ADFGVX is an encryption method used by the Germans in WW1. Some 20 ADFGVX radio messages from 1918 are still unsolved.

Click here for the complete top 50 list

In World War I cryptology plaid an important role. As cipher machines like the Enigma were still in its infancy, all nations involved in the war almost exclusively relied on manual cipher systems. Many of them proved laborious and insecure. The trouble virtually all WW1 armies had with their manual ciphers caused many engineers to invent encryption machines in the 1920s and 1930s. This resulted in the encryption technology of WW2 heavily relying on machines, which was a major improvement to WW1.

 

ADFGVX

In the last months of World War I, the German Army and diplomatic services used a manual cipher named ADFGVX to encrypt radio messages between Germany and its outposts and stations in the Balkans, the Black Sea, and in the Middle East. To use ADFGX one needs a table with 6×6 cells. The letters A, D, F, G, V and X mark the lines and the columns of this table (these letters were chosen because they can easily be distinguished from each other in their Morse code representations). The table is filled with the letters of the alphabet and the digits from 0 to 9. In the following example HOUSE is used as a key word to change the order of the alphabet:

    A D F G V X
A   H O U S E A
D   B C D F G I
F   J K L M N P
G   Q R T V W X
V   Y Z 0 1 2 3
X   4 5 6 7 8 9

Every letter and digit is represented by a bigram indicating its line and column: H=AA, O=AD, U=AF, S=AG, …

The cleartext I HAVE A DREAM encrypts to: DX AA AX GG AV AX DF GD AV AX FG.

In the second encryption step, the order of the letters is changed using a columnar transposition. To do so a second key word, say ROBIN, is needed. The columns are read out in alphabetic order:

ROBIN
-----
DXAAA
XGGAV
AXDFG
DAVAX
FG

As the B appears first in the alphabet, the B column is read out first: AGDV. It is followed by the I column (AAFA), the N column (AVGX), the O column (XGXAG) and the R column (DXADF). This results in the following ciphertext:

AGDV AAFA AVGX XGXAG DXADF

 

George Lasry’s decipherings

Hundreds of ADFXVX cryptograms from July to December 1918 were preserved by an US Officer named James Rives Childs. They appear in Childs’ treatise General Solution of the ADFGVX Cipher System, which was published as a book by Aegean Press Park in 2000.

Apparently, nobody was very much interested in these unsolved cryptograms until codebreaking expert George Lasry ran across them. George has solved many historical cryptograms in recent years. He is mentioned on Klausis Krypto Kolumne many times.

George-Lasry

George is clearly the best codebreaker I know when it comes to deciphering cryptograms encrypted with a known algorithm (i.e., the challenge is to find the key). For instance, George has broken the Double Columnar Transposition (Doppelwürfel) challenge, which many thought was unbreakable.

It comes as no surprise that George was very successful in breaking the ADFGVX messages mentioned in Childs’ book. He recently published his results in a Cryptologia article co-written with Ingo Niebel, Nils Kopal and Arno Wacker. In this article, the four authors present a new computerized method for the ciphertext-only cryptanalysis of ADFGVX messages, which they developed for that purpose. They also provide details on how all the keys were recovered and almost all the messages decrypted, despite the low quality of significant parts of the intercepted material. As co-author Ingo Niebel is a historian, the article also analyzes the messages in their historical context.

1 / 2 / Auf einer Seite lesen

Kommentare (56)

  1. #1 Thomas
    23. Februar 2017

    The (general) solution of ADFGVX ciphers is provided by Friedman: https://archive.org/stream/41761079080022#page/n85/mode/2up
    Have fun!

  2. #2 George Lasry
    23. Februar 2017

    Klaus, thanks for your kind words. I would suggest to also post the list of keys which I found. There are high chances that most of cryptograms were encrypted using one of those keys, which were used for a 3-day period each, but some error occurred, either at encryption time, at transmission or at interception. I was able with a lot of trial-and-error to recovery similar messages, where the problem was that some groups were either missing or superfluous. For those posted here I had no success. Good luck!
    George

  3. #3 Thomas
    23. Februar 2017

    Are there other cryptograms from Nov. 1st till Nov. 16th 1918 that are solved?

  4. #4 George Lasry
    24. Februar 2017

    Thomas@ – yes, there are quite a few.

    Here is a list of the keys recovered – hope this helps.

