Experiment 260
F13.4i (Alexis, #101: chase it). 0259 replicated K4's excess at B2 (the sculpture's table with a two-letter key, then a move) against 1,000 fresh shuffles, p 0.002, and showed it does not need width-19 moves (p 0.004 without them). Here the question is what in K4 carries it. A cyclic rotation keeps every adjacency of the carved letters (the width-7 doublet column and the width-21 repeats move together) but changes every place; the reversal keeps adjacency mirrored; cutting K4 into blocks of 4, 7 or 12 and shuffling the blocks keeps structure inside a block only. A planted B2 message is rotated the same way as a positive control: its excess should vanish when rotated.
What the method does with the 24 known letters.
K4the carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page. reads a little more like English than almost any scrambled copy of it under one particular two-step cipher, though the reading itself is gibberish. Is that because of where K4's letters are, which a hidden message would explain, or because of which letters sit next to each other, which a message would not need? We slide K4 around in a circle (every letter keeps its neighbours but changes its place), flip it, and chop it into pieces we shuffle, and see which versions keep the effect. A hidden test message treated the same way should lose it.
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.
F13.4i: K4 beats its shuffles under the table with a two-letter key, then a move. Does the effect follow K4's letters around, as a hidden message would?
K4 rotated 96 ways, reversed, and cut into shuffled blocks; three planted messages under the same cipher rotated 24 ways each.
Every rotation of every planted message stays above 95% of the shuffles; only 3 of K4's 96 rotations do. The excess belongs to K4's exact order against this search, not to a message enciphered this way.
What this experiment was built on, from the lineage page: 15 experiments before it, back to the root of its line. Its number is in bold.
This experiment produced no candidate: the search ruled everything out before a full text was built. Its product is the bounded negative under Result and bounds below.
Redo the method on a few letters with pencil and paper.
The same steps run by the code on the real letters.
F13.4i: rotate K4 and see whether its B2 excess survives.
K4 OBKRUOXOGHULBSOLIFBBWFLRVQQPRN...
rotated 1 BKRUOXOGHULBSOLIFBBWFLRVQQPRNG...
reversed RACKEUAUHUKGIDCJTXZKDGWKPFZMTT...
each through the same search (the table with a two-letter key, then any of 27,852 moves),
each scored against 500 scrambled copies of K4. A message at exact places loses its excess when rotated.
Family, key space, the constraint under test, and the clues it reads.
F13.4d/h: order B (periodic tableau key of length 2 on the sculpture's 8 table classes, then one of 0129's 27,852 moves), ciphertext only, 0238's frequency screen, climb and MIX8m rescoring. If K4's B2 excess comes from local structure of the carved letters that any arrangement of them by blocks keeps (letters that sit well together), rotations and large-block shuffles stay extreme; if it comes from a message enciphered at these exact places, only K4 in its own order is extreme, as only the unrotated plant is. Pre-registered: share of rotations above the shuffles' 95th percentile at B2 at least 0.5 means a position-free property (the excess carries no reading, F13.4h closed as a property of the carved letters); at most 0.1 means positional, and the reading is chased (judge, known words on the intermediate text). Clue entries C25 (the width-7/21 statistics), C1(a).
Written before the run; the result decides it.
Most rotations as extreme as K4 (share above the 95th percentile at least 0.5) and the reversal too: a property of the carved letters' neighbourhoods, kept by blocks of 7 and 12 more than by blocks of 4. The plants lose their excess when rotated (rotated plants at the shuffles' level). Zero candidates.
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
F13.4i: K4 B2 percentile 1.000; rotations above the 95th percentile 0.03 (median 0.46); reversal 0.736; blocks of 4/7/12 above 95th 0.04/0.05/0.11; plants unrotated 1.00, 1.00, 1.00, rotated above 95th 1.00, 1.00, 1.00; verdict positional
What was searched, as the harness recorded it.
NBENDJZERANDBCKNFILLBAADWRONNWIIHLGARTNRHHKORCULLHCNCSIDEPXMEDFCKRHALORKYXQTRGROWDOJOWSMSHDVNLYATHow this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.