31. Simulating the Limes. Challenges to computational modelling in Roman Studies

Session Chair: Philip Verhagen
Affiliation: Vrije Universiteit Amsterdam, the Netherlands

Session Abstract: The increasing availability of large digital data sets requires archaeologists and historians to develop or adopt new analytical tools in order to detect and understand socio-economic and cultural patterns and to compare these at wider spatial and temporal scales. Simulation and other types of computational modelling are rapidly becoming a key instruments for this type of research. They are used to bridge the gap between theoretical concepts and archaeological evidence. These models can be of an exploratory nature, or attempt to closely emulate historical dynamics, and enable us to understand the mechanisms underlying, for example, e.g. population changes or economic systems.

Despite having access to large amounts of high-quality data, Roman studies have so far been relatively slow in adopting computational modelling, and Limes studies are no exception. The Limes is a particular case since each border region has its own characteristics, environmental setting, cultural background and specific relationship with the ‘core’ but also shares common features derived from being at the ‘outskirts’ of political, economic and cultural life. The interaction between these two dimensions is highly complex. Thus, the Limes constitutes an arena where formal modelling methods have particularly high potential. However, key challenges to this approach are i) the proper integration of archaeological and historical data sets; ii) a good understanding of what proxies to use, and iii) the computational power needed for modelling at larger scales.

We invite papers that showcase examples of modelling within the broader thematic setting of the Limes, taking these challenges into account. Suggested topics of interest are the economy of the Limes, urbanisation and settlement dynamics, demography, military campaigns, and relationships between the Limes, the rest of the Roman Empire and the zones beyond the frontier. Statistical modelling, GIS, simulation (e.g., Agent-based modelling), network models and other types of formal approaches are all welcome. Comparative studies are especially welcomed.

Modelling the Social, Economic and Demographic Trends at the Roman Eastern Frontier
Iza Romanowska, Olympia Bobou, Rubina Raja, Aarhus University

For almost 300 years wealthy Palmyrenes commemorated their deceased with portraits set up in elaborate family tombs. Now we can use this wealth of information to reconstruct the historical trajectories of the city’s elite. We present an extensive analysis of over 3500 funerary portraits and other funerary data collected in the Palmyra Portrait Project and contrasted with other sources of data on the city’s wealth and status as well as the historical timeline of the region. We examine whether the trends in portrait production can serve as a proxy for i) demographic changes in the city population, ii) social transitions or iii) economic phenomena. The results of the analysis demonstrate that broad historical trends in the portraits concur with our understanding of Palmyrene history, but the more detailed patterns highlight the impact of particular historical events over other ones. In general, the funerary data reflects the continuously changing socio-economic circumstances of the Palmyrene elite more closely than its demographic history. Although the outline of Palmyra’s history is known thanks to written sources, large archaeological datasets can provide a backdrop to historical events by evaluating their impact on the communities involved. Thus combining historical and archaeological data enables us to gain a deeper and more nuanced understanding of the social and economic transitions that past societies underwent over century-long time scales.

Lost and Found on the Frontier: Modeling Forts and Landscape in Scythia Minor (4th-7th centuries A.D.)
Nathaniel Durant, Husson University

Between the 4th to 7th centuries A.D., the Roman province of Scythia Minor, located in modern-day southeastern Romania, was repeatedly overrun by Gothic, Hunnic, and other invasions from the north, which, according to literary sources, ravaged the countryside and even led to the capture and destruction of several frontier forts and settlements. Although this external stress may have not been on a daily basis, the sporadic yet destructive nature of the invasions coupled with as the repeated indications in the archaeological and historical records of foreign occupation of Roman land, suggest that the system of frontier forts were likely repeatedly modified and developed to combat these persistent threats. There is significant evidence that in setting up their frontier fortifications, the Romans took careful consideration of the surrounding landscape and opted for the most strategically viable locations. This study takes an interdisciplinary and spatial approach to determine which factors in the landscape played a major role in the placement of frontier fortifications during the 4th to 7th centuries A.D. by looking at the province of Scythia Minor as a case study. This project consists of a series of statistically generated predictive models, created from the spatial characteristics of extant frontier forts that reveal the most likely locations of “missing” forts in this province, focusing specifically on Vallis Domitiana and Ad Salices, two sites mentioned in the Itinerarium Antonini that are unattested archaeologically. Through these models and a number of other spatial methods including viewshed analysis, least-cost analysis, and remote sensing, this project addresses how the placement of Roman forts in Scythia Minor changed in this time period and what these changes reveal about the overall military strategies in this province.

