“The key letter behind each K4 letter: K4's carved lines laid over the tableau plate, straight or folded”
0017 slid the sculpture's texts along K4 as one long string. Here the key comes from the screen's geometry: each K4 letter sits at a carved line and column, and its key letter is the letter at the same line and column of another plate, or at the mirrored column when the screen is folded at its centre (the tableau reads correctly only from the back). Plate lines have different lengths, so these keys are not windows of 0017's strings.
On the sculpture the secret text and the big tablethe table The right-hand half of the sculpture, 26 rows of the alphabet each shifted one place further, starting from the word KRYPTOS. of alphabets stand side by side on one copper screen, and the letters are cut right through it. If you fold the screen in your mind, or look at it from behind, every letter of the fourth messagethe message The 97 letters of English that Jim Sanborn wrote and enciphered. lands on a letter of the table. That letter could be its key. We lay K4the carved letters The 97 letters cut into the copper at the CIA, shown at the top of every experiment page.'s four carved lines over every four lines of the table and of the other secret texts, straight across and folded, slide them a few columns either way, and check whether any lining-up gives all 24 known letters.
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 screen
On the sculpture the secret texts and the table of alphabets share one copper screen, cut right through. Seen from behind, or folded at the centre, each K4 letter lies on a letter of the table, which could be its key.
the check
A key read this way leaves no free letter, so all 24 known letters must come out right. Laid on the lower tableau plate with no shift, at most 2 of 24 do, straight or folded, under any of the 12 table rules.
the bound
Every window of four lines on the tableau and cipher plates, straight and folded, shifted up to four columns either way: at best 5 of 24, while the scrambled coppers reach 5-7.
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.
K4's O, B, K, R are columns 28-31 of line 11 of the lower-left plate; places 5-35 are line 12, 36-66 line 13, 67-97 line 14.
The lower-right plate holds tableau rows N to Z and the footer; its line 12 is row Y.
Place 22 is E on line 12, column 18. Straight across, row Y (label first) has key letter Y's 18th cell.
Folded at the centre, column 18 lands on column 32 - 18 = 14 of row Y.
With the sculpture's alphabet and Vigenere, E plus the key letter must give the carved F; count a hit if it does.
Repeat for all 24 known places, every window of four lines on every plate, both ways, shifts -4 to 4 and 12 conventions.
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.
K4 over the lower-right tableau plate (rows N-Z, footer), natural lines, no shift
sculpture's alphabet, Vigenere: carved = known + key
place known carved line col straight key -> folded key ->
22 E F 12 18 H Z D U
23 A L 12 19 I V C J
24 S R 12 20 J V B H
25 T V 12 21 L U A E
64 B N 13 29 K B K B
65 E Y 13 30 R F Z D
74 K K 14 8 G G W W
5
Hypothesis, for a cryptanalyst
Family, key space, the constraint under test, and the clues it reads.
Type 3 of directions.md: a key read off the sculpture by position. 01-facts.md: the screen's left half carries the ciphertext and the right half the tableau, each on two plates of 14 lines; the letters are cut through and the CIA says the tableau 'is intentionally flipped so it can only be read from the back'. K4 fills the last four cells of line 11 of the lower-left plate and lines 12-14. C6(a) ('the keys were concealed at the sculpture'), C9(b) (the tableau letters are the key), C11a, Sanborn's 'visual systems' (family 4). Key letter for K4 at plate line i, column c: the letter of a source plate at line i + d, column c + s (straight) or column 32 - c + s (folded), sources the two tableau plates (header, 26 rows with labels and the extra L, footer) and the two cipher plates, d over every window of four lines, s = -4..4, Vigenere, Beaufort and variant Beaufort with key and text numbered on the standard or KRYPTOS alphabet (12 conventions). Exact: a running key leaves no free letter, so the reading must hit all 24 known letters. Ten shuffled coppers set the chance level of the best hit count.
6
Prediction
Written before the run; the result decides it.
No overlay hits all 24 known letters; the best hit count on K4 lies inside the range of the ten shuffled coppers (about 6-7 of 24). Zero candidates.
7
Method
Lay out the four plates as carved (01-facts.md and the tableau rows with the extra L in row N).
Give each K4 letter its plate line and column.
For each source plate, line offset, straight or folded, column shift and convention, read the key letter under each known place and count the places where it turns the known letter into the carved letter.
Do the same on ten shuffled coppers.
Decrypt and return any overlay that hits all 24.
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
10,800 overlays. K4 best 5/24; controls best 5-7. Overlays hitting all 24: 0. Natural alignment (lower plates, no shift), best hits: {'tableau straight': 2, 'tableau folded': 2, 'cipher straight': 2, 'cipher folded': 3}.
What was searched, as the harness recorded it.
family
running key read by position off the carved screen
sources
tableau plates (header, 26 labelled rows, extra L in row N, footer) and cipher plates, 28 lines each
line windows
25
straight or folded
2
column shifts
-4, 4
conventions
12
trials
10,800
k4 best hits
5
controls best hits
5, 6, 6, 6, 7, 6, 5, 5, 5, 5
k4 hit histogram
0
4,872
1
3,828
2
1,595
3
387
4
98
5
20
natural alignment hits
tableau straight
2
tableau folded
2
cipher straight
2
cipher folded
3
k4 top
source
tableau
start line
6
folded
no
shift
4
convention
vig/kry/std
hits
5
source
tableau
start line
7
folded
no
shift
3
convention
vig/kry/std
hits
5
source
tableau
start line
8
folded
no
shift
2
convention
vig/kry/std
hits
5
source
tableau
start line
9
folded
no
shift
1
convention
vig/kry/std
hits
5
source
tableau
start line
12
folded
no
shift
2
convention
varb/kry/kry
hits
5
source
tableau
start line
13
folded
no
shift
1
convention
varb/kry/kry
hits
5
source
tableau
start line
14
folded
no
shift
0
convention
varb/kry/kry
hits
5
source
tableau
start line
15
folded
no
shift
-1
convention
varb/kry/kry
hits
5
source
tableau
start line
16
folded
no
shift
-2
convention
varb/kry/kry
hits
5
source
tableau
start line
25
folded
no
shift
4
convention
vig/kry/std
hits
5
source
cipher
start line
1
folded
no
shift
0
convention
beau/std/std
hits
5
source
cipher
start line
8
folded
yes
shift
1
convention
beau/std/std
hits
5
and 3 more
full 24
0
controls
10 shuffled coppers (seed 89), identical search
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.