Navigation – Plan du site

AccueilNuméros en texte intégralSupplément 3Édifices chrétiens et paysage mil...The Evidence for Late Roman Garri...

Édifices chrétiens et paysage militaire

The Evidence for Late Roman Garrison Churches

Three Case Studies from the African Frontier Zone
Les preuves de l’existence d’églises de garnison tardo-romaines. Trois études de cas de la zone frontalière africaine
Alan Rushworth

Résumés

Cet article examine les preuves de la présence d’églises dans les forts romains tardifs au cours du ive et du début du ve siècle, en se concentrant sur l’une des zones frontalières les moins étudiées, le limes nord-africain. Trois sites en particulier – Ala Miliaria (Bénian), Drah-Souid est et Gouea – où des églises ont déjà été identifiées dans des enceintes supposées être des forts, sont soumis à un examen détaillé pour déterminer s’ils fonctionnaient comme des chapelles de garnison. Pour ce faire, il faut prendre en compte un certain nombre de questions, notamment l’évolution de la disposition interne des forts romains, la relation chronologique des églises avec les phases d’occupation militaire des sites, en particulier lorsque le fort coexistait avec une communauté urbaine en développement, et l’émergence de complexes monastiques fortifiés. Les preuves littéraires et documentaires relatives au rôle de la conversion chrétienne dans la gestion des relations avec les communautés tribales maures le long de la frontière au début du ve siècle sont également évaluées. L’article conclut que le trio de sites était en fait un assemblage plutôt hétéroclite, chacun reflétant un aspect différent du monde frontalier nord-africain, et il y a peu d’indications que l’adoption précoce de la foi chrétienne par l’ensemble de la population africaine se soit traduite par la mise en place d’églises ou de chapelles pour desservir les garnisons militaires sur la frontière.

Haut de page

Entrées d’index

Index géographique :

Afrique du Nord

Index chronologique :

Antiquité tardive, Bas-Empire romain
Haut de page

Notes de l’auteur

This paper is based on a review of the published literature relating to the three sites. It has not yet been possible to visit and examine the sites on the ground. The sites of Bénian and Drah Souid East may be viewed on Google Earth as they have not been built over so far. The remains at Drah Souid East are covered by wind-blown sand, though Guey’s 1938 excavation trenches may still be discerned, albeit less distinctly than on the aerial photograph published by Baradez (1949, p. 88, pl. A). At Bénian any standing remains noted in the late 19th century appear to have been levelled and the excavations of the basilica have been backfilled, but the site of the basilica plus its enclosing ‘redoubt’ and the general extent of the fort/walled town may be traced. The remains at Gouea (Kherbet el-Djouhala) have not been examined since the late 19th century. Although the present-day village of El Gouea is evident, it is not possible to pinpoint the location of ruins of Kherbet el-Djouhala, in part due to the relative imprecision in the way they are marked on Gsell’s Atlas archéologique de l’Algérie (1911) map sheet 14 (as no. 60). However no trace is apparent in the vicinity of the village and it is likely that they have been entirely removed. Gouea and Drah Souid East were considered previously in Rushworth 2024, though the discussion has been further refined and amplified here.

Texte intégral

  • 1 Sozomen, Church History, 1.8.
  • 2 Lee 2007, pp. 184-185.
  • 3 Petrikovits 1971, p. 203
  • 4 Housesteads: Crow 1995, pp. 96-97; Rushworth 2009, pp. 321-322; South Shields: Bidwell and Speak 19 (...)

1One ancient source famously suggested that Constantine provided each regiment in his army with a tent designed to serve as a mobile chapel, complete with an associated complement of priests and deacons to perform the requisite religious services.1 The details, specifically the reference to tented chapels, would imply that it was the units of Constantine’s mobile field army that were being referenced here, rather than the regiments stationed on the frontier in permanent garrison stations, where stone-built chapels or churches might have been expected instead. However, Sozomen was writing in the mid-5th century and the accuracy of his account is now generally treated with scepticism, particularly in view of the failure of Eusebius, a contemporary observer, to mention the existence of such military chaplains. Indeed it has been concluded that a majority of the soldiery were relatively indifferent to the religious changes of the 4th century CE, with Christianization of the army making only slow progress before the 5th century, perhaps echoing the rate of conversion in wider late Roman society during the 4th century.2 Evidence for churches or features associated with Christian worship is not found in frontier forts before the late 4th century, and even then at relatively few sites. In the western provinces, examples include Zurzach, Kaiseraugst, Boppard, and Koblenz on the Rhine,3 and various sites on Hadrian’s Wall,4 though some of the latter may actually post-date the collapse of formal Roman authority in Britain.

  • 5 Frend 1952, pp. 84-111.

2Nevertheless, the African provinces formed one region of the western empire where Christianity had put down much deeper roots,5 with the result that a greater proportion of the local recruits conscripted into the African garrison forces may have already derived from a Christian background. It is, therefore, conceivable that such precocity would in turn be reflected in architectural terms by the provision of purpose-built chapels or churches in the permanent garrison stations of the African frontier during the 4th and early 5th centuries, and at a somewhat earlier stage than was the case on the western empire’s other frontiers.

The Case Studies

3To investigate this hypothesis further, this study will examine three sites in the North African frontier zone where buildings of basilical form interpreted as churches have been identified within the walls of known or suspected Roman forts (fig. 1), and will attempt to determine whether these structures may indeed represent garrison churches or whether alternative explanations are more convincing.

4The first of these sites is the fort of Ala Miliaria in western Mauretania Caesariensis, which took its name from the milliary ala garrisoned there from the beginning of the 3rd century. The second is the fortified complex of Gouea in central Caesariensis, which has been identified with Caput Cilani, headquarters of the limes Caputcellensis, a late Roman frontier district mentioned in the Notitia Dignitatum. The final example is the fort of Drah Souid East, located beside the frontier barrier of the limes Gemellensis in southern Numidia and excavated by Julien Guey in 1938.

Figure 1. Map showing the location of Ala Miliaria (Bénian), Gouea and Drah Souid East in Mauretania Caesariensis and western Numidia

Figure 1. Map showing the location of Ala Miliaria (Bénian), Gouea and Drah Souid East in Mauretania Caesariensis and western Numidia

Background map: Consortium of Ancient World Mappers, CC BY 4.0

5While the three basilicas discussed here have all been interpreted as churches, in two cases – Drah Souid East and Bénian-Ala Miliaria – it has also been argued that they could represent late imperial principia, at least in their initial phase, which may have been co-axially aligned with colonnaded streets, an arrangement observed at military installations on Rome’s other frontiers.

Bénian-Ala Miliaria

  • 6 AE 1902, 4.
  • 7 Site plans and descriptions: La Blanchère 1883, pp. 66-68, pl. V and Gsell 1899, pp. 8-9, fig. 2; 1 (...)
  • 8 AE 1936, 64: dedication of monument – probably a statue – by Atius Cresc[ens, ob hono]rem II vi(rat (...)

6The fort of Ala Miliaria was constructed c. 201 at Bénian,6 beside a loop in the Oued Taria, to house the Ala I Nerviana Miliaria, and was situated on the ‘new forward line’ (nova praetentura) established by the emperor Septimius Severus in Mauretania Caesariensis. The fort enclosure was depicted on plans published by René de La Blanchère and Stéphane Gsell, the latter giving dimensions of 235 × 195 m (4.58 ha), typical of a milliary ala fort (fig. 2.1).7 This is still evident today on satellite imagery (fig. 2.2). A civil settlement grew up around the fort and by the turn of the 4th century this had achieved municipal status.8 A prominent Donatist Christian community was present by the 5th century.

Figure 2. Ala Miliaria. Plan (1) and satellite photo (2)

Figure 2. Ala Miliaria. Plan (1) and satellite photo (2)

Credit: A. Rushworth, after Gsell 1901, vol. 1, p. 88, fig. 26, CC BY-NC-SA 4.0, satellite photo: SNES / Airbus

The Basilica

  • 9 Gsell 1899 and 1901, vol. 2, pp. 175-179; plans of the basilica: Gsell 1899, p. 9, fig. 2 and 1901, (...)
  • 10 Gsell 1899, p. 48
  • 11 Gsell 1899, pp. 45-46.

7A Christian basilica was located at the east end of the site opposite the surviving west gate at the other end of the walled site (fig. 3). The church was excavated by M. Rouziès in 1899 and published by Gsell in Fouilles de Bénian,9 who dated its construction to between 434 and 439, by association with the row of seven Donatist martyrs’ tombs situated just beyond the chevet of the church, most of which were furnished with funerary epitaphs dated by the Mauretanian provincial era10. The basilica was set within a small enclosure, measuring 34 m east-west by 35.10 m north-south, which Gsell interpreted as the town’s pre-existing citadel into which the church was subsequently inserted.11 The main entrance to this enclosure was closed by sliding across a large stone disc set in a grooved recess on the north side. Possible traces of two semicircular bastions flanking this entrance and protecting two lateral posterns were also identified. A range of two or more rooms was attached to the north wall of the enclosure.

