A 15th-century manuscript described as “the world’s most mysterious” contains a secret message, according to a new computer analysis.
The Voynich Manuscript is written in an unknown language and script - and the 240-page vellum book has defied dozens of attempts to decipher it, even by top World War II codebreakers.
Carbon dating suggests that it was written in the second half of the fifteenth century, but the book first "surfaced" in the seventeenth century. It appears to be a guide to plants, but almost all the illustrations show non-existent species.
The manuscript is highly controversial, with many experts dismissing it as a hoax - but a new analysis of the text appears to have found “patterns” of meaning which would have been impossible to fake in the 15th century.
The new research has also found "keywords", some of which seem to match to the strange, hand-drawn illustrations that surround the text. It could aid new attempts to crack the code.
[Related: Researchers crack Copiale cipher]
“The Voynich text has resisted all attempts to decipher it, even by top World War II cryptographers,” says Dr. Marcelo A. Montemurro of Manchester University. “However, the fact that it has been impossible to decode so far cannot be a proof that there is no message inside it.”
Other ciphers previously thought "unbreakable" have recently been cracked by computer technology - such as the Copiale Cipher, an 18th century German manuscript which was "broken" in 2011, revealing the secret rites of an occult society.
“For the past few years I have been studying the statistics of language - using methods from physics and information theory,” says Montemurro. “These methods allow the extraction of keywords (that is words that are closely relevant to the meaning of the text) even if the underlying language is unknown.”
Montemurro’s technique analysed the text at a large scale - looking for “clusters” of words as the text moved from one subject to another, rather than trying to understand the manuscript’s grammar.
“Over long spans of texts, words leave a statistical signature about their use,” says Montemurro. “When the topic shifts to a different one, other words are needed, and so on.”
Montemurro’s analysis found a range of “keywords” in the text - and found that the pattern of their use was similar to known languages. The researchers also found that clusters of keywords seemed to “match” the illustrations.
The knowledge required to put this level of detail into a hoax manuscript means it is less likely that a 15th century hoaxer could have faked it.
“It is not not an absolute impossibility that it is a hoax - but most if not all of these features were not known in the 15th century,” says Montemurro. “The hoax hypothesis is that it needs to explain all the levels of structure that are found in the text - and how they could naturally emerge from the hoaxing method.”
“I’m not a cryptographer, but I can see it as a step forward in the sense that now there are candidates among the text’s words to be those more closely connected with the meaning of the text,” says Montemurro. “There is still the question of what sort of method was used to encode the message and hide its message - making a connection between our analysis and a possible decoding mechanism will require more specialized research."