Introduction
From the late third century ad onwards, the north-western continental provinces of the Roman Empire underwent rapid change. The landscape of some regions was apparently depopulated (Heeren, Reference Heeren, Roymans, Derks and Hiddink2015), whilst settlement was fragmented in others (Lenz, Reference Lenz1999; Dodd, Reference Dodd2021). Frontier zones were in flux, with significant changes in the military footprint along the Rhine (Van der Meulen, Reference Van der Meulen2017) and a new system of hinterland defence developed, focused on the road network (Brulet, Reference Brulet1990; Brulet et al., Reference Brulet, Mertens and Leva1995).
The lowland zone of north-western Europe is dominated by fertile loess soils, stretching from northern France to the Rhine, bordered by coastal wetlands to the north and hillier landscapes to the south. This article examines one element of the Roman defensive infrastructure in the loess plain of the Lower Rhine region (Rheinische Löβbörden) (Figure 1). Archaeologically, this zone was one of the densest areas of first- to third-century rural settlements in northern Europe (Lenz, Reference Lenz1999; Gaitzsch, Reference Gaitzsch, Roymans and Derks2011; Jeneson, Reference Jeneson, Roymans and Derks2011) with a strong tradition of villa complexes (Roymans & Derks, Reference Roymans and Derks2011). This was complemented by several urban centres, including the colonia at Cologne as well as smaller towns such as Aachen, Zülpich, and Jülich. The region experienced significant dislocation during late antiquity, with several late third- and fourth-century abandonment episodes (Gechter & Kunow, Reference Gechter and Kunow1986; Lenz, Reference Lenz1999; Dodd, Reference Dodd2021), the early phases of this period being seen as representing a ‘crisis’ (see Fischer, Reference Fischer2012 for examples from the Gallic Empire). Most settlements north of the main Cologne-Bavay highway were abandoned (Heeren, Reference Heeren, Roymans, Derks and Hiddink2015, Reference Heeren, Roymans, Heeren and De Clercq2017; Dodd, Reference Dodd2021) and new defended sites appeared, both military and establishments seemingly related to earlier villa complexes. The smaller towns shrunk or were abandoned, with some transformed into fortified enclosures (at Jülich: Perse, Reference Perse, Bridger and Gilles1998; Aachen: Kyritz & Schaub, Reference Kyritz and Schaub2015; Zülpich: Gechter et al., Reference Gechter, Heimberg and Pahlen1979).
Figure 1. The study area, with site locations.
The appearance, use, and maintenance of defended enclosures, generally referred to as burgi, is a key element in this shifting pattern. Across both the lowland loess belt and the Eifel-Ardennes range, the number and scale of defended settlements increased during late antiquity. The development of these sites has been noted as far back as the mid-nineteenth century (e.g. Del Marmol, Reference Del Marmol1859) and associated with the retreat of populations to defended strongpoints during the fall of the Limes and the incursions of the Franks. This has long played an important role in our perceptions of defence, unrest, and ‘invasion’ in late antique northern Gaul.
Historically, the period under study covers the late third to the early fifth century, broadly equating with the late Roman Empire. The late third century saw a period of rapid and wide-ranging change (Esmonde Cleary, Reference Esmonde Cleary2013; see Millet, Reference Millet1990 for a wider view) with political fragmentation, especially in the west (Drinkwater, Reference Drinkwater1987). The most intense period in north-western continental Europe was between ad 250 and 270, under the so-called ‘Soldier-Emperors’ (Willems, Reference Willems1984: 273). Within our study area, there is evidence of episodes of significant depopulation, especially in the north and east (Heeren, Reference Heeren, Roymans, Derks and Hiddink2015, Reference Heeren, Roymans, Heeren and De Clercq2017; Dodd, Reference Dodd2021), with a lesser impact towards the south and west (Gechter & Kunow, Reference Gechter and Kunow1986; Lenz, Reference Lenz1999), whilst a breach of the Limes was postulated in the frontier zone (Mommsen, Reference Mommsen1894: 150–52; Van Es, Reference Van Es1981: 47–48; see Heeren, Reference Heeren2016 for a more recent critique). Rural settlements, especially villas, began to transform (Dodd, Reference Dodd2021) and there is widespread evidence for the demolition of Classical funerary monuments for spolia reuse, often in fortifications (Clemens, Reference Clemens and Boschung2009; see Perse, Reference Perse, Bridger and Gilles1998 for examples).
