A probabilistic framework and significance test for the analysis of structural orientations in skyscape archaeology.

Abstract

The assessment of the orientation of built structures is often hampered by the lack of historical or ethnographic records that could be used to support claims of alignments to celestial events such as the rising and settings of sun, moon or stars. This has led to an obsession with surveying large numbers of structures in order to iden- tify patterns in orientation that might betray intentionality - however, no first-principles framework has ever been proposed and the few statistical significance tests used are of limited applicability. This paper addresses these problems by laying out a probabilistic framework for the analysis of structural orientations and using it to develop a test of statistical significance. The framework is based on two simple premises: firstly, that the mea- surement of a structural orientation can be modelled as a probability distribution; and secondly, that in order to assess the likelihood of celestial alignments such a distribution needs to be coordinate-transformed in a man- ner not unlike that of radiocarbon calibration. A method that aggregates multiple structural orientations and quantifies their statistical significance in the form of a p-value is then introduced. Finally, the robustness of the presented methodology over previous approaches is demonstrated using real-world datasets of orientations of an- cient Egyptian temples and tombs and Scottish Recumbent Stone Circles, and the ensuing conclusions compared with past interpretations.