DAILY CRYPTOGRAM

Nobody invented it in one go. Substitution ciphers are ancient: Suetonius reports that Julius Caesar wrote his confidential letters "by so changing the order of the letters of the alphabet, that not a word could be made out", and tells the reader to substitute D for A to undo it.

What did get invented, and can be dated to the week, is the cryptogram as a public entertainment: a scrambled message printed in a paper for strangers to crack. That started on 18 December 1839, when Edgar Allan Poe told the readers of a Philadelphia weekly that they could not send him a substitution cipher he was unable to break.

He was showing off. He was also, mostly, right. Ninety years later the people who had never got over it founded an association, gave themselves code names, and started a journal that is still going.

Key Takeaways

• Poe issued the first mass cryptogram challenge in Alexander's Weekly Messenger on 18 December 1839, and printed solutions in almost every issue from 15 January to 29 April 1840.
• In 1841 he claimed only one of "perhaps, one hundred ciphers altogether received" had beaten him. The surviving file shows a smaller, more honest picture.
• The American Cryptogram Association was organised on 1 September 1930 and has published The Cryptogram since February 1932. Its members called the plain substitution puzzle "The Aristocrat of Puzzles".
• A cryptogram falls to a pencil because English letter counts barely move. Parker Hitt's 1916 army manual measured E at 1,277 times per 10,000 letters and Q at 8.

What is a cryptogram, exactly?

One alphabet, shuffled once, and used consistently.

The tidiest definition comes from the Yale scholar W. K. Wimsatt, in a letter reproduced by the American Antiquarian Society: a simple substitution cipher is "one in which the same symbol, be it of whatever sort, invariably stands for the same letter of the alphabet in the concealed message". That "invariably" is what makes the puzzle solvable rather than merely mysterious: the message keeps the shape of its own English underneath the disguise.

Solvers gave the family its own names long before anyone else did. The American Cryptogram Association calls a substitution puzzle with the word spaces left in an Aristocrat, and a foreign-language one a Xenocrypt, a department it opened in August 1933. For the many other ways a message can be hidden, the cipher identifier covers the rest.

Did Edgar Allan Poe invent the cryptogram puzzle?

He invented the dare, which is close enough that the credit has stuck to him for nearly two centuries.

Poe's opening move ran in Alexander's Weekly Messenger for 18 December 1839, under the heading "Enigmatical and Conundrum-ical". He argued at length for applying a rigorous and analytical method to puzzles of all kinds, then made it personal: if a reader sent in secret writing where arbitrary symbols stood for letters, no such cipher could be propounded which he would be unable to solve.

Describing it later in Graham's Magazine, he sounded slightly stunned. "This challenge excited, most unexpectedly, a very lively interest among the numerous readers of the journal. Letters were poured in upon the editor from all parts of the country."

They also cheated, cheerfully and constantly. Correspondents used foreign languages, ran words together with no gaps, and mixed several alphabets in one message. One reader, Poe wrote, jumbled together "no less than seven distinct alphabets, without intervals between the letters, or between the lines". The same essay carries the sentence people still quote at cryptographers: "human ingenuity cannot concoct a cipher which human ingenuity cannot resolve".

Two years later he turned the method into fiction. The Gold-Bug, printed in the Philadelphia Dollar Newspaper on 28 June 1843, has its hero break a cipher on the page while explaining exactly how, which taught frequency analysis to a far larger audience than an essay could.

What did Poe actually solve?

Fewer than the legend says, and the discrepancy is documented.

The legend is Poe's own. Writing in Graham's in July 1841, he reported that "out of, perhaps, one hundred ciphers altogether received, there was only one which we did not immediately succeed in solving", and that the one exception turned out to be random characters with no meaning at all.

For a century nobody could check, because no file of Alexander's could be found. Colonel William F. Friedman, of the United States War Department Signal Office, wrote that despite painstaking research by numerous Poe experts, "not a single issue containing any cipher solutions that Poe may have published as a result of his asserted challenge has ever been found".

