Skip to content
Above Top Secret
Unsolved Mysteries

The Voynich manuscript: undeciphered, and definitely medieval

Two hundred and forty pages of unknown script and impossible plants, in a book nobody can read. The carbon dating settled when it was made, which narrowed the question considerably.

ATS Field DeskSites and witnesses
2 min read
An old handwritten manuscript

The Voynich manuscript is held at Yale’s Beinecke Library and has been fully digitised.

The Voynich manuscript is a codex of about 240 surviving vellum pages, written in an unknown script, illustrated with plants that do not exist, astronomical diagrams, and pages of naked figures in what appear to be systems of pipes and pools.

Nobody has read it. Not because it is difficult, but because a century of professional and amateur effort — including by cryptanalysts who broke real ciphers in wartime — has produced no accepted decipherment.

What is documented

Illustration

It is named after Wilfrid Voynich, a book dealer who acquired it in 1912. Its earlier history is partial: a letter of 1665 accompanies it, and it can be traced with reasonable confidence to seventeenth-century Prague and the court of Rudolf II.

In 2009 the vellum was radiocarbon dated to roughly 1404 to 1438. Analysis of the inks and pigments is consistent with that period.

What the text is like

This is where it gets genuinely strange, and it is the reason the manuscript resists both decipherment and dismissal.

Statistically, the text behaves like language. Word lengths follow a natural distribution. Certain words recur in ways that track the subject matter of the surrounding illustrations — one set of words dominates the botanical pages, another the astronomical.

But it also has features no known language has. Words repeat consecutively far more often than natural language allows. There is very little of the variation in word structure that inflection produces.

The candidate explanations

A cipher. Attractive, and the difficulty is that no cipher of the period produces text with these statistical properties, and modern cryptanalysis has not broken it.

An unrecorded natural language, or a known one in an invented script. Various have been proposed and none has produced a consistent reading that other scholars can reproduce.

A constructed language, of the kind several European scholars attempted in this era.

Or elaborate nonsense — produced by a technique that generates language-like text without meaning, to be sold to a wealthy collector. This accounts for the statistics reasonably well and is currently the explanation with fewest difficulties.

Why it is worth keeping open

Because the nonsense hypothesis has a weakness that is not often stated: it requires somebody to have produced 240 pages of consistent, structured, illustrated gibberish, with internally coherent statistical properties, on expensive vellum, in the fifteenth century.

That is a great deal of disciplined work for a fraud, and the discipline is the part that resists explanation.

Join the discussion

Still open on this one:

  1. 01

    Does the text encode meaning at all?

    The most important question and possibly answerable statistically without decipherment. Work in this direction has been done and has not converged.

Join the discussion

Be the first to comment

Sources

Open them and check for yourself. Where a record is public, it is linked.

  1. Beinecke Rare Book and Manuscript Library

    Yale University

  2. The British Library

    British Library

0 comments