Figure 3. Ala Miliaria. Basilica and late fortified enclosure (1), with cross-section through the eastern end of the basilica (2)

Figure 3. Ala Miliaria. Basilica and late fortified enclosure (1), with cross-section through the eastern end of the basilica (2)

Credit: A. Rushworth, after Gsell 1899, p. 19, fig. 5, CC BY-NC-SA 4.0

The Variant Interpretations

  • 12 Lenoir 1986, pp. 659-662.

8More recently, however, the basilica and its function within the overall site has been the subject of considerable debate. A notable feature of the church was way the semicircular apse at its east end was raised some 1.50 m above the level of the three naves and was underlain by a crypt accessed from a small room to the north located beneath the sacristy. Maurice Lenoir argued that the apse and crypt beneath represented the raised apsidal tribunal or aedes and underlying strongroom belonging to a secondary phase of fort principia.12 Other elements that he considered might have formed part of the principia included what he vaguely termed the ‘basilical space’ to the west, plus the rooms exposed by Rouziès attached to the north wall of the citadel, which he interpreted as the offices and the armentaria traditionally associated with the principia. He postulated a matching row of rooms attached to the citadel’s south wall, an area not excavated in 1899. Lenoir dated this principia either to the third quarter of the 3rd century, or conceivably to the first quarter of the 4th century (contemporary with a possible refortification of the gates and curtain wall with round towers which he considered typologically characteristic of Constantinianic work). It would have replaced the original Severan principia of the fort which must have been more centrally located. The principia would have been transformed into a church after the departure of the garrison, perhaps at some point later in the 4th century, but at any rate some time before the addition of the Donatist martyr tombs from 422 onwards.

  • 13 Duval 1991, pp. 1087-1089.
  • 14 Gsell (1899, p. 46) mentions tombstones, cornice fragments and parts of olive presses.
  • 15 Duval 1991, pp. 1080-1081 and 1088, figs 4-6, pl. II; see Laporte 2015, p. 322.

9This interpretation was in turn called into question by Noël Duval in 1991,13 who pointed out that, although relatively rare in African church architecture, crypts are not unknown and apses elevated by 1 m or more are not at all rare. He conceded that the church might overlie an earlier principia in the same position, but was sceptical that the basilica occupied exactly the same situation and preserved structural components of that principia. Like Lenoir, he convincingly rejected Gsell’s argument that the church was erected immediately after the creation of the martyr tombs, and instead suggested that the church was probably built in the 4th century, following the removal or reduction of the garrison and the occupation of the enclosed area of the fort by the civil population. Further, it might be just one of several churches existing at Ala Miliaria and not necessarily the town’s cathedral. The seven tombs were added next to the east end of the pre-existing church from 422 onwards, when it became the focus of a Donatist martyr cult. Finally, the small enclosure can certainly be dated even later still. The incorporation of much reused material14 and use of stone discs set in grooved recesses to close gates are characteristically late features, typical of 6th- or 7th-century fortifications.15 The enclosure represents a small ‘redoubt’, added to protect the church, and falls into a pattern of small fortifications often erected in towns on local initiative to protect all manner of structures.

Ala Miliaria – Conclusions

10Duval’s qualifying and cautionary comments appear largely reasonable. Interestingly, Lenoir did not suggest that the principia might have been converted into a garrison chapel later in the 4th century nor did Duval suggest that the church might originally have served the garrison and only later been taken over by the urban community. Such transition from military to civil use would mirror the overall history of the site, but the uncertainty over exactly when the civil population replaced the military garrison as the principal occupants of the fort poses a major problem in determining how likely it is in the case of this basilica. It is noteworthy, however, that the urban community at Ala Miliaria had acquired the rank of municipium by 293/305, which suggests that it was already a town of some significance and prosperity by that stage. It is conceivable, therefore, that the garrison had already been withdrawn by the end of the 3rd century, perhaps redeployed to more turbulent areas in the east and centre of the province, although there are other examples where municipia appear to have coexisted alongside military forts for extended periods, such as Gemellae in Numidia.

  • 16 Gsell 1899, pp. 32 and 47-48; see Duval 1991, p. 1083.

11The significance of one detail noted by Gsell appears to have been somewhat overlooked. The chevet of the basilica (and therefore by implication the apse of Lenoir’s supposed principia) did not abut the inner face of the former fort’s curtain wall, as might be expected. Instead, the east end of the church sat directly on top of the projected course of the curtain wall, with a narrow gap left on either side between the church’s north and south walls and the surviving stretches of the curtain wall.16 This seems a rather odd arrangement, and surely inconceivable for the aedes of a fort headquarters building. Gsell explained this anomaly by suggesting that the row of Donatist tombs was initially erected just outside the standing curtain wall, which was then demolished to enable the construction of the church designed to facilitate the veneration of the martyrs. Thereafter, a small extension to the curtain wall, of reduced thickness, was built to enclose the tombs while allowing space to circulate around them. However, an alternative explanation presents itself. The church was located at the point where the fort’s east gate should have stood. Even if the steepness of the slope down to the Oued Taria rendered the use of that gate rather impractical, the construction of gateways at the top of unfeasibly steep slopes was far from unknown in Roman military castramentation of the 1st to 3rd centuries CE, as exemplified by the north gate at Housesteads on Hadrian’s Wall or the south-west and north-west gates at Gheriat el-Garbia in Tripolitania (fig. 4). Indeed it would be more difficult to find forts of this period when the full complement of four gates was not provided, whatever the topographic situation. It is quite likely, however, that the function of this gateway was fairly quickly reduced to that of a pedestrian-only passageway, like a postern, perhaps accompanied by partial blocking (as was the case with the north gate at Housesteads for example). Following the civil takeover of the fort, the gateway with its flanking towers was doubtless considered redundant and ultimately demolished, the site being repurposed with the construction of the basilica. The two passageways on the north and south sides of the church provided continuing access through the defences for minor traffic at this point, while the steepness of the slope may have been considered sufficient to compensate for any deficiencies in the defensive circuit. The tombs were added later, in the second quarter of the 5th century, when the site became associated with the Donatist martyr cult, and access through the defences at this point was finally closed off by an extension to the town’s enceinte enclosing the tombs. This scenario would strongly imply that the basilica neither represented a garrison church nor incorporated elements of a mid/late 3rd-century principia. Instead its initial construction was probably associated with the transformation of a frontier fort into an urban settlement at some point in the 4th century.

Figure 4. Plans of the Roman forts at Housesteads on Hadrian’s Wall and Gheriat el-Garbia (Tripolitanian pre-desert)

Figure 4. Plans of the Roman forts at Housesteads on Hadrian’s Wall and Gheriat el-Garbia (Tripolitanian pre-desert)

The gates are made largely redundant by the topography

Credit: A. Rushworth and T. Frain, The Archaeological Practice, CC BY-NC-SA 4.0

Gouea

  • 17 Gouea is the name of the locale, while Kherbet el Djouhala is probably that of the ancient remains. (...)
  • 18 Salama 1977, p. 583, map 3; see Hamdoune 2018, p. 198.
  • 19 Gsell 1901, vol. 2, pp. 198-200, fig. 122 and 1911, sheet 14, no. 60.
  • 20 Gui et al. 1992, vol. 1, pp. 48-49, no. 12 and vol. 2, p. 44, nos. 2-3; see Khatchatrian 1962, p. 2 (...)

12The second of the three sites, Gouea or Kherbet el-Djouhala,17 is located in rugged, hilly country about 75 km south of Algiers and 9 km west of the town of Berrouaghia, in what was the centre of Mauretania Caesariensis. This too has been identified as a military fortification.18 It is known from an 1883 unpublished note by Philippe, accompanied by a plan by Krolikowski de Poray. The latter was published, along with a description based on Philippe’s note, by Stephane Gsell in Monuments antiques de l’Afrique du Nord (fig. 5), while a further summary description appeared in his Atlas archéologique de l’Algérie, where the site is located on 1:200,000 map sheet 14 (Médéa).19 Gsell’s plan was subsequently reproduced in the Algerian volume of Basiliques chrétiennes d’Afrique du Nord, along with a schematic restoration of the church produced by Khatchatrian in 1962.20

Figure 5. Plan of Gouea by Krolikowski de Poray

Figure 5. Plan of Gouea by Krolikowski de Poray

Credit: A. Rushworth and T. Frain, The Archaeological Practice, after Gsell 1901, vol. 2, p. 199, fig. 122, CC BY-NC-SA 4.0

Site Description

13The plan shows an ecclesiastical basilica situated at the eastern end of an elongated and roughly rectangular, ridge-top complex, aligned north-east/south-west and consisting of two, long, rectangular courtyards leading towards the basilica. There is no scale on the plan, but Gsell notes the whole site covered 4,000 m² while the dimensions he gives for the church (23 m x 12.5 m) suggest that the whole complex was just over 200 m in length, but no more than c. 27.5-33.75 m wide for most of its length, narrowing markedly at its eastern apex where the church largely extends beyond the rest of the site. The basilica comprised a nave, with a flanking aisle on either side, and a vestibule to the south-west. In the chevet at the north-eastern end, the raised semicircular apse, which surmounted a vaulted crypt containing a stone sarcophagus, was encased within a solid quadrangular structure, flanked by a sacristy to the north and a small baptistery to the south. To the south-west of the basilica lay the inner and outer courtyards, perhaps with ranges of rooms or porticos along their north and south sides. A square chamber, which partially projected beyond the north wall of the outer courtyard, contained four rows of large amphora or jars and presumably served as a storehouse. On the north and west slopes below the main buildings were the remains of small huts, built of drystone walling, many of which were laid out in rows. Gsell suggested these huts served as the dwellings of the peasant cultivators of the domain.

