D'Agapeyeff Cipher
A 1939 numeric challenge with no generally accepted reconstruction of its method.
Unsolved challenge cipher
Interactive Exhibit
The widget loads a 130-digit excerpt of the 395-digit ciphertext (or paste the full cryptogram yourself) and will tally adjacent digit pairs. It counts overlapping pairs after removing spaces. Uneven counts can suggest hypotheses but do not establish a cipher family. This is an analytic helper, not a solver.
Why This Matters
Alexander d’Agapeyeff published a 395-digit challenge in Codes and Ciphers (Oxford University Press, 1939). Its method has no generally accepted reconstruction. Researchers report that he later could not recall how he had enciphered it; that report does not identify a lost key or establish that the printed text is error-free.
The digits are printed in five-digit groups. The widget’s default is the first 26 groups (130 digits), checked against the reproduced challenge page. Fractionation followed by transposition is one line of investigation, not a recovered description of the author’s method.
Nonuniform frequencies do not prove that a sequence encodes a meaningful message. A Polybius-style hypothesis must specify its coordinate alphabet, grouping, transposition and any treatment of the odd final digit. Dividing 395 by two does not establish a plaintext length, and no justified unicity-distance calculation is supplied here.
Gordon Rugg and collaborators investigated possible human error as well as computational attacks. Their report identifies a mistake in another worked example in the book; that is a reason to investigate transcription or enciphering errors, not proof that this challenge contains one. A convincing candidate needs explicit rules that reproduce the printed digits and explain any proposed correction.
| Hall | Hall XII · Unsolved |
| Region | London, England |
| Era | Pre-WWII Britain · 1939 |
| Discipline | Numeric fractionation (suspected) |
| Track | B (frequency analyzer) |
| Modern echo | Model assumptions · reproducible candidate testing |