difficult to conceive at any period, let alone
one before the development of long-range
artillery in the latter half of the nineteenth century . *■**
In practice, few fortifications built earlier than then, other than continuous physical barriers such as walls (fossata orclausurae), could have effectively blocked the path of an invading force, ' and even linear defensive works such as Hadrian' s Wall in
Britain or the Long Wall in Thrace were probably intended to serve equally as barriers and as
springing-off points, from which the defending garrisons could counter an impending attack before it ever reached the wall. 18 To speak therefore of the guiding principle followed in the siting of forts and fortresses in Byzantine Africa as 'le barrage des defiles, l'insolement des
massifs montagneux au moyens d'une chaine serree d'ouvrages fortifies', I9 a phrase inspired in fact by the poetic language of Corippus, 20 would seem most III. 6
implausible, unless it were meant to be 96
understood that it was not the fortifications themselves that formed the 'barrage' but the troops that were based in them, arrayed to meet the attackers in the open. Yet Diehl
maintains that such would have been contrary to normal Byzantine practice: ' Surtout on
comptait que 1' armee byzantine, peu
nombreuse et peu solide, rendrait de meilleurs et plus utiles services, ainsi repartie en petites garnisons dans une multitude de forteresses, qu'aventuree tout entiere sur un meme point en une bataille rangee'.21 Such a view is difficult to reconcile with what is known of the military history of the period.
One further consideration which would also make nonsense of Diehl's frontier system was that, as he himslef acknowledges, the Moors who were raiding Romano-Byzantine towns and estates in the sixth century came not only from the desert regions of Tripolitania, southern
Tunisia, the regions south of the Aures and the Hodna, but also from the Aures itself, the
Nementcha, the Tunisian Dorsal, Cap Bon, the Kroumirie and the mountains of the Edough and the Kabylies, to name but the more obvious
regions. 22 Courtois has shown that the
weakness of the Vandal regime had permitted the formation of large chiefdoms in these
regions in the fifth and sixth centuries,
comprising federations of smaller tribes. 23 The problem of defence in the sixth century was
therefore not only military, but also political;
and it was not confined to the southern limits of Roman dominion, but concerned the whole
prefecture. Because the political map of
Byzantine Africa had the appearance of a jigsaw puzzle with differently coloured pieces
representing the zones of Roman and Moorish settlement, the military effectives that were
available to protect the inhabitants of the former areas from those of the latter could not be
deployed to face any one particular quarter, like those along the Danube and, to a certain extent, in the Euphrates region; for the threat came from all quarters.
How then did the system of garrisons and fortifications develop in Byzantine Africa and how was it articulated ? It was Justinian's intention to re-occupy Roman Africa up to the boundaries (fines) that had existed before the Vandal and Moorish incursions 24 and to
garrison the limites with units of comitatenses and limitanei, comprising both infantry and
cavalry. 25 The word limes, however, means in this context not simply a fortified line, nor even a frontier zone distinct from the zone inhabited by civilian Africans, because, for the reasons stated, there was no distinction between civil and and military zones in Byzantine Africa. The references to the dux limitis Tripolitaniae
p_roumcia.26 ^d t0 the duces qui ordinandi sunt per Africanos limites 27 suggest that in sixth-century Africa, limes had become synonymous for all practical purposes with ducatus or
prouincia, the military circumscription of a dux.
2° There was therefore no geographical or administrative difference between fortifications of the limes and those in areas supposedly
protected by the limes, because all fortifications except perhaps those in Carthago Proconsularis belonged to limites just as they belonged to prouinciae.
29 The security of a province threatened by
Moorish attack depended less on the strength of its fortifications than on the ability of its garrison to counter the Moors in the field. The response to local disturbances was at first made locally;
the dux was to intervene using limitanean troops. ^0 If he was unable
III.