Interpretation

  • 21 Gsell 1901, p. 200, n. 1.
  • 22 Monasticism in the African provinces is relatively poorly understood, particularly in rural as oppo (...)

14It is evident that this site does not conform to the standard principles of Roman military architecture, either the square/rectangular playing-card-shaped forts of the Principate, or those of the later empire which typically adopted squarish plans with projecting towers and barracks set against the rear of the curtain wall. With its ridge-top location, the site was probably defensible. The smaller, squarish rooms shown at the south and east corners of the western, outer courtyard and the north and west corners of the inner courtyard might conceivably represent the bases of towers, as could the slightly projecting storage chamber, and Gsell suggested that the large projecting chamber on the south side of the church might represent a defensive redoubt or lodging.21 Gsell labelled the site a castellum and seems to have viewed it as a large, fortified, agricultural domain or estate centre. However, in view of the prominent position occupied by the church at the very apex of the site, with the two courtyards apparently forming a progression towards it, the complex is more convincingly interpreted as a fortified monastery, or perhaps the seat of a rural bishopric, erected at some stage during late antiquity.22 The rows of drystone huts on the slopes below the main complex could represent the cells of monks living there. There is certainly little to suggest it was a military site.

  • 23 Salama 1977, p. 583, no. 12, p. 594, carte 3.

15Why then has Gouea been identified as a military site, in particular by Pierre Salama, who included it amongst the fortifications of the Antonine frontier in his seminal article, ‘Les déplacements successifs du limes en Maurétanie Césarienne – Essai de synthèse’, labelling it a ‘forteresse’?23

  • 24 Itineraria provinciarum Antonini Augusti, 36.3-39.1.

16Firstly, its location is significant because it lay beside or at any rate very close to the course of the 2nd-century military highway, itemized in the Antonine Itinerary,24 which ran from east to west through Mauretania Caesariensis, linking many of the province’s garrison forts. It is located between the known ancient sites of Thanaramusa Castra (near modern Berrouaghia), 11 km to the east, and Sufasar (Amoura), situated beside the Oued Chélif, c. 30 km to the west, both of which are itemized in that particular iter.

  • 25 Gsell 1911, sheet 14, no. 60; Salama 1951, p. 122 and map; see Courtois 1955, p. 82; Matthews 1976, (...)
  • 26 Notitia Dignitatum, In partibus Occidentis, 25.14, 25.32, 30.9 and 30.18.
  • 27 Gsell 1911, sheet 14, no. 60; Courtois 1955, p. 82; Matthews 1976, p. 168; Gui et al. 1992, vol. 1, (...)
  • 28 Rushworth 2018, pp. 727-728 and 2024, pp. 322-323.
  • 29 Gsell 1911, sheet 24, no. 8.
  • 30 CIL VIII, 20847; Salama 1953, p. 236.

17Secondly, Gouea has been identified with the place called Caput Cilani listed in another iter itemized in the Antonine Itinerary.25 This was clearly a military centre during the late Roman period, the headquarters of the praepositus limitis Caputcellensis, listed in the Notitia Dignitatum under the disposition of both the comes Africae and the dux et praeses Mauretaniae Caesariensis.26 Gsell was the first to note the possibility that Gouea should be equated with Caput Cilani, but, despite the doubts he and later commentators expressed,27 the identification has stuck. There are, nevertheless, strong reasons to doubt its validity. As argued elsewhere,28 based on analysis of the itinerary routes in the area, a much better candidate to represent Caput Cilani is Boghar,29 overlooking the Oued Chélif, some 35 km south-east of Sufasar. Positioned on the 3rd-century strategic highway – the nova praetentura – and the site of a Severan fort,30 Boghar-Caput Cilani is typical of many district limes headquarters – an auxiliary fort located at a communications node where several routes converged (fig. 6). The place name Caput Cilani must reflect Boghar’s situation as furthest upstream post on the Flumen Cilanus, probably the late antique name of the Oued Chélif.

Figure 6: Map showing the location of Gouea and Caput Cilani in relation to the routes in the Antonine Itinerary

Figure 6: Map showing the location of Gouea and Caput Cilani in relation to the routes in the Antonine Itinerary

After Salama 1953; background map: Consortium of Ancient World Mappers, CC BY 4.0

  • 31 CIL VIII, 9286.

18Thus Gouea was simply an otherwise unknown, late antique, defensible monastic settlement. The surrounding region was no stranger to conflict in the decades following the collapse of Roman authority in North Africa, as evinced by the death of a bishop at the walled town of El Hadjeb, near Mouzaia (Elephantaria?), 40 km to the north-north-west, occisus est in bello Maurorum in 495.31 The defensive aspects of Gouea’s topography and architecture may therefore have been justified.

  • 32 Ward-Perkins and Goodchild 1953, pp. 50-54; Sjöström 1993, pp. 280-281, no. 606; Barker et al. 1996 (...)
  • 33 Barker et al. 1996, vol. 1, p. 337; Sjöström (1993, p. 281) suggest that the series of rooms, two f (...)

19A possible parallel is the site of Chafagi Aamer in the Tripolitanian pre-desert, which features a tower-like fortified church situated on the crest of a spur overlooking the headwaters of the Wadi Sofeggin.32 The church is associated with a rectangular tower or gasr (‘castle’) – the Arabic term given to the fortified dwellings of two or more storeys, which are typical of the region – situated to the south, with the remains of still more ruinous buildings visible, extending along the crest of the promontory towards the main plateau. This site, too, has been plausibly interpreted as a monastic settlement.33

Drah Souid East

  • 34 Gsell 1911, sheet 48, no. 69.
  • 35 Notitia Dignitatum, In partibus Occidentis, 25.6 and 25.24.
  • 36 Gsell 1911, sheet 48, no. 65; Baradez 1949, pp. 96-98, 100-104 and 107-108; Trousset 1977, pp. 559- (...)

20The third case study is the easterly of two forts identified at Drah Souid, beside the Gemellae fossatum, the length of ditched and embanked linear barrier known locally as the Seguia Bent el-Krass.34 During the late empire this area formed the limes Gemellensis, listed in the Notitia Dignitatum35 under the command of a praepositus and presumably centred on Gemellae, where there was a 2nd/3rd-century auxiliary fort – the base of the Ala I Pannoniorum – surrounded by a walled town (El Kasbat).36

  • 37 Baradez 1949, pp. 94 and 99.
  • 38 Gsell 1911, sheet 48, nos. 70-72; Guey 1939.
  • 39 Guey 1939, pl. 1.
  • 40 Guey 1939, pp. 214-219.

21In addition to numerous watchtowers and gateways along the course of the fossatum, Jean Baradez identified a series of fortlets or small forts set just behind the barrier, which typically measured anywhere between 50 and 80 m² (0.25-0.64 ha).37 Baradez’ records augmented the information previously collated by Gsell’s atlas and by Guey.38 The remains were generally covered with blown sand so details were difficult to make out, but the majority may well date back to the creation of the fossatum under Hadrian. For the most part, the descriptions make no mention of projecting towers and, on the aerial photograph published by Guey, the western fort at Drah Souid certainly appears to have rounded corners typical of forts and fortlets of the 2nd and 3rd centuries.39 The clear exception was the fort of Bourada where projecting towers were clear on aerial photographs even prior to excavation and were confirmed by Guey’s 1938 excavation. It was shown to be an addition to the barrier, erected during the latter part of the reign of Constantine (324/327).40 For its time the excavation was well published and the fort is consequently the best understood late Roman fort in North Africa.

Guey’s Excavations of the Fort and Basilica

  • 41 Guey 1939, pp. 195-196.
  • 42 Guey 1939, pp. 205-206.