    Period – Transposition Key (Length) + Substitution Key – Cryptograms Recovered
    Sep 19 – 21 – 12,2,7,20,10,19,1,13,9,18,3,17,21,8,14,4,6,16,11,22,5,15 (22)
    “D5613Q9KBNO0HY8EISJUTZFCW7VPML2ARG4X” 15
    Oct 4 – 6 – 4,13,3,14,1,16,9,15,5,19,10,18,6,17,7,20,11,21,8,12,22,2 (22)
    “YN87PJ3WRUCIEO1SKLZX0DFBH6MT9A2QV54G” 24
    Oct 28 – 31 – 6,15,12,16,5,7,14,4,13,8,11,1,17,2,10,3,18,9 (18)
    “HI20SXRUWQY8EK7O619CBJAP453FDZTGLMVN” 33
    Nov 1 – 3 – 3,16,4,15,7,12,18,6,17,8,19,1,13,10,2,14,11,9,5 (19)
    “UILOF9RCZVSX02G7QTD8WNB5JMHEKPY41A36” 100
    Nov 4 – 6 – 7,10,8,14,3,11,16,1,6,13,4,9,15,5,12,17,2 (17)
    “17WHFLJ5D2UPEXKVZ9O0Q3Y6R8ABGITCMS4N” 106
    Nov 7 – 9 – 6,12,7,15,1,11,16,5,8,14,3,18,9,13,2,17,20,10,19,4 (20)
    “PRMYUW3LZGES8C71QOV29ITB40-KXH-AJNDF” 93
    Nov 10 – 12 – 9,12,7,11,3,8,16,6,14,2,10,15,5,13,1,4 (16)
    “4ARUT1OIFSKN3-BZPVLD-JMXCWHQ2E-G0-Y-” 46
    Nov 13 – 15 – 13,8,6,16,7,18,1,14,9,20,10,15,17,2,3,11,5,19,4,12 (20)
    “JZLH-R–S-T-MKDWU-V-B-P–FAO-GIX-CNE” 13
    Nov 13 – 15 – 4,11,5,14,9,7,16,1,12,15,6,10,3,13,8,2 (16)
    “H–BMUF15PX0DJLR—-S6VONKZ-AWITEGC-” 52
    Nov 16 – 18/ Dec 7 – 9 – 7,12,1,14,8,16,13,9,19,3,15,4,10,18,6,2,11,17,5 (19)
    “WG-EITNHUB2R–FDZJS—PY-VQL-1OAXMKC” 36
    Nov 19 – 21 – 13,20,3,16,7,14,4,12,8,11,5,15,2,18,17,10,19,6,1,9 (20)
    “LC58QH7VI2YB9EURO60GX3MTFAKP1D4NJZSW” 12
    Nov 22 – 24 – 6,12,16,7,14,22,11,18,1,15,8,10,20,2,13,21,3,17,19,5,9,4 (22)
    “QNZ72XS4C0IJY3RBEKL9FD6GMTHUVWA5O8P1” 26
    Nov 25 – 28 – 21,9,6,14,10,20,1,16,18,7,15,4,11,22,5,17,23,2,12,8,19,3,13 (23)
    “HQ05DKZAOYM6BEIWTJ7PSCFLV94132NGURX8” 33
    Nov 28 – Dec 1 – 9,3,14,10,2,8,15,4,16,11,5,13,6,12,1,7 (16)
    “782GPY5OQHF91UDNI364TLVXEAR0JZBKMCSW” 29
    Total: 618

  5. #5 Klaus Schmeh
    24. Februar 2017

    @George:
    The keys are now available on page where the cryptograms are listed: http://scienceblogs.de/klausis-krypto-kolumne/unsolved-adfxvx-messages-from-world-war-i/

  6. #6 Thomas
    24. Februar 2017

    On (pencil and paper) cryptanalysis of the “RICHI” ADFGVX ciphers (by Childs and Friedman): https://ia801305.us.archive.org/30/items/41784789082381/41784789082381.pdf (pp. 14…)

  7. #7 Thomas
    25. Februar 2017

    Is there an error in the copy of page 171 (http://scienceblogs.de/klausis-krypto-kolumne/unsolved-adfxvx-messages-from-world-war-i/)? It seems that the cryptogram of page 176 (RICHI-224) partially covers the cryptogram of page 171 (RICHI-?).

  8. #8 Norbert
    26. Februar 2017

    @George:
    Two of the given substitution keys seem to have a length of 33 characters:

    (Nov 13-15) "H–BMUF15PX0DJLR—-S6VONKZ-AWITEGC-"
    (Nov 16-18) "WG-EITNHUB2R–FDZJS—PY-VQL-1OAXMKC"

    Looks a bit like MS Word’s auto correction (several consecutive hyphens replaced by a dash?). Clarifying much appreciated 🙂

    Another question, regarding the first cryptogram (Childs page 73): Was this message transmitted/intercepted on Oct 26th or Nov 8th?

