times in a row, and then two more that differ from it by a
letter or two. That is not rare in the manuscript. What it means, if it means
anything, is one of the questions the rest of this series is about.
### The scribes
Prescott Currier saw in 1976 that more than one hand had been at work. In 2020
the palaeographer Lisa Fagin Davis compared the letter forms page by page and
distinguished five scribes.[^hands] They wrote the same script, but each had
a recognisable way with certain letters, and they divided the book between them
roughly by section:
- **Scribe 1**, about 110 pages: most of the herbal in quires 1–7, and the
pharmaceutical section.
- **Scribe 2**, about 45 pages: the balneological section, and herbal bifolia
scattered through quires 4–7, between Scribe 1's.
- **Scribe 3**, about 35 pages: the recipes and part of the later herbal.
- **Scribe 4**, about 25 pages: the astronomical and zodiac pages and most of
the cosmological ones.
- **Scribe 5**, a handful of herbal pages.
The page counts follow the transcription, which leaves one page unassigned.
This is the standard view, and the first by a trained palaeographer, but it is
not settled. In 2026 Torsten Timm argued that the letter forms used to tell the
scribes apart occur almost everywhere, and that the five scribes largely restate
Currier's two kinds of text. No reply has yet appeared in print.
### Writing it down
To work with the text at all, it has to be turned into something that can be
typed and counted. Most work today uses EVA, a transcription alphabet designed
around 1998 by Gabriel Landini and René Zandbergen.[^eva] EVA gives each shape
a Latin letter chosen to look like it, not to sound like anything: the circle
is `o`, the *8* is `d`, the *4* is `q`, and the gallows are `k`, `t`, `p` and
`f`. The letter names in the list above are EVA. It deliberately records
strokes rather than deciding what counts as a letter, which is why the bench is
written as two letters, `ch`, and a run of minims as `iin`.
Here are the first five lines of the book, close up.
```figure
{ "image": "voynich/f1r-opening", "alt": "The first five lines of folio 1r, close up, in a looped, rounded script", "caption": "The first five lines of folio 1r. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
And here are the same five lines in the transcription by Zandbergen and Landini:
```text
fachys.ykal.ar.ataiin.shol.shory.[cth:oto]res.y.kor.sholdy
sory.ckhar.or,y.kair.chtaiin.shar.ase.cthar.cthar,dan
syaiir.sheky.or.ykaiin.shod.cthoary.cthes.daraiin.sy
soiin.oteey.oteo[s:r],roloty.cthiar,daiin.okaiin.or.okan
sair,y.chear.cthaiin.cphar.cfhaiin
```
A full stop is a space between words, a comma is a space the transcribers were
unsure of, and a bracket holds two readings of a letter they could not agree
on, such as `[cth:oto]` in the first line. Doubts of this kind are everywhere
in the manuscript. Spaces in particular are often narrow, and whether two
groups of letters are one word or two is a judgement call.
It looks like a language when it is typed out like this. It is important to
remember that it is only a labelling of shapes. Nothing about EVA says what any
letter sounds like or means.
### Writing in Latin letters
There is a little writing in the book in Latin letters, and it does not help
much. The month names on the zodiac pages, like the *mars* on the Pisces page,
are in a different hand and a different ink from the main text and were
probably added later. They are in a Romance language, often taken for French
or Occitan; where they come from has never been settled. The alphabet in the margin of
the first page has already come up. And on the last page, partly above and
partly inside the frame of the text, someone wrote four lines in Latin letters.
```figure
{ "image": "voynich/f116v-top", "alt": "Four lines of faded writing in Latin letters at the top of folio 116v, with words separated by small crosses", "caption": "The top of folio 116v, the last page. Four lines in Latin letters, with small crosses between some of the words. Beinecke Rare Book & Manuscript Library, Yale University.", "wide": true }
```
Parts of these lines look like Latin and parts like German. One phrase has been
read, following a suggestion of the art historian Erwin Panofsky, as *so nim
geismi[l]ch o*, "so take goat's milk, or …", which sounds like a recipe;
transcriptions of it differ. But no one has made the whole into sense, and it is not clear that
these lines have anything to do with the rest of the book.
## Why nobody has read it
Since 1912 the manuscript has been studied by professional cryptographers,
among them some of the best the American government employed in two world
wars, and by linguists, historians, botanists and statisticians. In the last
few years machine learning has had a go too. Readings are announced every year,
sometimes with some fanfare. Each one so far has failed under scrutiny: either
it could not produce the same reading twice from the same procedure, or it
produced something that no reader of the proposed language would accept as
that language.
