The Manuscript Nobody Has Ever Read

There is a book in a library at Yale that nobody can read.

It runs to around 240 surviving pages of vellum, covered in a flowing, confident script in an alphabet that appears nowhere else. It is illustrated throughout — with plants, with astronomical diagrams, with elaborate systems of pipes and pools, and with a great many small figures.

It has been studied by professional cryptographers, by medievalists, by linguists and by amateurs, continuously, for over a century. No one has produced a translation that anyone else accepts.

What is actually established

The vellum has been radiocarbon dated, and the result places the animal skins in the first half of the fifteenth century. That is a hard physical date for the material, though strictly it dates the skins and not the writing.

Analysis of the inks and pigments has found them consistent with materials available in that period. There is no evidence of anachronistic materials.

The script behaves statistically like a language rather than like random noise. Character frequencies and word structures show regularities — words repeat, some appear far more often than others, and the distribution has broad similarities to natural language. That is one of the strongest arguments against it being pure gibberish.

The book has a documented provenance going back to the seventeenth century, including correspondence about it between scholars of the time who were already unable to read it.

What is not established

Almost everything else.

Nobody knows what language or code underlies the script. Nobody knows who wrote it, or where. Nobody knows what it is for. The botanical illustrations do not reliably match known species, which has been read as evidence of anything from an unfamiliar regional flora to deliberate invention.

Claimed solutions appear regularly, often with a great deal of publicity. The pattern is consistent: a proposed decipherment produces readable text for a short passage and then fails to generalise to the rest of the manuscript, or the method turns out to be flexible enough to produce almost any desired result.

The two serious possibilities

The first is that it encodes a real text in a real language, through a cipher or a constructed script that nobody has cracked.

The second is that it is an elaborate fabrication — a book made to look like a valuable manuscript in order to sell it, with a script designed to look meaningful. That theory has been advanced seriously, and the statistical structure is the main objection to it: producing hundreds of pages of convincing pseudo-language by hand is a formidable amount of work for a forgery.

Both remain open. A third possibility, that it is a personal or mnemonic system meaningful only to its author, would explain the difficulty and is essentially untestable.

Why it keeps its grip

Undeciphered scripts are not rare. What makes this one unusual is that it is a complete, physical, beautifully made object that clearly meant something to somebody. Someone spent an enormous amount of time on it.

The honest position is that there is not currently enough information to settle it, and that the most likely route to an answer is not a flash of insight but an archival discovery — a letter, an inventory, a second document in the same hand.