The Zodiac's Ciphers: The Code That Took Fifty Years

In the summer of 1969, three newspapers in the San Francisco Bay Area — the Chronicle, the Examiner, and the Vallejo Times-Herald — each received a letter from a man claiming to have murdered several young people, with details of the crimes that only the killer could have known. Each letter came with one-third of a strange cryptogram: a grid of some four hundred symbols, a scramble of letters, astrological signs, triangles and filled circles. The writer demanded that the papers print his cipher on their front pages or he would go on a killing spree, and he promised that the code, once broken, would reveal his identity. The papers, after consulting the police, printed the panels. Within about a week the cipher was broken — not by the FBI, not by the police, but by a high-school teacher in Salinas and his wife, working at their kitchen table over a weekend.

That first solve is the true heart of the Zodiac cipher story, and it is worth holding up against everything that came after, because the Zodiac is remembered now as much for the codes nobody could crack and the “solutions” that turned out to be nonsense as for the ones that were genuinely solved. The killer was never caught. Into that empty space poured half a century of amateur codebreaking, some of it brilliant, much of it self-deceiving, and a great deal of it announcing, with total confidence, that it had at last decoded the murderer’s name. To understand why so many false solutions could seem so persuasive, you have to first understand what a real cipher solution actually looks like — and the Harden couple, in 1969, showed exactly that.

The cipher that fell in a week

The four-hundred-and-eight-character cryptogram, known to researchers as the Z408, was what cryptographers call a homophonic substitution cipher. In a simple substitution cipher each letter of the alphabet is replaced by one fixed symbol, which makes it easy to break by counting how often symbols appear — the commonest symbol is probably E, and so on. The Zodiac made his harder by giving common letters several different symbols each, so that E might be represented by any of half a dozen marks. This flattens the frequency counts and defeats the naive attack. It is a real technique with real teeth, and it is why the police cryptanalysts did not crack it first.

Donald Harden, a schoolteacher in Salinas, and his wife Bethye did. Their method was the kind of clean reasoning that codebreaking rewards. They guessed that a man this vain would begin with the word “I,” and that “kill” or “killing” would appear, given who was writing. Bethye Harden made the crucial suggestion to look for the ego at the front — to assume the message started “I like killing.” From those cribs they pried the whole thing open in a weekend. What it said was chilling and banal at once: “I like killing people because it is so much fun it is more fun than killing wild game in the forrest…” — the misspellings preserved — a rant about collecting slaves for the afterlife. And at the end, where the killer had promised to reveal himself, there was no name. Whether by design or incompetence, the message that was supposed to unmask him unmasked nothing.

This is the kernel, the documented, sourced, verifiable fact around which everything else grew: the Zodiac really did send genuine, solvable ciphers, and a real solution has a particular texture. It is grammatical. It reads as continuous language. It is confirmed by the method that produced it, so that anyone applying the same key gets the same plaintext, every time. The Hardens’ solve had all of that. Keep that texture in mind, because the false solutions never quite had it, and the difference between the two is the whole education this case offers.

The cipher that held for fifty-one years

In November 1969 the Zodiac sent the Chronicle a second cryptogram, three hundred and forty characters long: the Z340. It looked like the first. It was assumed to be another homophonic substitution and therefore breakable by the same means. It was not. Attack after attack failed. The FBI could not solve it. The best amateur and professional cryptanalysts in the world tried and were turned away. For fifty-one years the Z340 sat there, printed in every book about the case, taunting everyone who looked at it, and its silence became part of the killer’s dark mystique. Perhaps it was unbreakable. Perhaps it was gibberish, a hoax within the hoax. Perhaps it held the name at last.

The answer came in December 2020, and it came from three men working across three continents in their spare time: David Oranchak, a software developer in Virginia; Sam Blake, a mathematician in Australia; and Jarl Van Eycke, a warehouse worker and self-taught programmer in Belgium who had written his own cipher-solving software. The reason the Z340 had defeated everyone was that the Zodiac had added a twist on top of the substitution: the plaintext did not run in straight left-to-right lines but was written diagonally, in a scheme that broke the message into slanting sections and rearranged them — a transposition laid over the substitution. Once the team’s software tested that particular geometry, the language emerged. The message read, in part: “I hope you are having lots of fun in trying to catch me… I am not afraid of the gas chamber because it will send me to paradice all the sooner…”

Again — and this is the thread that binds the case — there was no name. The killer who had promised, at the start, that his cipher would identify him had produced only more of the same swagger. The FBI confirmed the solve. It behaved exactly as the Z408 solution had: grammatical, continuous, reproducible by anyone who applied the recovered key. That reproducibility is what makes it true. Fifty-one years of failure ended not because someone was cleverer than everyone before but because the right structural insight was finally tested against the machine.

