Virginia · a pamphlet of 1885 about events of 1819 to 1822
Are the three pages of numbers printed in an 1885 pamphlet from Lynchburg, a town in central Virginia, genuine ciphers pointing to buried gold and silver, or a hoax?
What happened. In 1885 a pamphlet called The Beale Papers was printed in Lynchburg, price fifty cents, and its author, who gave no name and is called the narrator here, told this story: in January 1820 a stranger named Thomas J. Beale took rooms at the Washington Hotel, kept by Robert Morriss, a merchant of the town, came back for the winter of 1822, and before leaving that spring handed Morriss a locked iron box. A letter from St. Louis dated 9 May 1822 told Morriss to open the box if no one called for it within ten years and said that a key to the papers inside would reach him in June 1832; no key came, and when Morriss opened the box in 1845 he found three pages of numbers and letters signed T. J. B. telling of thirty Virginians who had gone west in 1817, found gold and silver north of Santa Fe, then a Spanish town, and buried it in Bedford County, Virginia, the next county west of Lynchburg. In 1862 Morriss passed the papers to the narrator, who found that if the words of the Declaration of Independence, the founding document of the United States, are numbered from 1 upwards, each number on the second page stands for the first letter of the word with that number (a book cipher, in the language of code-breaking, with the Declaration as its key), and read that way the page describes 2,921 pounds of gold, 5,100 pounds of silver and jewels in a vault six feet down. The pamphlet calls the three pages papers 1, 2 and 3 and later writers call them ciphers 1, 2 and 3; ciphers 1 and 3 defeated the narrator, and after more than twenty years he published everything and invited the public to try.
Why it is a mystery. Ciphers 1 and 3 have never been read, no record of Beale, his party or the box has ever been found, and nobody knows who wrote the pamphlet: James B. Ward, who held the copyright, called himself the author's agent. The question stays open because cipher 2 is a working cipher with a real message, so either Beale's party existed or someone wrote a treasure story and enciphered part of it; those who take the story as true point to that message and to Carl Hammer, a computer scientist with the Univac company, who ran the ciphers through a computer in 1971 and found that the unread pages were not random numbers, which he and the Beale Cypher Association, a society of enthusiasts formed to work on the ciphers, took to mean that they encoded something. The sceptics' case was made in print by Jim Gillogly, a cryptanalyst, or code-breaker, who tried the Declaration as key on cipher 1 in 1980 and found one stretch of the result spelling the alphabet in order, the mark of a hoaxer writing from the numbered key; by Joe Nickell, an investigator of hoaxes and forgeries, who showed in 1982 that Morriss did not keep the Washington Hotel before 1823 and that the letters signed T. J. B. (the "Beale" letters) and the narrator's own account read like one hand; by John C. King, a cryptanalyst, who worked out in 1993 exactly how the Declaration had to be numbered for cipher 2 to decode cleanly and reported longer alphabetical runs in cipher 1 under that numbering; and by Louis Kruh, a cryptologic historian, in 1982 and 1988, and Todd Mateer, a cryptanalyst, in 2013. That work appeared in Cryptologia, a specialist journal of code-breaking, and in the enthusiasts' newsletter, and popular accounts still present the ciphers as an unsolved treasure puzzle: by January 1893 the Roanoke Times, a newspaper in the next city west of Lynchburg, could report that "the inhabitants of Bedford county have hunted for this treasure but so far in vain", and in 1972 a group of believers calling itself the Beale Cypher Study Committee was still paying for searches of the archives.
How it was judged. Everything was re-run from the pamphlet's own text, transcribed from the copy in the Library of Congress, the national library of the United States: cipher 2 was decoded with the numbering the pamphlet itself prints beside the Declaration (a number after every tenth word), that count goes wrong in five places, and each error matches a miscount made by the person who turned the message into numbers (the encoder, from here on), so the simpler reading is that whoever prepared the pamphlet held the encoder's own working key, although the story says that no key ever arrived. Cipher 1, decoded with that same miscounted numbering, holds three stretches of alphabet in order, seventeen letters at the longest, and the chance of that arising by accident is about one in twenty million by exact calculation under the most generous model of chance tested; cipher 3's numbers drift and climb in a way that enciphering a list letter by letter does not produce, and its 618 numbers cannot hold the thirty names with heirs and residences it is said to contain. Stylometry, the statistical comparison of writing habits, with 115 samples by 27 American authors for comparison, finds in the "Beale" letters and the narrator's account the involuntary habits of one writer; the letters use words that came into use decades after 1822; Morriss's own newspaper notices show that he took the Washington Hotel by a rental notice dated 20 September 1823 and printed on 30 September, to open on 1 October, and kept it until December 1825, so the pamphlet's date for Beale's first visit there is more than three years too early; and every documented American party that reached Spanish New Mexico, the territory around Santa Fe, between 1812 and 1819 was arrested. The verdict is a hoax, about 95 in 100 as a judgement that combines these lines, with a separate figure for each claim; a reading of cipher 1 as English that survived the statistical tests would overturn the verdict on the ciphers.
