Experiment 10

“The same two forced key stretches, with the tableau's alphabets mixed from the puzzle's own words”

The plain, cipher and key alphabets may each be a keyword-mixed alphabet built from any vocabulary word by any of the four Friedman mixings, not only the standard and KRYPTOS alphabets.

revertedstatus
experiment 9parents
4 mintime, all phases
0candidates
0/24known letters
1

The experiment method

What the method does with the 24 known letters.

The tablethe table The right-hand half of the sculpture, 26 rows of the alphabet each shifted one place further, starting from the word KRYPTOS. on the sculpture starts with the word KRYPTOS and then the rest of the alphabet. The old hand methods the designer taught allowed any word to start the alphabet, and four ways of shuffling it afterwards. So we repeat the last experiment with every word from the puzzle's world, PALIMPSEST, ABSCISSA, BERLIN, CLOCK and the rest, as the word that starts the alphabet, for the messagethe message The 97 letters of English that Jim Sanborn wrote and enciphered. letters, the carved lettersthe carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page. and the secret wordthe secret word The word or string a cipher needs besides the method. alike. If some choice makes the 13 pinned lettersthe forced letter Given a known message letter, its carved letter, and a way of using the table, the one letter of the secret word that would turn the first into the second. of the secret word read like English, that choice is probably the real one.

Words in the margin are defined on the glossary page.

1.1

The method at work

The carved letters go in at the top; the reading comes out at the bottom.

OBKRUOXOGHULBSOLIFBBWEASTNORTHEASTOTWTQSJQSSEKZZWATJKLUDIAWINFBBERLINCLOCKWGDKZXTJCDIGKUHUAUEKCARIN: THE COPPER, ALL 97 LETTERS, THE 24 KNOWN PLACES IN ACCENTLANGEYBCDFHIJKMOPQRSTUVWXZplain alphabet from LANGLEYLUCIDMEORYABFGHJKNPQSTVWXZcarved alphabet from LUCIDMEMORYIAKUNCMWVDOXSFPYBGQZLHREJTkey alphabet from INVISIBLE, by columnsEFORMESTATUforced under BERLINCLOCK: EFORMESTATUEnglish score, 5,905,872 stretches over every choice of the three alphabets-10-9-8-7-6-5-4-3random lettersEnglishshuffled K4, same search: OISTATTHEBY -3.80K4: EFORMESTATU -3.81OUT: NO READING; A GOOD-LOOKING STRETCH IS NOT A KEYEASTNORTHEASTBERLINCLOCK
the alphabets

The table can be rebuilt from any word: the word first, then the rest of the alphabet. The run tried every puzzle word for the plain and carved alphabets and four mixings of each word for the key alphabet.

the stretch

The best run, under BERLINCLOCK, reads EFORMESTATU and scores -3.81, inside the English band.

the control

The same search on shuffled K4 letters finds OISTATTHEBY at -3.80. With this many alphabets something always reads like English, so the stretch says nothing about the key.

2

Best candidate against the known letters

This experiment produced no candidate: the search ruled everything out before a full text was built. Its product is the bounded negative in section 7.

3

Walkthrough, by hand

Redo the method on a few letters with pencil and paper.

  1. Write the alphabet starting with PALIMPSEST, dropping repeats: PALIMSETBCDFGHJKNOQRUVWXYZ.
  2. Use it for the plaintext; keep the carved KRYPTOS alphabet for the ciphertext.
  3. Position 22 is plaintext E, index 6 in the PALIMPSEST alphabet; ciphertext F, index 11 in KRYPTOS; Vigenere gives key index 5.
  4. Do the same for positions 23 to 34 to get 13 key indices.
  5. Now choose a key alphabet, say ABSCISSA mixed by columns, and read off the letter at each index.
  6. Score the 13 letters by how common each run of four is in English.
  7. Repeat for every choice of the three alphabets.
  8. A stretch that scores like English, about -4.3 or better, is the signal.
the worksheet: the first steps by hand, on squared paper
THREE ALPHABETS, THEN THE FORCED KEY LETTERS UNDER EAST: VIGENERE, KEY NUMBER = CARVED NUMBER - KNOWN NUMBERplain, from PALIMPSESTPALIMSETBCDFGHJKNOQRUVWXYZcarved, KRYPTOSKRYPTOSABCDEFGHIJLMNQUVWXZkey, ABSCISSA by columnsADJOUZBEKPVSFLQWCGMRXIHNTY012345678910111213141516171819202122232425numberplaceknownplain no.carvedcarved no.key no.key letter22E6F126B23A1L1716C24S5R122H25T7V2215W26N16Q204U27O17Q203OFind the known letter's numberin the plain alphabet and thecarved letter's in the carvedone; the difference is the keynumber, read in the keyalphabet.The run tries every puzzleword for each alphabet: 63 x63 x 248 choices, 5.9 millionruns.
4

