Dual Indexing in the Z408: Letter-Side Counts, Grid Markers, and the Final Eighteen
Abstract
The final 18 characters of the Zodiac Killer’s Z408 cipher have lacked a
consensus reading since 1969. Many attempts have treated the final 18 as unread
semantic plaintext - often a name - without a clear fit to the cipher’s
recovered body or to the July 31, 1969, mailing that carried the three
sections. This article proposes a different account. It treats the Z408 cipher as
governed by two separable indexing systems: a linguistic constraint in
the July 31, 1969, letters (counts of the word cipher aligned to section
“error loads” under a stated accounting) and a positional pattern in the
cipher grid (final-line plaintext fragments recurring in structurally salient
locations). The final 18 symbols are read as the hinge between those systems - not
as a proven personal name. The linguistic alignment is post-solution and
testable; the positional pattern is credited to an independent observation by SolidSnake1990.
This article does not claim a semantic decode of the final 18 characters and
does not assert that the unsolved symbols lie outside homophonic substitution
as a mathematical fact.
1. Introduction
The Z408 cipher was distributed in three parts on July 31, 1969, to the
Vallejo Times-Herald, the San Francisco Chronicle, and the San
Francisco Examiner. The main body was solved quickly; the final 18
characters were not. Interpretations of the final line as a name, residual
message, or secondary cipher have not produced an accepted plaintext.
This article does not offer a new substitution key for those 18 symbols.
It offers a structural model of their role in the mailing-and-grid
package:
- Letter-side (linguistic) index - quantitative relationship between cipher counts in the three cover letters and section-level discrepancy loads in the recovered plaintext, under an explicit error unit accounting and a swapped newspaper-to-section assignment.
- Grid-side (positional) index - recurrence of fragments from the final line of the recovered plaintext body in corners and column positions within each third, as identified by SolidSnake1990.
2. Methods and definitions
2.1 Section division
The Z408 ciphertext is treated as three consecutive thirds corresponding to the
three mailed pieces.
Table: Division of the Z408 cipher into three mailed thirds. Traditional paper
assignment shown; the tested swap is defined in §2.2.
2.2 Traditional vs Tested Newspaper Assignment
2.3 Counting “cipher” in the July 31, 1969, letters
Counts follow the Zodiac’s letter texts (including orthographic variants such
as cyipher where present in the source images).
An error unit is one counted discrepancy between the recovered plaintext form in a cipher third and a stated orthographic target (standard English form used for this study). Units are counted per occurrence within that section’s plaintext span.
Error-unit protocol
Source text. Recovered plaintext by third follows the Harden solution as presented on zodiackiller.com (Tom Voigt). Where a letter form might be read more than one way in secondary copies, the token used for counting is the form given in the unit table below.
Section boundaries. Third 1: symbols 1–136. Third 2: symbols 137–272. Third 3: symbols 273–408.
Definition. An error unit is one occurrence, within a single third, of a plaintext token (word form) that is not the standard English orthography of the word required by context in this study.
Unit grain. Counts are by token, not by number of incorrect letters inside a word. One listed form = one unit even if several letters differ from the target. This protocol is not a restatement of any Harden numeric “error” tally that counts letter-level mistakes separately.
Counted under this protocol
Final eighteen. The final 18 ciphertext symbols are not counted as eighteen units. For letter-side load alignment only, they are treated as one structural tail block attached to the third third.
Not counted under this protocol
- FORREST (treated here as a real word in context, not as a listed nonstandard token)
- EXPERENCE (not listed as a counted unit)
- Any form not named in the table above
- Letter-level sub-errors inside a token already counted once
- The final eighteen as separate error units
Paradice. Counted as nonstandard relative to paradise for load-matching only. This does not assert that the spelling was unintentional; the same form appears in later Zodiac writing.
Intent. Error units are operational counters for load-matching. This protocol does not claim the forms were designed to match “cipher” counts; that remains interpretive (§2.6).
Loads under this protocol
- Third 1: 3 units
- Third 2: 5 units
- Third 3: 1 unit + one final-eighteen tail block
Figure. Recovered plaintext grid by traditional third, with counted error-unit loci highlighted. Final 18 (bottom) treated as one structural tail block. Highlighting by the author for this study.
2.5 Positional markers
Defined in §4; observation credited to SolidSnake1990; newspaper mapping from
markers not adopted.
2.6 What this study does not claim
- A plaintext
reading of the final 18 symbols.
- That those
symbols “do not participate” in the homophonic system as a proven
cryptanalytic fact.
- That error units were consciously designed solely to match cipher counts (that remains interpretive).
3.1 Mismatch under traditional assignment
Under traditional newspaper-to-third pairing, letter “cipher” counts (2
/ 5 / 3) do not match section error unit loads (3 / 5 / 1+tail).
3.2 Alignment under swap
After swapping Vallejo and Examiner assignments:
Table: Letter-to-third pairing under the tested swap and loads defined in §2.4.
3.3 Post-solution character
The alignment is not available as an assembly instruction before the plaintext
and discrepancy list exist. It functions as a post-solution index: a
check on section-to-letter pairing after decryption.
3.4 Role of the final eighteen in the linguistic system
Under this model, the tail is the structural complement that allows the third section’s load to correspond to the Vallejo letter’s two cipher references (one plaintext discrepancy unit + one tail block). That is a claim about accounting and pairing, not about reading eighteen letters of English from the symbols.
4. The positional system
4.1 Observation (SolidSnake1990)
A second pattern appears in the cipher grid. SolidSnake1990 (Zodiac Theory
Discord) identified that plaintext letters from the final line of the solved
Z408 body occupy specific structural positions within each third - referred
to in his account as letter markers:
- Section 1: Three of four
corners contain the last three plaintext letters of that final line (ITI).