22Drah Souid East was the second site to be excavated as part of Guey’s campaign with trenches being used to chase walls to determine the site plan, as was the case with Bourada, and, as at Bourada, these trenches are still evident on Google Earth satellite imagery (fig. 7). It comprises a rectangular fort enclosure, measuring c. 80 × 60 m (0.48 ha), with rooms set against the inner face of the circuit wall. The long axis of the fort is aligned east-west, with the single gateway being positioned in the middle of the west wall. The whole is surrounded by a ditch, with a counterscarp bank lying beyond that. Its distinguishing feature, however, is a basilica, which Guey interpreted as a church, set roughly in the centre of the enclosure, its long axis aligned with that of the fort itself and its apse set at its east end. The apse was judged to have had a raised floor, probably of timber, with a crypt beneath. The site was constructed using plastered mud-brick walls, while the central nave of the basilica was covered by a tile roof, the apse by a shell of pottery tubes coated with mortar, and the side aisles by palm trunks and reeds. The pillars of the choir, the nave, the altar, and the arch of the apse were all furnished with stone bases while the sill walls of the choir were constructed of gypsum.41 Again occupation was shown to be relatively late, with all the dating evidence deriving from the rooms in the south-east corner of the enclosure.42 This comprised five small and illegible late imperial bronze coins (nummi) and five ‘Christian’ lamps found on the ancient ground level, one, decorated with a winged Victory standing holding a long cross in her right hand, being dateable to the early 5th century.

Figure 7. Drah Souid East. Satellite photo of the fortress

Figure 7. Drah Souid East. Satellite photo of the fortress

Satellite photo: Maxar Technologies

  • 43 Baradez 1949, p. 88, pl. A.
  • 44 Guey 1939, p. 195, fig. 5.
  • 45 Lenoir 2011, p. 210.
  • 46 See Fentress 1979, pp. 105-108; Daniels 1987, pp. 260 and 262-263, fig. 10.19; Rushworth 2015, p. 1 (...)
  • 47 See Mattingly 1995, p. 99, fig. 5.8; Mattingly et al. 2013, p. 171, fig. 1; Sheldrick 2021, p. 47, (...)

23It is clear from the aerial photograph published by Baradez,43 after excavation had taken place, that Guey’s trenches traced the plan of the basilica, the rooms at the south-east angle of the site and one directly east of the basilica, and the eastern and south-eastern stretches of the counterscarp bank. However, Guey only published a plan of the basilica (fig. 8),44 and this means some of the details of what were found are a little unclear. Indeed, the position of the trenches shown on the aerial photograph is the best guide to the overall plan of the fort. Lenoir argued that the photograph seemed to show the fort was furnished with corner towers,45 but, whether or not this interpretation is correct, it is very likely, based on the position of Guey’s excavation trenches that any such towers did not project like those of late Roman forts.46 The contrast with Bourada is particularly instructive, as it was clear there, even on the aerial photographs taken prior to excavation that the angle and intermediate towers projected. Secondly, the south-east angle of the enclosure at Drah Souid East appears to be a sharp right angle, rather than a rounded corner typical of so many earlier fortlets or small forts.47 This does not rule out the possibility that the enclosure at Drah Souid East was a military site; fortlets with right-angled corners do exist, but they are rarer than those with rounded corners, typical of the Principate, and those with projecting towers, characteristic of the late empire.

Figure 8. Drah Souid East. Plan of the basilica

Figure 8. Drah Souid East. Plan of the basilica

Credit: A. Rushworth, after Guey 1939, p. 195, fig. 5, CC BY-NC-SA 4.0

  • 48 Guey 1939, pp. 194-195.
  • 49 Rebuffat 1977-1979, pp. 260-262.
  • 50 Carrié 1974, pp. 824-839; Fellmann 1976, pp. 173-191.
  • 51 Lenoir 1986, pp. 656-659 and 2011, pp. 210-211.
  • 52 Rebuffat 1977-1979, pp. 259-260 and 262; Lenoir 2011, pp. 210-211.
  • 53 Duval 1991, p. 1086, n. 18; see Gui et al. 1992, vol. 1, pp. 142-144, no. 44.

24Guey interpreted the Drah Souid East as a fort containing a Christian basilica, but didn’t really address the function of the church within the complex as a whole, other than dismissing the possibility that the site was a fortified monastery.48 In the late 1970s, René Rebuffat reinterpreted the basilica as a form of late Roman headquarters building or principia – co-aligned with the principal axis of the fort –, arguing that the crypt beneath the apse was in fact the strong room beneath the chapel of standards.49 He drew on the previous work of Rudolf Fellmann and also Jean-Marie Carrié examining the developmental sequence of 3rd-century and Diocletianic headquarters buildings, including those at Thamusida, Vindonissa, Drobeta, Palmyra, Iatrus, Dionysias (Qasr Qarun), and Luxor.50 On this basis, he suggested that the fort and its basilical principia were most likely to have been constructed during the second half of the 3rd century or the beginning of the 4th century. Rebuffat’s interpretation of the basilica was followed by Lenoir.51 The possibility was alluded to by both authors that the site might have been occupied by a civilian population by the first third of the 5th century, though with no significant break given the location of the finds on the original floors, and this might have involved repurposing the principia as a church.52 Rebuffat’s reinterpretation of the basilica did not meet with universal acceptance, however, being questioned by Duval, who favoured Guey’s original interpretation.53

Drah Souid East – Discussion

25Thus there are three possible interpretations of the site at Drah Souid East:

  1. It might represent a small fort, perhaps erected at some point during the mid- to late 3rd century and furnished, from the start, with a central principia which took the form of a basilica, as suggested by Rebuffat and Lenoir.
  2. Alternatively, the basilica might represent a later insertion in the fort, and represent either a garrison chapel or a church erected by a religious community that had taken over the site shortly after the troops stationed there had been withdrawn.
  3. Finally, it would be quite compatible with the available evidence if the entire complex was newly erected at some point around 400 or slightly later, as a fortified Christian ecclesiastical site, perhaps a small Christian monastery, copying the external form of the military forts and fortlets which were such a feature of this district. Guey, himself, raised this possibility, but dismissed it, citing the proximity of Drah Souid to the frontier barrier and the analogous example of Gouea as a castellum with a Christian basilica.Guey 1939, pp. 194-195.

In determining which of these possible interpretations of Drah Souid East’s chronology and function is the most plausible, certain aspects need to be addressed, notably details of the site’s morphology, construction, dating evidence, and wider setting.

  • 54 Augustine, Epistulae, 46-47.

26A major objection to the first option is the excessive size of the basilica-principia, in relation to the area of the enclosed courtyard. Lenoir calculated that the basilica only occupied a fourteenth of the total area of the fort, but it still appears to take up more room than the comparable principia analysed by Fellmann and Rebuffat in other late Roman forts. A smaller structure set against the east side of the courtyard, opposite the gateway, as at Qasr Qarun-Dionysias, would surely have been more appropriate for such a small fort. The comparison with Bourada is instructive. There, the central building probably combined multiple functions – accommodation for the commanding officer, the administrative and ceremonial functions of a principia (in the form of the inner U-shaped structure), and a small bathhouse – representing a more efficient and practical use of the courtyard space. In contrast, the basilica at Drah Souid East appears much more specialized. One could envisage it as a structure designed for the performance of oath-swearing ceremonies, binding the leaders of transhumant tribal groups to observe the peace within the provinces, a process reported by the frontier zone landowner, Publicola, one of Augustine’s correspondents.54 However, a simple altar, erected in any formal setting, would surely have sufficed for such ceremonial functions, and, on balance, Guey’s original interpretation of the structure as a church seems more plausible.

27It is more difficult to determine whether the basilica was added to an earlier fort or was erected at the same time as the enclosing curtain wall and attached rooms, and formed part of an integrated complex. However, on the basis of the dateable material recovered from the rooms in the south-east corner of the site, there is no indication that Drah Souid East was occupied before the late 4th or early 5th century. There is nothing like the chronological range of coinage recovered from Bourada, for instance, where the clearly dateable examples extend from Claudius II (270+) to Gratian (pre-383), although allowance should be made for the far greater extent of excavation at Bourada, which, hence, yielded a total of 180 coins.

  • 55 Guey 1939, p. 194.
  • 56 Lenoir 2011, p. 210.
  • 57 For the defences of Centenarium Aqua Viva see Leschi 1941, pp. 170-171 and 1943, pp. 49 (plan), 52, (...)

28Turning to the site’s topography, Guey noted that the basilica dominated the fort, being situated on a point elevated some 2 m above the surrounding enceinte.55 This might imply that the site was deliberately selected to accommodate the basilica, which was intended to be a prominent feature visible over a wider area. Lenoir also argued that construction of the basilica and the surrounding enceinte, plus attached rooms, were so similar that the structure must be of a single date.56 This argument is perhaps rather less compelling, however, since such mud-brick construction was the most practicable option at this location, and was also used in the nearby fort of Bourada, for example. Finally, with its enclosing ditch and circuit wall, Drah Souid East was clearly defensible, but the apparent lack of the kind of projecting towers found at known late Roman forts, like Bourada or Centenarium Aqua Viva (Ain Namia),57 would be unusual for a military fort newly built sometime around 400.

