Experiment 102
0100 left one cell above all 50 coppers: Quagmire IV at period 3 after a move, 12 distinct equation sets against 0-11. Here the four period-3 cells (one alphabet or a pair, key after or before the move) are counted against 400 plain shuffles, with a look-elsewhere correction across the four, and against 400 copies of K4 whose 21-wide grid has its columns shuffled, which keeps every pair of letters 21 apart and so K4's width-21 repeats.
What the method does with the 24 known letters.
In experiment 100 one count stayed higher on K4the carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page. than on fifty scrambled copies. K4 has a known oddity: letters 21 places apart repeat in pairs more than chance. Because 21 is a multiple of 3, that oddity could raise this count by itself. So we compare K4 with 400 ordinary scrambles and with 400 scrambles that keep the oddity, by moving whole columns of a 21-wide grid around.
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.
Experiment 100 left one count higher on K4 than on fifty scrambles. K4 also repeats letter pairs 21 apart more than chance, and 21 is a multiple of 3, which could raise a period-3 count by itself.
So K4 is compared with 400 ordinary scrambles and with 400 copies that move whole columns of a 21-wide grid, which keeps every pair 21 apart.
K4's 12 beats the ordinary scrambles with p 0.043, but against copies that keep the pairs p is 0.10, and allowing for four cells it is 0.048 even under scrambles. Not significant.
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.
K4 in rows of 21, and one copy with its columns shuffled (13 short-row columns among themselves, 8 among themselves)
OBKRUOXOGHULBSOLIFBBW
FLRVQQPRNGKSSOTWTQSJQ
SSEKZZWATJKLUDIAWINFB
NYPVTTMZFPKWGDKZXTJCD
IGKUHUAUEKCAR
KHBUUOXROLOBGSBIBWLOF
RGLQKRPVQSFSNOJTSQWTQ
EJSZKAWKZLSUTDFWNBAII
PPYTKZMVTWNGFDCXJDZKT
KKGHCUAUUAIRE
letter pairs 21 apart that repeat: K4 11, the copy 11
Family, key space, the constraint under test, and the clues it reads.
Follow-up of 0097 and 0100. Mechanism to test: carved places 21 apart share a key slot mod 3, and K4 has 11 repeated letter pairs 21 apart (0056), so a period-3 key after a move may fit K4 more often for that reason alone. Null A: 400 letter shuffles (seed 102). Null B: 400 column permutations of K4 written 21 wide, the 13 columns of the short last row permuted among themselves and the other 8 among themselves, so every vertical pair is kept. Cells: Quagmire III or IV, key after (A) or before (B) the move, period 3; type-1 library plus no move; distinct canonical equation sets; linear test mod 2 and mod 13 as in 0097. Report K4's p under each null and, under null A, the share of shuffles whose smallest p over the four cells is at most K4's.
Written before the run; the result decides it.
Under plain shuffles K4's cell is near p 0.02, with a look-elsewhere p above 0.05; under the column permutations K4 is unremarkable (p above 0.2), so the excess comes from the width-21 repeats. Zero candidates.
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
III A: K4 0, shuffles mean 0.01 max 1 p 1.000, 21-column permutations mean 0.01 max 1 p 1.000; III B: K4 0, shuffles mean 0.0 max 0 p 1.000, 21-column permutations mean 0.0 max 0 p 1.000; IV A: K4 12, shuffles mean 4.81 max 23 p 0.043, 21-column permutations mean 6.21 max 26 p 0.100; IV B: K4 0, shuffles mean 0.0 max 0 p 1.000, 21-column permutations mean 0.0 max 0 p 1.000. Look-elsewhere over the four cells under shuffles: p 0.048.
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.