Preface

Context: A recently reported breakthrough claims that Anthropic’s Claude Fable 5.1 decrypted a mysterious 373‑year‑old Scottish cipher in under an hour. This article summarizes the background, how the solution was reached, and the remaining questions. The purpose is to present the evidence and reasoning behind the claimed solution, and to note why independent verification is still required. The account emphasizes the unexpected nature of the key: rather than relying on external codebooks or advanced cryptanalysis, the solution reportedly used the book itself as the key, a simple but previously overlooked possibility.

Lazy bag

Key takeaway: Claude Fable 5.1 is reported to have solved the Cyphral Distich by mapping numbered positions to specific passages in the same book, extracting initial letters to form a pro‑royalist prayer. The method is straightforward but required patient, exhaustive reading rather than advanced cryptographic techniques.

Main Body

In the mid‑17th century, Scottish nobleman Thomas Urquhart left behind a puzzling pair of number sequences appended to his 1653 work Logopandecteision. Each line contained 32 numbers, and those two lines together—known to later readers as the Cyphral Distich—became an enduring cryptographic mystery. Over centuries, scholars and amateur codebreakers applied common cryptanalytic tools: frequency analysis, substitution ciphers, homophonic substitutions and other standard techniques, yet none produced a convincing, self‑consistent plaintext.

The sequences in question read (numbers separated by periods):

5.3.27.38.32.14.21.8.66.8.70.39.5.9.12.18.2.3.56.5.1.7.3.2.13.19.3.25.9.3.16.6.

25.15.13.6.11.20.5.1.2.12.1.20.20.49.20.20.35.33.4.6.8.35.5.33.5.5.18.10.3.11.32.42.

Vals AI, an AI evaluation firm, recently reported on the claimed solution produced by Anthropic’s Claude Fable 5.1: in roughly 44 minutes and using about 176,000 tokens of computation, the model returned a plaintext that reads as a prayer of loyalty to King Charles II. Vals AI called the result “embarrassingly simple” for humans in the sense that it did not require exotic cryptanalytic insight; rather, it required noticing and testing a decoding rule that many previous solvers apparently overlooked.

The crucial observation is structural. Urquhart prefaced the cipher with 32 short passages he labeled Proquiritations, and he emphasized the significance of the number thirty‑two. The presence of thirty‑two Proquiritations plus two 32‑number cipher lines, together with phrasing promising that the reader could find their heart’s desire in the text, suggests a natural one‑to‑one mapping. The decoding rule that Claude Fable reportedly used is simple: for the i‑th number in a cipher line, go to the i‑th Proquiritation and take the nth word indicated by that number, then use the first letter of that word. In other words, the book itself provides the lookup table.

Applying that rule to the Distich yields the English lines:

O GOD UPHOLD KING CHARLS THE SECOND AND

MAKE HIM THE SUPREME RULER OF THIS LAND

Translated, this is a plea that God support King Charles II and make him the supreme ruler of the land—precisely the sort of royalist sentiment Urquhart is known to have held. The proposed solution meets several self‑consistency checks: each decoded line contains exactly 32 letters, the two lines form a distich with rhyming cadence at line ends ("and" / "land"), and the political content aligns with Urquhart’s biography and likely motives for hiding a loyalist sentiment. Those three independent alignments make accidental or random coincidence unlikely.

Encouraged by that success, Claude Fable then tackled a larger Urquhart puzzle from 1652 titled The Jewel. The Cyphral Octastich there involves 285 numbers mapped to 284 pages of text; the reported decode method was similar but uses page positions rather than Proquiritation entries: the k‑th number points to a specific page and a word on that page, and the first letter of that word yields the poem’s letters. The returned plaintext allegedly forms an eight‑line poem ending with an "Amen," again a prayerful declaration supporting Charles II across England, Scotland, and Ireland.

However, important caveats remain. The result currently rests on Vals AI’s public write‑up and Anthropic’s internal model outputs. The solution has not yet been independently authenticated by cryptographers or historians of cryptography. In the Octastich case, Vals AI notes specific unresolved issues: nine positions among the 285 remain unreadable by any straightforward mapping; there is a blockwise offset beginning at position 159 that suggests a duplicated or misnumbered page in the surviving copies; and crucially, public scans of the 1652 edition covering the ambiguous pages are not available on major digital archives, preventing external verification against the original typeset pages.

Vals AI also disclosed that their team deliberately guided the model’s search: they asked Claude Fable 5.1 to examine its own prior solutions to mathematical puzzles as demonstration material, encouraged creative thinking, and constrained the task away from problems with known or multiple accepted solutions and away from world‑class unsolved ciphers (for example, Kryptos K4). According to Vals AI, no other frontier models produced verifiable answers to similarly meaningful unsolved puzzles during their experiment period.

Why this matters: the episode highlights how large language models can surface historically plausible solutions by exploring nonstandard but literarily grounded decoding rules—solutions that human researchers may have missed because they assumed a conventional cryptographic key external to the text. It also underscores the ongoing need for independent, cross‑disciplinary verification when AI proposes answers to longstanding scholarly problems. A model may find an elegant, self‑consistent explanation, but scholarship demands corroboration through primary sources and domain expertise.

Next steps include: locating or digitizing authoritative scans of the 1652 and 1653 editions to confirm page layouts and wordings; independent replication of the decoding process by historians and cryptanalysts; and careful analysis of the unresolved Octastich positions to determine whether printing anomalies or later alterations account for the discrepancies. Until such external checks are completed, the Claude Fable solution stands as a compelling, but provisional, claim.

Key Insights Table

Conclusion: The Claude Fable result is an intriguing demonstration of how AI can uncover simple, text‑internal keys that human researchers overlooked. It offers a plausible decryption consistent with historical context, but proper scholarly confirmation requires locating primary source page images and independent replication by subject specialists.