The Lower Rhine was stabilized and reorganized under the Tetrarchs ([Nixon et al., Reference Nixon, Rodgers and Mynors1994] Panegyrici Latini VIII, 5), who carved out a new provincial structure based around a senior official at Trier (Wightman, Reference Wightman1985: 202–03). This shift corresponds with an upswing in fortification: new defensive circuits are evident, with most cities fortifying a reduced settlement core, for example at Jülich and Heerlen (Butler, Reference Butler1959; Mertens, Reference Mertens1977; Johnson, Reference Johnson1983); some sites were abandoned completely, as at Voorburg (Forum Hadriani). The widespread development of new fortified sites was coupled with new military networks along the rivers and the Via Belgica (Brulet, Reference Brulet1990, Reference Brulet, Roymans, Heeren and De Clercq2017; Heeren, Reference Heeren, Diarte-Blasco and Christie2018).
From the late fourth century onwards, the region slowly drifted out of direct Roman control. Military engagement was increasingly transferred to allied groups of foederati (Claudianus [Platnauer, Reference Platnauer1922], De Bello Gothico: 419–29; Roymans, Reference Roymans, Roymans, Heeren and De Clercq2017: 66–67), whilst incursions from Germanic groups may have broken the Middle Rhine Limes in ad 405–406 (Kulikowski, Reference Kulikowski2000: 325–45). The early fifth century saw the region effectively leave the Roman orbit: surviving Romanized settlement patterns broke down and new Frankish groups repopulated the landscape (Dierkens & Périn, Reference Dierkens, Périn, Taayke, Looijenga, Harsema and Reinders2003; Roymans & Heeren, Reference Roymans, Heeren, Kars, van Oosten, Roxburgh and Verhoeven2015: 557–58).
This article focuses on one facet of the wider defensive pattern of this period: the proliferation of burgi, or burgus-like features across the loess belt of the Lower Rhine region in late antiquity (Figure 1). It provides an initial overview of the sites, their chronology, and morphology and puts forward ideas about their function within the defence scheme. As a first survey of these installations, it classifies and organizes sites identified as burgi by type and builds a picture of their development in the Rhineland from the late third century onwards. The data is presented holistically and the assessment of chronological, regional, and typological variations should be viewed as complementing work on the much better studied defended hilltop settlements (Gilles, Reference Gilles1985; Brulet, Reference Brulet, Steuer and Bierbrauer2008; Prien & Hilbich, Reference Prien and Hilbich2012).
Early Investigations
The study of defended settlements in north-western Europe was originally rooted in the historical sources. Antiquarian excavations at a range of sites were strongly influenced by these historical sources, which painted a dark picture of late third-century collapse (Aurelius Victor [Gruendel, Reference Gruendel1966], De Caesaribus; cf. Drinkwater, Reference Drinkwater1987) and late fourth- and early fifth-century unrest (in Zosimus’ Historia Nova, for example [Mendelhsson, Reference Mendelhsson1963]). This coloured our concept of late third- to fifth-century defended sites, defining site types and patterns with a distinct focus on sites identified as ‘military’ (e.g. Heidenburg-Hüchelhoven: Hagen, Reference Hagen1928; Brunhaut-Liberchies: Breuer, Reference Breuer1931; Heumensoord: Holwerda, Reference Holwerda1933) or hilltop refuges (e.g. Samson: Del Marmol, Reference Del Marmol1859; Nismes-Roche Trouée: Bequet, Reference Bequet1887–88). The excavation of these sites naturally influenced the wider narrative, with fortified sites, both military, such as the road forts of the Via Belgica, and the hilltop sites of the highland zone, reinforcing the story of an unstable, defended landscape.
