Rayleigh Test
Non-uniformity · primary test
Tests whether joint azimuths differ from uniform. Statistic Z = n · R̄². A p-value < 0.05 rejects randomness and confirms a preferred direction exists at the site.
Applied to pooled per-site data at all 5 sites
V-test
Directional · targeted test
One-tailed Rayleigh variant that tests clustering toward a specific hypothesised bearing. Use to formally test whether data clusters toward Najd (~135°) or Red Sea (~050°). Strengthens the directional argument beyond Rayleigh alone.
Useful where the Rayleigh test is borderline
Von Mises Fitting
Distribution · per family per site
Circular analogue of the normal distribution. μ = mean direction; κ = concentration (higher = tighter cluster). Fit separately to each family at each site to characterise rather than just detect the preferred direction.
10 fits: 5 sites × 2 families
R̄ Mean Resultant Length
Concentration metric · range 0–1
Length of the mean resultant vector. R̄ ≈ 1.00 = all measurements in the same direction. R̄ ≈ 0.07 = scattered. The Najd family yields R̄ = 0.98–1.00 across sites; this consistency is the key finding that validates the family as a reliable index input.
Core diagnostic separating the two families