  9. #9 Norbert
    26. Februar 2017

    Also this one is too short (34 chars):
    (Nov 13-15) "JZLH-R–S-T-MKDWU-V-B-P–FAO-GIX-CNE"

  10. #10 George Lasry
    27. Februar 2017

    The first one:
    “H BMU”,
    “F15PX0”,
    “DJLR “,
    ” S6VO”,
    “NKZ AW”,
    “ITEGC “,

  11. #11 George Lasry
    27. Februar 2017

    (Ignore the previous message)

    Here they are again, this time with + for unknown

    “H++BMU”,
    “F15PX0”,
    “DJLR++”,
    “++S6VO”,
    “NKZ+AW”,
    “ITEGC+”,

    “WG+EIT”,
    “NHUB2R”,
    “++FDZJ”,
    “S+++PY”,
    “+VQL+1”,
    “OAXMKC”,

    “JZLH+R”,
    “++S+T+”,
    “MKDWU+”,
    “V+B+P+”,
    “+FAO+G”,
    “IX+CNE”,

  12. #12 George Lasry
    27. Februar 2017

    Regarding the message on page 73 – this is not clear. The interception log is marked as Oct 26, the message itself is marked as Nov 8. I don’t know what to say.

  13. #13 Armin
    1. März 2017

    The unsolved message of page 100 can be decrpyted if the third last ciphertext block FGDDF is replaced by DGFGDDF, i.e. adding DG at the beginning of the block. Then we get this:

    KEINESTOERUNGDURCHFEINDXMITTAGS2FEINDLXDIVXIMMARSCHAUFBELGRADX

  14. #14 Narga
    1. März 2017

    Gratulations Armin! Excellent work! What did you use for the substitution and transposition key?

  15. #15 Norbert
    1. März 2017

    Some ciphertext modifications that yield readable text passages, still needing optimization.

    Page 105:
    Group 10: Cancel X
    Group 20: Cancel DG

    resulting ciphertext:
    GAGVA AXADF XVAXV XGVDA XDGAF ADFFG FXXXV DAGDX DXDXX VVGGA VDAGG GFDGG AFXDX AGAAF FDGDX GDVDV GGAFX DXAFA VDXFV VXDDF FXDAD DGVGD GDXVX XGDGD DGFFF GVADV XDVXF XDGXA ADDVX XGVDA XXVAX DXGFD DVVFV DAVXA DDGFA FDGAV XAGVD GAFFD DDGAG AGDGF DFXFD XGGFD DGVGA FAXFG AVDGG VFXGD GDFDX VXXAX GDXGG GVGDX GXXGV DFVXA AVGDX DXDGA DXGVA ADFXG GVDGG AX
    key = 3,16,4,15,7,12,18,6,17,8,19,1,13,10,2,14,11,9,5 (19) UILOF9RCZVSX02G7QTD8WNB5JMHEKPY41A36

    plaintext:
    GERMANIAATAPPEKONSTANTINOPELXRFUERMITTEHMEERDIVIC9
    VNZUXTEL9VRX62XDOV2TTELEGRXN0X58XISTELTX32141XQ4XA
    XVOMX4CNOVEMBERNULEDIGTXA8MIRALSTAFX32398XB

  16. #16 Max Baertl
    1. März 2017

    @Norbert

    Gratulation zu dieser Leistung !

    Korrigiert und mit Leerzeichen lautet der Text vermutlich wie folgt:

    GERMANIA ETAPPE KONSTANTINOPEL XX FUER MITTELMEER DIVISION ZU X TEL X NR X 62 X DIV X TELEGR X NR X 58 X ERSTELLT X 32141 X Q4 X AN X 32141 X Q4 X A X VOM X 4 X NOVEMBER ERLEDIGT X ADMIRALSTAB X 32398 X B

  17. #17 Max Baertl
    1. März 2017

    @Norbert und Armin

    Gratulation zu eurer Leistung !

    Der von Norbert entschlüsselte Text lautet korrigiert und mit Leerzeichen vermutlich wie folgt:

    GERMANIA ETAPPE KONSTANTINOPEL XX FUER MITTELMEER DIVISION ZU X TEL X NR X 62 X DIV X TELEGR X NR X 58 X ERSTELLT X 32141 X Q4 X AN X 32141 X Q4 X A X VOM X 4 X NOVEMBER ERLEDIGT X ADMIRALSTAB X 32398 X B

  18. #18 Norbert
    1. März 2017


    page 109:
    Group 16-19 (XXAXA DDVVV XXFXV DVDDG): reduce to XXAVV XXDDG

    resulting ciphertext:
    XGDFG GADFD GAXGF VFVGG VDDGV GADXF DGVGX AAGDF DADGF AGXAG GADDV FGDXD FGVVG XXAXA DDVVV XXAVV XXDDG VDAAD DXDAA VFXAX GDVGG GDAVG FDVGD GVFGF VDVVX DFFDD VVXDX VDADX XVVXV VDFGF GVVVV AGXGG XGFDA FGADV VDAFG AGDAD GDXXV GGXVA AADGV XGGDX GAGGD VGGXV FAFAX GDVGV FADGD FAXAD GVFDD DAVGV FAGAG VGD

    key = 3,16,4,15,7,12,18,6,17,8,19,1,13,10,2,14,11,9,5 (19) UILOF9RCZVSX02G7QTD8WNB5JMHEKPY41A36

    plaintext:

    OXKXMXABECDPELUNGXS4V4XUMBJRGA8GLETZTEVTVILEBEEVUE
    TX1WEITEDEFDUXNIV5IMMARSCH4UFBELGRADERKAVV2XSONSTK
    EINEEDMILNISSEXXASOXK511

  19. #19 Norbert
    1. März 2017

    This one looks good:

    page 146

    Insert FVXAA before group 10
    Reduce fourth/third to last group (XAAXV XAFDX) to XAAXX

    resulting ciphertext:
    FVFAF DXAXX GXDXA GVVXX FFFAX DXDXF AGVFV AXAXV VAFXA FVXAA XFGVD XGAXA VAVVX XVDVA AXAFA FFFXV DVXAV AXFGA DFFGA XAADG DAVAX GVGVA VXVGX AXXFX XVFAX XXAVV FAAVA FXAVX VGAGX FGFFV AFAVD FXADA DVAVX XXVGA FXDGX AAAFD AVAAF GVVFA DXVFX AFFVA FVFGX DVFDA XGXXF VFAVD XAAAV XAAXV XAAXX AXDGF GXXV

    key = 7,10,8,14,3,11,16,1,6,13,4,9,15,5,12,17,2 (17) 17WHFLJ5D2UPEXKVZ9O0Q3Y6R8ABGITCMS4N

    plaintext:
    FUNKSTELLEKERTSCHERHAE9TRB12X1FXR4FNAMENRICHARDEMI
    LKARLXFUNKSTELLENDORTIGENBEREICHSBENACHRICHTIGENXN
    ACHRICHTENCHEF4BG7834X

  20. #20 Norbert
    1. März 2017


    page 171 (from transcript, as scan is corrupted)

    group 10: replace first char by X
    group 19: first (unknown) char: X
    group 25: replace ADDVV by ADDFX
    group 27/28: reduce GDVFA ..FFA to GDFFA
    (old) group 30: replace G.VVD by XDVVD
    (old) group 34: replace GXVXV by VXXVV
    (old) group 50: after this group (DDDAX) insert DDDAA

    resulting ciphertext:
    AAVGF DDADX XDAAV GXAGX AFVAA ADDXD AXDDD DGDDD AAGDV XGVXD GDVDV GVVVV VVGXV VDFVG GVVDV DDVGX DVXVX XGAXA XDDDX XDDXD XXAXD VDXFX ADAGD AGXGA ADDFX XGXDD GDFFA XGDDX XDVVD VVXDF GVVDG VGVAG VXXVV VVVVD VXVGG VGGXG GVVGD GXXVG GVVXX DXDDG VDVVV DDDGG XVVGD DXXDD GGVDX XVVVD GGXGV DGVFA DDDAX DDDAA ADDGD VGDDX XVDVD DVVDV XGVGX XVXXV XXGVA GAGFD XDDAD AXGXD AXXAV DXXGX DXAX

    key = 6,12,7,15,1,11,16,5,8,14,3,18,9,13,2,17,20,10,19,4 (20) PRMYUW3LZGES8C71QOV29ITB405KXH6AJNDF

    plaintext:
    INUKRAINEUNDPOLENRUBELKURSETARKSTEIGENDINFOLGEBRUC
    HESZWISCHENDEUTSCHLANDUNDSOWJETREGIERUNGUNDERWARTU
    NGDERWIEDERHERSTELLUNGRUSSLANDSDURCHDEUTSCHLANDUND
    ENTENTE

  21. #21 George Lasry
    1. März 2017

    Norbert, Armin

    Great job! It is so great to see progress on those elusive messages.

    Did you use a particular/automated method, or mainly trial and error (this is what I did for other corrupted messages).

    Congratulations!
    George

  22. #22 Thomas
    1. März 2017

    @Norbert, Armin: Fantastic!

    Do your algorithms calculate the index of coincidence of the transposition results or directly apply a monoalphabetic substitution solver with n-gram fitness function after the transposing steps?

  23. #23 Norbert
    1. März 2017

    Not a very sophisticated code on my part: Just trying to add or remove up to 5 characters at two different positions, performing an exhaustive search with a n-gram value function. Could be optimized in several details, but the more messages we can sort out by a first run, the better 🙂

  24. #24 Norbert
    2. März 2017


    page 187:
    Scan and transcript show different messages. I assume the scanned one is not intended to be there and has been solved already, because no extra character is needed:
    group 13: unknown char is probably D: VADXX
    group 21: XXXDV should probably read XXDXV

    key = 4,11,5,14,9,7,16,1,12,15,6,10,3,13,8,2 (16) H23BMUF15PX0DJLR4789S6VONKZQAWITEGCY

    plaintext:
    SELLVXGENXKOMX9XAXKXBRESLAUXXBITTEWEGENDRINGENDER
    NOTLAGEINBEKLEIDUNGX4000XPAARSTIEFELZUNAECHSTNACH
    XODERBERG

