Michael Mackensen

The Roman fort of Gheriat el-Garbia is located 280 km south of Tripoli in the pre-desert, above an oasis, on the central route to the land of the Garamantes. Along with Bu Ngem and Ghadames, it is the largest of the three new vexilliation forts of the limes Tripolitanus, advanced far to the south under emperor Septimius Severus. By autumn AD 201 at the latest, its defences were finished by a detachment of legio III Augusta from Lambaesis. Since the English surveys (1953, 1981), the fort with its spectacularly well-preserved main gate has provided a valid basis for architectural reconstructions of 3rd century forts.
Within the framework of an interdisciplinary project of the Archaeology of the Roman Provinces at the University of Munich, funded by the Excellence Initiative, two campaigns each year were conducted in 2009 and 2010. The focus was laid on surveys and excavations within the fort, which had been overbuilt by a Berber village (gates, headquarters building, barracks), and in one of the sanctuaries. In addition to scientific investigation methods (geodesy, geophysics, archae-ometry, radiocarbon analyses), terrestrial 3D laser scanning was used. The ancient place-name Myd(---) is recorded on a pedestal for a statue. Surprisingly, in addition to the occupation of the fort until AD 275/280, there was a late antique reoccupation: from AD 390/400 until the middle of the 5th century, a unit of limitanei, the milites munifices, was garrisoned in the castra Madensia of the Notitia Dignitatum Occ. XXXI 30.
The first volume presents the history of research in the Tripolitanian frontier zone in north-western Libya and southern Tunisia from 1819 to 1996, as well as the location and topography of Gheriat el-Garbia. This is followed by the geodetic recording of the topography of the fort and temple plateau based on radar data from the German earth observation satellite TerraSar-X. The focus is on the archaeological surveys: numerous architectural elements from the interior area of the fort result in seven orders of columns, which suggest a sophisticated architectural decoration of the central buildings of the fort. The extensive material from the ceramic survey provides information on the supply of the garrison with African red slip ware and cooking wares as well as with wine, olive oil and fish preserves transported in various types of amphorae. In addition, bronze small finds, coins and radiocarbon analyses are presented; the latter form the basis for dating the resto-ration of the fort between AD 390 and 425.

The study presents a summary of the development over 500 years of the high-quality North African red slip ware, a fineware which was mainly used as tableware. The masterful handicraft of several large pottery production centres in one of Rome’s economically most important provinces, Africa Proconsularis, is particularly well reflected in the exceptional relief-decorated wares. The study focuses on the rare appliqué-decorated north Tunisian red slip ware of the late 2nd/early 3rd century and also the wide range of forms and decorations of the appliqué-, relief- and stamp-decorated red slip ware of the early 3rd to mid 6th century, produced in Sidi Marzouk Tounsi, the most important pottery production centre in central Tunisia which operated for at least 350 years. Production at this site included lamps and terracotta figurines in red slip ware as well, but in particular a wide variety of plain and decorated red slip ware vessels made not only for a regional market, but for Mediterranean long-distance trade.
Research is based on the exceptionally extensive collection accumulated by K. Wilhelm mainly between 1960 and 1995. The material is presented in a catalogue containing 138 vessels and 104 fragments, including several unique vessel forms and a series of unknown appliqués and stamp types. For comparison, some important, but partially unpublished vessels and fragments in several European, North American and North African museums and private collections were additionally taken into account.
For the first time, the decoration schemes and also the vessel forms of an appliqué-decorated pottery produced in a north-eastern and in another central Tunisian production centre throughout the late 2nd and 3rd century, are classified and clearly represented. The research also succeeded in defining the late phase of the so-called El Aouja ware and highlights the stylistic development leading to the appliqué-decorated red slip ware of the 4th century. Moreover, the analysis of the appliqué and relief decoration with pagan, allegoric and early Christian motifs emphasises an iconographic transformation over the centuries based on the visible change in the figural scenes. Of utmost importance is the fact that the production centre at Sidi Marzouk Tounsi clearly survived the Vandal rule of the provinces Africa and Byzacena (429-533/534). Alongside the late relief-decorated red slip wares, there is now also evidence for the production of mould-made, relief-decorated red slip ware platters and several specific forms with mainly Christian motifs, far exceeding the previously suggested end of production (c. 430/440) and suggesting an end around the middle of the 6th century.

