In 1935 US president Franklin D. Roosevelt received an encrypted letter from an anonymous person. The solution of the main part is not known.

Click here for the complete top 50 list

In 1935 an unknown person sent the following letter to US president Franklin D. Roosevelt:


A strange kind of marketing?

This letter is depicted in the book The Friedman Legacy: A Tribute to William and Elizebeth Friedman. This book mentions the solution of the second part of the cryptogram:


If we only read every second letter of the first line we get: DID YOU EVER BITE A LEMON?

Blog reader Rich SantaColoma posted an interesting explanation for this message. It may have referred to the fact that Roosevelt suffered from poliomyelitis (or a similar disease that could not be distinguished from it at that time). According to a theory that has meanwhile been debunked, lemon juice can cure poliomyelitis. Sending the message DID YOU EVER BITE A LEMON? OR ELSE YOU DIE!! to the president in an easy-to-decipher code might therefore have been an unusual attempt to make this treatment more popular.



The first part is still unsolved

Now, the question is what the first part of the cryptogram means.


Here’s a transcription:

1721’01’51’71’9 2’12’68 3’2 3’32’03’7 000000 1’62’71’23’43’8
2’8 2’2 3’9 4’04’14’24’84’4 000000 9 31’3000 1’8 42’32’4000
4’62’93’5 5’154’34’700 0 61’13’65’05’23’04’94’52’53’11’4 –

I blogged about this mystery for the first time two years ago (in German), but no reader could solve it. At least, blog reader Marc Gutgesell made some interesting observations:

  • If we only look at the digit pairs that are separated by an apostroph (1’0, 1’5, 1’7, 1’9, 2’1, …) no  pair appears twice.
  • The digits that are not a part of such a pair are 1, 2, 3, 4, 5, 6, 7, 8, and 9, as well as 18 zeros.

I’m absolutely clueless what kind of encryption this could be. Does a reader know more?

Further reading: How FBI codebreakers found out what “K1, P2, CO8, K5, P2” means


Subscribe to Blog via Email

Gib Deine E-Mail-Adresse an, um diesen Blog zu abonnieren und Benachrichtigungen über neue Beiträge via E-Mail zu erhalten.

Kommentare (17)

  1. #1 Thomas Ernst
    8. Dezember 2017

    If you take every numerical pair joined by an apostrophe, and leave the rest out, you get a sequence of 43 numbers:

    10 15 17 19 21 26 32 33 20 37 16 27 12 34 38
    28 22 39 40 41 42 48 44 13 18 23 24
    46 29 35 51 43 47 11 36 50 52 30 49 45 25 31 14.

    In this group, every number from 10 to 52 occurs once. This pattern – in line 2, you even have four subsequent numbers in a row – is too obvious not to constitute the core structural element. There are other ways of adding or dividing the numerical pairs. However, the sequence 10 – 52 structurally is the most cohesive pattern. Since this is the case, I cannot find a “cipher” in this at all. Despite the “Secret Service”, and the signature “621” at the top. I cannot think of any form of substitution – addition, multiplication, Polybios, what-have-you – that would yield exactly every number from 10 to 52 once. Let’s call this one a “doodle”, if not a fake. And the shuffling of plaintext letters underneath is simplistic enough not to expect much more from the numbers above. 10 to 52 it is. Why? Who knows? Put a fork into it, it’s done.

  2. #2 Thomas
    8. Dezember 2017

    This seems to be a sequence of the numbers from 1 to 52 (1 to 9 written without preceding zeros). It might be a homophonic substitution, cipher numbers only assigned to the letters occuring in the plaintext. But obviously even Friedman hasn’t managed to solve the first part.

  3. #3 Thomas Ernst
    9. Dezember 2017

    My central argument is: if within the complete extent of a numerical text a sequence of 43 numerically sequential characters can be detected (while only dropping a very small percentage of the symbols), the presence of these 43 sequential characters is indicative that this is not a cipher. Whether the Friedmans could solve it not is irrelevant. A pattern like “10 – 52”, each value only once, is not cryptological. Even if one reads “1 – 52” into the text, the fact of “10 – 52” is self-sufficient to preclude a cipher, however you wiggle and waggle the numbers.