  25. #25 Thomas
    2. März 2017

    Who can explain the abbreviations of the radio stations? I’ve found only these ones:

    LP: Berlin
    NKJ: Nicolaev (Ukraine)
    UKS: Sevastopol (Krim)
    OSM: Constantinople
    COS: Tiflis (Georgia)
    POT: 2 ships in the port of Poti (Georgia)

    What does “ZÖN” mean? (station in Silesia?, see Norberts decryption of page 187)

  26. #26 Norbert
    2. März 2017


    page 176 ("missing 10 letters")
    after group 03 (DFGGV): insert five letter group FVFFF
    (old) groups 25/26 (AFDV. ADV..): replace by AFFGA ADVAX DAVVV

    Quite a few additional minor changes required. The used key is:
    9,12,7,11,3,8,16,6,14,2,10,15,5,13,1,4 (16) 4ARUT1OIFSKN35BZPVLD6JMXCWHQ2E7G08Y9

    Suggested plaintext:
    DURCHBRUCHVORBEREITETXDURCHBRUCHSRICHTUNGNACHNORDEN
    ODERNORDOSTENERFOLGENWIRDXKANNJETZTNOCHNICHTBEEURTEILTWERDENX

  27. #27 Thomas
    2. März 2017

    Page 132:
    Key Nov 4-6 yields plaintext fragment “WIEDERHOLE TELEG..” The rest is silence.

  28. #28 George Lasry
    2. März 2017

    Thomas – this is a good start. It might be possible to extract more by playing with missing/superfluous symbols.
    OSM, OS = Constantinople
    COS = Tiflis
    POZ = Nauen (Germany)
    SP – Varna
    RAT – Odessa
    NKJ or SEW – Nikolaiev (main relay station)
    SFA – Sofia
    ASO, RKI – Sevastopol
    VC or CVS – Constanza
    RAK – Kiev
    UKS or MAC – Bucharest Mackensen HQ (OKM)
    also uses ZON
    POT is Poti

    Let me know if I missed anything you need.
    George

  29. #29 George Lasry
    2. März 2017

    Norbert@ – Page 187 – mistake, this is not the correct page for the mutilated message (even though they start with the same group, which is the source of my error).

    Given your descriptions and Armin’s, have we got anything left?
    🙂

    If needed, I can send a copy of the book.

    Thanks to all for making progress!
    Regards,
    George

  30. #30 Thomas
    2. März 2017

    @George
    Thank you!
    Do you have also the keys of the western front ( for the CHI-messages)?

  31. #31 George Lasry
    2. März 2017

    I think I have one key.

  32. #32 Norbert
    3. März 2017

    page ?? (between “page 176” and “page 187”)

    This message uses the key from Nov 22-24.
    After group 6 (D.F..) insert new five-letter group and after group 43 (FDFFX) as well.

    resulting plaintext:
    ABSXMI.I5XNA6M.LD.XS4NXS4.XN.R.DAKER.SC.XBRINGTENT.N.E
    FLHTTEZWO.IVI.IONENX.EUSRELAEND.RXENGLXUXF.ATZ.XMITX
    N..OHLXKOR1S..OSCHX3Y52

    This looks promising. Some potential plaintext fragments are:
    XEILMELDGX, or maybe XEILMELDGX24NX24NX (24N: from Nov 24?), KERTSCHXBRINGTENTENTEFLOTTEZWODIVISIONENX, LAENDERXENGLXUXFRANKRXMITX, and OHLX (OHL – oberste Heeresleitung).
    Kertsch (Kerch) is a seaport in the east of the Crimea; Entente troops landed in the Crimea in November 1918 (source).
    I am stuck at this point…

  33. #33 Thomas
    3. März 2017

    Some messages are about the withdrawal of the german occupation troops from Belgrade and the Crimea. I suppose George has deciphered messages that have the same subject matter and could help clarify some uncertainties.

  34. #34 George Lasry
    3. März 2017

    If anyone needs/wants additional material, I can send via email. You can find my coordinates here:
    http://www.uni-kassel.de/eecs/fachgebiete/ais/mitarbeiter/george-lasry-bsc-mba.html
    I will be happy to share other decryptions.

    • #35 Norbert
      3. März 2017

      Thanx for the offer – just sent you an email 🙂

  35. #36 Klaus Schmeh
    3. März 2017

    @Norbert:
    Congratulations! Great job! Apparently, most (or all?) of the cryptograms are solved now. It went so fast that I could hardly read all the comments. Thanks also to Armin, Max Baertl, Thomas and George. I plan to write a follow-up article, as soon as I have read and understood all the comments.

  36. #37 Norbert
    11. März 2017

    I found out something interesting about “AAXAX FXAVF” (page 132, key Nov 4-6). The sender made a mistake during the first stage of enciphering: at one point, he encrypted the plaintext letter “G” by “VVV” instead of “VV”. This lead to the “impossible” message size of 155 characters. I hope the following rectangle makes it clear – the superfluous “V” is marked, and suggestions for the two missing letters are in italics.


