Experiment 127
F1.4, replication (order-of-work rule 5). 0126 ran out of time: the skip schedule (B) was tested on part of its 3,729 distinct partitions and schedules C-F not at all (their zeros mean untested). In what was done, B with two free alphabets gave K4 461 consistent partitions against 305-449 on ten coppers. Here all 3,729 B partitions are tested on K4 and 30 fresh shuffles (seed 127), all four tests, with the search's node limit lowered to 100,000 so no case can stall, undecided cases counted, coverage recorded, texts run in parallel.
What the method does with the 24 known letters.
In the last run, one hand rule (skip a key letter every so often) fitted K4the carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page. with two unknown alphabets in more ways than it fitted ten scrambled copies. That run was cut short, so here the rule is tested completely, against thirty new scrambled copies, to see whether K4 really stands out.
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.
In 0126 the skip rule with two unknown alphabets fitted K4 in more ways than ten scrambled copies. The rule says: replicate before anything else.
All 3,729 groupings of the known places, the four exact tests, K4 and 30 new scrambled copies, nothing cut short.
K4 1,135 against 681-1,300: rank 6 of 31. The settings that put places 28 and 66 (both R under P) together are the only ones where K4 leads, as 0124 explained.
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.
skip schedule: key word of 7 letters, one key letter skipped every 5 letters; KRYPTOS on both sides, Vigenere
place known carved key letter no. key value
22 E F 4 1
23 A L 5 10
24 S R 6 21
25 T V 0 18
26 N Q 2 1
27 O Q 3 15
28 R P 4 2
29 T R 5 23
30 H N 6 5
31 E G 1 2
32 A K 2 19
33 S S 3 0
34 T S 4 2
64 B N 5 11
65 E Y 6 17
66 R P 1 2
67 L V 2 5
68 I T 3 15
69 N T 4 11
70 C M 5 9
71 L Z 0 8
72 O F 1 7
73 C P 2 20
74 K K 3 0
key letter numbers with clashing values: [0, 1, 2, 3, 4, 5, 6]: dead
Family, key space, the constraint under test, and the clues it reads.
Follow-up of 0126 (C20, C5). Same model and exact tests: key value at place i = W[f(i) mod n], f(i) = i + floor((i + phi)/m), m 2-30, phi 0..m-1, n 2-26; distinct partitions of the 24 known places; tests: 8 standard/KRYPTOS tables, one free alphabet V and B, two free alphabets (CRT decider of 0119/0123). The partitions are heavily nested (different m, phi, n giving related slot patterns), so counts are strongly correlated and a single coincidence in the known letters can lift many at once, as R under P at 28 and 66 did in 0124 (that pair shares a slot whenever f(65) - f(27) is a multiple of n). Report K4's rank among 30 fresh coppers and, after the run, whether K4's excess partitions are those that put 28 and 66 in one slot.
Written before the run; the result decides it.
K4's count for B with two free alphabets falls inside the range of 30 fresh coppers (rank no better than 3rd), so the 0126 excess was the small control set; the standard/KRYPTOS tables give 0 consistent partitions on K4. If K4 tops all 30, stop and report.
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
B skip, two free alphabets: K4 1135, coppers 681-1300, K4 rank 6 of 31. Tables: K4 0. Coverage 3729/3729; undecided K4 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.