Within this early framework, burgi were somewhat understudied. Early excavations at Heidenburg-Hüchelhoven and Brühl-Villenhaus assumed that these sites were part of the larger road surveillance network (Hagen, Reference Hagen1931) and there was little understanding of the role played by other burgus-like structures in the region. Moreover, the term began to be used more broadly to include widely differing sites, with ‘burgus’ being applied to stone towers or forts defended by ditches along the late Roman Limes and hinterland across Europe (cf. Von Petrikovits, Reference Von Petrikovits1971).
Recent Approaches
By the end of the Second World War, this initial phase of investigation had given way to more systematic analysis (e.g. Barfield, Reference Barfield and Barfield1968; Heimberg, Reference Heimberg1977; Mertens, Reference Mertens, Hanson and Keppie1980), establishing the long-term development of these sites, their relationship to earlier settlements (Brulet, Reference Brulet1974), and the transformation of defensive architecture more generally (Lander, Reference Lander, Hanson and Keppie1980; Reddé, Reference Reddé1995). The initial development of fortified sites, both in the loess belt and the Eifel range, was dated by coinage of the third-century Gallic Emperors and Carausius (Von Petrikovits, Reference Von Petrikovits1971; Johnson, Reference Johnson1983; Brulet, Reference Brulet1990). Studies of defended hilltop settlements (Gilles, Reference Gilles1985) established a base for future work, and excavations of Belgian refuges provided an important dataset for the study of these sites (see Brulet, Reference Brulet1978; Mertens & Remy, Reference Mertens and Remy1973 for examples).
Partly influenced by the broad concepts of ‘defence-in-depth’ developed in the 1970s (Luttwak, Reference Luttwak1976), scholars considered the defence of Germania Secunda and Gallia Belgica ‘solved’ and that new data would merely fill in the gaps in a pattern of defended highways and flexible defence in a partially depopulated landscape (Brulet, Reference Brulet, Otte and Willems1986, Reference Brulet1990; Van Ossel, Reference Van Ossel1995; Van Ossel & Ouzoulias, Reference Van Ossel and Ouzoulias2000: 143–45; Reddé et al., Reference Reddé, Brulet, Fellmann, Haalebos, von Schnurbein and Aupert2006; Deschieter, Reference Deschieter2016). This model has been challenged in recent years, with new data entering circulation. The defended hilltop settlements have undergone a great deal of further analysis, re-evaluation of older excavations (Hunold, Reference Hunold2011), and elaboration of new approaches (Böhme, Reference Böhme, Steuer and Bierbrauer2008; Brulet, Reference Brulet, Steuer and Bierbrauer2008; Prien & Hilbich, Reference Prien and Hilbich2012; Bayard & Fourdrin, Reference Bayard and Fourdrin2019). In the loess belt, systematic excavation from the late 1970s onwards in the Rhenish lignite mining area (Gaitzsch, Reference Gaitzsch, Roymans and Derks2011) has increased the number of known burgi exponentially (e.g. Gaitzsch & Haarich, Reference Gaitzsch and Haarich2012), whilst aerial photography from the 1960s onwards has identified further potential sites, especially in the Zülpich area (Scollar, Reference Scollar1963; Heimberg, Reference Heimberg1977; Krüger & Zantopp, Reference Krüger and Zantopp1992).
Despite this, there has been little holistic examination of burgi in the region's loess belt and their place in the landscape. Work has tended to focus on individual excavations or broad, thematically linked groups of sites (e.g. the road forts of the Via Belgica: Brulet et al., Reference Brulet, Mertens and Leva1995; Bazelmans et al., Reference Bazelmans, Bakels and Kocken2004; or fortified villa complexes: Van Ossel, Reference Van Ossel1992; Dodd, Reference Dodd2021). Modern approaches have, however, begun to be applied: typologies are being developed (Henrich, Reference Henrich2010, Reference Henrich and Henrich2015, Reference Henrich2017), with several projects underway. Some work has also targeted defensive architecture (Henrich, Reference Henrich and Henrich2015; Brulet, Reference Brulet, Bayard and Fourdrin2019) as well as wider issues of integration (Heimerl, Reference Heimerl2021: 117–29), making it possible to take further steps in analysis.