- Section 2: Five plaintext
letters in the middle column correspond to middle letters of that final
line (TEMET), slightly out of order.
- Section 3: Three of four corners contain letters from the first part of that final line (EBE).
SolidSnake1990 also suggested that marker placements are distinct from the cipher’s spelling-error pattern. Under the token-level error protocol in §2.4, that claim needs a narrower form. In the first and third thirds, the counted error tokens do not occupy the corner marker cells used for ITI and EBE. In the middle third, three counted error tokens intersect the middle column in which the TEMET letters sit (two overlapping at the first letter of the word, one at the last). Markers and error units remain different systems: markers are final-line echoes in fixed structural positions; error units are orthographic tokens for load-matching. Whole-word highlighting can place an error token in the same column as a marker without making the marker cell the designated “error letter.” This article therefore does not treat markers and error units as the same phenomenon, and it does not claim that error tokens never fall in the middle column.
Figure: Positional letter markers in the three Z408 sections (final-line plaintext echoes in corners and middle column). Pattern identified by SolidSnake1990 (Zodiac Theory Discord). Diagram redrawn by the author. Colored whole-word cells mark error tokens under §2.4 for cross-reference; in the middle third some error tokens intersect the TEMET column. One word character error from an earlier diagram supplied in connection with SolidSnake1990's work was corrected independently. This figure does not assign sections to newspapers.4.2 Interpretations not adopted
SolidSnake1990 has further suggested that the Zodiac may have chosen final-line
letters to serve as markers; that the final 18 are therefore deliberate message
content rather than filler; and that marker groups may assign sections to
specific newspapers. Those proposals are his. This article adopts the positional
observation (recurrence of final-line fragments in section-salient positions) but does
not adopt his newspaper mapping. In particular, it does not assign the
final 18 to the Chronicle on the assumption that “In this cipher is my
identity” places a personal name in the tail.
Figure: SolidSnake1990’s written outline of the
letter-marker observation and his further interpretive suggestions (newspaper
assignment; final 18 as message). Included for attribution; interpretive lines
are not adopted in this article.
4.3 Function in this article
On the reading used in this article, the pattern functions as a positional
annotation - echoing plaintext fragments to label sections visually on the
grid.
In that role, the pattern:
- depends on
plaintext layout and grid formatting;
- does not by
itself determine newspaper assignment;
- does not
replace the letter-side cipher-count test.
It operates as a reinforcing grid-side index, complementary to the
linguistic index in the July 31, 1969, letters.
In this article the positional pattern is not used to assign newspapers.
Newspaper pairing is fixed by the letter-side test in §3. The markers are
treated only as grid-side section annotation; any sense in which they “confirm”
a three-part structure is secondary to that letter-side result and is not
required for the load alignment.
5.1 Linguistic role
Under §3, the tail is the block that completes the Vallejo-paired section’s
load (1 unit + tail block ↔ two cipher references).
5.2 Positional role
Under §4, material tied to the final line of the recovered body supplies the
fragments that recur as section markers. The unsolved eighteen remain at the
end of the ciphertext stream; the model treats the final-line environment
as marker-relevant without claiming a full English reading of those eighteen
symbols.
5.3 Dual system partition (11 vs 7)
Of the characters associated with the final line, eleven participate in the ITI
/ TEMET / EBE marker roles defined in §4; seven do not.
Figure: Partition of final-line-associated characters into
marker-related (11) and non-marker (7) under the §4 definition. Relative to marker
roles only; not required for the letter-side alignment in §3.
5.4 What is not claimed
The tail is not established here as non-homophonic or non-linguistic in
principle - only as unsolved and structurally accounted for under
dual indexing.
6. “In this cipher is my identity”
The Chronicle letter states that identity is in the cipher. That
line has often been taken to mean a personal name in the unread tail.
This model does not require that reading. The July 31, 1969, package ties
the cipher to identity language (offender claim, proof facts, the Chronicle
line). The recovered body refuses a name while developing other themes. Tease
and refusal can coexist. A post-solution ordering rule can coexist with
identity language without turning the tail into a verified signature.
This article does not argue that “identity” is technical cryptographic
jargon. It argues that identity need not mean “legal name in the final
18,” and that sectional alignment after decryption is a concrete
structural alternative focus.
7. Limitations and alternative explanations
Alternatives to the linguistic match
- Discrepancies
are ordinary mistakes; agreement with cipher counts after swap is
coincidental.
- Different
orthographic targets (especially excluding Paradice) change the
Chronicle total.
- Different
section boundaries or transcripts change loci.
Alternatives to the tail’s role
- Unsolved
plaintext, nulls, noise, or incomplete encryption remain possible.
- Positional
recurrences could be accidental given language and grid size.
Scope
This study is limited to the Z408 and the July 31, 1969, letters. It does not analyze
other Zodiac ciphers.
8. Conclusion
Under a stated error accounting, the July 31, 1969, letters’ cipher
counts align with section loads when Vallejo and Examiner assignments are
swapped, with the final 18 treated as one structural tail block paired
with the single third-section discrepancy unit. Independently, SolidSnake1990’s
letter-marker observation shows final-line plaintext fragments recurring in
section-salient grid positions without disturbing those discrepancy loci.
Together, the two systems support a dual-indexing model of the Z408 mailing:
letter-side order-checking after solution, grid-side section annotation, and a
tail that participates in structural accounting rather than in a confirmed name
payload.
The model is offered for testing and refinement. It clarifies which
assumptions about the final 18 are optional and which patterns can be
reproduced under explicit rules. Credit for the positional observation belongs
to SolidSnake1990; the letter-side load alignment, the operational error list,
and the refusal to equate “identity” with a legal name in the tail are the
contribution of this article.