Astronomical Correspondence in the Textual Architecture of the Qurʾān: Occurrence Ranks, Verse Coordinates and the Metonic Cycle

Version 3.0 is a corrected version. Read CORRECTIONS.md first. This preprint evaluates a reported correspondence between the Qurʾānic co-occurrence of the nouns shams (sun), qamar (moon) and arḍ (earth) and the Metonic cycle (19 tropical years ≈ 235 synodic months). Exactly four verses name all three bodies (Q 7:54, 22:18, 29:61, 39:5), the same four in each of the seven classical verse-counting systems and in the digital texts of eight transmissions of four of the ten readings. At the third of them, the running counts of verses naming the sun and the moon read 19 and 12, and 19 × 12 + (19 − 12) = 235; these readings are the same in all seven systems. The sum of sūra and verse numbers of the four verses is 235 in five systems (Kūfī, both Madanī, Makkī, Ḥimṣī) and 234 in two (Baṣrī, Damascene). Under a dependence-preserving joint null model, the probability that both indices agree on the same lunisolar cycle by chance lies between 3.3 × 10⁻⁵ and 2.7 × 10⁻³ (4.0σ–2.8σ) if the coordinate function is taken as fixed in advance, and between 8.9 × 10⁻⁴ and 3.9 × 10⁻² (3.1σ–1.8σ) if all seven coordinate functions examined are counted. Because the finding was made in exploration, two further analyses were registered before they were run (https://doi.org/10.5281/zenodo.23228564). Corrected for the choice of words among eleven nouns, the probability is 3.4 × 10⁻³ (2.7σ) with the coordinate function fixed and 2.4 × 10⁻² with all seven counted; under two null models that preserve the clustering of verses it is 6.3 × 10⁻⁵ and 5.1 × 10⁻⁶, and for the coordinate index alone 3.6 × 10⁻³. With every documented freedom priced at once it is not significant. The study makes no claim of design. What changed from versions 1 and 2. Several statements in the earlier versions were incorrect and are withdrawn or corrected here: the pre-registration they cite does not exist (the DOI given belongs to an unrelated record); the coordinate index is not 235 in all counting systems but 234 in the Baṣrī and Damascene systems, two of the five traditions the earlier versions name; the joint probability 6.3 × 10⁻⁶ (4.4σ) and the permutation result p < 2 × 10⁻⁵ are withdrawn; the data deposit they cite did not contain the data. CORRECTIONS.md lists every change with its reason, and VERIFICATION.md states what was checked against primary sources and what was not. Contents: the article (PDF and Word); occurrence lists; verse numbers in the seven counting systems (nQuran, 21 transmission entries; the Ḥimṣī system from Spitaler 1935 and al-Dimyāṭī's Itḥāf); all scripts with their outputs, including the repetition of the counts on the King Fahd Complex texts of eight transmissions and the registered analyses of Table 9. The journal submissions named in the descriptions of versions 1 and 2 are closed.