Defining the Burgus
Morphologically, the term burgus is difficult to define. Arguably a form of crisis architecture (cf. Driessen, Reference Driessen1995), what it represents is problematic. The word was in use from the late second century onwards and appears to apply to small installations or garrisons; its definition is limited (see Vegetius [Lang, Reference Lang1872], Epitoma Rei Militaris IV, 10 for an ancient example; Brulet, Reference Brulet, Reddé, Brulet, Fellmann, Haalebos, von Schnurbein and Aupert2006: 157) and the term seems to be Germanic in origin (Visy, Reference Visy2009). Burgi have generally been interpreted as evolving from early imperial timber watchtowers, with examples known as early as the first century ad (see Woolliscroft & Hoffmann, Reference Woolliscroft and Hoffmann2006 for Flavian Scotland). They were common along the Limes (e.g. Duisburg-Baerl-Dachsberg: Bechert & Willems, Reference Bechert and Willems1995: 49) and began appearing in greater number and variety from the third century ad onwards, with examples both in stone and wood. They became increasingly frequent in the hinterland, with the classic example of the Tongres-Bavay roadside forts (Brulet, Reference Brulet1990; Brulet et al., Reference Brulet, Mertens and Leva1995).
The term encompasses a range of sites generally viewed as ditched enclosures, sometimes with towers (Darvill, Reference Darvill2008: 68; cf. Van Ossel, Reference Van Ossel1992: 164; for related storage towers, see Hiddink, Reference Hiddink2022: 143–49). Archaeologically, burgi are defined primarily by the presence of V-shaped ‘defensive’ ditches, i.e. earthworks that would halt or delay attackers; some had interior features, others did not. In this study, this is used as a starting point for definition. The key issue is how ‘military’ these sites were. Scholarship has amalgamated unambiguous military installations—defined by stone architecture and multiple V-shaped ditches, coupled with explicit weapon finds, for example at Brunhaut-Liberchies (Breuer, Reference Breuer1931; Brulet et al., Reference Brulet, Mertens and Leva1995: 45–49)—with much simpler and more enigmatic establishments such as Weisweiler 32 (Schwellnus & Hermanns, Reference Schwellnus and Hermanns1980), resulting in a complex terminology, now deeply embedded in the research landscape. Although the differences are not necessarily clear, it is important to note that unambiguously military sites are morphologically much more elaborate. Within the study area, there is very little evidence of this, despite some sites (Heidenburg-Hüchelhoven and Brühl-Villenhaus) having long been identified as military installations connected to the defence of the road.
To address these issues, the burgi will be considered in a morphological framework (Table 1) that avoids a simple military vs civilian dichotomy and treats defence as a sliding scale. Burgi come in a variety of forms, all of which increase greatly in number in the loess plain of the Lower Rhine region during late antiquity (Figure 2).
Table 1. Categories of burgi found in the loess belt of the Lower Rhine region (where data are available).
Figure 2. Chronology of occupation of the burgi, where data are available. Sites are divided into certain, probable, and possible categories, based on finds and other evidence.
The Dataset
The dataset for this study consists of thirty-eight sites identified as burgi, spread across the region (Table 2). The sites have either been excavated and their excavation has revealed defensive-type enclosures (Heimberg, Reference Heimberg1977; Gaitzsch & Haarich, Reference Gaitzsch and Haarich2012), or they have been surveyed or photographed from the air (Scollar, Reference Scollar1963; Krüger & Zantopp, Reference Krüger and Zantopp1992; Frank & Wippern, Reference Frank and Wippern1999; Wessel & Wohlfarth, Reference Wessel and Wohlfarth2008; Song, Reference Song2018). All sites were dated in some way to between the third and fifth centuries ad. Although the inclusion of sites located by non-invasive techniques reduces our ability to date and characterize them, it provides a more complete geographical perspective on these sites; we should, however, be aware that some sites, such as Euskirchen-Borr (Scollar, Reference Scollar1963), have been misidentified in the past.