The Roman site Submuntorium/Burghöfe is situated on a plateau of a promontory above the valleys of the rivers Danube and Lech, about 30 km north of the provincial capital of Augusta Vindeli-cum/Augsburg. It represents one of the most important military sites of the province of Raetia during the early and late Roman periods. From 2001–2007, four excavation seasons, including a geophysical and a field survey, were conducted on the so-called east plateau, a spur of the promontory, as well as on the tip of the spur by the University of Munich. Research focussed on the late Roman fortification and its period of occupation. However, it also offered new information on the early Roman military sites and the extension of the civil settlement (vicus) in the 1st c. AD. One of the most important results is the verification of a fortification of 1.5 ha, consisting of a curved system of two defensive ditches and possibly an accompanying turf or earth and timber rampart. They pre-date the well-known auxiliary fort and should belong to the early 40s of the 1st c. AD. After the abandonment of the fortification and the construction of the auxiliary fort the so-called eastern vicus developed in this area. It was destroyed by fire around AD 80.
The late Roman fort and its 1.5 m wide opus caementicium wall were documented in various areas on the tip of the spur. The fortification dates to the late 290s AD and belongs to a group of forts that were built around AD 300 as part of an official Tetrarchic building programme in Raetia. The Notitia Dignitatum mentions the deployment of a mounted unit and a detachment of the legio III Italica at Submuntorium. During the Constantinian period, an artisanal settlement of combined workshops and living quarters processed ferrous and non-ferrous metals. It was fortified around AD 350 with a defensive ditch and an earth or turf rampart. The small finds provide evidence of the period of occupation of the late Roman fortification and the adjacent vicus as well as of the military and civil population. Two gold coins of Leo I probably attest a military presence during the third quarter of the 5th c. AD. It seems likely that the security of the frontiers and military structures of the provinces Raetia prima et secunda, both part of the Italian diocese Italia annonaria, were main-tained until the death of Aetius in AD 454. Raetia may have consequently acted as a type of defensive glacis for the Italian prefecture until the abdication of the last Roman emperor of the western empire, Romulus Augustulus.

The publications of J.W. Hayes, J.W. Salomonson and the “Atlante delle forme ceramiche I” (EAA) remain the main sources of reference for any study of ARS ware of the 3rd century. The classification of fabrics, as instituted by A. Carandini, L. Saguì and S. Tortorella in 1981 on the basis of the material from excavations at Ostia, is not always straightforward in its application to North African sigillata of fabrics A (A1, A2, A1/2) and A/D, as well as C (C1, C2, C3) and C/E. As the centres of production of types A and A/D, presumed in the hinterland of Carthage, remain undiscovered, any judgement on the development of sigillata in the 3rd century is made more difficult. Yet it remained beyond doubt that the high-quality, plain or appliqué-decorated sigillata of fabric C1-C3 (including A/C, the so-called el-Aouja ware)  was produced in central Tunisian workshops.
Notable progress was achieved by the British-Tunisian survey, which located several new production centres in central Tunisia. Yet the repertoire of forms and decorations of the workshops at Djilma and Sidi Marzouk Tounsi, presented in 1990, is by no means representative, especially for the 3rd century. Furthermore, the outset of production was not defined, while the organisational structures as well as the reasons for the foundation of these pottery production centres remain unclear.
Different approaches with varying methodology, however, enable a more extensive appraisal and presentation of the range of forms and decorations of the 3rd century, which is to be attempted here. Utilizing archaeometric investigations such as quantitative chemical analyses of main and trace elements, it was possible to form reference groups specific to each pottery workshop. Earlier A/D forms as well as C1 and C2 forms are in evidence for Djilma, while no C1 forms with appliqués were found. For Sidi Marzouk Tounsi, on the other hand, there are only the characteristic open C1 and C2 forms, as well as few thin-walled closed el-Aouja forms (C1). Both chemical analysis and the use of C1 appliqués on the C3 goblet Löffler 591 have shown Sidi Marzouk Tounsi as the only production centre for appliqué decorated C1 and C2 sigillata (el-Aouja ware) in central Tunisia. This means that plain and appliqué decorated sigillata was produced in different workshops (officinae) at this site since the early 3rd century – and not since the mid 3rd century as previously believed. It is reasonable to propose continuous development throughout the 3rd century. Obviously a wide range of forms was produced here, while the manufacturing potential of the site enabled cutting edge development, perfection, modification and combination of various techniques of decoration – from feather-rouletting and appliqué decoration to incised and stamped decoration.

