Henty & Silva/From the Ground, 4.Analysis of Structures’ Orientations in Archaeoastronomy: Methods for the Quantitative Statistical Assessment of Peaks in Composite Probability Distributions

The study of structures’ orientations in archaeoastronomy often involves the analysis of the empirical sum of probability densities (SPD) for azimuths and declinations. A first logical step is to test the null hypothesis A of a random distribution of azimuths. This paper applies two methods: (1) the confidence envelope method with the azimuth SPD; and (2) binomial distribution with a probability parameter found from geometric estimates. Some empirical SPDs for which A is rejected can be composed of several subpopulations of structures with different grouping criteria. To assess the statistical significance of peaks or attractors potentially targeted to specific azimuths or declinations, a null hypothesis B is formulated. In its strong version, this hypothesis negates the existence of any attractor, while its weak version focuses on the local region of the peak without hypothesising about the rest of the SPD. This paper presents two methodologies for assessing B: (1) the method of the confidence envelope, applied to the strong version of B; and (2) a geometric method with the binomial distribution and a probability parameter estimated from the normalised frequency around the peak in the SPD (for the weak version of B). The methodology is illustrated with a case study of churches dedicated to the Virgin of the Assumption in southern Spain