For decades, members of the American Cryptogram Association have tried to decipher an encrypted note found in Denmark – but to no avail.

Click here for the complete top 50 list

Sometime after the Second World War – probably at the end of the 1950s – an employee of a military museum in Copenhagen came across an encrypted message. It was noted on a piece of paper that had been hidden on the back of a painting from 1835, showing a Danish general. Here’s the message:

Danish-General-Cryptogram

The staff of the military museum could not decipher this cryptogram. So, they sent it to the American Cryptogram Association (ACA), the world’s most important organisation of hobby cryptologists (I’m a member of it, by the way). However, ACA members who tried to break the cipher were not successful, either.

In 2015 blog reader Kent Ramliden from Florida, a long-standing member of the ACA, thankfully provided me a scan of the note. I named it the “Copenhagen cryptogram”.

I wrote my first blog post about the Copenhagen cryptogram in 2015 (in German). I received a few comments, but nothing that led to new insights.

As far as I know, the Copenhagen cryptogram has never been covered in The Cryptogram, the newsletter of the ACA. Probably, not very many ACA members have occupied themselves with this mystery.

to my regret, I don’t have any additional information about the Copenhagen cryptogram. I neither know the name of the museum where it was found nor the year of the finding. The cryptogram probably doesn’t have a relationship to the painting where it was hidden or the depicted general. The name of the general isn’t known to me, either. The cryptogram is probably much younger than the painting. Maybe a reader can estimate the time when these lines were written. The writing looks quite modern to me.

Can a reader find out more about this strange note?


Further reading: Marie-Antoinette’s encrypted letters to Axel von Fersen

Linkedin: https://www.linkedin.com/groups/13501820
Facebook: https://www.facebook.com/groups/763282653806483/

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Kommentare (11)

  1. #1 Bill Briere
    USA
    18. August 2017

    My quick analysis:

    It’s not transposition or polyalphabetic substitution. It is almost certainly a simple substitution cipher.

    The message is not in English. The logical language to test first would be Danish. Three vertical lines delineate four short sentences.

  2. #2 Paolo Bonavoglia
    Venezia
    18. August 2017

    A simple statistical analysis gives a presence function consistent with a mono-alphabetic substitution; the number of different symbol is a bit too high, 25 out of a 107 char text, this can be consistent with a language like Danish which has 29 letters; unfortunately my knowledge of Danish is zero, and 107 chars is a bit too short for statistics, so for now I give up!

  3. #3 Braunschweiger
    Brunswick/Germany (or nearby)
    19. August 2017

    The danish language consists of the 26 latin standard letters, and additionally Å, Æ, and Ø, which are considered part of the alphabet, thus 29. But also, these 3 could be replaced by AA, AE, or OE respectively, so they are not really needed. Additionally, the letters C, Q, W, X, and Z are only needed for foreign non-danish words, so in a text with pure danish the possibly might not appear. Also, all these latter letters and J, Y, Æ, and Ø do have a frequency with less than 1% in danish texts (see any DA frequency table for this), thus it is easy to assume they may or may not appear in a short 107-letter text.

    Hence, possibly there have been used less than 25 alphabet letters in the text and we have space for some possible extra symbols, like colons or marks, or even digits. What we don’t know is which of these are really used in the cipher text. What would a text look like that uses only some, not all, of the 10 digits? The sorted frequency table of the cipher does not exactly match the one of danish language (besides we only have 107 symbols). All provided, this is a one-to-one substitution.

    Danish has no double vowels except for possible AA in writing, but it knows some double consonants like bb, dd, kk, ll, mm, nn, or tt. There are many two-letter and three-letter words in the language, like in others; the shortest word consists of one vowel, like “ø” for island. Here are some common danish short words, found in many texts:

    jag, du, han, hun, det — I (me), you/thou, he, she, it (resp.)
    vi, i, di — we, you (pl.), they (resp.)
    den, det, dette — the, that, this (not 1:1)
    en, ett, nogen, noget — (one, a), someone, something
    og, eller, (ikke, ej), ingen, inget — and, or, (not), no-one, nothing
    som, vad, vem — what/which, what, who/whom
    er, har, se, hør, vid, skrive — is/be, has/have, see, hear, know, write
    man, fru, pojke, pige — man, woman, boy, girl
    hus, by, stad, vej — house, village, town/city, way/street
    idag, igen, ifald, fordi — today, again, if/when, because

    Mange tillykke — goog luck!

