Experiment 104
0003 chained plaintext autokey, where the key for place i is the message letter a fixed distance back. Here the key is the message letter at a + b i mod 97 for every a and b, K3's counting rule, the mirror (b = -1) included. Because the rule is a permutation of the 97 places, each known letter fixes every letter of its cycle, so the test is exact and, where it passes, it decrypts.
What the method does with the 24 known letters.
Here the messagethe message The 97 letters of English that Jim Sanborn wrote and enciphered. is its own key: to encrypt a letter, you add another letter of the message, picked by K3the earlier messages The three solved parts of the sculpture.'s counting rule. Because that rule visits places in a loop, one known letter gives the next one in its loop, and so on all the way round, until the loop comes back to another known letter. If the two disagree, the setting is impossible. We try every start, step and table rule.
Words in the margin are defined on the glossary page.
The carved letters go in at the top; the reading comes out at the bottom.
Here the message is its own key: each letter is shifted by another message letter, picked by K3's counting rule. The rule runs in loops, so one known letter forces the next letter round its loop.
When a loop comes back to a known letter, or closes on itself, the forced letter must agree. With the mirror rule every loop is two places long and gives a check at once.
Every start and step, 12 table rules: no setting survives on K4, and none on ten scrambled coppers either. The idea is ruled out by the known letters alone.
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.
Redo the method on a few letters with pencil and paper.
The same steps run by the code on the real letters.
a = 0, b = 96 (the mirror): place i takes its key from the message letter at place 99 - i
sculpture's alphabet, Vigenere: carved = message + key
place known carved key place message letter there = carved - known
22 E F 77 R
23 A L 76 D
24 S R 75 U
25 T V 74 M
64 B N 35 E
65 E Y 34 L
Family, key space, the constraint under test, and the clues it reads.
Companion to 0103 (the copper keyed by itself). Model: c_i = E(p_i, p_{f(i)}), f(i) = a + b i mod 97, b = 1-96 (97 is prime, so f is a permutation), a = 0-96, f without fixed points among the known places; E in 8 classes (plain and cipher alphabets standard or KRYPTOS; Vigenere, Beaufort; key letters numbered on the plain alphabet), plus variant Beaufort via the sign. Propagation: if p_i is known, p_{f(i)} = the key that turns p_i into c_i; if p_{f(i)} is known, p_i = D(c_i, p_{f(i)}). Each cycle of f containing a known place is fixed completely; a contradiction with another known letter on the same cycle kills the setting. Survivors are decrypted on every fixed place and scored. C6, C13 ('mirrors'). Ten shuffled coppers.
Written before the run; the result decides it.
No setting survives where a cycle carries two or more known letters; the survivors, if any, are settings whose cycles each hold one known letter, as often on the coppers, and they decrypt to noise. Zero candidates.
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
7,031 maps x 12 rules. Settings with no contradiction: K4 0, controls [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]. K4 survivors' checks: {}. Full decryptions returned: 0.
What was searched, as the harness recorded it.
How this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.