The Voynich Manuscript sits in the Beinecke Rare Book and Manuscript Library at Yale, catalogued as MS 408. It is roughly 240 pages of vellum, illustrated with plants no botanist can identify, naked women bathing in green pools, and zodiac wheels labeled in a script no living person can read. Carbon dating places the vellum between 1404 and 1438. The text has resisted every cryptanalyst who has tried it.

That list is long. William Friedman, the United States Army cryptologist who broke the Japanese Purple cipher during World War II, spent decades on the Voynich and died believing it was an artificial language. His wife Elizebeth Friedman, herself a celebrated codebreaker, agreed. Neither could prove it.

The manuscript surfaced publicly in 1912, when the Polish-American book dealer Wilfrid Voynich bought it from a Jesuit college near Rome. A letter tucked inside claimed it had once belonged to Emperor Rudolf II of Bohemia, who reportedly paid 600 ducats for it in the late 1500s. The trail before that is thin.

For most of the twentieth century, theories swung between two poles. Either the manuscript was a sophisticated cipher concealing Latin, Hebrew, or some other known language, or it was a deliberate hoax — beautiful nonsense produced to extract money from a credulous nobleman.

That does not mean the manuscript has been read. It means it probably is not nonsense.

The hoax theory gained ground in 2004 when Gordon Rugg, a researcher at Keele University, demonstrated that the statistical quirks of the Voynich text could be reproduced using a sixteenth-century device called a Cardan grille. His point was modest: it was possible to fake the manuscript. Whether anyone actually did remained open.

More recent work has pushed back. In 2020, the linguist Claire Bowern at Yale ran the Voynich text through analyses developed for genuinely undeciphered languages. The word-frequency distributions, she found, matched natural language better than random gibberish. Letters cluster in ways consistent with phonological rules. Words repeat at intervals that suggest grammar.

That does not mean the manuscript has been read. It means it probably is not nonsense.

Several research teams have since proposed identifications. In 2019, Gerard Cheshire at the University of Bristol announced the script was "proto-Romance," a hypothetical spoken Mediterranean language. His paper was widely criticized by medievalists and quietly distanced from by the university's press office. Linguists pointed out that proto-Romance, as Cheshire described it, does not appear to have existed.

A separate effort by the cryptographer Greg Kondrak at the University of Alberta used machine learning trained on hundreds of languages and suggested Hebrew as the closest match, with letters possibly reordered within each word. His team produced a tentative reading of the opening line that referenced "recommendations" and "priests," but acknowledged that without a longer parallel text, the result could not be confirmed.

The most cautious researchers now describe the manuscript as a constructed cipher built on a real underlying language, possibly an early modern European vernacular, encoded by a system that compressed or substituted letters in a way that destroyed obvious patterns while preserving deeper ones.

The illustrations add their own puzzle. Some plants resemble known species — one bears a passing resemblance to a sunflower, which would be impossible if the dating is correct, since sunflowers were not known in Europe before Columbus. Others appear to be composites of real plants. The astronomical drawings include what may be a depiction of the Andromeda galaxy, though that identification rests on enthusiasm more than evidence.

What seems likely, after a century of attention, is that the Voynich Manuscript was produced by someone who knew what they were doing. The script is consistent across hundreds of pages. The same scribal hand recurs. The vellum was expensive, the binding skilled. Whoever made it invested considerable effort.

Whether that effort concealed medical recipes, herbal lore, mystical philosophy, or an elaborate joke remains, for now, unknown. The manuscript has been fully digitized and is freely available online, which means anyone with patience can try. Many do.

Yale's curators report receiving proposed solutions roughly once a month. None has yet survived peer review.