Then the American Antiquarian Society acquired a complete 1840 file, Clarence S. Brigham reprinted the lot in 1943, and Wimsatt counted. Poe's cipher articles run from 18 December 1839 to 6 May 1840, with solved puzzles in almost every issue between 15 January and 29 April. Across fifteen 1840 articles he refers to thirty-six ciphers received, and Wimsatt itemises them like this:

What Poe did with it	Ciphers
Printed the text and the solution	9
Printed the solution but not the text	14
Stated he had solved it, without publishing either	3
Said he had not solved it	6

That accounts for thirty-two of the thirty-six; the rest are mentioned without being itemised. The six failures are itemised though, and they are mundane: one defaced and written in pencil, one using 51 characters and so not a simple substitution at all, two he judged deliberate impositions rather than real ciphers, one he had no time to look at, and one he had lost.

The famous seven-alphabet monster shrinks under inspection too. Poe reveals the plain text as "The Siege of Belgrade", the alliterative alphabet poem that begins "An Austrian army, awfully arrayed", and Wimsatt reads the alliteration as the thing that gave the puzzle away. Friedman's point was that seven alphabets used one per line are just seven substitution ciphers in a row, which is nothing like the cyclic ciphers that change alphabet every letter.

None of that makes Poe a fraud. It makes the hundred-to-one boast a magazine flourish written eighteen months after the fact, and the real record something better: a puzzle column with its failures printed in it.

Why can a substitution cipher be cracked with a pencil?

Because English is predictable in bulk, and a shuffled alphabet does nothing to hide that.

Captain Parker Hitt set it out plainly in the Manual for the Solution of Military Ciphers, published at Fort Leavenworth in 1916: for any given language "the frequency table is almost as fixed as the alphabet". Count ten thousand letters of English and you get very nearly the same answer every time.

His table, taken from military orders and reports, counts the top of English like this.

Letter	Times per 10,000
E	1,277
T	855
O	807
A	778
N	686
I	667
R	651
S	622
H	595

At the other end sit Q at 8, Z at 17 and X at 27. The usual rule of thumb runs E, T, A; on Hitt's military sample O edges ahead of A into third, which is a fair warning that the tail of any frequency order shifts with the kind of text it was counted from.

Pairs and triples are even more lopsided. Hitt found TH the commonest pair by a distance, and THE turning up 89 times in 10,000 letters, well clear of AND at 54. He also noted a rule worth more than any table: "It is practically impossible to find five consecutive letters in an English text without a vowel."

A scrambled alphabet moves that structure onto different symbols rather than removing it. The commonest symbol still stands where E stands, the commonest three-symbol group is still almost certainly THE, and once those land the rest unravels. That is the reasoning a cryptogram solver automates. Two special cases are easier still: the Caesar cipher shifts every letter by the same amount, so there are only 25 alphabets to try, and the Atbash cipher simply flips the alphabet end to end, leaving nothing to guess at all.

Poe's famous frequency list was wrong

Not on E, but on nearly everything after it.

In The Gold-Bug, Legrand recites the order of English letters: "Now, in English, the letter which most frequently occurs is e. Afterwards, the succession runs thus: a o i d h n r s t u y c f g l m w b k p q x z."

Count them and you get twenty-four. J and V are missing from the list entirely.

Set that beside Hitt's counted order of ETOANIRSHDLUCMPFYWGBVKJXZQ and the top of the list barely agrees with itself.

Letter	Poe's rank, 1843	Hitt's rank, 1916
E	1	1
A	2	4
O	3	3
I	4	6
D	5	10
H	6	9
T	10	2

Poe demotes T to tenth, where a real count puts it second behind only E. His list is a character's speech in a short story rather than a tally, and it works in the tale because the E is right and E is what opens the door. Solvers today start from a measured table, for the reason Hitt gives: the tally is stable, and impressions are not.

It is worth keeping Poe's reputation in proportion. Friedman later took the harder key-phrase ciphers Poe solved in Graham's and put them in front of students of elementary cryptography, who found them straightforward. Poe was a very quick amateur working in public, which is a different and more interesting thing from being ahead of the field.

Who keeps the cryptogram going?