      1   1   1 1     1     1   1 1
    7 0 8 4 3 1 6 1 6 3 4 9 5 5 2 7 2

    A V A V F A V A F A V X A X F G F
     f   f   e   r   e   i   l   v   e
    A X G F A F D A F V X F A D F F A
       s   e   x   w   i   e   d   e
    V A A G G A A X F A X A X A F A A
     r   h   o   l   e   t   t   e   l
    X F A V V<V>V A F D F G G A X X F
       e   g     r   x   v   o   n   v
    G V X F A V A X A F A V A D X F A
       i   e   r   t   e   r   p   e
    V A V X G A D F F A V X X X A V D
     r   i   o   d   e   i   n   f   u
    V X A X X A V X A F A V A F D V V
       t   n   f   t   e   r   x   g
    F A V G F A X A X G G A V A A G F
     e   b   e   t   s   o   r   h   e
    A V A D D X F V X X X F D F G F G
       r   5   m   i   n   x   v   v
    F G
     v

    ffereilvese x wiederhole ttelegr x
    von vierter periode in futnfter [fuenfter] x
    gebetsorher [gebetsorder] 5 min x vvv

  37. #38 Thomas
    11. März 2017

    Interesting finding. What do you think is meant by “fünfter Gebetsorder”? Any idea?

  38. #39 Max Baertl
    11. März 2017

    @Norbert
    Hier die Entschlüsselten Texte mit Leerzeichen und korrigierten Fehlern.

    GERMANIA ETAPPE KONSTANTINOPEL XX FUER MITTELMEER DIVISION ZU X TEL X NR X 62 X DIV X TELEGR X NR X 58 X ERSTELLT X 32141 X Q4 X AN X 32141 X Q4 X A X VOM X 4 X NOVEMBER ERLEDIGT X ADMIRALSTAB X 32398 X B

    IN UKRAINE UND POLEN RUBELKURSE STARK STEIGEND INFOLGE BRUCHES ZWISCHEN DEUTSCHLAND UND SOWJETREGIERUNG UND ERWARTUNG DER WIEDERHERSTELLUNG RUSSLANDS DURCH DEUTSCHLAND UND ENTENTE

    FUNKSTELLE KERTSCHER HAT BETRIEB X 1F X RUFNAMEN RICHARD EMIL KARL X FUNKSTELLEN DORTIGEN BEREICHS BENACHRICHTIGEN X NACHRICHTENCHEF 4B X 7834 X

    O X K X M X ABENDMELDUNG X S4V4 X UNTERBRINGUNG LETZTER TEILE BEENDET X 1 WEITERER DU X DIV 5 IM MARSCH AUF BELGRADER KAVV X 2 X SONST KEINE EREIGNISSE XX ASO X K511

    STELLV X GEN X KOM X 9 X A X K X BRESLAU XX BITTE WEGEN DRINGENDER NOTLAGE IN BEKLEIDUNG X 4000 X PAAR STIEFEL ZUNAECHST NACH X ODERBERG

    DURCHBRUCH VORBEREITET X DURCHBRUCHS RICHTUNG NACH NORDEN ODER NORDOSTEN ERFOLGEN WIRD X KANN JETZT NOCH NICHT BEURTEILT WERDEN X

    ABS X MIDIV 5 X EILMELDG X 24N X 24N X ARMADA KERTSCH X BRINGT ENTENTE FLOTTE ZWO DIVISIONEN X NEUSEELAENDER X ENGL X U X FRANZO X MIT X NUR OHL X KORPS FRISCH X 3Y52

    FUER EILVESE X WIEDERHOLET TELEGR X VON VIERTER PERIODE IN FUENFTER X
    GEBETSORDER 5 MIN X VVV

  39. #40 Norbert
    12. März 2017

    @Thomas: Naja, ich dachte, dass die Gebetsorder der vierten Periode (was auch immer darunter zu verstehen ist) in der fünften wiederholt wurde. GEBETSORHER könnte allerdings auch “Gebet vorher” heißen und wäre dann vom resultierenden Geheimtext her sogar etwas näher dran (s = XG, v = FG, also nur ein Zeichen zu ändern statt zweien bei “Gebetsorder”). Die Order schien mir nur naheliegender.