The failure goes deeper than having no key. It is not even known what kind of
thing the text is. It could be a natural language written in an invented
alphabet. It could be a cipher hiding a known language, or a language written
in a way that leaves most of its vowels out. It could be an artificial language
invented for the book, or a text with no meaning at all, produced by some
procedure to look like writing. Each of these has serious defenders. The
evidence that would settle the question is the text itself, and the text is the
one thing nobody can read.
A genuine, well-dated medieval object, a large amount of fluently written text,
and no agreement even about what sort of text it is: that combination is what
makes the Voynich manuscript one of the best-known unsolved problems of its
kind. It is not a puzzle with a missing piece, where we know what the answer
will look like. Nobody yet knows what kind of answer to expect.
## What comes next
The [next](/posts/voynich-readers/) [two](/posts/voynich-measured/) posts tell the story of the attempts to read the
manuscript, and then to measure it, from its first owners to the present: who
tried, what they claimed, and why each claim fell. After that comes a post on method: the measurements any
explanation of the text has to reproduce, and how to take them so that they
can be checked, repeated and fairly compared. A question narrower than "what
does it say?", and one that can be answered, is what sort of process could have
produced text like this.
The manuscript is in the public domain, and the Beinecke has made every page of
it freely available. All the images in this post come from its digital copy,
and the photographs of pages can be clicked to open at twice the size. They are
worth the time.
[^history]: René Zandbergen's site [voynich.nu](https://www.voynich.nu/) is the
reference for the manuscript's history, and this post follows its account of
the provenance, the Baresch letter, the sale to Kraus and the gift to Yale:
see [*The Voynich MS — Its history*](https://www.voynich.nu/history.html).
The translation of Marci's letter quoted above is the one usually given;
voynich.nu quotes excerpts of it.
[^torlonia]: The case for Villa Torlonia rests on the archival work of Francesca
Potenza in the Vatican Library, as reported in Zandbergen's history pages
cited above.
[^physical]: The physical description follows R. Clemens (ed.), *The Voynich
Manuscript* (New Haven: Beinecke Rare Book & Manuscript Library and Yale
University Press, 2016), the facsimile with the library's own essays, and
the description of the quires and binding on
[voynich.nu](https://www.voynich.nu/descr.html).
[^mccrone]: J. G. Barabe, *Materials Analysis of the Voynich Manuscript*
(McCrone Associates, 2009),
[published by Yale University Library](https://library.yale.edu/sites/default/files/2026-06/voynich_analysis.pdf).
The analysis was commissioned by the makers of a documentary film.
[^carbon]: G. Hodgins, *Dating Results for Voynich Manuscript Parchment Leaves
8, 26, 47 and 68*, letter to Yale University Library, 17 February 2009,
[reproduced at voynich.nu](https://voynich.nu/papers/Carbon_GH_2009.pdf).
The combined result is 516 ± 18 radiocarbon years before present. The
letter itself makes the point that this dates the parchment and not the
writing.
[^guzy]: S. Guzy, "Book Transactions of Emperor Rudolf II, 1576–1612: New
Findings on the Earliest Ownership of the Voynich Manuscript", *CEUR
Workshop Proceedings* 3313 (2022). The seller was Carl Widemann, a collector
of alchemical and medical manuscripts.
[^fagin]: L. Fagin Davis, ["Multispectral Imaging and the Voynich
Manuscript"](https://manuscriptroadtrip.wordpress.com/2024/09/08/multispectral-imaging-and-the-voynich-manuscript/),
*Manuscript Road Trip*, 8 September 2024. The images were taken in 2014 by
the Lazarus Project, of ten pages only.
[^beinecke]: Beinecke MS 408 in the [Yale digital
collections](https://collections.library.yale.edu/catalog/2002046). The
images here are courtesy of the Beinecke Rare Book & Manuscript Library,
Yale University.
[^hands]: L. Fagin Davis, "How Many Glyphs and How Many Scribes? Digital
Paleography and the Voynich Manuscript", *Manuscript Studies* 5/1 (2020),
164–180, [doi:10.1353/mns.2020.0011](https://doi.org/10.1353/mns.2020.0011). The assignment of pages to scribes used in the list is the one
recorded in the Zandbergen–Landini transcription. The challenge is T. Timm,
*One Hand, Five Labels* (2026), a self-published deposit,
[doi:10.5281/zenodo.19169450](https://doi.org/10.5281/zenodo.19169450).
[^eva]: G. Landini and R. Zandbergen, EVA, the European Voynich Alphabet
(1998–2000); see [voynich.nu](https://www.voynich.nu/transcr.html). The
transcription quoted is Zandbergen and Landini's file ZL3b, and the word and
page counts in this post are taken from it by
`scripts/export-voynich-stats.mjs` in this site's source.