Table 2. Key information on burgus sites studied. The estimated date ranges in fifty-year blocks are recorded as certain, rather than possible or probable.
The burgi are not evenly distributed; they cluster for various reasons, some indicative of late Roman settlement patterns, others reflecting excavation bias (e.g. in the Rhenish lignite mining area). There seems to be no real discernible pattern: some sites are located along major routes, such as Heidenburg-Hüchelhoven (also known as Quadrath-Ichendorf; Brulet, Reference Brulet, Roymans, Heeren and De Clercq2017), whilst others are sited at some distance, such as Weilerwist-Groß-Vernich. Although some burgi, such as Euskirchen-Nettersheim (Hepa et al., Reference Hepa, Forrest and Ortisi2010), are strategically located at road junctions or river crossings, others are away from transport arteries, such as Hambach 133. Seven sites cluster in the Hambach and Weisweiler mining zones. Although this is probably the result of excavation bias owed to intense landscape-wide excavation since the 1970s, the presence of these sites also points towards the defence of the local glass industry, a high-value product in late antiquity (Gaitzsch et al., Reference Gaitzsch, Follmann-Schulz, Wedepohl, Hartmann and Tegtmeier2000).
The dates of the sites broadly span late antiquity (Figure 2). They are most abundant in the fourth century, particularly in the second half. There is clear evidence that these sites were used over the course of the late third to late fourth centuries, in line with long-held assumptions on the use of burgi as a whole (Brulet, Reference Brulet1990: 297–99; Brulet, Reference Brulet, Reddé, Brulet, Fellmann, Haalebos, von Schnurbein and Aupert2006: 156). Although there is some evidence of early third-century burgi, their primary use-phase begins in the late third century, when burgi structures are abundant, with an eighty-one per cent increase in their appearance and use. This early shift in the late third century is worth further exploration: some of the earlier sites, for example Euskirchen-Flamersheim, are known from survey and may belong to a transitional phase between earlier watchtowers and true burgi. At the other end of the date range, there is a distinct decline after ad 400. More than half the sites do not survive into the fifth century, with little evidence for continuity beyond the early decades. Very few sites show demonstrable evidence of occupation, with a few, such as Rheinbach-Flerzheim, represented by early fifth-century coinage. This suggests that the majority of surviving sites did not last long into the fifth century and raises questions as to whether there was a continuing need for such defences beyond this point. Although this does not necessarily explain who occupied late fourth- and fifth-century burgi, it is clear that these sites were no longer important by the mid-fifth century.
Fortification Morphology
Typological variation
Burgi have been traditionally difficult to categorize. Superficially, they are small, enclosed sites, defended by ditches in broadly rectangular, square, or oval configurations. Some of these structures have internal timber, stone, or earth features such as palisades, towers, or ramparts, correlating with Henrich's (Reference Henrich2017: 263–64) Types 1a, 1b, 2a, and 2b.
Recently, Henrich (Reference Henrich2010, Reference Henrich and Henrich2015, Reference Henrich2017), based in part on previous work (Van Ossel, Reference Van Ossel1992: 163–65), has attempted to define typological groups for fortified settlements. These site groups are focused on the development of fortifications on villa complexes, which does not necessarily demonstrate the proliferation of burgus features across the region. There are clear examples of ex novo foundations as well as rebuilding or reuse at villa complexes. This division between new foundations and reused buildings or sites forms the basis of the typological division employed here (Table 3). New foundations are defended enclosures established on virgin ground, without known antecedent sites, whilst reused buildings or sites represent burgi which use older settlements, often ruined, as locations for defensive structures.
Table 3. Typology of selected sites in the study area.