    It is always easier to prove than to disprove a possible cipher. Voynich and “Dorabella” being prime examples of non-ciphers difficult to explain as such. However: the “Roosevelt” is an easy, give-away fake.

    Previous error of mine: the “signature” is 261, not 621.

  4. #4 Andreas
    9. Dezember 2017

    lässt sich die leicht durchgedrückte Schrift im Hintergrund irgend wie herausfiltern? Dies könnte evtl. Rückschlüsse auf den Verfasser der Nachricht geben.
    An manchen Stellen wurden anscheinend Buchstaben über(oder unter?) die Zahlen geschrieben. Wäre interessant zu wissen ob dies von Anderen, die versucht haben die Nachricht zu entschlüsseln, oder sind dies notizen von Roosevelt, die er beim aufschlüsseln der Nachricht, als Gedankenstütze, aufschrieb. Wenn letzteres der Fall wäre, könnte sich doch damit der Rest auch aufschlüsseln lassen.
    Und das was oben rechts bereits als – 26′ – identifiziert wurde, könnte m.E. auch – 26x – bedeuten.

  5. #5 Thomas
    9. Dezember 2017

    I only wanted to show that it could be a cipher, I’m aware there is no evidence. An example of homophonic substitution, in which the numbers 1 – 50 only occure once:

    19 45 49 1 40 5 34 46 16 28 23 2 41 9 35 25 42 47 50 17 29 43 30 22 48 10 6 11 24 32 12 36 31 37 15 13 38 7 20 26 3 27 8 44 4 21 33 18 14 39

    Key: A 1 – 3, C 4, D 5 – 8, E 9 – 14, F 15, H 16 – 18, I 19 – 21, L 22, M 23/24, N 25 – 27, O 28 – 31, P 32/33, R 34 – 39, S 40 – 44, T 45 – 47, V 48, W 49/50.

  6. #6 Gerd
    9. Dezember 2017

    Thomas, your example works, but in the end it shows that a message of length 50 can be coded with a key of length 50. So this is like a one time pad, and such an example cannot be broken.
    In other words, we could provide any key of length 52 to find any message in the Roosevelt cryptogram.
    If the cryptogram makes any sense at all, it gives a sequence of 52 (known) items that lead to the solution. I think the clue is to find which 52 items are behind it.


  7. #7 Klaus Schmeh
    9. Dezember 2017

    Richard SantaColoma via FaceBook:
    Thanks for the mention! I’ll look at that again, with a 100% chance of the same 0% results as last time, of course… but its always fun to try.

  8. #8 Norbert
    9. Dezember 2017

    Somehow reminds me of the way you convert a keyword into a numerical transposition key, e.g. XMAS -> 4, 2, 1, 3. Of course, this process is not unambigously reversible, especially with short keywords, which means you cannot “decrypt” a numerical sequence to the original keyword (4213 -> REAL? SEAL? TIES? …). With sequences as long as 52 numbers, the case could be different. Don’t know – maybe worth a try (but I do think Friedman would have checked that possibility).

  9. #9 Gerd
    9. Dezember 2017

    Here is a transcription includuding the one-digit numbers:
    1 7 2 10 15 17 19 21 26 8 32 33 20 37 16 27 12 34 38
    28 22 39 40 41 42 48 44 9 3 13 18 4 23 24
    46 29 35 51 5 43 47 6 11 36 50 52 30 49 45 25 31 14.
    Just some remarks:
    – the series starts with “1”. why? Norbert’s idea would not require that.
    – the zeroes occur only in groups of three or six.
    – the shortest sequence enclosed in zeroes is “9 3 13” in the second row.
    – including the zeroes, we have 70 numbers, of which 52 are non-zeroes.

  10. #10 Thomas
    9. Dezember 2017

    I also guessed that the numbers might stand for points in a coordinate system (f. e. 19 = 1 abcissa, 9= ordinate), but the connecting lines don’t reveal anything recognizable. Hence, if there is anything meaningful at all, it must be a cipher. But a homophonic cipher of that length cannot be broken, unless there is a pattern in the assignment of numbers (e.g. alphabetical order in a Polybius square). The zeroes might stand for word breaks.