Subsistence, surplus and trade along the Lower Germanic limes: modelling rural-urban socio-economic interactions
Philip Verhagen, Vrije Universiteit Amsterdam, Faculty of Humanities

Over the past 10 years, several research projects have focused on modelling the rural subsistence economy of the Dutch part of the Lower Germanic limes in order to answer the question whether the local population could provide the Roman military with surplus food. For this, various modelling techniques have been used, including GIS-based carrying capacity models (Van Dinter et al. 2014), agent-based models (Joyce 2019) and cellular automata approaches (De Kleijn et al. 2018). All these models have pointed to the possibility, if not the plausibility, of surplus agrarian production in the region, further supporting the hypothesis that food supplies for the Roman army were also obtained locally. In these models the investigation of socio-economic relations between the rural population and the urban centres has not played a significant role, even when a strong economic dependency between the two can be assumed, given the widespread occurrence of imported goods in the countryside and the spatial configuration of the rural settlement patterns. The specific mechanics of economic interaction thus remain poorly understood, leading to divergent hypotheses on the reasons for economic and demographic growth and decline. In this paper I will present a first exploratory model that can be used to simulate rural-urban socio-economic interactions in the Lower Germanic limes. It departs from, on the one hand, understanding and modelling the processes of production and trade for some of the core goods involved, in particular agrarian produce, pottery and building materials. It then focuses on the decision-making processes of producers, traders and consumers: what were the socio-economic goals of these groups, what positions did they occupy, and to what extent could they adapt their economic behaviour to changing circumstances? In the end, this should allow us to identify to what extent dependencies and inequalities arose from internal economic dynamics, or were governed by larger political, social and environmental developments. The model will be applied to the town and surrounding countryside of Forum Hadriani (modern Voorburg), using the economic and demographic evidence and hypotheses presented by Buijtendorp (2010) and De Bruin (2017).

Are all nodes born equal? The application of spatial network analysis to assess the role of rivers in the Roman-era transport network of the Meuse basin
Toon Bongers, University of Ghent

Spatial network analysis has successfully been applied by de Soto in his fundamental study of the Roman-era transport network of Britain and the Iberian and Italian peninsulas.1 De Soto focussed on the road network to deduce past political decisions, visualise chronological differences, and propose economic consequences. Since then, de Soto’s method has been applied to the Roma-Era Scheldt basin to assess the role of the rivers in the transport network.2 This study compared transport costs between several important nodes (i.e. agglomerations) and identified areas with a high degree of accessibility. The Scheldt basin was a well-connected region, facilitating exchange between northern Gaul, the North Sea coast, Britannia, and the Rhine frontier. Rivers constituted seasonal south-north oriented corridors essential in minimising transport costs. Nonetheless, the accessibility of sites depended primarily on access to the road network. This paper turns to the transport network of the Meuse basin. The results of this study connect to those of the Scheldt basin, allowing for a more comprehensive regional analysis and a comparison between two distinct regions of the northern Roman empire. The reconstruction of travel times will augment cost distance and accessibility analysis for the early, middle, and late Roman periods. Including settlement-, epigraphical-, and economic data will allow us to validate our results. This paper will produce regional maps that display transport costs, time, and overall accessibility within the network. These results cannot always be taken at face value and constitute (proposed) reconstructions of the past transport system and its economic characteristics. Although based on scientific data, our results are influenced by a general lack of data on Roman roads and Roman-era rivers, resulting in dead ends (i.e. dangles) and poorly documented regions. Furthermore, detailed data on both riverine ports (i.e. nodes) or waterborne goods often remain unpublished.

Exploring Economic Regionalism through Maritime Mobility: The Roman Red Sea and Beyond
Nicholas Bartos, Stanford University

Throughout the Roman period, lucrative maritime networks extended from the Red Sea to the Gulf of Aden and beyond, enmeshing people and goods from the Mediterranean, East Africa, Arabia, Western Asia, and South Asia into mutually constituted webs of production, exchange, and consumption. The development of economic communities at various scales and times across this space was conditioned not only by the differential involvement of regional participants, but also by the preferential opportunities and limitations of seaborne connection. Ancient and historical sources such as the Periplus Maris Erythraei and the 15th-century writings of Ibn Majid provide a glimpse into the dynamic marine topographies and diverse sailing technologies that mediated interaction. Yet despite the inaccuracies and incomplete coverage of these accounts, scholarship on the subject remains heavily reliant on the textual canon. Geographic Information System models provide an additional and underutilized perspective to address the various factors that shaped movement and social formation across the Roman Red Sea frontier. This paper will present a new cost-surface analysis for Roman-period sailing in the Red Sea and the Gulf of Aden, to my knowledge the first attempt to use geospatial software to formally model sailing times in this area and period. It will outline the methodological challenges and potential of a case study that utilizes modern coastal/off-shore wind and current data alongside a proxy for ancient vessel performance at different points of sail. It will then relate the results of voyage simulations between contemporaneous ports with the extant material record, especially ceramic networks. While these combined datasets in part serve to highlight probable missing links and archaeological lacunae, they ultimately reveal the complex human geographies and socioeconomic entanglements at this cosmopolitan edge of empire.