With most burgi reusing older settlements, and in some cases renovating and fortifying older buildings (e.g. at Jülich-Kirchberg-Auf dem Steinacker/Weisweiler 112: Päffgen, Reference Päffgen and Horn2000), these choices have influenced the morphological development of burgi in our study area. In some zones, such as the Hambacher Forst, the overwhelming majority of burgi are located on or near abandoned or reused villa complexes, sometimes incorporating them into the defences, for example at Hambach 158 and 139. Others, especially towards Zülpich, appear to be ex novo foundations with little direct evidence of earlier settlements at a range of sites, both excavated and surveyed. There is, however, limited evidence that these siting decisions were taken on the basis of assessable criteria. Instead, the picture is highly mixed (Figure 3).
Figure 3. Distribution of new and reused burgi. Numbers refer to the sites listed in Table 2.
Architectural morphology
The traditional approach to architectural morphology at burgi sites has been based on function rather than form. This created two general groups: ‘regular fortifications’, denoting sites with a military presence, and ‘rural fortifications’, implying emergency measures taken by local populations, devoid of central planning and organization, with a variety of terms used to elaborate on this (Brulet, Reference Brulet1990, Reference Brulet, Reddé, Brulet, Fellmann, Haalebos, von Schnurbein and Aupert2006). With respect to form, three major types of sites stand out within our dataset: enclosures without internal structures, enclosures with wooden internal structures, and enclosures with stone internal structures (Table 1). All three forms are found widely throughout the loess belt, both in larger ‘military’ installations and in sites traditionally identified as ‘civilian’ (Johnson, Reference Johnson1983: 138–41; Brulet, Reference Brulet, Roymans, Heeren and De Clercq2017: 49–51). These categories represent a development within the specific site-based lens of Henrich's (Reference Henrich2017: 263–64) Types 1a, 1b, 2a, and 2b.
The dataset is dominated by Type 2 features: enclosures with timber internal structures. Primarily these are palisades, i.e. lines of stakes forming a wooden barrier. It is also notable that where stone buildings are present, palisade structures often form part of the defence, for example at Rheinbach-Flerzheim and Mechernich-Satzvey (Song, Reference Song2018: 27, fig. 5). Timber defences are therefore represented at all levels in the dataset, both temporally and spatially. Early sites, especially those dating to the late third century and the Gallic Empire, are primarily constructed from wood, often in multiple phases. Two key sites stand out, Brühl-Villenhaus and the first phase at Heidenburg-Hüchelhoven. Both sites are roadside forts, primarily timber and earth constructions, and can be viewed within the milieu of the Via Belgica defence. Timber defences are also known at later sites, such as Köln-Widdersdorf Burgus 1, where internal structures and a possible palisade were recorded. This suggests a long-term continuity in design across the late third and fourth centuries, with building practice following designs that were widely disseminated across the region.
This homogeneity is evident in the construction of the ditches. Almost every site where evidence was available had near identical ditches. The similarity of their profiles is remarkable: in nearly all cases it is a deep V-shaped ditch, in some cases tapering to a flat base. There are no examples of more elaborate features such as ankle-breakers or lines of stakes, elements which are present in the military installations along the Via Belgica (Brulet et al., Reference Brulet, Mertens and Leva1995). The wide-ranging, but poorly dated, tower structures suggest similar designs being continually used in the loess belt, with the layout of Rheinbach-Flerzheim (Gechter, Reference Gechter1987) reflecting wider developments in defensive architecture. Notably, the best published stone tower, the fortification at Vettweiß-Froitzheim (Barfield, Reference Barfield and Barfield1968), has a more complex style, with multiple rebuilding phases.
These issues may relate to building methods and materials. Labour needed to be marshalled and resources acquired. The optimal construction times suggested by Shirley (Reference Shirley2000) for the first-century fortress at Inchtuthil in Scotland have been accepted as the norm for military construction; no parallel has been suggested for ‘civilian’ sites, but it is likely that the process was much slower. Whilst it may be possible to see emergency construction in poorly built, quickly erected defences, the startling similarity of the deep V-shaped ditches rather suggests that building a burgus was a planned operation. It does not appear to represent a short-term solution to perceived vulnerability but rather the result of longer-term planning.