29Overall, therefore, the evidence suggests that Drah Souid East may have had an ecclesiastical function focussed on the church at its centre. The quasi-military aspects of its form may simply represent a suitable adaptation to the arid steppe environment and the potential insecurity of its location. Indeed, the form of the enclosure is not too dissimilar from the larger courtyard gsur of the Tripolitanian pre-desert, which were rural fortified farms and estate centres rather than military fortlets. The ‘fort’ may have been designed to house a small religious community, perhaps monks seeking the solitude of the desert or priests charged with the duty of converting and ministering to transhumant tribal groups which regularly crossed the frontier barrier.

  • 58 Augustine, Epistulae, 199.12.46; discussed by Decret 1985.
  • 59 Augustine, Epistulae, 185.
  • 60 Augustine, Epistulae, 220.
  • 61 Wijnendaele 2015, pp. 32-40.

30A contemporary reference sheds light on events which may be relevant to the creation of this fortified frontier church. In a letter to Hesychius, the bishop of Salona, written in 420/421, Augustine, specifically mentions the conversion of certain of the barbarian tribes in Africa, admittedly very rare and few in number (rarissimi atque paucissimi), after their pacification and the replacement of their kings by prefects appointed by the imperial authorities a few years earlier.58 It is likely that this pacification and subsequent conversion occurred as a result of operations led by Boniface, the military commander active in the Numidian frontier zone during the previous decade. Augustine corresponded with Boniface in 417,59 when he was engaged in operations there at the head of a unit of Gothic foederati, and a few years later, perhaps c. 421, visited him at Tubunae, another limes headquarters, located 80 km to the north of Gemellae.60 The bishop would thus have been relatively well informed regarding the results of the campaigns of 416/417.61

  • 62 Guey 1939, p. 191.
  • 63 Gsell 1911, sheet 48, no. 72; Guey 1939, p. 191.
  • 64 Gsell 1911, sheet 48, nos. 70-72; Baradez 1949, pp. 94 and 99.
  • 65 See Gsell 1911, sheet 48.

31The geographical location of Drah Souid East within the wider Gemellae frontier zone may also be significant (see fig. 9). The church complex lay only 900 m to the east of the fort of Drah Souid West62 and some 3.5 km west of Bourada, while yet another fort, Esdeit,63 was situated only 2.2 km west of Drah Souid West. This was a remarkable concentration of military installations within the space of 6.6 km. By contrast, the fortlets/small forts identified by Gsell and Baradez along the remainder of the Seguia Bent-el-Krass, to the west of Esdeit, are typically spaced 7-8 km apart.64 The reason for such a density of sites is not difficult to divine. Even today the main road (N3) and the parallel railway line cross the line of the fossatum in this area before continuing down the Oued Rhir towards the major Saharan oases centres of Touggourt and Ouargla. These follow the course of a piste shown on early 20th-century 1:200,000 maps of the Biskra region.65 This piste doubtless represented an ancient caravan and transhumance route, forming the most direct and practicable passage northwards from several major Saharan oases. Monitoring this route and the people travelling along it was probably one of the major responsibilities of the garrisons of the Gemellae fossatum, and this surely explains the particular concentration of military posts in the Esdeit-Bourada zone. Just as this was the obvious zone to reinforce with a new fort in the early 4th century, probably in order to intensify control of frontier crossing, so it was also the best location to establish a church around a century later, perhaps with the deliberate aim of fostering religious contact with and possible conversion of transhumant groups regularly transiting through the frontier barrier.

Figure 9. Map of the Gemellae fossatum in south-west Numidia, showing the location of Drah Souid East and the various fortlets identified by Gsell, Guey and Baradez along its course

Figure 9. Map of the Gemellae fossatum in south-west Numidia, showing the location of Drah Souid East and the various fortlets identified by Gsell, Guey and Baradez along its course

Credit: A. Rushworth and T. Frain, The Archaeological Practice, CC BY-NC-SA 4.0

  • 66 Arce 2009, 2015, 2017, and 2018.
  • 67 Arce 2009 and 2018.
  • 68 Arce 2017.
  • 69 Codex Theodosianus, 7.15.1; see Rushworth 1992, pp. 27-59 and 197-229 and 2017, pp. 155-158.

32The crucial role that fortified ecclesiastical sites could play in the dynamics of a frontier zone under certain conditions has been illustrated by a series of studies by Ignacio Arce, relating to another desert frontier, the limes Arabicus.66 There, late Roman quadriburgia, such as Qasr el-Hallabat67 and Deir el-Kahf,68 mostly erected c. 300 as part of a Diocletianic reinforcement of the frontier zone and its garrison, were subsequently transformed to house monasteries (coenobia), along with praetoria – administrative and palace sites furnished with reception halls – under the aegis of federate Ghassanid phylarchs in the mid- to late 6th century. The Ghassanids employed more mobile forces, rather than relying on permanent garrisons installed in defensible forts. However they did require centres of logistical support, which a network of ideologically supportive Monophysite monasteries could provide, and spaces where the phylarchs could present themselves to their assembled followers in suitably impressive settings, both of which the old forts could accommodate. These changes in the form and use of the Arabian forts, of course, reflect the specific circumstances resulting from the political and military evolution of the eastern frontier zone, and cannot be directly transferred to the North African frontier. Nevertheless, there may be some parallels. Certain tribal communities (gentiles) in the North African frontier zone had long been integrated in the defence of the frontier districts (limites) and the maintenance of their defensive infrastructure, such as the fossata, as confirmed by an imperial edict of 409 preserved in the Theodosian Code,69 and it would be interesting to know if a site like Drah Souid East played any role in the transformation of such communities into the Moorish polities which emerged along the periphery of the Vandal kingdom later in the 5th century.

*

33In sum, our trio of sites, which initially appeared to have the potential to represent a group of late Roman garrison chapels, turn out to be a rather miscellaneous assemblage, each reflecting different aspects of the North African frontier world.

34The basilica at Ala Miliaria-Bénian was certainly constructed inside the walls of an important Roman fort, but by the time the church was erected the ala garrison may already have been withdrawn or at the very least reduced in size so that it only occupied part of a fort enclosure. By the 4th century, it is likely that the fort defences had been transformed into the town wall of the municipium of Ala Miliaria. It is conceivable that the church was erected on the site of a mid- to late 3rd-century principia, but if so those remains must lie buried beneath those of the basilica, which gives every appearance of being a fairly conventional Christian basilical church, contained within a small and doubtless much later, defensive compound, as argued by Duval. In other words, the creation of this basilica occurred in the context of the development of an urban community which, after initially surrounding the fort, later completely absorbed it.

35At Gouea, on the other hand, all the morphological evidence points to the site being a rural monastery and its suggested military function can be completely dismissed, even if the complex was clearly defensible, given its location on a steep ridge and the possible towers which punctuated its perimeter wall.

36At Drah Souid East, it has been suggested above that the site was probably newly created early in the 5th century, purpose-built as a protected ecclesiastical complex. A plausible historical context for this kind of development was outlined based on information in the letters of St Augustine. However, in this case, the evidence is not sufficient to rule out the alternative possibility that the basilica was a later addition to a pre-existing fortification, perhaps marking a shift from a military function to a Christian monastic one. The excavation undertaken there in February-March 1938 was simply not sufficiently extensive to be confident that the entire chronological timespan and various possible uses of the site were fully represented by the small assemblage of dateable material recovered. The latter derived only from a few trenches in the south-east corner, with the basilica itself apparently being devoid of diagnostic finds.

  • 70 Pringle 1981, p. 164.
  • 71 Lassus 1981, pp. 147-167; see Pringle 1981, p. 235.

37At any rate it seems unlikely that any of the three churches discussed here represent garrison chapels. Despite the widespread diffusion of Christianity in North Africa during the 3rd and 4th centuries, clear evidence for the construction of such chapels is not to be found in the region until the widespread fort construction programme initiated by the reoccupying East Roman army in the 6th century.70 The best-known example is the chapel inside the Justinianic fort at Thamugadi (Timgad), excavated during 1938-1956.71

38While further excavation would be required to clarify our understanding of Drah Souid East and confirm or refute the hypothesis presented above, it is clear, as Guey foresaw, that the sites in the section of the limes Gemellensis between Esdeit and Bourada have considerable potential to illuminate the development of the frontier. Further work in this sector could help to build a much greater understanding of long-term change in the frontier zone.

39Similarly, Bénian – which remains as yet undeveloped – would be an ideal site to investigate the evolution of the frontier zone in western Caesariensis during late antiquity, including the fate of the Roman fort and its garrison, the development of associated civic community and the subsequent history of the site during the succeeding Mauro-Roman period.

