“F3.4: Scheidt's patent with two keyed tables and three repeating keys (N = 2), tables from the KRYPTOS order and its inverse, key lengths 1-10 each: exact”
F3.4. 0173 decided the patent's text/key relation with one table between two keys. Here N = 2, as the patent writes it: v = d + P2[b + P1[u + a]] mod 26, three repeating keys a, b, d of length 1-10 each, P1 and P2 each the KRYPTOS order or its inverse (a standard table collapses to 0173's form), the message letter numbered by the standard, KRYPTOS or inverse order with either sign (u), and the carved letter read the same way (v): 144 classes. Each known place ties one slot of each key; any two fix the third, so a compiled search with propagation decides every cell exactly.
Scheidt's patent can stack the shift-and-look-up step: shift, mixed alphabet, shift, mixed alphabet, shift. With three short secret words as the three shifts, each known letter ties a letter of each word together, and any two decide the third. A computer search tries every short length for the three words and every way of choosing the two mixed alphabets, on K4the 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
Scheidt's patent can stack its step: shift, mixed alphabet, shift, mixed alphabet, shift. Three short secret words would give the three shifts; Sanborn spoke of 'triple matrixes'.
the check
Each known letter ties one letter of each word; any two decide the third. A compiled search with that rule decides every choice of the two mixed alphabets and every length triple up to 10 exactly.
the bound
Where the known letters give 4 or more checks, nothing fits K4 or a scrambled copy. With fewer checks K4 fits as often as the copies. F3.4 closed; the few-check cells are too weak to decide.
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.
Choose key lengths, say 5, 6 and 7, and the KRYPTOS order for both tables.
Place 22: E must become F. Guess the first key's slot 1 (21 mod 5) and the second key's slot 3 (21 mod 6); the third key's slot 0 (21 mod 7) is then fixed.
Every other known place sharing two guessed slots fixes its third slot, and so on.
A contradiction sends the search back to change a guess; if every guess dies, the cell is dead.
Every length triple and table choice, on K4 and 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.
key lengths 5, 6, 7: the three slots each known place ties together
place 22: E -> F: first key slot 1, second 3, third 0
place 23: A -> L: first key slot 2, second 4, third 1
place 24: S -> R: first key slot 3, second 5, third 2
place 25: T -> V: first key slot 4, second 0, third 3
place 26: N -> Q: first key slot 0, second 1, third 4
place 27: O -> Q: first key slot 1, second 2, third 5
place 28: R -> P: first key slot 2, second 3, third 6
place 29: T -> R: first key slot 3, second 4, third 0
place 30: H -> N: first key slot 4, second 5, third 1
place 31: E -> G: first key slot 0, second 0, third 2
place 32: A -> K: first key slot 1, second 1, third 3
place 33: S -> S: first key slot 2, second 2, third 4
place 34: T -> S: first key slot 3, second 3, third 5
place 64: B -> N: first key slot 3, second 3, third 0
place 65: E -> Y: first key slot 4, second 4, third 1
place 66: R -> P: first key slot 0, second 5, third 2
place 67: L -> V: first key slot 1, second 0, third 3
place 68: I -> T: first key slot 2, second 1, third 4
place 69: N -> T: first key slot 3, second 2, third 5
place 70: C -> M: first key slot 4, second 3, third 6
place 71: L -> Z: first key slot 0, second 4, third 0
place 72: O -> F: first key slot 1, second 5, third 1
place 73: C -> P: first key slot 2, second 0, third 2
place 74: K -> K: first key slot 3, second 1, third 3
slots used: 18; checks: 6
5
Hypothesis, for a cryptanalyst
Family, key space, the constraint under test, and the clues it reads.
US 6,075,865 (Scheidt and Wack): O = a_N + pi_N[a_N-1 + ... + pi_1[I + a_0]] mod W; Sanborn's 'triple matrixes' (reports/Scheidt patents and field ciphers.md, issue #9); C5(b), C6. Model as in 'changed'. A cell (class, La, Lb, Ld) is live if some a, b, d satisfy all 24 known places; checks: 24 minus the key slots the known places use. Statistic: live cells by checks, K4 against 20 shuffled coppers; plants as a builder check.
6
Prediction
Written before the run; the result decides it.
No cell with 4 or more checks is live on K4; K4's live cells at each number of checks fall inside the coppers' range. Zero candidates.
7
Method
For each class: the message number u and the carved number v at the 24 known places.
For each key-length triple 1-10: depth-first search over key slots with propagation (any two slots on a known place fix the third), node limit 2 million (undecided counted).
Count live cells by checks on K4 and 20 coppers; three planted texts must be live in their own cell.
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
live cells with >=4 checks: K4 0 (coppers 0-0); >=1: K4 124 (coppers 98-130); undecided K4 0; plants live 3/3
What was searched, as the harness recorded it.
family
patent N = 2: v = d + P2[b + P1[u + a]], P1, P2 in {KRYPTOS, inverse}, u and v on std/KRYPTOS/inverse with signs (144 classes), key lengths 1-10 each, exact
classes
144
length triples
916
cells per text
131,904
live at least checks
1
k4
124
coppers min
98
coppers max
130
2
k4
3
coppers min
1
coppers max
7
4
k4
0
coppers min
0
coppers max
0
6
k4
0
coppers min
0
coppers max
0
8
k4
0
coppers min
0
coppers max
0
12
k4
0
coppers min
0
coppers max
0
live by checks k4
1
121
2
3
undecided
k4
0
coppers
0, 0, 0, 0, 0
max nodes
990,419
k4 live
class
kry/1/kry-inv/kry-inv/kry-inv/-1
lengths
9, 8, 5
checks
2
class
kry/-1/kry-inv/kry/std/-1
lengths
8, 5, 9
checks
2
class
kry-inv/-1/kry/kry/kry-inv/1
lengths
10, 3, 9
checks
2
class
std/1/kry/kry/std/-1
lengths
4, 10, 9
checks
1
class
std/1/kry/kry/std/-1
lengths
10, 7, 6
checks
1
class
std/1/kry/kry/kry/1
lengths
10, 4, 9
checks
1
class
std/1/kry/kry/kry-inv/1
lengths
4, 9, 10
checks
1
class
std/1/kry/kry-inv/std/1
lengths
4, 9, 10
checks
1
class
std/1/kry/kry-inv/kry/-1
lengths
10, 9, 4
checks
1
class
std/1/kry/kry-inv/kry-inv/1
lengths
8, 9, 6
checks
1
class
std/1/kry-inv/kry/std/-1
lengths
6, 10, 7
checks
1
class
std/1/kry-inv/kry/kry/-1
lengths
10, 9, 4
checks
1
and 28 more
plants
class
std/-1/kry/kry/kry-inv/-1
lengths
4, 4, 5
checks
11
status
1
class
kry/1/kry/kry-inv/kry/-1
lengths
7, 5, 5
checks
7
status
1
class
kry-inv/1/kry-inv/kry/kry/-1
lengths
5, 8, 8
checks
3
status
1
copies
5 of 20 shuffled coppers run (seed 1770)
seconds
1,772
exhaustive
yes
How this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.
Rendered from the record on 2026-10-18 12:26 UTC; updated after every experiment.