A hoax, about 95 percent as a composite judgement. The figure is a judgement rather than a measurement, and each claim beneath it carries its own figure.
The hoax reading rests on counting that anyone can repeat from the pamphlet's text, on probabilities computed from the cipher's own numbers and on Morriss's own newspaper notices, and it agrees with the cryptanalysts who worked on the ciphers in print between 1979 and 2013. Every percentage is a judgement, and the stylometric likelihood ratio (how many times more likely the letters' closeness to the narrator's account is if one person wrote both than if two did) runs from 3 to 25 depending on the measure (2.7 for Burrows Delta, the measure with the best accuracy on the comparison authors).
The 95 is a composite. It combines the two firmest findings, that the three alphabetical strings in cipher 1 are deliberate (beyond reasonable doubt under every model of chance tested: a probability of 4.9 × 10⁻⁸ by exact calculation under the most generous of them, and no such run in a million random reorderings of the cipher's own numbers) and that the pamphlet's printed numbering matches the encoder's count at each of his five miscounts (established by counting, and anyone can repeat it from the pamphlet's text), with the simpler reading of that second finding, the Morriss hotel dates, the Santa Fe record and the language of the letters. Each line has an alternative explanation; for the story to be true, every one of them would have to hold at once. The case is a replication and extension of the argument made by the cryptanalysts Gillogly (1980), Stephen Matyas (1979), Peter Holst, writing in the Beale Cypher Association's newsletter (1987), and King (1993), and by J. A. Hill in an unpublished paper (1989), with statistical, stylistic and archival checks added.
The figure moves with the reading of the printed key: if the narrator reconstructed the encoder's miscounts himself, that finding no longer bears on the story, and the verdict rests on the alphabet in cipher 1, the Morriss dates and the Santa Fe record. It moves with the stylometric measure chosen, since Burrows Delta, the measure with the best accuracy on the comparison authors, alone gives about 73 percent for common authorship of the letters and the narrator's account, and with the prior odds, the view taken before the texts are compared, which are a judgement. A reading of cipher 1 that survived the statistical tests would overturn the verdict on the ciphers.
The strongest case for the other side is the alternative King left open: that the Declaration is the true key to cipher 1 and a further layer of encoding hides the text. Statistics cannot exclude it, and nothing supports it. Cipher 2 has no second layer, the pamphlet says all three papers work the same way, and the alphabetical runs exist specifically in the miscounted numbering of the Declaration and use the front of its lists of available words (the Declaration offers many words beginning with each letter, and the runs take theirs from the first such words in the numbered text, the same habit as the encoder of cipher 2), so whoever wrote them was working from the Declaration numbered as for cipher 2; a hidden layer would have to be one that no one has found in the 140 years since the pamphlet appeared. The hoax explanation of the strings is the best supported, at about 99 percent as a judgement.
The three rows are separate claims, each with its own figure, and none is a share of a total. Two findings are firmer than any row: the three alphabetical strings in cipher 1 are deliberate, beyond reasonable doubt under every model of chance tested (a probability below 5 × 10⁻⁸ by calculation, and none in a million reorderings), and counting establishes, with no percentage attached, that the printed key and instructions cannot produce the printed translation and that the numbering matches the encoder's count at each of his five miscounts. The 90 for the letters is 73 to 96 percent with even prior odds, depending on the stylometric measure; one claim is weaker than any row, that the narrator was James B. Ward, which is not established.
The evidence is the pamphlet's own text, transcribed from the copy in the Library of Congress, the national library of the United States: the three ciphers, the Declaration of Independence exactly as the pamphlet prints and numbers it, its printed translation of cipher 2, the three letters signed T. J. B. and the anonymous narrator's account. To that were added the Lynchburg newspapers of the 1820s, the records of Spanish New Mexico as searched in 1972, the files on the ciphers released by the National Security Agency, the United States code-breaking agency, and 115 samples of writing by 27 American authors of the period, used for comparison.
