Exploring Resilience through Connectivity
Hollow Ways and Social Networks in the Khabur Valley

Abstract

he social, economic and political networks of past human societies offer critical insights into their organisation, interactions and resilience. Network analysis provides archaeologists with a robust toolbox of theories and methods to quantitatively analyse these networks, thereby integrating archaeological data with formal and statistical approaches. While archaeological network research has gained popularity, the application of advanced statistical models remains rare. This is largely due to the high and often unknown proportions of missing data in archaeological networks, which exceed those of modern datasets and challenge the performance of such models.

This study addresses these challenges by focusing on one of the best-preserved ancient transport networks in the world: the hollow way system of Northern Mesopotamia. Despite its exceptional preservation, significant gaps remain in both the hollow way network and the associated settlement record. To mitigate these issues, I develop and implement two computational procedures: 1) connecting the fragments of the hollow ways and 2) predict potential settlement locations based on characteristics of the hollow ways. These procedures enhance the datasets and enable a more comprehensive reconstruction of the prehistoric social networks of the Khabur Valley.

The second part of this thesis applies MCMC-MLE temporal exponential random graph models (MTERGMs) to analyse the reconstructed network. These statistical models are used to identify the social processes underlying the formation of hollow ways, treating settlements as nodes and hollow ways as edges. The results reveal key factors influencing connectivity, including proximity, preferential attachment and the influence of existing relationships on the establishment of new connections. For example, the analysis confirms that larger settlements acted as hubs, forming central nodes in the network, and that closer sites were more likely to be connected.

By tracing settlement evolution in the Khabur Valley during the Bronze and Iron Ages, the study highlights broader patterns of urbanism and resilience. Those patterns are mapped on Holling's adaptive cycle by using network metrics as quantitative representations of connectivity and potential.

This research contributes methodological advancements for dealing with incomplete archaeological networks and demonstrates the potential of statistical network models to uncover hidden social processes. It provides new perspectives on the prehistoric social networks of the Khabur Valley and offers a replicable framework for investigating connectivity, urbanism and resilience in other regions and contexts.