# Tengri 137 — Page 23: evidence-led interpretation **Date:** 25 July 2026 **Tier:** investigator-derived from primary material: two PGP-signed Pastebin records and the arithmetic printed on Page 23. **Reproduction:** solve/verify_derivation.py, gaps/c1-02-files/classical_battery.py, and solve/lexicons/turkish_score.py. ## Claim Page 23 is not supported as a conventional hidden ciphertext. Its defensible payload is a crude Turkish rhyme concealed in row initials, while the complete matrix is a deliberately constructed pseudo-Turkish object built on a rigid phonotactic template. Read the first two or three letters of each of the eleven rows from top to bottom, using prefix lengths: ~~~ 2 2 2 3 2 3 3 2 2 2 3 BUNUYAZANTOSUNOKUYANAKOSUN ~~~ This is conventionally read as “Bunu yazan tosun, okuyana koşun,” a vulgar Turkish toilet rhyme. The detailed interpretation must remain bounded: it shows a constructed acrostic, not an author-supplied endpoint or a general decryption rule. ## Evidence ### Rigid phonotactic template Eight of the eleven rows share the exact consonant/vowel pattern **CVCVCVCCVCVCCV**; rows 3–4 share **CVCCVVCCVCVCCV**; row 7 has **VCVCVCVCVVCVCV**. The dominant form syllabifies as CV-CV-CVC-CV-CVC-CV: six syllables per row and 66 in total. This constraint explains why unconstrained cipher tests fail: most matrix letters are consumed by the construction template and cannot independently carry a substitution or transposition payload. ### The acrostic is not an arbitrary selection All 2,048 assignments of two- and three-letter row prefixes were tested. The published prefix sequence ranks fifth under a character 5-gram model and first under the Turkish lexicon-aware model; it is the only exact concatenation of the relevant Turkish words. A matched null tested 60 random matrices with the same consonant/vowel pattern and letter pool. None achieved the acrostic score (0/60; approximate p = 0.016). The next character-model candidates are grammatically damaged strings such as TOSU for TOSUN and KORSUN for KOSUN. The lexicon-aware control therefore supplies the relevant ranking. ### OKUZ is a deliberate template break OKUZ occurs at the beginning of row 7 and aligns to two complete template slots. It is vowel-initial where the dominant template expects a consonant. That placement causes the only template break in the whole matrix. It is consistent with a deliberately placed Turkish animal word, but it does not justify a wider plaintext claim. ### Incidental substrings are not decoded layers Searches found strings such as SNAKE, AFRIKA, KONUS, and ROMERO. Unlike OKUZ, they cross construction-syllable boundaries. Rows built from the dominant template necessarily generate pronounceable Turkish-like fragments. These strings are therefore downgraded to template artifacts. ### Amino-acid interpretations fail The proposed amino-acid reading contains implausibly high proportions of non-standard residues, including U, O, B, and Z, and lacks several ordinary amino-acid letters. It cannot represent a plausible protein sequence. This rejects the amino-acid/selenoprotein interpretation independently of other repository claims. ### The arithmetic is a presentation mechanism The eleven imposing factorisations reduce to digit blocks over repunit-like denominators. For example, several reduce to 28 threes, several to a 28-digit repunit, and others to simple factors of 10^28−1. The selected 28-digit block is divided by a repunit-like denominator and then the numerator and denominator are factorised for presentation. This demonstrates how the Page-23 matrix is generated. It does not establish that the periodic-table conversion is a cryptographic key for another layer. ### Tested channels with no supported payload The choice among element-symbol alternatives provides nominal binary choices, but divisibility sharply constrains them; one row has only a single admissible combination. Exhaustive simple bit, byte, mask, singleton, and 29-letter mapping variants produced no readable result. Periodic-index, Kasiski, route/columnar transposition, Vigenère, Beaufort, affine, and 2 × 2 Hill tests likewise produced no controlled result beyond the unmodified matrix. ## Status ### Established - The 11 × 14, 154-letter matrix is deterministically derived from the eleven fractions on Page 23. - The signed forward and reversed variants are authenticated under key 0xD152D6C5666AB731. - The dominant consonant/vowel template is real and measurable. - The selected Turkish prefix sequence is the unique lexicon-valid concatenation in the tested prefix family. - OKUZ is aligned to the row-7 template boundary and creates its unique break. - A 47 × 137 layout for themessage.txt is arithmetically impossible; the signed text has 6,448 letters. ### Rejected - Amino-acid, protein, and selenoprotein readings. - The tested families of classical ciphers. - The claim that Page 23 was decoded from the second audio file. - A 47 × 137 grid for themessage.txt. - Incidental words such as SNAKE, AFRIKA, KONUS, and ROMERO as deliberate payloads. ### Open - Whether any untested, author-selected transform exists after the acrostic. - Whether the constrained element-symbol alternatives have a separate, independently justified reading rule. - Whether the author treats the rhyme as the intended answer; later material still describes Page 23 as unsolved while hinting at an animal name. ## Conclusion The matrix is best understood as the carrier of the acrostic rather than as a solved multi-stage cipher. It was constructed to look like a mystical, genetic-encryption object — through a constrained 19-letter surface, periodic-table derivation, and nucleobase diagrams — while hiding a crude reader-directed joke in its row prefixes. That conclusion is evidence-led and bounded: it does not establish an Eternal Library instruction, a subsequent key, or a real-world attribution.