Haut de page

Bibliographie

Abbreviations

AE: Année épigraphique

CIL: Corpus Inscriptionum Latinarum

Works

Alberti M. and Birley A.R. 2024, ‘Sub-Roman and Post-Roman Christianity on Hadrian’s Wall: The Remarkable New Evidence from Vindolanda’, in H. Van Enkevort, M. Driessen, E. Graafstal, T. Hazenberg, T. Ivleva, and C. Van Driel-Murray (eds), Living and Dying on the Roman Frontier and Beyond. Proceedings of the 25th International Congress of Roman Frontier Studies 3, Leiden, Sidestone Press, Archeologische Berichten Nijmegen 11, pp. 281-288, DOI: https://doi.org/10.59641/1b090en.

Arce I. 2009, ‘Qasr Al-Hallabat (Jordan): Transformation of a limes Arabicus fort into a monastic and palatine complex’, in A. Morillo Cerdán, N. Hanel and E. Martín Hernández (eds), Limes XX. XX Congreso internacional de estudios sobre la frontera Romana [XXth International Congress of Roman Frontier Studies] León, España, septiembre, 2006, 3 vols, Madrid, Polifermo, Anejos de Gladius 13, pp. 155-180.

Arce I. 2015, ‘Severan Castra, Tetrarchic Quadriburgia, Justinian Coenobia, and Ghassanid Diyarat: Patterns of Transformation of Limes Arabicus Forts During Late Antiquity’, in R. Collins, M.F.A. Symonds and M. Weber (ed.), Roman Military Architecture on the Frontiers. Armies and Their Architecture in Late Antiquity, Oxford, Oxbow Books, pp. 98‑122.

Arce I. 2017, ‘Transformation Patterns of Roman Forts in the Limes Arabicus from Severan to Tetrarchic and Justinianic Periods. The Case of Deir el-Kahf (Jordan)’, in N. Hodgson, P.T. Bidwell, and J. Schachtmann (eds), Roman Frontier Studies 2009. Proceedings of the XXI International Congress of Roman Frontier Studies (Limes Congress) held at Newcastle upon Tyne in August 2009, Oxford, Archaeopress Roman Archaeology 25, pp. 121-130, DOI: https://doi.org/10.2307/j.ctv177tk1m.25 [restricted access, accessed in April 2025].

Arce I. 2018, ‘The Anastasius Edict Project: A preliminary report. Part 2 – The Archaeological, Architectural and Historical Contexts’, in C.S. Sommer and S. Matešić (eds), Limes XXIII. Proceedings of the 23rd International Limes Congress of Roman Frontier Studies, Ingolstadt 2015, 2 vols, Mainz, Nünnerich-Asmus, Beiträge zum Welterbe Limes 4, pp. 682-690.

Baradez J. 1949, Fossatum Africae. Recherches aériennes sur l’organisation des confins sahariens à l’époque romaine, Paris, Arts et métiers graphiques.

Barker G., Gilbertson D., Jones B. and Mattingly D. (eds) 1996, Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, 2 vols, Paris, UNESCO, London, Society for Libyan Studies.

Bidwell P.T. and Speak S. 1994, Excavations of South Shields Roman Fort. Volume I, Newcastle upon Tyne, Society of Antiquaries, Monograph series 4,.

Birley A.R. 2019, ‘Vindolanda’, in R. Collins and M. Symonds (eds), Hadrian’s Wall 2009-2019. A Summary of Excavation and Research Prepared for the Fourteenth Pilgrimage of Hadrian’s Wall, 20-28 July 2019, Kendal, Cumberland and Westmorland Antiquarian and Archaeological Society, pp. 164-177.

Carrié J.-M. 1974, ‘Les Castra Dionysiados et l’évolution de l’architecture militaire romaine tardive’, Mélanges de l’École française de Rome. Antiquité 86, pp. 819-850, DOI: https://doi.org/10.3406/mefr.1974.991.

Courtois C. 1955, Les Vandales et l’Afrique, Paris, Arts et métiers graphiques.

Crow J.G. 1995, Book of Housesteads, London, Batsford, English Heritage.

Daniels C.M. 1987, ‘The Frontiers: Africa’, in J.S. Wacher (ed.), The Roman World I, London-New York, Routledge, pp. 223-265.

Decret F. 1985, ‘Les gentes barbarae asservies par Rome dans l’Afrique du ve siècle. Remarques d’Augustin d’Hippone sur un point d’histoire sociale et religieuse à la veille de l’invasion vandale’, Bulletin archéologique du Comité des Travaux historiques. Nouvelle série 19, fasc. B, pp. 265-271.

Duval N. 1991, ‘La basilique de Bénian (Ala Miliaria) est-elle un remploi de principia militaires ? (Etudes d’archéologie chrétienne nord-africaine, XXII)’, in A. Mastino (ed.), L’Africa romana. Atti dell’VIII convegno di studio, Cagliari, 14-16 dicembre 1990, Sassari, Gallizzi, Pubblicazioni del Dipartimento di storia dell'Università di Sassari 18, pp. 1079-1089.

Fellmann R. 1976, ‘Le « Camp de Dioclétien » à Palmyre et l’architecture militaire du Bas-Empire’, Cahiers d’archéologie romande 5. Mélanges d’histoire ancienne et d’archéologie offerts à Paul Collart, Lausanne, pp. 173-191, DOI: https://doi.org/10.5169/seals-835538.

Fentress E.W.B. 1979, Numidia and the Roman Army. Social, Military and Economic Aspects of the Frontier Zone, Oxford, British archaeological reports, BAR. International series 53.

Frend W.H.C. 1952, The Donatist Church. A Movement of Protest in Roman North Africa, Oxford, Clarendon Press.

Gsell S. 1899, Fouilles de Bénian (Alamiliaria), Paris, E. Leroux, ARK: https://gallica.bnf.fr/ark:/12148/bpt6k54963737.

Gsell S. 1901, Monuments antiques de l’Algérie, 2 vols, Paris, A. Fontemoing, online: https://bibliotheque-numerique.inha.fr/idurl/1/8653 and https://bibliotheque-numerique.inha.fr/idurl/1/8654 [accessed in April 2025].

Gsell S. 1911, Atlas archéologique de l’Algérie. Édition spéciale des cartes au 200.000e du Service géographique de l’armée, Algiers, A. Jourdan, Paris, A. Fontemoing, online: https://bibliotheque-numerique.inha.fr/idurl/1/21113.

Guey J. 1939, ‘Note sur le limes romain de Numidie et le Sahara au ive siècle’, Mélanges d’archéologie et d’histoire 56, pp. 178-248, DOI: https://doi.org/10.3406/mefr.1939.7303.

Gui I., Duval N., Caillet J.‑P., Chevalier P. and Lorquin A. 1992, Basiliques chrétiennes d’Afrique du Nord. Inventaire et typologie I. Inventaire des monuments de l’Algérie 1, 2 vols, Paris, Institut d’études augustiniennes, Collection des études augustiniennes 129-130.

Hamdoune C. 2018, Ad fines Africae Romanae. Les mondes tribaux dans les provinces maurétaniennes, Bordeaux, Ausonius, Scripta Antiqua 111.

Khatchatrian A. 1962, Les baptistères paléochrétiens, Paris, EHESS, Collection chrétienne et byzantine.

La Blanchère M.R. de 1883, ‘Voyage d'étude dans une partie de la Maurétanie Césarienne’, Archives des missions scientifiques et littéraires. Troisième série 10, pp. 1-129, online: https://gallica.bnf.fr/ark:/12148/bpt6k56197658/f6.item.

Laporte J.-P. 1989, Rapidum. Le camp de la cohorte des Sardes en Maurétanie Césarienne, Sassari, Università degli studi di Sassari-Dipartimento di Storia, Pubblicazioni del Dipartimento di storia dell’Università di Sassari 12.

Laporte J.-P. 2015, ‘Quelques sites notables au sud-ouest de l’Aurès (Sadouri, Doucen, Sidi Khaled, etc.), Aouras 8, pp. 295-337.

Lassus J. 1981, La forteresse byzantine de Thamugadi. Fouilles à Timgad 1938‑1956, Paris, CNRS, Études d’antiquités africaines.

Lawless R.I. 1969, Mauretania Caesariensis. An Archaeological and Geographical Survey, 3 vols, PhD thesis, University of Durham (unpublished).

Lee A.D. 2007, War in Late Antiquity: A Social History, Malden (Mass.)-Oxford-Victoria, Blackwell, Ancient World at War.

Lenoir M. 1986, ‘Une martyre près des Principia. À propos du camp et de la basilique d’Ala Miliaria’, Mélanges de l’École française de Rome. Antiquité 98/2, pp. 643-664, DOI: https://doi.org/10.3406/mefr.1986.1517.