  4. #4 Braunschweiger
    19. August 2017

    There are also some common prepositions:
    at, på, af, op, ned, før, her –to, on/at, of, up(on), down, for, here.

  5. #6 Paolo Bonavoglia
    Venezia
    19. August 2017

    The presence function, given the length of a text, returns the expected number of different chars in the text, and was studied and treated by Sacco in his Manuale di Crittografia; I’ve also implemented a web page about it (sorry, in Italian!):
    http://www.crittologia.eu/critto/presenze.html?lingua=en&N=107.
    I find it very useful for a first approach.

    Unfortunately Danish was not among the languages analyzed by Sacco. Any way a 25 value for a text-length of 107, is quite unusual for languages like German, English, French and latin 26 letters alphabet: the expected value is between 19 and 20 … such a high value could suggest Russian!! Or a poly-alphabetic cipher but this seems unlikely for other reasons. Or a misinterpretation of signs, with all those dots.

    Of course it is a statistical function, deviations are possible, but since the cryptogram comes from Copenhagen the most plausible conjecture is Danish language with 29 letters; otherwise it could also be Norwegian or Swedish or Russian or … who knows. If it is Danish it’s likely to have also space encrypted; the frequency distribution matches very well with the one of Danish+space; the two most frequent chars, 3 and x (or n?) should then be space and E (or vice versa) but then the language obstacle arises.

    Of course the web is full of dictionaries, Google translate and language tools, and it is easy to build statistical tables of any language, but I think only a Danish mother language person would have a decisive advantage here … if it is really Danish …. it could be another language after all, Russian for instance …

    If the ACA people couldn’t find a solution, it is a sign the matter is not so simple as it may appear at first glance!

    So for now I give up …

  6. #7 Thomas
    19. August 2017

    @Paolo Bonavoglia
    Moreover, the index of coincidence, if significant in spite of the shortness of the cryptogram, seams to speak for a monoalphabetic substitution. In addition, some patterns (e.g. the beginning of line 1) catch the eye, but sufficient language proficiency (Danish?) is needed to interpret them correctly. By the way: Is the arctan-function which you present on your website as suitable as the index of coincidence for distinguishing monoalphabetic from polyalphabetic ciphers? (I hope you will keep providing cryptoanalytic “news” concerning your grandfather – most interesting!)

  7. #8 Nick Pelling
    http://ciphermysteries.com
    19. August 2017

    CryptoCrack has a Danish datafile & dictionary, so I transcribed the cipher here (there’s a little bit of interpretation, sure, but it’s pretty close):

    wnanabscdnosdevsbseygrnsrsfgfhiapsfnrpbnozinezrdrjjsdpkbjsdenoasypdexpsbnonazndjnqsqnonsalsgdkqrsf

    Raw patristocrat simple substitution suggests:

    winindejligelskedesforierevovqhnaeviradigthistrlrbbelaudbelsignefalspaedigintilbimemigienceolumrev

    Removing the most frequent letter, CryptoCrack suggests:

    xrrnepsdestjenetfolelebobygraeblandigtilslkkesamnkestdrefasthaendriskueuderveosmule

    Removing the second most frequent letter, it suggests:

    cenenlyserskflkbitetgighunogetolerduekdtstaasomlaskernboskjolerendesaepperenvismptg

    Removing both the first and the second most frequent letter, it suggested:

    yttigesenminolaaplpqutrparishunhaeaddervidenstorenfristhedkkstblevkap

    Still, it was worth a try. 🙂

  8. #9 Thomas
    19. August 2017

    @Nick
    CryptoCracks English patristocrat solver yields a higher best fitness score than the Danish database, but also nothing but gibberish

  9. #10 Marc
    20. August 2017

    I’ve tried swedish but without success

  10. #11 Dana
    Orlando
    21. Juli 2019

    Hi! After 12 hours, I think I may have cracked the first 5-10%. Email me.