THE VOYNICH MANUSCRIPT

THE CLAIM

That a book written six centuries ago in a script found nowhere else on Earth contains a message no one has ever read — or, in the rival telling, contains nothing at all, and is the most durable practical joke in the history of writing. Either way, the codebreakers of two world wars looked at it and went home empty-handed.

WHAT IS KNOWN

The manuscript takes its name from Wilfrid Voynich, the rare-book dealer who bought it in 1912 from a Jesuit collection in Italy. Its earlier paper trail runs through seventeenth-century Prague — a 1665/66 letter accompanying it claims Emperor Rudolf II once paid 600 ducats for it — before the record goes dark for two centuries. In 2009, University of Arizona radiocarbon testing dated the vellum to between 1404 and 1438, ruling out the popular Roger Bacon attribution and most theories of a modern forgery.

The book itself is roughly 240 vellum pages written in a flowing, confident, unknown script, organized into sections modern scholars label by their illustrations: botanical (plants matching no known species), astronomical (zodiac roundels, star charts), balneological (small figures bathing in networks of tubs and tubes), pharmaceutical, and pages of dense text. Statistical study has found two distinguishable "dialects" of the script, and the text obeys patterns real languages obey — including a Zipf-like word-frequency distribution — while violating others: words repeat in odd stutters, and the information structure is unlike any known natural language.

The failure record is distinguished. In 1921 William Romaine Newbold announced it was Roger Bacon's work in microscopic shorthand; his method was demolished within the decade. William Friedman, America's premier cryptologist, studied it for years without result. The twenty-first century has added a steady traffic of announced solutions — a 2019 "proto-Romance" reading, AI claims that it is enciphered Hebrew, and others — each dismissed by medievalists and linguists, usually within days, for producing translations that only the translator can reproduce.

LEADING THEORIES

• An enciphered natural language. The classical view: real content — herbal, medical, alchemical — hidden by a cipher or invented alphabet. Against it: fifteenth-century ciphers known to history are far simpler than anything that would defeat modern cryptanalysis.
• A constructed or private language. An invented script recording an idiosyncratic language, glossolalia, or a personal shorthand — which would explain why cryptanalysis fails: there is no cipher to break, only a lost key of one mind.
• A meaningless hoax — but a period one. Researchers have shown that quasi-Voynichese can be mass-produced with simple medieval-compatible techniques, suggesting a fraud built to sell an emperor a wonder. The text's language-like statistics are the standing objection: gibberish rarely organizes itself this well.
• Meaningful but never decipherable. The bleak middle position: too short a corpus, too strange an encoding, no bilingual crib — a real message with no surviving path back to it.

ASSESSMENT

The OSA notes with professional respect that the manuscript has survived every generation's best tool: the philologist, the wartime codebreaker, the statistician, and now the language model, each announced as its inevitable end, each defeated. The carbon date is the case's one hard fact; everything else is inference balanced on 240 pages of beautiful noise. Until a decipherment arrives that a second, independent scholar can apply and reproduce, the file remains what it has been since 1912: open, and enjoying itself.

FURTHER READING

• Wikipedia: Voynich manuscript
• Wikipedia: Wilfrid Voynich
• Wikipedia: Beinecke Rare Book & Manuscript Library (the manuscript is fully digitized in Yale's collections)