  11. #11 Gerd
    9. Dezember 2017

    as we need a set of 52 items, I tried two alphabets a..z a..z with no result, and then I tried two alphabets connected as the “did you bite a lemon” sentence:
    azbyc … cybza. Also no useful cleartext, but there is one thing I noticed. Around three of the “zeroes” I then get the same letter before and after, i,e:
    37 0 0 0 0 0 0 16 => 37=53-16
    44 0 0 0 0 0 0 9 => 44=53-9
    47 0 0 0 6 => 47=53-6
    Don’t know what this means, but it’s not accidentally…

  12. #12 Thomas Ernst
    9. Dezember 2017

    I noticed the deciphering attempts underneath the numbers, too. Those seven or eight letters appear to occur always underneath number pairs framed by apostrophes. Thus, initially, I wrote out number pairs preceded, then followed by apostrophes, counting longer stretches without apostrophes as nulls, and came up with sth. like:

    21 01 51 71 92 12 83 23 32 03 70 01 62 71 23 43 8
    2 82 23 94 04 14 24 84 40 31 30 01 42 32 40
    04 62 93 55 154 34 70 61 13 65 05 23 04 94 52 53 11.

    That’s the “followed by apostrophe” version; probably has mistakes in it, because I did it on the quick. One might quibble over the initial “17”, the split “82”, the final “4”. Yes, there are a few repetitions, tempting a frequency count. While the “preceded by” as well as the “followed by” numerical pairs invite all sorts of possibilities – multiplication, addition, Polybios etc. – the fact that the “split” pairs yielded exactly 10 – 52, each one only once, anulls – at least for me – “preceded by” and “followed by” pairs. Yes, as Thomas has proven, you can squeeze a cipher with homophones out of the “apostrophe-split” version, a “one time pad”, as Gerd has observed (which, btw, Trithemius already envisioned in his “Clavis Polygraphiæ”). However, given the remainder of the text, and assuming that the words “Secret Service”, the diagonal, i. e. important looking “261” were written by the same person, I can’t see anything else here but a make-belief cipher, perhaps concocted by a disgruntled employe.

    Regarding the whole document as a whole: there is type written text on the back, and the date stamp of APR241935AM is a receiving stamp, as was, for instance, common with registered mail: sending postmark on the front, receiving postmark/stamp on the back. We are looking at the back side of a typewritten note of sorts, which constitutes the actual, and main body of text (typed letters are sufficiently clear, especially in the top section, though difficult to read). The round reception stamp above (or beneath the date) seems to contain the letters “[…]LATI[…] BUREAU. Am guessing “FOREIGN RELATIONS BUREAU”. The pseudo-cipher on the back of this typewritten page was added, I should think, by someone having a lazy morning, and definitely not liking Roosevelt too much. Because the lemon-juice-comment and “or else you die” certainly are derisive in tone.

    Adding more guesswork: perhaps this fake-cipher, certainly created at a later date than the stamped reception date, was created with a specific recipient in mind. After all, in 1621 some unknown forger concocted a whole manuscript “Steganographia” for Duke August of Braunschweig-Wolfenbüttel. This pseudo cipher may have been concocted with the Friedmans in mind, on the back of an actual FRUS document received before noon on April 24, 1935. Which brings me to Andreas’ question: I think – but haven’t looked into it – there is an online archive of declassified FRUS documents. Which – with a little luck – may enlighten us about the provenance of the document.

