Point Pattern Settlement Analysis

Complete spatial randomness (CSR) describes a pattern in which points are placed independently and every location in the study window is equally likely — formally a homogeneous Poisson process. It is the natural null hypothesis because it represents the absence of any attraction or repulsion between sites: deviations from CSR are what indicate that some process is concentrating sites (clustering) or spacing them apart (dispersion). All the core statistics work by quantifying how far the observed pattern departs from CSR. As Hodder and Orton stress, CSR is a yardstick, not a claim that real settlement is ever truly random; it lets us measure and test structure objectively.

The nearest-neighbour ratio summarizes the pattern with a single number based only on each point's closest neighbour, so it captures small-scale spacing but can miss structure at other distances. A distribution can be clustered at short range yet regular at longer range, and a single statistic averages this away. Ripley's K (and its L-function form) evaluates the pattern across a whole range of distances at once, plotting clustering or dispersion as a function of scale. Conolly and Lake recommend it precisely because it exposes scale-dependent organization — for instance tight site clusters that are themselves evenly spaced — that nearest-neighbour analysis cannot detect.

A great deal, and this is the most important caveat. Point pattern statistics assume the mapped points are a faithful sample of the real distribution within the window. If survey was uneven — intensive in some areas, cursory in others — the recovered sites will appear clustered where coverage was good and sparse where it was poor, producing patterns that reflect fieldwork rather than past behavior. Differential preservation and visibility have the same effect. Because of this, analysts must document survey intensity, restrict analysis to comparably surveyed areas or model the bias explicitly, and treat patterns from patchy data with caution before offering any settlement interpretation.