Connectivity and the Roman army on the Lower Danube – a GIS approach
Ioana A. Oltean, D Ciprian Lungescu, University of Exeter

The Lower Danube Roman limes represents a complex archaeological landscape, with numerous military sites along the border of the empire set within a distinct cultural setting and a challenging ecology. Instead of watching over a neat river line, its edges clearly defined and visible to all, the Roman bases were stretching along the border of the Danube floodplain – a flat, up to 30-kilometres wide world of swamps, marshes and lagoons, rushes and reeds, floating islands and wooded patches of land interspersed with winding river channels constantly exposed to silting. This study uses Geographic Information Systems (GIS) spatial analysis, including Least Cost Path and Viewshed analyses, to better understand the positioning and the connectivity potential of Roman military bases in Dobrogea (Romania). The results allow us to assess the ability of the Roman army to function as a coherent system and the challenges to be overcome in order to efficiently control this sector of the limes. Moreover, through a range of scenarios considered, our modelling approach will highlight areas along this limes sector where additional sites may be confirmed by future research.

Modelling supply logistics and strategic aspects of the Roman military presence in the Middle Danube region during the Marcomannic wars
Marek Vlach, Balázs Komoróczy, Institute of Archaeology of the Czech Academy of Sciences, Brno

Campaigning and expeditionary form of deployment of any army inevitably generates a wide range of requirements and needs. In the ancient world, particularly the Roman army excelled in supporting and sustaining substantial armed bodies dislocated and dispatched to all kinds of environmental settings. Roman-Germanic relations in the Middle Danube region during the Roman Period are characterized through the wide range of forms of interaction, oscillating between violent confrontations and diplomatic and economic relations. Despite the conflict periods cover a considerably lesser proportion of time, some of them had far-reaching implications towards the overall relations and geopolitical situation. The most significant Roman-barbarian conflict has occurred during the reign of emperor Marcus Aurelius and is traditionally called the Marcomannic wars. The present state of knowledge gradually tends to comply with the surviving narratives describing the large-scale occupation of the barbarian region. A simple emulative agent-based model was established to address some of the featuring aspects of the Roman military intervention into the barbarian territory during the conflict. Its explicit nature comprises formalized properties of the modelled environment (e.g. geomorphology, movement friction, local population density) and several types of agents (army units, garrisons, on-land/river transportation means). Their movement is solved both using the least-cost path and network structure. Based on the existing data and proxies (archaeology, Roman historiography, etc.) it aims to test assumption about conditions of the Roman military presence in the hostile territory and its occupation, amongst others the capacities to control and supply logistics.

The Weight of Roman Reign
Kira Lappé, University of Vienna

In reconstructing the Limes, new data sets need to be looked at that have been scarcely used or completely overseen in former research. In the interdisciplinary project “The Anthropocene Surge”, the case study of Vindobona – Vienna has been chosen to study the human impact on the ground from Roman times until today, using uncommon data sets and approaches. Vienna offers a huge data basis with more than 63,000 well cores drilled over the past 190 years, providing deep insights into the city’s 2,000 years of history. The knowledge of archaeological strata of the area is complemented by more than 1,200 digitised records of the excavations produced by the Urban Archaeology Division Vienna. As a novelty, both data sets were combined to give numbers to the anthropogenic ground, including thickness, volume, and mass. Challenges include the preprocessing and integration of those heterogenous data sets and the handling of information that have not been recorded by archaeological professionals (i.e. the description of material and layers of the well cores by drillmasters) and the incoherence in detail and accuracy that often accompanies large data sets. Geostatistical methods (primarily Sequential Gaussian Simulation) are used to interpolate the lower boundary of archaeological-anthropogenic strata, which builds the basis of a 3D model of human-made deposits. By including the materials documented in the well core logs, the change over time is sought to be explored, the layers of the Roman period are thus placed in relation to the later periods of the city and urban development. Although during the last fifty years the accumulation of anthropogenic ground has increased rapidly, the imprint of the Roman reign is still clearly visible in the city’s subsurface.

Take me home, Roman road: A model of the Roman terrestrial transport network, Challenges of data collection along the Limes
Adam Pažout, Tom Brughmans, UrbNET Centre, School of Culture and Society, Aarhus University, Pau de Soto, Universitat Autònoma de Barcelona

Itiner-e is an open online gazetteer of ancient roads. It aims to become an online platform for a community of scholars to explore, query, download and edit historical road data, leading to a continually improving resource. The first Roman road data is currently being added within the context of projects Viator-e (https://viatore.icac.cat) and MINERVA (https://projectmercury.eu). These projects aim to draw on all available historical and archaeological data to develop a model of the Roman road system in high detail across the entire Empire. Existing digital models of the entire Roman imperial transport system exist at a coarse level of detail that is not representative of our current knowledge of Roman roads or of geographical structuring. However, much more detailed regional and local summaries of the historical evidence for Roman roads exist, and the remaining challenges rely on digitizing and integrating these. The resulting dataset can be used to perform GIS and network analyses of specific regions or the empire as a whole and can be improved with up-to-date scholarship by a community. In this presentation, we will introduce Itiner-e and the challenges related to collecting and digitizing empire-wide Roman road data, using the Limes in the Roman east as an example. Particular attention will be paid to exploring best practice of the collection and integration of historical, epigraphical and archaeological datasets. These are highly diverse in their temporal and spatial resolution and in the level of information they provide, ranging from traveller’s accounts of the 19th century to the high-resolution surveys of recent decades. We will explore the limitations presented by primary and secondary literature and evaluate the contributions of topographical maps and remotely sensed data.