The Celestial Dog (v2.1) — Phase 2.5 FINAL: landscape-target null model + F2a operationalization audit + revised R5.8 manuscript (OSF parent 3c8y4 supplement)

This data package accompanies the manuscript "The Celestial Dog: A Neolithic Stellar Alignment System in Southeastern Sicily (5000-4300 BC)" by Gaspare Russo. It contains the deterministic Python implementation of the confirmatory statistical test, the cross-validation of stellar azimuths between two independent astronomical libraries (Stellarium 24.3 and Astropy 6.0), the deposited geographic coordinates and terrestrial azimuths, the alignment map (KMZ), and the supplementary documents supporting the pre-registration and the independent archaeological assessment of the selected sites. Primary statistical result on the two pre-registered axes (Axis A: Nymphaeum Terrace -> T1 Scoglio dei Cani; Axis C: Anaktoron of Pantalica -> T1 Scoglio dei Cani), 1,000,000 Monte Carlo iterations, deterministic with seed 42, identifies three confirmatory matches at threshold 0.25 degrees: Mintaka at 5000 BCE on Axis A, Sirius at 4300 BCE on Axis A, Betelgeuse at 4300 BCE on Axis C. Bonferroni correction is reported explicitly at five distinct levels: x18 conditional (P = 2.88e-4, 3.44 sigma), x54 unconditional recommended (P = 8.64e-4, 3.13 sigma), x72 full pre-registration pool (P = 1.15e-3, 3.05 sigma), x108 axis-aware (P = 1.73e-3, 2.92 sigma), x144 ultra-conservative (P = 2.30e-3, 2.83 sigma). Sensitivity analysis shows the result remains significant across match-thresholds from 0.20 to 2.0 degrees. All four temporal negative-control epochs (5500, 4500, 4000, 3500 BCE) yield zero matches at the adopted threshold. The package also documents the public OSF pre-registration (DOI 10.17605/OSF.IO/3C8Y4, 10 March 2026; underlying Technical Report v5.0 dated 7 January 2026, included in the OSF deposit) and the independent archaeological assessment of the selected sites by Dr. Paolo Scalora (Italian National Register of Archaeologists, Tier I, n. 9532). All scripts are deterministic and reproducible bit-for-bit using only the Python standard library. Replication: python3 PAPER_I_MonteCarlo_v10.py (3-5 minutes runtime on a standard laptop).