“F1.6: one carved letter wrong at a known place, or one letter missing or extra between the known runs, run through the exact tests for one and two unknown alphabets at every key length”
F1.6, first part. 0119 and 0123 decided one and two unknown alphabets with a periodic key exactly, on the copper as carved. K2 lost a letter and K1-K3 carry slips; F1.1's added form (K1 and K2's own system) dies only because S sits under S at 33 and K under K at 74. Here every single-letter error the exact tests can see is tried: each of the 24 carved letters at the known places replaced by each of the other 25 letters (600), and one letter missing or extra between the two known runs (4, the shift read on either side; an error before EAST or after CLOCK shifts every known place together and changes no periodic check, so it is the copper as carved). For each variant: one free alphabet, added and Beaufort forms, and two free alphabets, key lengths 1-26, decided exactly. Ten scrambled coppers get the same 604 variants.
The earlier messagesthe earlier messages The three solved parts of the sculpture. contain slips: a dropped letter, misspellings. If one carved letter of the fourth messagethe message The 97 letters of English that Jim Sanborn wrote and enciphered. is wrong, a cipher already ruled outdead Said of a period, grid or method when the known letters contradict it: no secret word of that kind can produce the carved letters. might have been ruled out unfairly. In particular K1 and K2's own method, with a scrambled alphabet, fails only because two carved letters equal the known lettersthe known letters Sanborn has confirmed that positions 22 to 34 of the message read EASTNORTHEAST and positions 64 to 74 read BERLINCLOCK. above them. We change each carved letter at the 24 known places to every other letter, and add or drop one letter between the two known words, and rerun the exact tests for one and two scrambled alphabets, on the carved textthe carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page. and on scrambled copies.
Words in the margin are defined on the glossary page.
1.1
The method at work
The carved letters go in at the top; the reading comes out at the bottom.
the idea
K2 lost a letter; K1-K3 carry slips. If one carved letter of K4 is wrong, an exact test may have ruled a cipher out unfairly; K1 and K2's own system with a scrambled alphabet dies only on S under S at 33 and K under K at 74.
the check
Each carved letter at the 24 known places changed to each other letter, and one letter added or dropped between the known words: 604 variants, each decided exactly for one and two scrambled alphabets, key lengths 1-26, on K4 and 10 scrambled copies.
the bound
Fits at the copies' rate. One cell stands out by meaning: with the S at 33 changed to any other letter, K1 and K2's system fits at key length 19; 0148 judges it by English.
2
Best candidate against the known letters
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.
3
Walkthrough, by hand
Redo the method on a few letters with pencil and paper.
Take the carved S at place 33 and suppose Sanborn cut the wrong letter: try each of the other 25 letters there.
With one scrambled alphabet on both sides and the key added (K1 and K2's method), write the 24 equations: place of carved letter = place of known letter + key value of its slot.
A carved letter equal to its known letter forces key value 0 in its slot; with the S changed, place 33 no longer does.
Decide, for every key length, whether any alphabet and key satisfy all 24 equations.
Do the same for every other carved letter at a known place, for an added or dropped letter between the two known words, and on scrambled copies.
the worksheet: the first steps by hand, on squared paper
4
Worked example, generated
The same steps run by the code on the real letters.
one alphabet on both sides, added form (K1 and K2's method): place(carved) = place(known) + key value of the slot
as carved: place 33 is S under S and place 74 is K under K, so their slots need key value 0;
at every key length 1-26 another known letter shares one of those slots and is not its own carved letter: dead.
variant: the carved S at 33 replaced by R (one slip of the chisel)
place 33 no longer forces 0; the slot of 74 still does (K under K)
the 24 equations are decided again at every key length, and the same for every other letter at every known place
5
Hypothesis, for a cryptanalyst
Family, key space, the constraint under test, and the clues it reads.
C17(a)/(b) and family G (knowledge/06-hypotheses.md: 'every family should be tested once more with one letter inserted or deleted at each position'), C27(c), C1(b); Scheidt 2005 (knowledge/sources/wired-2005-zetter-questions-for-kryptos-creator.txt): 'since he's the one who had the chisel in his hands, there could be some changes'. Models: 0119's one unknown bijection on both sides, added form x(c) = x(p) + k or Beaufort x(c) = k - x(p); 0123's two bijections C(c) = P(p) + k. Deciders: 0119's (CRT mod 13 and 2, exhaustive mod-13 enumeration) and 0123's, witness re-checked; node limit 20,000 for the two-alphabet search, undecided cases counted. Statistic fixed before the run: per model and length, the number of variants that fit; cells where K4's count is above every copper's, against each copper's count against the rest. Also reported: every variant that revives the added form on K4.
6
Prediction
Written before the run; the result decides it.
Changing the carved letter at 33 or 74 revives the added form at some lengths (it removes the self-encryption that kills it), on K4 as on the coppers' corresponding variants; overall fit counts per (model, length) sit inside the coppers' range, K4 above every copper in no more cells than the null's maximum. No reading; any revived cell joins F1.5's undecided cells for the English decider.
7
Method
Build the 604 variants of K4 (and of each of 10 coppers): substitution at each known place, and the known runs pulled together or pushed apart by one.
For each variant, model and key length 1-26: the 24 equations, decided exactly.
Count fitting variants per (model, length) per text; compare K4 with the coppers; list K4's added-form fits.
8
Result and bounds
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
0candidates returned
0passed every gate
0/24known letters, best
–lowest loss
nonechampion before
K4 fitting (variant, length) pairs: {'1V': 114, '1B': 1421, '2': 5097}; added form revived on K4 in 114 pairs; K4 above every copper in 2 of 78 cells, null 0-11; stop False; undecided K4 {'1V': 0, '1B': 0, '2': 8}
What was searched, as the harness recorded it.
family
one carved letter wrong at a known place (600) or one letter missing/extra between the runs (4: two readings each), through the exact one- and two-free-alphabet periodic tests
models
1V
one alphabet, added form (Quagmire III, K1/K2's system)