What is it about the Voynich Manuscript—that cryptic, illustrated 15th century text of unknown origin and meaning—that has so fascinated and obsessed scholars for centuries? Written in what appears to be an invented language, with bizarre illustrations of otherworldly botany, mysterious cosmology, and strange anatomy, the book resembles other proto-scientific texts of the time, except for the fact that it is totally indecipherable, “a certain riddle of the Sphinx,” as one alchemist described it. The 240-page enigma inspires attempt after attempt by cryptologists, linguists, and historians eager to understand its secrets—that is if it doesn’t turn out to be a too-clever Medieval joke.
One recent try, by Nicholas Gibbs, has perhaps not lived up to the hype. Another recent attempt by Stephen Bax, who wrote the short TED Ed lesson above, has also come in for its share of criticism. Given the investment of scholars since the 17th century in cracking the Voynich code, both of these efforts might justifiably be called quite optimistic. The Voynich may forever elude human understanding, though it was, presumably, created by human hands. Perhaps it will take a machine to finally solve the puzzle, an artificial brain that can process more data than the combined efforts of every scholar who has ever applied their talents to the text. Computer scientists at the University of Alberta think so and claim to have cracked the Voynich code with artificial intelligence (AI).
Computer science professor Greg Kondrak and graduate student Bradley Hauer began their project by feeding a computer program 400 different languages, taken from the “Universal Declaration of Human Rights.” While “they initially hypothesized that the Voynich manuscript was written in [ancient] Arabic,” reports Jennifer Pascoe, “it turned out that the most likely language was [ancient] Hebrew.” (Previous guesses, the CBC notes, “have ranged from a type of Latin to a derivation of Sino-Tibetan.”) The next step involved deciphering the manuscript’s code. Kondrak and Hauer discovered that “the letters in each word… had been reordered. Vowels had been dropped.” The theory seemed promising, but the pair were unable to find any Hebrew scholars who would look at their findings.
Without human expertise to guide them, they turned to another AI, whose results, we know, can be notoriously unreliable. Nonetheless, feeding the first sentence into Google translate yielded the following: “She made recommendations to the priest, man of the house and me and people.” It’s at least grammatical, though Kondrak admits “it’s a kind of strange sentence to start a manuscript.” Other analyses of the first section have turned up several other words, such as “farmer,” “light,” “air,” and “fire”—indeed the scientists have found 80 percent of the manuscript’s words in ancient Hebrew dictionaries. Figuring out how they fit together in a comprehensible syntax has proven much more difficult. Kondrak and Hauer admit these results are tentative, and may be wrong. Without corroboration from Hebrew experts, they are also unlikely to be taken very seriously by the scholarly community.
But the primary goal was not to translate the Voynich but to use it as a means of creating algorithms that could decipher ancient languages. “Importantly,” notes Gizmodo, “the researchers aren’t saying they’ve deciphered the entire Voynich manuscript,” far from it. But they might have discovered the keys that others may use to do so. Or they may—as have so many others—have been led down another blind alley, as one commenter at IFL Science suggests, sarcastically quoting the wise Bullwinkle Moose: “This time for sure!”
You can find the Voynich Manuscript scanned at Yale’s Beinecke Rare Book & Manuscript Library. Copies can be purchased in book format as well.