  13. #13 Thomas Ernst
    9. Dezember 2017

    Trying to get a better idea of the contents of the invisible front page of our document, I found the relevant section in the unpaginated, 2014 Kindle edition of “The Friedman Legacy […]”, which – by chance – is available on amazon com. – Unfortunately, all Friedman has to say about this “cipher” and its facsimile is this:

    “Here’s an authentic example of a transposition cipher; I call it authentic because it was sent to President Roosevelt, and the Secret Service asked me to decipher it (fig. 2). Imagine my chagrin when I had to report that it says ‘Did you ever bite a lemon?'” – Follows facsimile of our “cipher”. By nonchalantly only pointing out the “lemon”-comment, Friedman bypasses the question of the numerals altogether. However, they were not part of his argument: he meant to illustrate transposition, not substitution. But I should like to point out one important observation: Friedman comments and insists on authenticity, in this case supported by “sent to President Roosevelt”, and “the Secret Service asked me to decipher it”. There is a bit of cryptological swashbuckling here, “President Roosevelt” and “Secret Service” appealing to Friedman’s (deserved) cryptological vanity. It is a pity, though, that Friedman does not comment at all about the front of the page, or even the time when he was given the document. Was this too mundane for him?

    The FRUS online archive, four volumes for the year 1935, all available as e-books, can be consulted at: – However, not being able to make out a single typewritten word, this will be a chase for a needle in a haystack.

    The only other possibility to correlate front and back of our document would be to look at the original, probably at the Friedman archives.

    By now I am convinced that the answer to the number-scramble lies in the easily decipherable “Did you ever bite a lemon?” The “lemon” is no reference to Roosevelt’s poliomyelitis. It refers to the fake cipher. Although in contemporary American (and British) English, the term “lemon” usually refers to a defective appliance, in particular a car, the OED informs us that the term is much older, and earlier had wider applications. I think the person intended to “bite a lemon” – i. e. to be a “sucker”, by taken in by something fraudulent, by something that didn’t live up to expectations, was William Friedman himself. The “lemon”, in this case, was the numerical pseudo-cipher. Biting on which Friedman, as the few attempts at deciphering attest to, must have spent at least some time. – The “Or else you die”, I think, was added for drama, and to distract from the lemon-question. Which, by itself, might have alerted Friedman that someone was having the better of him.

    Nota bene: when life gives you a lemon, you can’t always make lemonade …

  14. #14 Thomas Ernst
    10. Dezember 2017

    Should like to add a final “turn of the screw”. If cryptologically minded, anyone can design a cipher. However: how do you proceed to design a non-cipher? A fake cipher guaranteed not to yield a plaintext. Not even by chance. I am mentioning chance because there is a part in the fake-Steganographia mentioned above, where the forger actually created sense although his intention was to create non-sense. This happened because the forger worked without a system. A watertight “non-cipher”, however, demands as much of a system as a “true cipher”. In order to create the former, I would have proceeded as our Anonymous did: use sequential numbers only once – perhaps he stopped with 52 because that was his age, who knows – and then pad your “only-once-numbers” with some cryptologically alluring tinsel: in this case, a few bunched zeros, and a few triplets. The perfect “non-cipher”, which looks like a cipher. The person who designed this little send-off certainly knew whom he was up against. Friedman was able to debunk all the Baconians et alia, in his “Shakespearian Ciphers Examined”, by proving their lack of a coherent cryptological system. Here, someone faced him with the opposite: a coherent non-cipher. Quite smart …

  15. #15 Aginor
    11. Dezember 2017

    My first impulse would be to try and see if the almost-plaintext part can be used as a keyword for some sort of transposition.
    The 26′ is maybe important as well. 26 is the number of letter in the alphabet. Could be a hint (or some kind of signature).

  16. #16 Breaker
    19. Dezember 2017

    @Thomas Ernst regarding this comment:

    “We are looking at the back side of a typewritten note of sorts, which constitutes the actual, and main body of text (typed letters are sufficiently clear, especially in the top section, though difficult to read). The round reception stamp above (or beneath the date) seems to contain the letters “[…]LATI[…] BUREAU. Am guessing “FOREIGN RELATIONS BUREAU”. The pseudo-cipher on the back of this typewritten page was added, I should think, by someone having a lazy morning, and definitely not liking Roosevelt too much. Because the lemon-juice-comment and “or else you die” certainly are derisive in tone.”
    Lemon Juice is the method of concealing something with hidden text, as you can faintly see the other numbers added there, it shows that their meaning relates to this being a key to a massive matrix……a numerical one….take note of the themes forming here and the timing surrounding them in the early 30’s.

    The Backside stamp method is key to reverse the text.