Where the honest work forks into fantasy

Now the fork. The Zodiac left two further short ciphers, the thirteen-character “my-name-is” cryptogram (the Z13) and a thirty-two-character one attached to a map (the Z32). These are almost certainly too short to be soundly solved. A cipher that brief simply does not contain enough constraint to force a single answer; many different keys will produce many different plausible-looking words, and there is no way to prove which, if any, is real. Cryptographers understand this as a hard mathematical limit. It is the same reason a two-letter word could be “at,” “it,” “on” or “of,” with nothing but context to decide — and thirteen scrambled symbols supply almost no context at all.

This limit is exactly where the false solutions bloomed, and they bloomed by the thousand. Over the decades countless people announced that they had cracked the Z13 or the Z32 and that it named the killer — and, with dismal regularity, that it named whichever suspect the solver already believed in. Someone convinced the murderer was a particular man would find, in the thirteen symbols, that man’s name. This is the machinery of apophenia, the mind’s hunger to see pattern in noise, running at full power against a target too small to resist it. When the plaintext is under-determined, the solver’s own expectation supplies the missing constraint, and the “solution” is really a mirror. The tell is always the same: it cannot be reproduced by a neutral method, it works only for the person who wants it, and it collapses the moment someone with a different pet suspect applies the same freedom to reach a different name.

The most infamous of these was a 2021 claim, widely publicised, that the unbreakable Z340 had “really” contained a coded name pointing to a specific long-suspected man — announced within months of the genuine Oranchak team solve, and flatly contradicting it. Here the two textures sat side by side for anyone to compare. One solution was confirmed by the FBI, reproducible by software, and read as fluent English. The other was a claimed hidden layer that the actual decipherment did not support and that no independent method could recover. The case handed the public a rare, clean laboratory: this is what a real cryptographic solution looks like, and this is what wishful pattern-matching looks like, and here they are on the same page.

What the codes were really for

Strip away the codebreaking and a plainer question remains: why did a murderer send cryptograms at all? The answer is that the ciphers were never really about hiding information. A man who wanted to stay hidden would have stayed silent. The Zodiac wrote to newspapers, demanded front-page placement, threatened carnage if he was ignored, and wrapped his boasting in codes precisely so that a whole nation would spend its evenings trying to think his thoughts. The cipher was a technology of attention. It made strangers labour over his words, printed him on front pages, and gave an otherwise pathetic and cowardly killer the one thing he plainly craved above the killing itself: to be studied, feared, and never solved.

That craving is what the false solutions accidentally fed, and it is the deepest reason to be careful with them. Every confident, unearned “I have decoded his name” gave the Zodiac exactly the immortality he was fishing for — the sense of a mind so cunning that experts and amateurs would wrestle his symbols for generations. The dignity of the real work, the Hardens over a weekend and the Oranchak team over years, lies in its refusal of that game: it took the parts that could genuinely be solved and solved them, and it said plainly of the tiny ciphers that they cannot be solved, because thirteen symbols do not carry a proof. Admitting the limit is the harder discipline. It is the same discipline demanded by any case where the evidence runs out before certainty does, the same honest shrug required of anyone standing in front of a book nobody can read or an assassination the state itself could never quite close.

The Zodiac wanted his codes to make him unsolvable and therefore permanent. Two of them were solved, and each time the promised name turned out not to be there. The ciphers never gave up an identity. What they gave up was a temperament — vain, self-mythologising, desperate to be decoded — and the surest way to deny such a man his victory is to hold the line between what has genuinely been read and what we merely wish we could read. The people who cracked the Z408 and the Z340 did that. The people who “named” him from thirteen symbols only wrote him the legend he asked for.