Claimed Solutions

We recovered the exact claimed plaintext of every Z13 and Z32 solution we could find in print, in academic preprints, in the research forums and in the press, then put each through the same mechanical test.

What is being tested

Two things the ciphertext genuinely requires of any symbol-by-symbol reading. First, length: a substitution decode has to be exactly as long as the cipher. Second, equality: a shape the Zodiac drew twice must decode to the same letter both times. Fail either and the reading is refuted without argument.

Z13 — 98 distinct claimed readings

Consistent with the ciphertext

Ordered by how readable the text is. Consistency is the minimum bar, not evidence.

Refuted by the ciphertext

Not judgeable by this test

---

Z32 — 21 distinct claimed readings

Consistent with the ciphertext

Refuted by the ciphertext

Not judgeable by this test

---

The famous ones

ALFRED E NEUMAN, published by Robert Graysmith and argued by Craig Bauer, is refuted by one constraint: positions 3 and 11 carry the same drawn symbol and the reading needs F at one and M at the other.

DR EAT A TORPEDO, Ryan Garlick's reading through the genuine Z340 key, is consistent in the spelling DREATATOTPEDO and refuted in the spelling DREATATORPEDO. The word TORPEDO cannot in fact be placed anywhere in Z13 — none of its seven possible positions is legal — which is why the misspelling is doing the work.

MARVIN MERRILL, the 2026 claim reported as endorsed by former NSA cryptographers, violates all four constraints.

EARL VAN BEST JR, from Gary Stewart's book, violates all four.

None of this settles who the Zodiac was. It settles only that these particular strings are not what those thirteen symbols say, which is a smaller claim and a firmer one.

Three claims worth knowing in detail

Fayçal Ziraoui, 2021

Widely reported as having cracked both remaining ciphers. His Z13 pipeline applies the Z340 key, then a numeric step, then a Trifid cipher — and the Trifid stage has six possible key assignments. Ziraoui published all six outputs: UAMW, KAYR, XNBS, DNZI, G OE, Q CJ. He then selected the one that resembled the suspect he already favoured.

That is the whole difficulty with these ciphers in one gesture. When a method emits six answers and the analyst picks the one matching a prior belief, the cipher has contributed nothing. A commenter on the same thread observed that the identical intermediate can be steered toward “Theodore Ka” just as easily. There is also a chronology problem: the key he starts from was not recovered until December 2020, half a century after the letter.

Ryan Garlick

Garlick's is the most careful of the well-known Z13 proposals, and he is candid about its limits: applying the genuine Z340 key leaves two symbols unresolved, and he acknowledges that this leaves 676 possible completions. Our own independent enumeration produced exactly that number, and ranks his choice 108th of the 676 by readability.

His promoted reading also asks the same circled symbol to be T, T and R. In the uncorrected form, DREATATOTPEDO, it is fully consistent — but the word TORPEDO cannot be placed anywhere in Z13 at all, so the misspelling is not incidental to the reading, it is load-bearing.

Alex Baber, 2026

Reported as endorsed by former NSA cryptographers, and claiming the name MARVIN MERRILL. It violates all four of Z13's constraints — every single one. Several accounts of the claim state that it fails three; it fails four.

The methodology has not been published, the filter parameters were never disclosed, and the candidate name list was never released. Reporting on it noted that the group appeared to have reached a conclusion and then constructed an explanation for it.

---

Check your own

The same test runs in your browser on the Test a Solution page. It will never tell you a reading is correct, because nothing can.