In plan, key differences between the different types of defended enclosures begin to appear. Oval and sub-oval plans, such as at Rommerskirchen-Steinbrinkerhof, contrast with rectangular or square ditched enclosures, for example at Zülpich-Rövenich. Figure 4 illustrates these changes, grouping sites by categories (see Table 1). A limited number of sites exist without internal features (Type 1), for example at Weisweiler 32 (Schwellnus & Hermanns, Reference Schwellnus and Hermanns1980). These sites superficially represent more developed burgi, especially in the form of their ditches, but their lack of internal defensive architecture such as palisades suggests that these features would be indefensible; hence it may be worth considering other functional options. There appears to be a correlation between more oval ditch circuits and a lack of internal features, but this is difficult to substantiate; much more work on the publication and excavation of burgi is needed to test this.
Figure 4. Forms of enclosed defensive settlements.
Six sites within this dataset have multiple ditches, displaying evidence of elaborate defences (see Figure 4 for two examples). Multiple-ditched sites overwhelmingly have some form of central tower feature, either in wood or stone, representing greater expenditure of time and effort than more rudimentary Type 1 or Type 2 enclosures. Unlike burgi in the Rhineland-Palatinate, where there are significant examples of developed burgi with multiple ditches, stone defences, and considerable cost and time invested in their construction (see Krier, Reference Krier and Kremer2009; Henrich, Reference Henrich2017 for examples), equivalent sites in the loess belt are not frequent; they are clearly not representative of the wider defended landscape and generally something of an anomaly within the loess belt.
Geographically, the differences between burgi types are difficult to discern. There are very few clear patterns in the distribution of the burgi in the loess belt, with a wide spread of different types across the region. Nonetheless, one trend is discernible (Figure 5). By and large, stone-built burgi are located closer to the road network, with a distribution tied much more to transport links, arguably playing a role in road surveillance. Their morphological similarity to the roadside forts of the Via Belgica between Tongres and Bavay hints at a similar design aesthetic, although further work is needed to publish these sites fully (Brulet et al., Reference Brulet, Mertens and Leva1995 for an overview; Vilvorder & Verslype, Reference Vilvorder and Verslype2019 for Taviers). Further work is also needed to identify the nature of poorly understood sites, or sites revealed by aerial photography (as has been undertaken by Wessel & Wohlfarth, Reference Wessel and Wohlfarth2008); this would enhance our understanding of the role of these fortifications in the landscape and provide a better picture of their internal features.
Figure 5. Distribution of morphological variations in burgi sites. Numbers refer to the sites listed in Table 2.
Functionality and Burgi in the Loess Plain of the Lower Rhine Region
The data presented demonstrates the complexity of burgus sites in the loess plain of the Lower Rhine region. The broad typological differences between the types of burgus, and the challenges offered by their distribution, highlight some of the issues related to the spatial, typological, and chronological divisions in terms of the function of these sites. Without significant modern excavation, coupled with an intensive publication programme of previously excavated burgi, we cannot be certain about the relationships between these sites and the landscape.
The role of the military community in the construction, operation, and maintenance of burgi is an obvious target for enquiry. Several sites have been claimed as military, based on a perceived typology, including Vettweiß-Froitzheim and Rheinbach-Flerzheim, whilst others are thought not to be military (Reddé et al., Reference Reddé, Brulet, Fellmann, Haalebos, von Schnurbein and Aupert2006). This division appears arbitrary; although there is a clear lack of defensive architecture at Weisweiler 32, the ditch profiles are overwhelmingly similar and suggest a design developing from a similar source. Separation between military and non-military is not necessarily demonstrable in the material culture either. The finds from burgi sites are usually scarce (Heimberg, Reference Heimberg1977; Woolliscroft & Hoffmann, Reference Woolliscroft and Hoffmann2006 for finds issues at small Roman installations in Scotland) and, if present, ‘military’ finds tend to be somewhat ambiguous. There is some limited evidence that more developed burgi yielded overtly military objects (Barfield, Reference Barfield and Barfield1968: 92–110) but this is not universal since some less elaborate burgi have also yielded objects of a possibly military nature (Spiegel, Reference Spiegel2002: 727–29).