A central Tunisian red slip ware lamp from Sabratha, published by D.M. Bailey in 1994, should be classified as Type Salomonson I/Atlante I. Its discus decoration is a personification of Autumn, a standing female figure with a basket of fruit as a seasonal attribute and a Cupid (Eros) sitting on it. The prototype of the moulded decoration motif appears as an appliqué on a Hayes 171/El Aouja sigillata jug of C1 quality. The lamp, which probably dates from the third quarter of the 3rd century or the late 3rd century, was subjected to chemical analysis and comparison with recently published reference groups from central Tunisian pottery-making centres showed that it was made at the central Tunisian fine-ware potteries at Henchir el Guellal near Djilma. A/D and C1-C4 (= C/E and E!) sigillata as well as Type Atlante IV A, VI B, VII A1, VII A2 and VIII C1a lamps were produced there from about the second quarter of the 3rd century until the mid-5th century.

The most outstanding oasis fort of the limes Tripolitanus, Myd(---)/Gheriat el-Garbia, saw the depoloyment of various military units between AD 198/201 and 275/280. Archaeological research focussed on the fortifications, particularly the main and rear gates, the northern angle tower and two interval towers. The towers of the main gate have been reconstructed at an overall height of c. 12.5 m with two storeys and a crenellated platform. Of the headquarters building two of the rear rooms, including the regimental shrine (aedes), were preserved. A monumental inscription and corresponding architectural elements suggest the existence of a groma building. Within sight of the fort’s main gate lies a precinct with several temples on a plateau, where a Severan temple of the so-called ‘à cour’ type was excavated. Late Roman building activities within the fort as well as repairs were attributed to the deployment of a unit of limitanei between AD 360/380 + and 430/455. The official military nature of the late Roman occupation is confirmed by a Latin inscription and some ostraca written in a new variant of late Punic. The reoccupied fort with its restored central building complex might be identified with the so far unknown castra Madensia, where the milites Munifices were garrisoned (Notitia Dignitatum Occ. 31, 30).

Archaeological excavations at the oasis fort on the limes Tripolitanus, Gheriat el-Garbia, were conducted between 2009 and 2010 and provided for the first time evidence of late Roman building activities. The interior buildings of the Severan fort show a bad state of preservation. However, parts of the headquarters building (principia) could be observed including the aedes as well as an adjacent room (officium?). A large rectangular building was added to the northeastern side of the aedes and officium probably during the 2nd half of the 4th c. and may have incorporated the rooms of the Severan principia. Another late antique building of almost quadrangular shape, possibly dating to the 1st half of the 5th c., was located in the (former) praetentura of the Severan fort. Further late antique building activities within the fort comprise repairs and the blocking of the gates. It remains uncertain, however, if they are contemporary with the late antique buildings in the area of the principia and in the praetentura. Moreover, a stratigraphic sequence of the early 6th c. near the porta praetoria attests a late antique settlement, but no corresponding structural remains have been excavated.
The evidence of late Roman occupation within the fort (castra) may be attributed to the deployment of a unit of limitanei between A.D. 360/380 and 430/455 as well as to the presence of a Libyan chieftain and his tribe during the second half of the 5th c., continuing thereafter until A.D. 540+.

In der vorliegenden Rekonstruktion der porta praetoria wurde versucht unter Berücksichtigung aller verfügbaren Daten und Maße eine von möglichst wenigen Hypothesen belastete maßgenaue Zeichnung der Fassade des severischen Vexillationskastells Myd(---)/Gheriat el-Garbia (NW-Libyen) zu erstellen. Für das frühe 3. Jahrhundert n. Chr. steht somit eine vergleichsweise verlässliche Rekonstruktion für das Kastell einer legionaren Vexillation zur Verfügung. Abgesehen von regionalspezifischen Charakteristika wie den fünfeckigen Grundrissen der Tortürme und den für Tripolitanien typischen zinnenbekränzten Turmplattformen und unter Berücksichtigung der offenbar von Kastell zu Kastell unterschiedlichen Geschosshöhen in den Tortürmen und deren Erschließung vom Erdgeschoss (und nicht zusätzlich vom Wehrgang) aus, lassen sich viele Beobachtungen und Details bei der Rekonstruktion von römischen Kastellen des späten 2. und der ersten Hälfte des 3. Jahrhunderts n. Chr. in den Nordwestprovinzen in die Diskussion einbringen.