Lenoir M. 2011, Le camp romain. Proche-Orient et Afrique du Nord, Rome, École française de Rome, Bibliothèque des Écoles françaises d’Athènes et de Rome 345.

Lepelley C. 1981, Les cités de l’Afrique romaine au Bas-Empire. Tome II. Notices d’histoire municipale, Paris, Études Augustiniennes, Collection de études augustiniennes. Série Antiquité 81.

Leschi L. 1941, ‘Centenarium quod Aqua Viva appelatur…’, Comptes rendus de l’Académie des Inscriptions et Belles-Lettres 85/1, pp. 163-176, DOI: https://doi.org/10.3406/crai.1941.77409.

Leschi L. 1943, ‘Le « centenarium » d’Aqua Viva près de M’doukal (Commune mixte de Barika)’, Revue Africaine 87, pp. 5-22

Matthews J.F. 1976, ‘Mauretania in Ammianus and the Notitia’, in R. Goodburn and P. Bartholomew (eds), Aspects of the Notitia Dignitatum. Papers presented to the conference in Oxford, December 13 to 15, 1974, Oxford, British Archaeological Reports, BAR Supplementary Series 15, pp. 157-186.

Mattingly D.J. 1995, Tripolitania, London, B.T. Batsford.

Mattingly D., Sterry M. and Leitch V. 2013, ‘Fortified Farms and Defended Villages of Late Roman and Late Antique Africa’, Antiquité tardive 21. Mondes ruraux en Orient et en Occident II, pp. 167‑188, DOI: https://doi.org/10.1484/J.AT.5.101410 [restricted access, accessed in March 2025].

Pringle D. 1981, The Defence of Byzantine Africa from Justinian to the Arab Conquest, 2 vols, Oxford, British Archaeological Reports, BAR International Series 99.

Petrikovits H. von 1971, ‘Fortifications in the North-Western Roman Empire from the Third to the Fifth Centuries AD’, Journal of Roman Studies 61, pp. 178-218, DOI: https://doi.org/10.2307/300017 [restricted access, accessed in April 2025].

Rebuffat R. 1977-1979, ‘Sur le fort oriental de Drah-Souid’, Bulletin d’Archéologie Algérienne 7/1, pp. 259-270.

Rushworth A. 1992, Soldiers and Tribesmen. The Roman Army and Tribal Society in Late Imperial Africa, 2 vols, PhD thesis, University of Newcastle upon Tyne (unpublished).

Rushworth A. (ed.) 2009, Housesteads Roman Fort – The Grandest Station. Excavation and Survey at Housesteads, 1954-95, by Charles Daniels, John Gillam, James Crow, and Others, Swindon, English Heritage Archaeological Report, online: https://archaeologydataservice.ac.uk/archives/view/eh_monographs_2014/contents.cfm?mono=1089086.

Rushworth A. 2015, ‘Castra or centenaria? Interpreting the later forts of the North African frontier’, in R. Collins, M.F.A. Symonds and M. Weber (ed.), Roman Military Architecture on the Frontiers. Armies and Their Architecture in Late Antiquity, Oxford, Oxbow Books, pp. 123-139.

Rushworth A. 2017, ‘Soldiers or tribesmen: who guarded the frontiers of late Roman Africa?’, in N. Hodgson, P.T. Bidwell, and J. Schachtmann (eds), Roman Frontier Studies 2009. Proceedings of the XXI International Congress of Roman Frontier Studies (Limes Congress) held at Newcastle upon Tyne in August 2009, Oxford, Archaeopress Roman Archaeology 25, pp. 151-159, DOI: https://doi.org/10.2307/j.ctv177tk1m.29 [restricted access, accessed in April 2025].

Rushworth A. 2018, ‘Roads, itineraries and limites in the African frontier zone’, in C.S. Sommer and S. Matešić (eds), Limes XXIII. Proceedings of the 23rd International Limes Congress of Roman Frontier Studies, Ingolstadt 2015, 2 vols, Mainz, Nünnerich-Asmus, Beiträge zum Welterbe Limes 4, pp. 723-730.

Rushworth A. 2024, ‘Principia or monasteries? Two fortified basilicas in the North Africa frontier zone’, in H. Van Enkevort, M. Driessen, E. Graafstal, T. Hazenberg, T. Ivleva, and C. Van Driel-Murray (eds), Living and Dying on the Roman Frontier and Beyond. Proceedings of the 25th International Congress of Roman Frontier Studies 3, Leiden, Sidestone Press, Archeologische Berichten Nijmegen 11, pp. 319-327, full volume online: https://doi.org/10.59641/1b090en.

Salama P. 1951, Les voies romaines de l’Afrique du Nord, Algiers, Imprimerie officielle du gouvernement général de l'Algérie.

Salama P. 1953, ‘Nouveaux témoignages de l’œuvre des Sévères dans la Maurétanie Césarienne’, Libyca 1, pp. 231-261.

Salama P. 1955, ‘Nouveaux témoignages de l’œuvre des Sévères dans la Maurétanie Césarienne’, Libyca 3, pp. 329-365.

Salama P. 1977, ‘Les déplacements successifs du limes en Maurétanie Césarienne (Essai de synthèse)’, in J. Fitz (ed.), Limes. Akten des XI. Internationalen Limeskongresses. Szekesfehervar, 30. 8.-6. 9. 1976, Budapest, pp. 577-595.

Seston W. 1934, ‘Le monastère d’Aïn-Tamda et les origines de l’architecture monastique en Afrique du Nord’, Mélanges d’archéologie et d’histoire 51, p. 79-113, DOI: https://doi.org/10.3406/mefr.1934.7245.

Sheldrick N. 2021, Building the Countryside. Rural Architecture and Settlement in Tripolitania During the Roman and Late Antique Periods, London, Society for Libyan Studies, Open Access Monograph 2, DOI: https://doi.org/10.2307/j.ctv2m7c4wz.

Sjöström I. 1993, Tripolitania in Transition: Late Roman to Islamic Settlement. With a Catalogue of Sites, Aldershot, Avebury, Worldwide Archaeology Series 4.

Trousset P. 1977, ‘Le camp de Gemellae sur le limes de Numidie d’après les fouilles du colonel Baradez (1947-1950)’, in J. Fitz (ed.), Limes. Akten des XI. Internationalen Limeskongresses. Szekesfehervar, 30. 8.-6. 9. 1976, Budapest , pp. 559-576.

Ward-Perkins J.B. and Goodchild R.G. 1953, The Christian Antiquities of Tripolitania, Oxford, C. Batey, Archaeologia 95.

Wijnendaele J.W.P. 2015, The Last of the Romans. Bonifatius – Warlord and Comes Africae, London, Bloomsbury, Bloomsbury classical studies monographs.

Haut de page

Notes

1 Sozomen, Church History, 1.8.

2 Lee 2007, pp. 184-185.

3 Petrikovits 1971, p. 203

4 Housesteads: Crow 1995, pp. 96-97; Rushworth 2009, pp. 321-322; South Shields: Bidwell and Speak 1994, p. 103; Vindolanda: Birley 2019, pp. 166-167; Alberti and Birley 2024, pp. 281-288.

5 Frend 1952, pp. 84-111.

6 AE 1902, 4.

7 Site plans and descriptions: La Blanchère 1883, pp. 66-68, pl. V and Gsell 1899, pp. 8-9, fig. 2; 1901, vol. 1, pp. 87-88, fig. 26 and 1911, sheet 32, no. 93; Lawless 1969, vol. 2, pp. 104-114.

8 AE 1936, 64: dedication of monument – probably a statue – by Atius Cresc[ens, ob hono]rem II vi(ratus) A(lae) M(iliariae) during the Tetrarchy – 293-305 CE –; see Lepelley 1981, pp. 520-521.

9 Gsell 1899 and 1901, vol. 2, pp. 175-179; plans of the basilica: Gsell 1899, p. 9, fig. 2 and 1901, vol. 2, pp. 176, fig. 117.

10 Gsell 1899, p. 48

11 Gsell 1899, pp. 45-46.

12 Lenoir 1986, pp. 659-662.

13 Duval 1991, pp. 1087-1089.

14 Gsell (1899, p. 46) mentions tombstones, cornice fragments and parts of olive presses.

15 Duval 1991, pp. 1080-1081 and 1088, figs 4-6, pl. II; see Laporte 2015, p. 322.

16 Gsell 1899, pp. 32 and 47-48; see Duval 1991, p. 1083.

17 Gouea is the name of the locale, while Kherbet el Djouhala is probably that of the ancient remains. The 1:200,000 map sheet in Gsell’s atlas (1911) marks El Goueah, which is also the name of the current village.