The end of the burgi provides another, contradictory, perspective on occupation. The majority are abandoned by, or in, the early fifth century, in line with villa occupation in the area (Dodd, Reference Dodd2021, 90-94). The ‘burgi abandonment horizon’ perhaps indicates some military function. New arrangements along the Lower Rhine, where groups of foederati (Roymans, Reference Roymans, Roymans, Heeren and De Clercq2017) had an impact on troop deployments in the loess belt, coincide with burgi being abandoned by the early fifth century. Regardless of the reasoning behind this, the key point is that the smaller fortifications of the loess belt could not remain viable without the overarching framework of Roman political-military control, especially given their large number in periods of relative stability such as the early fourth century. This naturally suggests that the burgi were used, in part, by members of the military community, or at the very least, by groups that relied on the Roman state for their social, political, or military existence.
Morphologically, it is clear that some burgi were effectively indefensible. The lack of obvious defensive architecture, beyond V-shaped ditches, at Weisweiler 32 and Rommerskirchen-Steinbrinkerhof suggests that these sites were difficult or impossible to defend. What they represent instead is more challenging. Options for their roles within the landscape may include corals for livestock, which might be tested through chemical analysis or landscape-scale examination. The location of some sites, along with several Type 2 and Type 3 burgi, in the Hambacher Forst suggests that they may have played a role in the glass industry. The late antique Hambach glass industries are well studied (Brüggler, Reference Brüggler2009; Rehren & Brüggler, Reference Rehren and Brüggler2020) and the defence of this key industry probably accounts for the cluster of burgi in this zone, either for processing, surveillance, or protection. Although Type 1 enclosures are undefended in a traditional sense, a V-shaped ditch may have acted as a sufficient deterrent to store glass securely before transport and, when coupled with local Type 3 defences, may have deterred would-be thieves. Much further work is needed to establish the relationship between the burgi and the glass production and processing sites in the lignite mining area.
Conclusion
The disparate evidence presented here suggests that burgi sites, bar a few probable military exceptions such as Brühl-Villenhaus and Heidenburg-Hüchelhoven, were multi-functional. Military and civil activities across the continental north-western Roman provinces were increasingly merging in the fourth century (Brulet, Reference Brulet, Bayard and Fourdrin2019) and it is appropriate to see independent ‘militia’ as well as military detachments occupying defended sites (see Brulet, Reference Brulet, Roymans, Heeren and De Clercq2017 for similar themes on urban defences). The location of the sites, both on and away from important transport arteries, supports this theory: some burgi were located in zones with little or no interest to the Roman state, whilst architectural similarities between the sites suggest that they are related to one another in terms of design and form. Furthermore, although burgi probably had elements of ‘private’ operation and cost (Henrich, Reference Henrich and Henrich2015 for an overview, Reference Henrich2017: 263–69), it is likely that designs spread quickly from military to civil contexts, perhaps along with military support or labour.
The data illustrate a web of multifunctional usage, military design schemes, different meanings, and occupation by different groups. The key obstacle to further understanding the burgi is simply a lack of publication. Most site records form part of the backlog held in store by the heritage authorities; since many sites are located in the lignite mining area, the polluter-pays principle does not apply (Gaitzsch, Reference Gaitzsch, Roymans and Derks2011). Publication beyond preliminary reports would help refine chronological and morphological resolution and allow us to better understand the sites’ construction, development, and longevity. Beyond that, integrating the burgi into the wider context will place them into the fortified landscape at a regional level and comparisons with other regions, for example Pannonia, are essential. The burgi in the loess plain of the Lower Rhine region are a key element in the defensive architecture of north-western Europe in Roman times. This article is an initial step in the study of these sites.
Acknowledgements
This work was supported by the UCLouvain Fonds Spéciaux de Recherche (F.S.R.) under Grant 419.052.272. I would like to thank the anonymous peer-reviewers and the editors for their comments on this paper.
