Quantifying ancient architecture: Stone-level machine learning reveals architectural variation and settlement growth at Shivta.

Built environments provide durable evidence for how ancient settlements were organized, expanded, and transformed over time. Yet quantitative approaches to documenting architectural variability are generally labor-intensive, limited to individual structures, or dependent on aggregate volumetric estimates. We present a scalable, machine-learning-based framework that shifts the unit of analysis to the individual building stone and combines two complementary datasets: low-altitude drone imagery of surface remains and close-range photographs of standing walls. Applied to the Late Antique settlement of Shivta in the Negev Desert, the method automatically digitized approximately 235,000 surface stones and 12,000 stones in 604 walls, quantifying stone size, shape regularity, and masonry arrangement across walls, households, insulae, and the settlement as a whole. Higher masonry scores cluster around churches and parts of the settlement boundary, while the spatial distribution of masonry variation reveals a previously unidentified earlier boundary and provides evidence for outward expansion from a southern core. Combined with architectural sequencing, these patterns indicate the use of larger and more regularly shaped stones during Shivta's principal phase of growth. The framework combines analytical flexibility with demonstrated robustness, supporting broader application across diverse stone-built archaeological contexts while complementing architectural energetics and enabling settlement-wide and cross-site comparisons of construction practices and settlement change.