The Beale ciphers

Nothing outside the pamphlet confirms its story. Joe Nickell searched Virginia records in 1982 and found no Thomas J. Beale who fits; he found that Robert Morriss, who was a real Lynchburg innkeeper, did not take over the Washington Hotel until 1823, three years after the pamphlet has Beale lodging there, and that the letters attributed to Beale use "stampeding" and "improvised" in senses not recorded in English until after 1830⁠[4]⁠[7]. Louis Kruh noted that the numbering of the ciphers does not match the order in which the story needs them, asked how Beale could have expected cipher 2 to be solved first, and compared the Beale letters with the narrator's prose⁠[5]⁠[8].

Treasure hunters have dug across Bedford County since the 1890s, most intensively around Montvale (the old Buford's), on private land and sometimes without leave; Virginia law gives found treasure to the finder, which has not reduced the digging⁠[9]. No deposit has been found. The afterlife of the story in fiction and television is large and adds nothing to the record⁠[10].

The second cipher's decoding is reproducible and is not in doubt; what is tested is whether the other two are ciphers at all. Carl Hammer of Sperry Univac ran the numbers through computers in the 1960s and 1970s and reported that ciphers 1 and 3 showed non-random structure consistent with enciphered text; the Beale Cypher Association he led circulated "Historical and Analytical Studies" in 1970 and 1972, now in the NSA release⁠[2]⁠[3]. Gillogly's 1980 test cut the other way: decoding cipher 1 with the same Declaration key that works for cipher 2 produces, among gibberish, the run "abcdefghiijklmmnohpp" and other alphabetical stretches. The probability of that by chance is negligible, and the natural explanation is that whoever made cipher 1 was picking numbers off a numbered Declaration while running his eye down the alphabet, not encoding a message⁠[6]. It has been argued that the doubled letters in the run fall where the pamphlet's numbering of the Declaration miscounts, which would tie the construction of cipher 1 to the same flawed key⁠[6]⁠[3].

Later statistical work has tested the ciphers against Benford's law, against the distributions expected of book ciphers on English, and, in 2026, against families of composite constructions, with a reproducible pipeline and explicit cautions against over-reading anomalies⁠[11]⁠[12]⁠[13]. The NSA release shows that agency cryptanalysts also examined the ciphers and left no solution⁠[2].

The pamphlet's own reading is that the ciphers are genuine, that cipher 1 names a place and cipher 3 thirty names and addresses, and that a key text exists for each⁠[1]. It predicts that a key will eventually turn them into English, and that the names in cipher 3 will match real people of 1820s Virginia. Its difficulty is that no key has been found in 140 years of searching, that the book-cipher method is sophisticated for the setting, and that Gillogly's strings are hard to reconcile with honest encoding⁠[7]⁠[6].

The hoax reading holds that the pamphlet is a work of fiction written in the 1880s, most likely by Ward or someone near him, in the mode of Poe's "The Gold-Bug", with cipher 2 built to be solvable and ciphers 1 and 3 built from the same numbered Declaration to look like ciphers. It predicts exactly what has been found: no documentary Beale, anachronisms in the letters, one author's style throughout, and alphabetical runs when the Declaration key is applied to cipher 1⁠[4]⁠[5]⁠[6]. It does not explain why ciphers 1 and 3 might show non-random structure, if they do, which is why the recent statistical papers matter⁠[12]⁠[13].

A middle reading is that a real local legend or a real box of papers existed and that the 1885 author dressed it up; it predicts that some element (Morriss, Buford's, a Beale family in the region) will be found in the record, as Morriss has been, without any of the treasure story being true⁠[4].

A decoding of cipher 1 or 3 with a published key that yields coherent English would settle it in favour of genuine ciphers, whatever is then found in the ground. Against: a document from the 1880s showing the pamphlet's composition, a manuscript draft, a printer's record or a letter by Ward, would close the question of authorship, and an archival census of Thomas Beales in Virginia and Missouri from 1810 to 1830 would close the question of the hero. Short of those, the most useful work is statistical: a published, reproducible test of whether ciphers 1 and 3 carry the structure of real book ciphers or of numbers drawn from a numbered text, which is what the 2020 to 2026 papers attempt⁠[11]⁠[13]. Digging settles nothing, because the pamphlet gives no spot to dig.