Comparison of sequence repetition rates

From Zodiac Killer Ciphers Wiki
Jump to: navigation, search

This is a comparison of repeated sequences in both the 408 and 340 ciphers.

Note: I am using the Webtoy's transcription scheme.

These sequences are derived from reading the cipher as normal text (from left to right, top to bottom).

Digrams

Repeated digrams in the 408 Total repeated digrams Repeated digrams in the 340 Total repeated digrams
#B (6), q_ (4), qE (3), VW (3), OR (3), LM (3), BP (3), A9 (3), 9% (3), 9# (3), r\ (2), r9 (2), qW (2), qG (2), eG (2), cZ (2), c+ (2), _T (2), ^R (2), Xq (2), Wq (2), WV (2), WI (2), VE (2), UI (2), Tt (2), TY (2), S) (2), RU (2), PO (2), P# (2), L( (2), Ik (2), I( (2), HM (2), GY (2), EH (2), DR (2), BV (2), B% (2), AP (2), @X (2), 7L (2), 75 (2), 6q (2), /9 (2), +q (2), (S (2), %% (2) 112 (49 unique digrams) M+ (3), G2 (3), FB (3), ++ (3), p7 (2), UZ (2), Op (2), O+ (2), Np (2), I5 (2), Fl (2), By (2), Bc (2), 8R (2), 5F (2), +R (2), +& (2), )L (2), () (2), (# (2), #O (2) 46 (21 unique digrams)

Trigrams

Repeated trigrams in the 408 Total repeated trigrams Repeated trigrams in the 340 Total repeated trigrams
9#B (3), qEH (2), eGY (2), _TY (2), UIk (2), POR (2), P#B (2), EHM (2), #BV (2), #BP (2) 21 (10 unique trigrams) I5F (2), FBc (2) 7 (2 unique trigrams)

4-grams

Repeated 4-grams in the 408 Total repeated 4-grams Repeated 4-grams in the 340 Total repeated 4-grams
qEHM (2), 9#BP (2) 4 (2 unique 4-grams) None 0 (0 unique 4-grams)

No repeats exist in either cipher for n-grams where n is greater than 4.

N-grams, in multiple rotations and mirrored arrangements

This n-gram analysis was performed upon various rotations. First, the cipher is rotated 0 and 90 degrees clockwise. Each resulting ciphertext's repeated n-grams are tallied. For the second part of the analysis, the cipher is first flipped horizontally, and then rotated 0 and 90 degrees clockwise. The resulting ciphertext's repeated n-grams are again tallied.

Rotations to 180 and 270 degrees are excluded for brevity, because the repeated n-gram counts are identical to the 0 and 90 degree cases, respectively.

Maximum non-zero repeated n-gram counts are in boldface for each combination of cipher and value of N.

N Flip/Rotation Total repeated n-grams in the 408 Total unique repeated n-grams in the 408 % of ciphertext covered by repeats Total repeated n-grams in the 340 Total unique repeated n-grams in the 340 % of ciphertext covered by repeats
2 No/0 112 49 27% 46 21 14%
2 No/90 34 17 8% 40 17 12%
2 Yes/0 106 47 26% 54 25 16%
2 Yes/90 42 20 10% 41 18 12%
3 No/0 21 10 5% 4 2 1%
3 No/90 0 0 0% 0 0 0%
3 Yes/0 20 10 5% 4 2 1%
3 Yes/90 0 0 0% 2 1 1%
4 No/0 4 2 1% 0 0 0%
4 No/90 0 0 0% 0 0 0%
4 Yes/0 2 1 0% 0 0 0%
4 Yes/90 0 0 0% 0 0 0%

Observations:

  • The maximum numbers of repeated digrams and trigrams occur for the original 408 cipher, when it is unaltered by flips and rotations.
  • However, the maximum number of repeated digrams occurs for the 340 cipher when it is flipped horizontally.
    • Also, the 340 cipher's repeated trigram count is unchanged by the horizontal flip.

Remaining questions:

  • What is the significance of the increase in the maximum number of repeated digrams when the 340 cipher is flipped horizontally?
    • Experiment: Using the known 408 key, generate random encipherments to fill the 408's cipher block with encipherments of random sentences. Then, count the number of times that the encipherment of a random sentence results in ciphertext arrangements that produce maximum repeated digram counts when the ciphertext is rotated and/or flipped.