Worked example, generated

The same steps run by the code on the real letters.

plain alphabet  PALIMPSEST/keyword  PALIMSETBCDFGHJKNOQRUVWXYZ
cipher alphabet KRYPTOS             KRYPTOSABCDEFGHIJLMNQUVWXZ
key alphabet    ABSCISSA/columnar   ADJOUZBEKPVSFLQWCGMRXIHNTY

pos  plain  cipher  key index  key letter
 22    E      F         6         B
 23    A      L        16         C
 24    S      R        22         H
 25    T      V        15         W
 26    N      Q         4         U
 27    O      Q         3         O
 28    R      P        10         V
 29    T      R        20         X
 30    H      N         6         B
 31    E      G         7         E
 32    A      K        25         Y
 33    S      S         1         D
 34    T      S        25         Y

stretch: BCHWUOVXBEYDY   score -8.44 per quadgram (English about -4.3 at this length, noise about -8.3)
5

Hypothesis, for a cryptanalyst

Family, key space, the constraint under test, and the clues it reads.

0009 showed that under the standard and KRYPTOS alphabets the key stretches forced by the two crib runs read as noise. Scheidt taught the 1900-1945 hand systems (knowledge/03-scheidt-said.md), whose tableaux use keyword-mixed alphabets: a keyword then the rest of the alphabet, or the same written into a block and read by columns, by keyed columns, or alternately vertically (Friedman's mixings, kryptos/alphabets.py). K1 and K2 used KRYPTOS; K4 may use another word from the puzzle's world (C5, C6, C8, C29). Here the plain and cipher alphabets range over standard, KRYPTOS and the keyword mixing of each vocabulary word, and the key alphabet over all four mixings of each word, under Vigenere, Beaufort and variant Beaufort. For each triple the 13-letter stretch under EASTNORTHEAST and the 11-letter stretch under BERLINCLOCK are scored as English against the 0009 calibration. A key alphabet is a bijection on the stretch, so this is the test 0009 could not do. Bounds: exact copper, periods 27 and above, every triple of alphabets drawn as stated.

6

Prediction

Written before the run; the result decides it.

About 3 million stretch pairs are scored, so the best will sit around 4 standard deviations above noise by chance alone and still below the English mean. No stretch scores above -4.3 per quadgram. Zero candidates pass.

7

Method

  1. Build the alphabet pools: plain and cipher from standard, KRYPTOS and the keyword mixing of each vocabulary word; key from all four mixings of each word plus the two basics.
  2. For each plain alphabet, cipher alphabet and mode, compute the forced key index at each crib position of both runs.
  3. For each key alphabet, map the indices to letters and score the two stretches by mean log10 quadgram probability.
  4. Keep the best 20 overall and the best per mode; compare with the English and noise distributions from 0009.
8

Result and bounds

The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.

0candidates returned
0passed every gate
0/24known letters, best
–lowest loss
nonechampion before

Diagnostic, no candidates. 5905872 stretches scored over 63^2 x 248 x 3 alphabet triples.