    @Max Baertl: Very impressive! I think the one solved by Armin is missing:
    KEINE STOERUNG DURCH FEIND X MITTAGS 2 FEINDL X DIV X IM MARSCH AUF BELGRAD X

    @Klaus Schmeh: Thanks to Max Baertl’s listing, it’s easy to count 🙂 Eight messages have been solved by now (the ninth was the scan of a message already solved by the Kassel team which mistakenly appeared on the list). We are not even half-finished …

  40. #41 Norbert
    13. März 2017

    VFGAG DDFFX (page 164: “5 TL Richi-177”): Decrypted with key from Nov 7-9, the text adumbrates SONST KEINE EREIGNISSE VON BEDEUTUNG. Here is a suggested partial completion of the ciphertext:

    VFGAGD DFFXF DVDDA XA[.]DV FGDa. ..xVX DXGGA VDGGG
    VGVDD DDXDX VDDDV VDGGV FXGFA DGDAF FVFFX DGFVX
    FvXXVA GGGVA Fvgv. ..VVD GVGVA XXXXFg XVGVV gADVG
    VFXXD VVXGX VGXXG GVXXV VGXVv vVVDD .vx.g xv...
    vdfaf d..dv ddgx. ..

    Original characters (Childs) always uppercase, my suggestions lowercase.
    A: delete character
    f: suggestion for missing character
    Af: suggested correction
    [f]: insert new character here

    ELXDIEHOEHEX828HOEHELX9IR.OX
    SONSTKEINEEREIGNISSEVONBEDEUTUNGXX
    .KM.1A..31.XO.XR..SF.2O.D.

    If the first ciphertext character was G instead of V, the plaintext would begin with TL, i. e. the station shortcut quoted by Childs, but I don’t know if this was usual.

  41. #42 Norbert
    14. März 2017

    F.G.X FGXXD (page 164: 3 TL Richi-242, November 9th)
    was apparently also encrypted with the Nov 7-9 key. The actual problem with this message is the sheer amount of unknown ciphertext letters (btw, the transcript lacks one complete line of unknown letters). However, it turned out that in many cases, one half of the adfgvx bigram is given which means we can choose from at most six plaintext letters. In other words, it is pure anagramming fun 😉 Here is my suggestion for a partial ciphertext completion. No guarantee, of course, but I think most of it should fit.

    FdG.X FGXXD DAFxVD AX.GD XXDGV XXGVA VAADF DVVVV
    FVDVg dgvVVx VFG.x GdXvvvx .dx.. .gfxa axv.a a...d
    xaddd .gx.. .xvdd gdgag .a.ad dgadv xv.vv dvvvv
    f.vg. XGADG Ggd.v XVDAG XGXXG VGDGV VVFVV GFVFA
    VDGXG XXDAV XFDDV VVGVV VVVDD VXFFF XVDGG VXAGV
    GXVVG VVDGA VDVXV VXXGD DGAXF DxaDG VGVXX VAGDGd
    DD

    (same notation convention as in my last post)


    ... X . X IN X TEMESVAR X BEFRIEDNIS X E...5.V.P.T..FN. X UND M..1..R.. X DIE VON X MIRCO NACH WESTEN UND SUEDEN WEG X LEIDER WEGE VOM GEGNER BESETZT X

  42. #43 Norbert
    14. März 2017

    Addendum: “DIV VON X MIRCO” would probably make more sense (but requires one more ciphertext letter correction). A historian may help.

  43. #44 Norbert
    14. März 2017

    Addendum to page 164, VFGAG DDFFX:
    Mysterious MIRCO seems to appear in this message as well, so one of the following readings should be closer to the original text than the previously posted one:

    EL X DIE HOEHE X 828 X (or: EL X D.. HOERE X 828 X)
    O X H X L X MIRCO X
    SONST KEINE EREIGNISSE VON BEDEUTUNG XX (etc.)

  44. #45 Max Baertl
    14. März 2017

    Möglicherweise steht “Mirco” für Prinz Mirko von Montenegro siehe: https://de.m.wikipedia.org/wiki/Mirko_von_Montenegro

  45. #46 Max Baertl
    14. März 2017

    Ich Vermute ” … X . X ” am Anfang der Nachricht von Seite 164 lautet Vollständig ” VON X 9 X ” da die Nachricht vom 9. November stammt.

  46. #47 Norbert
    14. März 2017

    @Max Baertl
    Wie ich gerade sehe, könnte der Anfang sogar lauten

    ... X 9 X 11 X TEMESVAR

    Hierbei würde das F in der dritten Fünfergruppe des Geheimtextes unverändert bleiben (eine “Korrektur” weniger). Damit das erste Wort VOM, VON oder NOV lautet, müsste man wiederum einen anderen Geheimtext-Buchstaben “korrigieren” (Auswahl für das zweite Klartextzeichen WSOBHF, also möglich, für das dritte jedoch 8C71QO).

    Ich frage mich aber inzwischen, ob es überhaupt sinnvoll bzw. üblich war, das Sendedatum in die Nachricht einzupflegen. Militärische Absender-/ Adressaten- und Ortskürzel kann ich mir gut vorstellen, aber das aktuelle Datum war doch eh klar und konnte vom Empfänger notiert werden, ohne dass der Sender es erwähnen musste, oder?

  47. #48 Thomas
    15. März 2017

    @ Norbert

    Das Sendedatum würde ich im Text auch nicht vermuten. Möglicherweise aber das Datum eines bestimmten Ereignisses in Temesvar (statt “heute”).

