UC Berkeley

This dissertation analyzes changing rural settlement patterns in the Lower Danubian Plain from the Late Iron Age through Late Antiquity in order to elucidate the role played by garrison settlements in the economic strategies of peasants living near the Roman frontier. The military cordon on the northern frontier has been seen both as a stimulus to provincial economic development and as an oppressive burden preventing growth in its vicinity with no consensus forthcoming. I approach the question from the perspective of the rural producers, using the landscapes in which they chose to live as evidence for common goals and conditions. In order to isolate the role of garrison settlements from other features in the landscape, I employ a novel method of comparative multivariate logistic regression analysis. This allows me to test different hypothesized relationships against known settlement patterns while controlling for other influences on location. The result is a quantitative measure of how well each hypothesis fits empirical data.

The first chapter reviews the state of the question on Roman peasant economies, frontier economies and the military community. Having concluded that current interpretations based on documentary and artifactual evidence have failed to settle the issue of military-rural relations, I propose the quantitative analysis of archaeological landscapes as a promising way forward. Here, landscape refers to the embodied perception of a meaningful environment. Each settlement anchors the movements of the people who live there, so the locations of ancient settlements, combined with modern topographic and climatological data provide a foundation for the reconstruction of landscapes as experienced by their ancient inhabitants. I finish the chapter by describing a method of comparative modeling using logistic regression analysis for hypothesis testing. The goal of most locational analysis of this sort is a single mathematical model that predicts or explains settlement location using environmental variables. I suggest that multiple models be created using variables that have been constructed according to competing hypotheses and the goodness of fit between each model and known data be compared to the others. The model with the closest fit contains the variable that best reflects ancient reality. In this way, it is possible to assess the empirical support for each hypothesis and to select the best one.

The second chapter discusses the Lower Danubian Plain in modern Bulgaria, ancient Moesia Inferior. This frontier zone has not figured prominently in discussions of Roman frontier society. This is unfortunate because the area has a unique history of conquest that sets it apart from other, better-known frontiers: unlike the frontier in Western Europe, it was not heavily garrisoned until relatively late and, after the beginning of the fourth century, it was quite close to the imperial capital at Constantinople. In this chapter, I describe the natural environment of the study area in the middle of the Danubian Plain and the local economy prior to Roman conquest before discussing the history of the area from first century BCE through the sixth century CE. I show that the pre-Roman population of the area, though sparse, was well suited to integration with the Roman military community. I then describe the history of violence in the area and the ways in which different violent episodes impacted the countryside. Next I trace the construction and maintenance of the physical infrastructure of Roman power—forts, cities, and roads—from conquest to collapse before investigating the changing origins of the resident population. I conclude the chapter with an examination of the evidence for the economic status of garrison settlements in Moesia Inferior.

In the third chapter, I describe the process of systematizing the diverse and varied record of archaeological research in the study area. The result is a database that includes ancient places of various functions grouped into chronological phases stretching from Pre-Roman to Late Antique. I also describe how I reconstructed archaeological landscapes for each settlement and how these landscapes, grouped chronologically, were analyzed. First, the immediate territories around settlements are compared to territories around random locations to determine if there are factors that are more or less abundant in one group than the other. Then, the accessibility of traffic routes and possible market centers is compared. As a result of this, I show that Roman settlements are located in very different landscapes than either Pre-Roman or Late Antique settlements. There is little consistency in Pre-Roman landscapes, but Roman landscapes are ideal for intensive agricultural production, and Late Antique landscapes offer greater defensive capabilities. I then use logistic regression analysis to create baseline models of settlement location to which I add Market Potential variables to test the various hypotheses on which they were constructed. The primary result is that settlements from the Middle Roman period (second to third centuries CE), avoid forts and cluster around non-military centers.

In the final chapter, I discuss the strengths and weaknesses of quantitative landscape analysis and comparative modeling before assessing the impact of these results on our understanding of the local economy and the role of garrison settlements in both central Moesia Inferior and the Empire in general. I end by outlining next steps, both for improving the methodology and expanding the scope of investigation.

This dissertation reaches the following main conclusions: 1) Settlement-centered landscapes contain valuable evidence for the behavior of people who are not well-represented by traditional archaeological and historical evidence. 2) While no clear tendencies emerge from the Pre-Roman settlement pattern, Roman settlements show a strong preference for landscapes best suited to intensive agricultural production. In contrast, after the late fourth century, rural settlements prefer locations with access to defensible refuges, demonstrating the value of the security previously provided by the Empire. 3) The rural economy of central Moesia Inferior flourished during the Roman period so military demand did not depress the local economy. 4) At the same time, peasants in this particular frontier zone were not using garrison settlements as frequent markets for their produce. They may have supplied the frontier indirectly or infrequently, but most would have had few opportunities to visit the army bases themselves. This means that peasants were not in a position to exploit soldiers’ demand for local produce to supplement their rations. 5) Nothing in the material or literary record would have suggested that the military communities were isolated in this way, so further investigation along these lines in other frontier zones is warranted.