Bayyinah al-Munāsabāt: A Multi-Corpus, Multi-Pipeline Falsification Study of Canonical-versus-Chronological Long-Range Autocorrelation in the Uthmanic Quran

This paper reports a multi-round, multi-corpus falsification investigation testing whether the structural autocorrelation properties of the canonical Uthmanic Quran (114 surahs) are properties of the underlying revelation chronology, of the canonical's approximate length-sorted compilation structure, or of intra-quartile structural ordering decisions distinct from both. The canonical Quran is tested against the Cairo 1924 standard chronology, the verbatim Nöldeke-Schwally 1860 chronology, within-length-quartile shuffles, and four negative-control corpora drawn from the same broad linguistic and religious tradition: three canonical Sunni hadith collections (Sahih al-Bukhari, Sahih Muslim, Muwatta Malik), the Hebrew Bible at the book level under both Protestant 39-book and Tanakh 24-book orderings, and twenty-eight diwans of pre-Islamic Arabic poetry including all seven Muʿallaqāt poets. The pipeline (TF-IDF normalized cosine similarity, Fibonacci-additive growth metric, multi-lag autocorrelation) is applied uniformly across three independent tokenization regimes (Tashaphyne morphological roots, Qalsadi citation-form lemmas, character 3-grams without morphological processing). The headline finding: approximately 70 to 82 percent of the canonical Quran's apparent autocorrelation gap over random shuffles is attributable to its approximate length-sort, not to deeper intra-quartile ordering; the remaining residual (+0.15 to +0.25 at lag 20 depending on tokenization) survives Bonferroni-corrected family-wise significance at all four tested lags under the orthogonal character 3-gram pipeline and at lags 5 and 20 under the Qalsadi pipeline. The Hebrew Bible in Tanakh ordering shows distinct short-range genre clustering that decays to noise by lag 10, qualitatively different from the Quran's sustained pattern out to lag 30. A numerical discrepancy with Cross-Text Computational Linguistics §3.3 (which reports +0.53 at lag 20 where the present paper reports +0.80) is documented and preserved as unresolved per the program's reconciliation discipline. Status: candidate-finding under Cross-Text §6 Caveat 1; independent replication by a research group not the present author, with pre-registered lag thresholds and pipeline choices, is invited. All corpora are publicly available; all scripts run reproducibly in approximately five minutes on a typical laptop.