    RBEIVTEEUAOLYEDMIODN is left to ponder

    Took a guess that it is a columnar grid cipher layout with rearrangements…..removal of pairs as there was one in the first line….the EE

    The use of 4 letters =


    Then relaying the text out horizontally as you read top to bottom in columns….


    Here both the “LATI BUREAU” (intentional) and the Matching Pairs DD, EE, & II are removed to see a hint of the Skull and Bonesmen’s Coding Systems……OPUS DEI and the Count of letters is 6 for the next rearrangement in columns


    Removing the OE and OE as the patterns….and the next hint to look for the Old English or Box Cipher that relays the messages through a sort of word search system boxing in segments ……

    We restructure to 4 letter columns as the count dictates the structure again by the pairs removed, and then add the boxes around the words using the OE “Old English” Method….


    Here we see the Phrases:

    N 10 MILES

    Guess where this is?

    Virginia….these are instructions to find the Second Trove of Gold Stores from the Beale Operation……

    It was known that he was on Oak Island as an investor in the “Money Pit”….lol

    And also was rampantly digging up the White House after initiating a “Renovation” excavating tons for the intent of seeking the “Hidden Treasure” in the Old Cornerstone……LOOL

    This was of course in the NorthDakota DepartmentOfInteriors hint in the first line of the columns above…..NDOI…….just a hint in each one it seems….but all holding the same theme of a Foreign Govt telling the president to go dig up gold for someone….almost like he was puppeted by a higher order to continue their affairs from the Beale?

    Beverly… James Beverly Ward…….a listed “Agent” of the Confederates in the Beale Papers. Ward was a Cartographer and Map Maker…..deriving the most accurate Mileage Charts of the times….the Beale Cover Story lists two dates that are actually mileages…..this is derived from the system of Persher or Purser Coding Systems that George Beale (A Ship’s Accountant) encrypted his documents. He later sold the compressed 5 disc wheel that he invented to the Mexican Army and then it was adopted by the CSA and KGC/OAK…..the numbers more than likely are this very wheel system used…..but the words in the threats hold the images of an occult illusion….

    The Ciphers in the Beale Papers show that running them backwards, through the same systems, as Ken Bauman showed from Brad Meltzer’s Decoded, and taking notice of their relation to the ERE FENN DUE RED KNEE comment he found in the Beale revealing a similarity to the second operation out west at Wounded Knee.

    I discovered the Beale Cipher’s plot was involving the CSA and KGC/OAK operations in Arizona…..but that small phrase that Brad Meltzer showed was a mere small reveal compared to this.

    Now it seems that this is the cipher that carried their operations to the mines in ND for the Second Half and that this was some form of correspondence that was made to look like a threat to the president, but encoded using their Skull and Bones Methods of encryption…..showing the locations of the mines and the stores that were stolen and buried after the massacre by their men.

    This “Threat” was the message that would definitely get through to him to examine through the Secret Service….By making it look fake it would be overlooked….

    But it possibly holds a major piece of the Beale Mystery….

    I jumped the second I made the Breaks…..!

    This one is awesome and will make a lot of people happy as they have been talking a lot recently about Beverly Ward’s Journals and names there in the vicinity of Otay Rd…..I found the maps !

  17. #17 Breaker
    19. Dezember 2017

    Also confirmed is the relation to the Beale Ciphers to find Layer Two……using the DOI clue in the first Line…..

    “1721’01’51’71’9 2’12’68 3’2 3’32’03’7 000000 1’62’71’23’43’8
    2’8 2’2 3’9 4’04’14’24’84’4 000000 9 31’3000 1’8 42’32’4000
    4’62’93’5 5’154’34’700 0 61’13’65’05’23’04’94’52’53’11’4 ”

    This is the DOI Book Code…..find the corresponding word and the first letter is the plaintext.

    The Hint in the Beale also applies to a certain numerical sequencing……DO NOT ALLOW ANY FALSE OR IDLE PUNCTILLIO FROM INTERFERING WITH THE COLLECTION OF YOUR SHARES….this phrase was used to derive a key in the first Beale derivatives…..but there is clearly another layer…..