Das Kastell Gholaia/Bu Njem wurde im Jahr 201 im östlichen Tripolitanien von einer Vexillation der legio III Augusta aus Lambaesis erbaut. Als Befehlshaber des Kastells sind bis 238 legionare centuriones inschriftlich bezeugt, nach Auflösung der Legion und bis zur planmäßigen Räumung um 260 dagegen auxiliare decuriones als praepositus der vexillatio Golensis und eines numerus. Das Wohngebäude des Kommandeurs (praetorium) entspricht in etwa dem Kopfbau eines centurio im Legionslager Lambaesis, erreicht aber nur 60% der Fläche. Für die acht Mannschaftsunterkünfte in der praetentura läßt sich die Soll-Stärke aufgrund der Anzahl und der Innenfläche der contubernia mit 192 Mann kalkulieren; diese Teileinheiten wurden von Unteroffizieren (principales) befehligt, denen die Kopfbauten als Wohngebäude dienten. Die Soldaten wurden gleichmäßig den centuriae verschiedener Kohorten der Stammeinheit entnommen.

Aus einer spätantiken zweiteiligen Gipsmatrize aus Zentraltunesien konnten Tonstatuetten eines Reiters ausgeformt werden. Die Ausrüstung gibt Hinweise auf die Zeitstellung und Herkunft des behelmten Reiters, der wohl mit einem Kettenhemd und langen Hosen bekleidet ist und einen schmalen Leibgurt ohne Nebenriemen, aber mit großer Ringschnalle trägt. Der ungewöhnliche, eng gerippte konische Helm mit anhängendem eisernen Kettengeflecht unterscheidet sich durch die Form der Kalotte und die Helmbrünne deutlich von spätrömischen Kammhelmen ebenso wie von frühbyzantinischen Spangenhelmen des Typs Baldenheim und den Lamellenhelmen des Typs Niederstotzingen. Eine Datierung in das späte 5. Jh. und die erste Hälfte des 6. Jhs. liegt nahe, als Provenienz des dargestellten Reiters kommt der sasanidische Bereich jenseits des Euphrats in Frage. Es ist vorstellbar, daß das Aussehen und die Ausrüstung derjenigen östlicher (barbarischer) Reiter im byzantinischen Expeditionsheer entsprach, das unter Führung des Belisar im Jahr 533 das Vandalenreich eroberte.

Die Tonmatrize aus dem zentraltunesischen Töpfereizentrum Sidi Marzouk Tounsi, einem der Hauptproduktionsorte nordafrikanischer Sigillata und Lampen vom frühen 3. bis Mitte des 6. Jhs., zeigt die Abformung eines fünfeckigen kerbschnittverzierten Gürtelbeschlags mit sog. Randtieren. Entsprechende Gürtelgarnituren mit Kerbschnittdekor (Böhme Typ B) sind vorwiegend in den Donauprovinzen in militärischem Kontext verbreitet, in Nordafrika aber bislang unbekannt. Der in das späte 4. und das erste Drittel des 5. Jhs. datierbare bronzene Gürtelbeschlag ist als Zeugnis personengebundener Mobilität zu sehen. Ein Zusammenhang mit der Verlegung comitatensischer Einheiten aus dem mittleren Donauraum nach Nordafrika oder dem Einsatz von Expeditionsheeren wird in Erwägung gezogen.

The late Roman frontier in Raetia runs from the lower end of Lake Constance via Bregenz to Kempten and along the river Iller and the Upper Danube to the confluence of the river Inn. New excavations as well as a review of epigraphic, numismatic and archaeological evidence have yielded some astonishing results, as a result of which a reassessment of frontier development in this province during the last quarter of the 3rd and the early 4th century can be offered. The small forts at Bürgle, Burghöfe and probably Neuburg and Straubing, as well as the larger forts at Konstanz, Kellmünz and Martinsbühel belong to the same Tetrarchic building programme, which involved the construction of military fortifications of varying size, and differing interior layouts and disposition of barracks or other interior buildings according to need; the same applies to Eining, the only auxiliary fort on the Raetian Danube to be refortified and strengthened at this time, although reduced in size. At sites which provided ample coin-evidence, the construction dates can be shown to not predate AD 295/300. Characteristic for Valentinianic building activities are not only the watch-towers (burgi) but also the fort at Bregenz-Leutbühel next to a natural harbour on Lake Constance, and additional storage facilities (horrea) at military sites such as Eining and in the hinterland at Goldberg, Lorenzberg, Innsbruck-Wilten and Schaan etc. as well. Numismatic and in particular archaeological evidence of rouletted Argonne red slip ware and chip-carved and punch-decorated fittings of late Roman military belts suggest that military units must been stationed at several forts until AD 420/440 and beyond. Flavius Aëtius and his units of comitatenses campaigned in Raetia against the Juthungi in AD 430/431. Probably at Burghöfe and certainly at Passau-Niedernburg evidence is provided for the continued presence of small military garrisons after the middle and during the third quarter of the 5th century.