Not everything could be read. Of the earlier writers on the ciphers, Jim Gillogly's paper of 1980 was read, and John C. King's paper of 1993 only in the fragments that the full-text search of the Internet Archive, the digital library, returns; the writings of Matyas, Holst, Hill, Hammer, Nickell, Kruh and Mateer and the writer William Poundstone's book of 1993 were not read, and Nickell's findings are known only through Wikipedia's summary. The Lynchburg newspapers of 1819 to 1821, 1862 to 1865 and 1885 to 1892 are not digitised, so silence there is untested, and the 1885 copyright record book was not accessible.
Cipher 2 was decoded with the numbering the pamphlet prints beside the Declaration, and with a plain count, to see where the printed numbering goes wrong. The chance that the alphabet in cipher 1 arose by accident was computed exactly and checked against a million reorderings of the cipher's numbers. The three ciphers were compared as sequences of numbers, with a capacity bound for cipher 3: the letters its promised list would need. Stylometry, the statistical comparison of writing habits, measured the letters and the narrative against the comparison authors; the words in the letters were dated; and the story's dates were checked against the newspapers and archives.
Figure 1 plots, along cipher 2, how far the word numbers used by the encoder, the person who turned the message into numbers, depart from a plain count of the Declaration; every step falls at one of the five places where the pamphlet's printed count misnumbers the text. Table 1 shows positions 184 to 210 of cipher 1 decoded with a plain count and with the encoder's miscounted numbering; under the miscounted numbering the stretch reads seventeen letters in alphabetical order. Figure 2 shows the numbers in the three ciphers in ranges of fifty: ciphers 1 and 3 share cipher 2's preference for small numbers but reuse them far less.
The verdict is a hoax, about 95 in 100, a judgement combining these lines rather than a measurement. The pamphlet's printed numbering reproduces the encoder's five miscounts, so the simpler reading is that whoever prepared the pamphlet held the encoder's key, although the story says that no key ever arrived; and the chance of the alphabet in cipher 1 arising by accident is about one in twenty million under the most generous model of chance tested. The figure moves with the reading of the printed key and the stylometric measure chosen; a reading of cipher 1 as English that survived the statistical tests would overturn the verdict on the ciphers.
Already established by others. Gillogly (1980) found the alphabetical passage in cipher 1, showed its improbability and drew the hoax inference, with the caution that it is strong evidence rather than proof. Matyas (1979), Hammer (1979), Holst (1987) and King (1993) worked out the renumbering of the Declaration that decodes cipher 2 and the encoder's errors; King, and Hill (1989) independently, reported the three longer strings in cipher 1 under the reconstructed key. George Love, an enthusiast who published his analysis on the web in the 2000s, observed that cipher 2 works only with the pamphlet's counts, and Mateer (2013) concluded from the encoding process that the ciphers are likely a hoax. The objection that 618 numbers cannot hold thirty names with heirs and residences was made in 1927 and by Kruh (1982). Nickell and the linguist Jean Pival (1982), Kruh (1988) and Poundstone (1993) argued common authorship of the letters and the narrator's account, excluded Edgar Allan Poe, whom some have proposed as the author, and placed Morriss at the Washington Hotel no earlier than 1823. Viktor Wase (in a conference paper of 2020; a second paper of his, in Cryptologia in 2021, was not accessible) and Leonardo Campanelli (in a Cryptologia paper published online in 2022, in the volume dated 2023), both analysing the digits statistically, found the digit anomalies in ciphers 1 and 3.
What historians have said. The hoax reading was argued in print by Gillogly (1980), King (1993), Mateer (2013), Nickell (1982) and Kruh (1982, 1988); King left three explanations of the alphabet in cipher 1 open. The position is not unanimous: Hammer (1971, 1979) and the Beale Cypher Association held the unsolved ciphers to be genuine, and popular accounts treat them as an unsolved treasure puzzle.
The verdict is the cryptanalysts' verdict, re-run on the pamphlet's own text and given a confidence per claim. The pamphlet's numbering reproduces the encoder's miscounts, and the alphabetical strings in cipher 1 are a second line of evidence, visible even with a standard text of the Declaration; the verdict on the ciphers rests on both lines, and the verdict on the letters on two more, the stylometry and the newspaper record. The page agrees with the hoax reading; it does not show that the hoax reading is the field's settled position.