What was searched, as the harness recorded it.

family
periodic tableau, period 27+, alphabets mixed from vocabulary; forced key stretches scored as English
plain cipher alphabets
63
key alphabets
248
modes
3
stretches scored
5,905,872
calibration from
0009
best overall
score
-3.813
stretch
EFORMESTATU
run
64-74
plain
LANGLEY/keyword
cipher
LUCIDMEMORY/keyword
key
INVISIBLE/columnar
mode
beau
z vs random
4.15
z vs english
0.85
top twenty
  1. score
    -3.813
    stretch
    EFORMESTATU
    run
    64-74
    plain
    LANGLEY/keyword
    cipher
    LUCIDMEMORY/keyword
    key
    INVISIBLE/columnar
    mode
    beau
    z vs random
    4.15
    z vs english
    0.85
  2. score
    -3.88
    stretch
    YDAMETHINGR
    run
    64-74
    plain
    CIA/keyword
    cipher
    BERLINCLOCK/keyword
    key
    CARNARVON/keyword
    mode
    varbeau
    z vs random
    4.09
    z vs english
    0.72
  3. score
    -4.002
    stretch
    KHISTOSEARL
    run
    64-74
    plain
    PALIMPSEST/keyword
    cipher
    VIRTUALLYINVISIBLE/keyword
    key
    DESPERATELY/keyword
    mode
    beau
    z vs random
    3.97
    z vs english
    0.48
  4. score
    -4.006
    stretch
    ERSTORSHINGHC
    run
    22-34
    plain
    WEBSTER/keyword
    cipher
    HILL/keyword
    key
    WORLDCLOCK/alt_vertical
    mode
    beau
    z vs random
    4.37
    z vs english
    0.53
  5. score
    -4.053
    stretch
    THEHELECANF
    run
    64-74
    plain
    DIGITAL/keyword
    cipher
    TUTANKHAMUN/keyword
    key
    WEST/keyword
    mode
    varbeau
    z vs random
    3.93
    z vs english
    0.38
  6. score
    -4.053
    stretch
    THEHELECANF
    run
    64-74
    plain
    DIGITAL/keyword
    cipher
    TUTANKHAMEN/keyword
    key
    WEST/keyword
    mode
    varbeau
    z vs random
    3.93
    z vs english
    0.38
  7. score
    -4.066
    stretch
    DERATOPLANG
    run
    64-74
    plain
    SANBORN/keyword
    cipher
    DESPERATELY/keyword
    key
    EAST/columnar
    mode
    vig
    z vs random
    3.92
    z vs english
    0.36
  8. score
    -4.066
    stretch
    DERATOPLANG
    run
    64-74
    plain
    SANBORN/keyword
    cipher
    DESPARATLY/keyword
    key
    EAST/columnar
    mode
    vig
    z vs random
    3.92
    z vs english
    0.36
  9. score
    -4.066
    stretch
    IFTHESTCESA
    run
    64-74
    plain
    SECRET/keyword
    cipher
    LOCATION/keyword
    key
    ILLUSION/columnar
    mode
    vig
    z vs random
    3.92
    z vs english
    0.36
  10. score
    -4.071
    stretch
    KHISTOSERAL
    run
    64-74
    plain
    PALIMPSEST/keyword
    cipher
    VIRTUALLYINVISIBLE/keyword
    key
    DESPARATLY/keyword
    mode
    beau
    z vs random
    3.91
    z vs english
    0.35
  11. score
    -4.086
    stretch
    ITTHASEDEDF
    run
    64-74
    plain
    NORTH/keyword
    cipher
    BURIED/keyword
    key
    CIA/alt_vertical
    mode
    varbeau
    z vs random
    3.9
    z vs english
    0.32
  12. score
    -4.1
    stretch
    NDENDTHENGY
    run
    64-74
    plain
    LUCID/keyword
    cipher
    UNDERGROUND/keyword
    key
    BERLINWALL/keyword
    mode
    varbeau
    z vs random
    3.88
    z vs english
    0.29
and 8 more
null control best
  1. -3.801, OISTATTHEBY
  2. -3.94, MOFSINTHEDB
  3. -3.944, BUSIVEREDEC
null control
same search on three shuffles of the K4 letters; best score and stretch each

How this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.

Rendered from the record on 2026-10-18 12:26 UTC; updated after every experiment.
view on GitHub