An association with a constitution, code names and a house journal.

The American Cryptogram Association describes its own founding without much self-importance. It was "organized originally to place the 'cryptogram' on an equivalent basis with chess, thus contributing to the happiness of mankind". The trigger was a 1920s cryptography column by M. E. Ohaver in Detective Fiction Weekly. Dr C. B. Warner and some friends were taken with it and formed the ACA on 1 September 1930.

Meanwhile George Lamb was running a column called "Secret Corner" in a newspaper in Burton, Ohio. Lamb belonged to the National Puzzlers' League and brought its habit of pen names across, choosing DAMONOMAD for himself. Members have used noms ever since, on the principle that "only cryptography counts". He and Warner agreed to publish a journal: The Cryptogram first appeared in February 1932 and, by the association's own account, now runs to 32 pages, six times a year.

The association wrote the textbook too. Helen Fouché Gaines, who solved under the nom PICCOLA, edited Elementary Cryptanalysis out of the correspondence lessons members circulated among themselves. It appeared in 1939 and was reprinted by Dover in 1956 as Cryptanalysis. Ninety-odd years on, the kit has not changed: a pencil, an eraser, paper, and a message somebody scrambled on purpose.

Ready for a different kind of hidden message? The codes and ciphers collection has the tools for every alphabet swap, number code and dot-and-dash system on the site.

Frequently asked questions

What is the difference between a cryptogram and a code?

A cryptogram hides a message letter by letter, one symbol standing invariably for one letter, so the length and shape of every word survives intact. A code replaces whole words or phrases, and needs a codebook rather than reasoning to undo. Frequency analysis works on the first and is close to useless on the second.

Did Edgar Allan Poe really solve every cipher sent to him?

He said he solved all but one of about a hundred. The surviving 1840 file, reprinted in 1943, shows thirty-six ciphers referred to and six that Poe openly says he did not solve, one of them defaced, one lost, one using 51 characters. The boast was generous; the column was genuine.

What is an Aristocrat in cryptogram solving?

It is the American Cryptogram Association's name for a plain substitution puzzle with the word spaces left in, the type that started the hobby. Members applied the label because they regarded it as "The Aristocrat of Puzzles". A foreign-language cryptogram is a Xenocrypt, carried in the association's journal since August 1933.

Which letter should you guess first in a cryptogram?

E, on the evidence. Parker Hitt's 1916 count of ten thousand letters put E at 1,277 occurrences against 855 for T, so the commonest symbol in a decent-length puzzle is nearly always E. Then look for a repeated three-symbol group ending in it: THE turns up 89 times in Hitt's ten thousand letters, more than any other trio.

Are cryptograms actually good for anything?

As cryptography, no. Hitt's 1916 manual was already teaching junior officers that no practicable military cipher is safe once intercepted, and a single scrambled alphabet is the easiest case of all. As a puzzle it survives precisely because it is breakable: the pleasure is in the deduction, not the secrecy.

Sources

• Edgar Allan Poe, A Few Words on Secret Writing, Graham's Magazine, July 1841, 19:33-38 - Edgar Allan Poe Society of Baltimore, retrieved 10 August 2026
• Clarence S. Brigham, Edgar Allan Poe's Contributions to Alexander's Weekly Messenger (1943), introduction, pp. 3-11, including W. K. Wimsatt's letter itemising the 1840 ciphers and quoting Col. William F. Friedman - Edgar Allan Poe Society of Baltimore, retrieved 10 August 2026
• Edgar Allan Poe, The Gold-Bug, Dollar Newspaper (Philadelphia), 28 June 1843 - Edgar Allan Poe Society of Baltimore, retrieved 10 August 2026
• Parker Hitt, Manual for the Solution of Military Ciphers, Press of the Army Service Schools, Fort Leavenworth, 1916 - introduction and Tables I, IV and V, full text on the Internet Archive, retrieved 10 August 2026
• American Cryptogram Association, History - retrieved 13 August 2026
• Suetonius, The Lives of the Caesars, Julius 56, Loeb translation on LacusCurtius, University of Chicago - retrieved 10 August 2026