Cryptography: The codes that got away

Journal name:
Date published:
Published online

Andrew Robinson takes on a compendium of past and current ciphers ripe for decoding.

Unsolved! The History and Mystery of the World's Greatest Ciphers from Ancient Egypt to Online Secret Societies

Craig Bauer Princeton University Press: 2017. ISBN: 9780691167671

Buy this book: US UK Japan

Leemage/Corbis via Getty

The Phaistos disc, discovered in Crete in 1908, remains untranslated.

The concluding words of Unsolved! are a call to action. Craig Bauer, a US mathematician and editor-in-chief of the journal Cryptologia, ends his hefty history of cryptography by noting that even as he was compiling the book, “unsolved ciphers from decades, sometimes centuries, in the past were coming to light on a regular basis”, along with a plethora of new puzzles. For cryptography fiends, it's a thrown gauntlet.

Unsolved! spans a huge arc of time and space, from Julius Caesar's simple substitution cipher to composer Edward Elgar's 1897 Dorabella Cipher — a still-unsolved letter to Dora Penny, a dedicatee of his Enigma Variations. Uncracked ciphers from the twentieth century are associated with the Irish Republican Army, a series of grisly murders in California — and messages 'detected' from Mars.

Bauer's compelling chapter on the medieval Voynich manuscript occupies one-sixth of the book. In his 1967 The Codebreakers, cryptography historian David Kahn called the manuscript “the longest, the best known, the most tantalizing, the most heavily attacked, the most resistant, and the most expensive of historical cryptograms”. Its weird colour illustrations and indecipherable calligraphy attract 16% of online traffic to the library at Yale University in New Haven, Connecticut, where it is held (A. Robinson Nature 539, 2829; 2016). Bauer speculates as to whether the manuscript is written in a monoalphabetic substitution cipher (MASC) — each plaintext letter substituted with a letter from a single scrambled alphabet. A crackable MASCed text in English reveals the principles. But, as he shows, the Voynich manuscript has too much redundancy (order) to be MASCed English, French, German, Italian, Spanish or Japanese. (Wisely, Bauer offers no theories of his own.)

Unsolved! digs into the riches of ancient Viking, Roman, Greek and Egyptian cryptography. Egyptologists tend to avoid tackling the latter because of its sheer complexity. Bauer reveals how Caesar's cipher worked, substituting each plain-text letter with a letter a fixed number of places away in the alphabet. Inexplicably, however, he relegates to an endnote the undeciphered Phaistos disc found on Crete in 1908 — the only example of its much-discussed script (A. Robinson Nature 453, 990991; 2008). Nor is there even a passing reference to Michael Ventris, celebrated for his 1952 decipherment of the script Linear B as a form of archaic Greek, or to the exciting solution of Central America's Mayan script, launched by Soviet linguist Yuri Knorozov.

Perhaps the most successful chapter centres on ciphers by the notorious, never-captured 'Zodiac Killer', who murdered at least five people in California in 1968–70 (dramatized in David Fincher's 2007 film Zodiac.) The murderer sent taunting letters to local newspapers, featuring four ciphers offering clues to his identity. The first was broken by husband-and-wife amateurs Donald and Bettye Harden. She guessed that a self-centred person might begin his message with 'I'; that 'KILL' might feature in it more than once; and even that the phrase 'I LIKE KILLING' might appear. This proved the key to translating the simple MASC to meaningful, if misspelt, English, although no sense could be made of the killer's signature, EBEORIETEMETHHPITI.

The Zodiac's other ciphers have proved resistant. Bauer hazards that a nine-letter 'word' with some resemblance to the ten-letter 'CALIFORNIA' may mean just that, although a letter is missing — possibly due to “lousy spelling”. But would a killer with poor spelling, as opposed to someone like the wordplay-loving Elgar, be attracted to ciphering? Later, Bauer guesses that such misspellings were “likely intentional”.

The level of decoding skill needed for Unsolved! varies significantly. A willingness to grapple with plain text and cipher text is necessary, but some parts require undergraduate-level mathematics. One is the section on RSA, unveiled in the 1970s as one of the first practical public-key cryptosystems. The book's combination of convincing logic and sometimes-convincing speculation is a familiar mix to those of us interested in undeciphered writing, such as the script of the Indus civilization (A. Robinson Nature 526, 499501; 2015) and the rongorongo script used on Easter Island.

As science-fiction writer Arthur C. Clarke commented when I published Lost Languages (McGraw-Hill, 2002): “Many, it seems likely, will never be deciphered — which raises an interesting question. If we cannot always understand messages from our fellow humans — how successful will we be when we receive the first communication from Outer Space?” And Clarke was talking about ordinary writing systems. For all the clues analysed in Unsolved!, there is plenty of Earthly decoding to do before we tackle any extraterrestrial communiqués.

Additional data