18 Salama 1977, p. 583, map 3; see Hamdoune 2018, p. 198.

19 Gsell 1901, vol. 2, pp. 198-200, fig. 122 and 1911, sheet 14, no. 60.

20 Gui et al. 1992, vol. 1, pp. 48-49, no. 12 and vol. 2, p. 44, nos. 2-3; see Khatchatrian 1962, p. 28, fig. 216.

21 Gsell 1901, p. 200, n. 1.

22 Monasticism in the African provinces is relatively poorly understood, particularly in rural as opposed to urban settings. In the wider region of Gouea, Ain Tamda (Gsell 1911, sheet 14, no. 83), 12 km west of Rapidum, has been interpreted as a rural monastery (Seston 1934, pp. 86-113). It comprised ranges of rooms set around a rectangular courtyard, with a basilica on its west side, all enclosed by a stout wall. However the monastic interpretation was questioned by Frend (1952, p. 45, n. 3), who suggested it was a fortified communal storehouse plus church, citing comparable sites on the high plains of Numidia. Lawless (1969, vol. 2, pp. 203-205, site no. 84), Laporte (1989, pp. 157-161) and Gui et al. (1992, vol. 1, pp. 49-52, no. 13) were also doubtful. Chafagi Aamer, in the Tripolitanian pre-desert, may provide a better parallel for Gouea (see below).

23 Salama 1977, p. 583, no. 12, p. 594, carte 3.

24 Itineraria provinciarum Antonini Augusti, 36.3-39.1.

25 Gsell 1911, sheet 14, no. 60; Salama 1951, p. 122 and map; see Courtois 1955, p. 82; Matthews 1976, p. 168 and map p. 187; Hamdoune 2018, pp. 198 and 381; Itineraria provinciarum Antonini Augusti, 31.2.

26 Notitia Dignitatum, In partibus Occidentis, 25.14, 25.32, 30.9 and 30.18.

27 Gsell 1911, sheet 14, no. 60; Courtois 1955, p. 82; Matthews 1976, p. 168; Gui et al. 1992, vol. 1, p. 48.

28 Rushworth 2018, pp. 727-728 and 2024, pp. 322-323.

29 Gsell 1911, sheet 24, no. 8.

30 CIL VIII, 20847; Salama 1953, p. 236.

31 CIL VIII, 9286.

32 Ward-Perkins and Goodchild 1953, pp. 50-54; Sjöström 1993, pp. 280-281, no. 606; Barker et al. 1996, vol. 2, pp. 289-290, Sf110.

33 Barker et al. 1996, vol. 1, p. 337; Sjöström (1993, p. 281) suggest that the series of rooms, two floors high, attached to the north side of the church, were monks’ cells.

34 Gsell 1911, sheet 48, no. 69.

35 Notitia Dignitatum, In partibus Occidentis, 25.6 and 25.24.

36 Gsell 1911, sheet 48, no. 65; Baradez 1949, pp. 96-98, 100-104 and 107-108; Trousset 1977, pp. 559-564.

37 Baradez 1949, pp. 94 and 99.

38 Gsell 1911, sheet 48, nos. 70-72; Guey 1939.

39 Guey 1939, pl. 1.

40 Guey 1939, pp. 214-219.

41 Guey 1939, pp. 195-196.

42 Guey 1939, pp. 205-206.

43 Baradez 1949, p. 88, pl. A.

44 Guey 1939, p. 195, fig. 5.

45 Lenoir 2011, p. 210.

46 See Fentress 1979, pp. 105-108; Daniels 1987, pp. 260 and 262-263, fig. 10.19; Rushworth 2015, p. 125, fig. 10.2 and p. 127.

47 See Mattingly 1995, p. 99, fig. 5.8; Mattingly et al. 2013, p. 171, fig. 1; Sheldrick 2021, p. 47, fig. 4.7, p. 49, fig. 4.8, and p. 52, fig. 4.10; for a representative range of examples from Tripolitania. The suggested fortlets/outposts of El Medina Ragda, Ain Wif 2/Thenadassa and Henchir Krannfir have right-angled corners.

48 Guey 1939, pp. 194-195.

49 Rebuffat 1977-1979, pp. 260-262.

50 Carrié 1974, pp. 824-839; Fellmann 1976, pp. 173-191.

51 Lenoir 1986, pp. 656-659 and 2011, pp. 210-211.

52 Rebuffat 1977-1979, pp. 259-260 and 262; Lenoir 2011, pp. 210-211.

53 Duval 1991, p. 1086, n. 18; see Gui et al. 1992, vol. 1, pp. 142-144, no. 44.

54 Augustine, Epistulae, 46-47.

55 Guey 1939, p. 194.

56 Lenoir 2011, p. 210.

57 For the defences of Centenarium Aqua Viva see Leschi 1941, pp. 170-171 and 1943, pp. 49 (plan), 52, and 54 (aerial photograph).

58 Augustine, Epistulae, 199.12.46; discussed by Decret 1985.

59 Augustine, Epistulae, 185.

60 Augustine, Epistulae, 220.

61 Wijnendaele 2015, pp. 32-40.

62 Guey 1939, p. 191.

63 Gsell 1911, sheet 48, no. 72; Guey 1939, p. 191.

64 Gsell 1911, sheet 48, nos. 70-72; Baradez 1949, pp. 94 and 99.

65 See Gsell 1911, sheet 48.

66 Arce 2009, 2015, 2017, and 2018.

67 Arce 2009 and 2018.

68 Arce 2017.

69 Codex Theodosianus, 7.15.1; see Rushworth 1992, pp. 27-59 and 197-229 and 2017, pp. 155-158.

70 Pringle 1981, p. 164.

71 Lassus 1981, pp. 147-167; see Pringle 1981, p. 235.

Haut de page

Table des illustrations

Titre Figure 1. Map showing the location of Ala Miliaria (Bénian), Gouea and Drah Souid East in Mauretania Caesariensis and western Numidia
Crédits Background map: Consortium of Ancient World Mappers, CC BY 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-1.jpg
Fichier image/jpeg, 556k
Titre Figure 2. Ala Miliaria. Plan (1) and satellite photo (2)
Crédits Credit: A. Rushworth, after Gsell 1901, vol. 1, p. 88, fig. 26, CC BY-NC-SA 4.0, satellite photo: SNES / Airbus
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-2.jpg
Fichier image/jpeg, 569k
Titre Figure 3. Ala Miliaria. Basilica and late fortified enclosure (1), with cross-section through the eastern end of the basilica (2)
Crédits Credit: A. Rushworth, after Gsell 1899, p. 19, fig. 5, CC BY-NC-SA 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-3.jpg
Fichier image/jpeg, 111k
Titre Figure 4. Plans of the Roman forts at Housesteads on Hadrian’s Wall and Gheriat el-Garbia (Tripolitanian pre-desert)
Légende The gates are made largely redundant by the topography
Crédits Credit: A. Rushworth and T. Frain, The Archaeological Practice, CC BY-NC-SA 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-4.jpg
Fichier image/jpeg, 183k
Titre Figure 5. Plan of Gouea by Krolikowski de Poray
Crédits Credit: A. Rushworth and T. Frain, The Archaeological Practice, after Gsell 1901, vol. 2, p. 199, fig. 122, CC BY-NC-SA 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-5.jpg
Fichier image/jpeg, 402k
Titre Figure 6: Map showing the location of Gouea and Caput Cilani in relation to the routes in the Antonine Itinerary
Crédits After Salama 1953; background map: Consortium of Ancient World Mappers, CC BY 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-6.jpg
Fichier image/jpeg, 573k
Titre Figure 7. Drah Souid East. Satellite photo of the fortress
Crédits Satellite photo: Maxar Technologies
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-7.jpg
Fichier image/jpeg, 820k
Titre Figure 8. Drah Souid East. Plan of the basilica
Crédits Credit: A. Rushworth, after Guey 1939, p. 195, fig. 5, CC BY-NC-SA 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-8.jpg
Fichier image/jpeg, 66k
Titre Figure 9. Map of the Gemellae fossatum in south-west Numidia, showing the location of Drah Souid East and the various fortlets identified by Gsell, Guey and Baradez along its course
Crédits Credit: A. Rushworth and T. Frain, The Archaeological Practice, CC BY-NC-SA 4.0
URL http://journals.openedition.org/frontieres/docannexe/image/3975/img-9.jpg
Fichier image/jpeg, 110k
Haut de page

Pour citer cet article

Référence électronique

Alan Rushworth, « The Evidence for Late Roman Garrison Churches »Frontière·s [En ligne], Supplément 3 | 2025, mis en ligne le 17 juillet 2025, consulté le 16 mai 2026. URL : http://journals.openedition.org/frontieres/3975 ; DOI : https://doi.org/10.4000/14dyh

Haut de page

Auteur

Alan Rushworth

Director of the Archaeological Practice Ltd, Associate Fellow in Newcastle University

Haut de page

Droits d’auteur

CC-BY-NC-SA-4.0

Le texte seul est utilisable sous licence CC BY-NC-SA 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search