    Zu #41: “TL” steht vermutlich für “Teile” (bei mehrteiligen Nachrichten wegen der beschränkten Zeichenzahl in einer Sendung). Die Stationskürzel stehen, wenn ich das richtig sehe, nur in den Sender- und Empfängerangaben jeweils in der ersten Zeile (…. v ….), hier taucht “TL” nie auf, auch nicht in Georges Liste #28.

    Was für eine Puzzlearbeit! Bestimmt wärst Du mittlerweile in der Lage, ein Handbuch der ADFGVX-Verschlüsselungsfehler zu schreiben

  48. #49 Norbert
    30. April 2017

    I have located the “page ??” messages in Childs’s book.

    Childs page 215:
    22nd NOVEMBER, 1918
    C) UKS v BSL (08.42 p.m., November 22nd)
       1905 (11/22)
       
       RICHI-248  VGADA FAXAD etc.

    Childs page 187:
    (The scanned “FXVVF” message appears on page 191)
    13th NOVEMBER, 1918
    A) SU v LP (07.52 a.m., November 13th)
       10.50 MK (13/11) 2 TLE
       
       1 TL    FFVXV GVVXA etc.
       CHI-212
       
       2 TL       AFAFF AXDVA etc.
       RICHI-142

    The two-part message of November 13th is remarkable: It seems very unlikely that both parts were encrypted with different keys (CHI/RICHI). Could it be possible that the sender erroneously encrypted both parts with the CHI key of that day? Then the radio operator might have noticed the discrepancy only after transmitting the first part, and probably changed for the second one the CHI indication to RICHI (mistakenly) because the receiving station “SU” was in East Europe?

    @George: Have you already tried to apply your algorithm to exactly these two message parts (FFVXV and AFAFF)?

  49. #50 George Lasry
    3. Mai 2017

    Very well guessed!
    This is a new key which does not exist in the list:
    This is the combined decryption

    russischetundpolnheeresvergehrvollerfassenkwichtimesbesondersauspolnvergehrueberohlstationverziffertfungensoweitfernschreiberverbdmnithtarbeicetkrestschriftlichknachchefjkjx

    and the transposition key (length 16)

    CMBLAKOHIDENFJGP

    Can you try and format the German text (word spplit, maybe correct garbles). I am interested to understand the context.

    Congratulations for the accurate gut feeling!

    (also happy to see that my program works on 173 plaintext = 356 ciphertext letters for length 16).

  50. #51 George Lasry
    3. Mai 2017

    This is my best take:

    russische und polen Heeres verkehr voll erfassen wichtiges besonders aus polen verkehr ueber ohl station entziffert funken soweit fern schreiber verbdmnithtarbeicetkrest schriftlich
    Nachrichten chef

    Does that make sense, and what about verbdmnithtarbeicetkrest?

  51. #52 George Lasry
    3. Mai 2017

    fern schreiber verbd ? = ? fernschreiber verbindung

  52. #53 Norbert
    3. Mai 2017

    I am glad to see that my guess was not so bad 🙂

    I think the substitution key can be improved:
    change g to k, m to g, and k to x. j and x might be numbers.

    This yields:

    russischetundpolnheeresverkehrvollerfassenx
    wichtigesbesondersauspolnverkehrueberohlstationverziffertfunken
    soweitfernschreiberverbdgnithtarbeicetx
    restschriftlichxnachchef 0 x 01 (or any digits)

    Apparently, this is exactly the same message as sent a few hours before (Childs p. 186, not on your list for Klausis Krypto Kolumne, but already deciphered by your team), but with a different key. What a mess!

    With a few corrections:
    Russischen und poln[ischen] Heeresverkehr voll erfassen. Wichtiges besonders aus poln[ischem] Verkehr ueber OHL Station verziffert funken soweit Fernschreiberverbdg [=Fernschreiberverbindung] nicht arbeitet. Rest schriftlich. Nach[richten] Chef 0.01 (or any digits)

    An amateurish translation (without garantee!):

    Capture complete radio traffic of Russian and Polish armies. If relevant, especially from Polish traffic, transmit encrypted via OHL station unless radioteleprinter (?) is working. Rest follows in written form (or: send the rest in written form). Head of intelligence 0.01 (or any digits)

  53. #54 Thomas
    4. Mai 2017

    @Norbert, George: Congratulations!

  54. #55 George Lasry
    Israel
    4. Mai 2017

    Great work Norbert, happy to see two more down 🙂

    Maybe we could summarize all those findings (in this comment thread) in a table form, so that Klaus may publish them as a new blog entry.

  55. #56 Norbert
    17. September 2017

    @George:
    I am currently working on such a table 🙂

    I think you definitely should try your algorithm on Childs page 153 “VFVAX” and “AXVAA” as well (combined and separately), because these two have also a “CHI” identifier, just like p.187 “FFVXV”. Maybe too few ciphertext letters, but worth a try!