6 to deal effectively with the probelm, however, 97 comitatenses, either belonging to the standing garrison of the province or detached from the field-army, would have to be used. If even these forces were unequal to the task, the field-commander, who later in the sixth century
would have been the magister militum Africae or the exarch, might himself take charge of
operations. 31 Only if the situation got out of hand at any point did provincial forces retreat behind their fortifications until reinforcements could arrive, and then with their help drive out the invaders= This is precisely what happened in 546, when the whole comitatensian garrison of Byzacium found itself cooped up in
Hadrumentum, unable to prevent the ravaging of the countryside by the Moors. *^
As shown in the preceding section, the initiative for planning and siting fortifications would have resided with the military
commanders of the diocese and provinces,
though control would also have been exercised by the prefect and the emperor. The principles guiding the duces, however, would have
produced somewhat different results from the lines of defence identified by Diehl and so
readily accepted by other historians.
Fortifications had to be placed in such positions that the garrisons from a number of them could be concentrated with the greatest speed and ease in the places most under threat. In general, these areas would have been the ones of richest Romano-Byzantine settlement, where the
pasture was good and the plunder—principally agricultural produce, slaves and valuables—was best to be found. This explains in large part why, almost without exception, sixth-century fortifications in Africa were sited either in
localities that were or had earlier been Roman towns, or in areas showing evidence for
extensive agricultural exploitation. 33
Placing fortifications in the midst of areas of population, however, also made sense for other reasons. It gave the inhabitants a
refuge to flee to, if they did not have fortified buildings of their own. It allowed the
garrisons to be used for certain administrative functions, such as collecting taxes. 34 It
made the supply of the annona to the troops
easier to arrange. 35 More importantly, it made the best strategic sense, since towns were sited, for a variety of reasons, in the best positions for controlling their territoria.
They were also usually placed on major roads or cross-roads, and were thus ideally situated to be used as points for massing troops or dispatching them to other parts of the province. 36
Before discussing how these factors are reflected in the archaeological evidence of
fortifications in Africa, a word needs to be said about chronology. Diehl's maps illustrating the distribution of Byzantine fortifications in Africa not only include a large number of sites whose fortifications may not have been Byzantine at all, but also fail to distinguish between Byzantine fortifications of different periods. An attempt has been made to rectify these shortcomings in Maps 3-5. As will be seen in Maps 3 and 4, the majority of datable official fortifications are
Justinianic, completed during one or other of Solomon's terms as governor. This being so, and in view of the prodigious efforts made by Justinian in building fortifications elsewhere, it would seem reasonable to assume that the majority of the undated fortifications are also Justinianic and so regard them all, until evidence to the contrary is forthcoming, as having formed part of the system of defence established within two decades of the reconquest. Later
fortifications, shown in Map III. 6
98
324, seem to have been planned simply to complement existing arrangements, without
modifying greatly the overall dispositions. In Map 5, the locations of a number of undated or insecurely identified fortifications are shown
(together with one of Heraclius, whose date falls outside the sixth century). Map 7 illustrates the relationship of all the fortifications shown in Maps 3 and 4, and some from Map 5, to the
Roman road-system as reconstructed by Salama.
These maps form the basis for the discussion of the strategic planning to be seen in the siting of fortifications that will be given, province by
province, in the following section.
7. Strategy and the siting of fortifications (2).
The archaeological evidence
In the province of C arthago Proconsularis, the number of recorded fortifications is small in
comparison with that of the other provinces.
Only two town walls, at Carthage itself (M3, 1)
*■ and Vaga (M3, 46), remaned lustiniana and Theodoriana respectively, and a single fort at Thugga (M3, 44) are dated securely to the reign of Justinian. Undated forts are known, however, at Aih Metouia (M3, 48), Mustis (M3, 58) and Clupea (M3, 51); and an imperial arms-depot is attested at Thabraca (M3, 61).2 It would
appear, therefore, that during the reign of
Justinian, the troops of the field-army stationed in Africa would have been based almost
exclusively in Carthage itself. In the later sixth century, however, some new fortifications were erected in the province. The fortress at
Thubursicu Bure (M4, 70) and probably the fort at Tignica (M4, 71) were built under Justin II; a fort was built at Tubernuc (M4, 76) under
Tiberius II, and another at Agbia (M4, 77) probably under Maurice. This minor flurry of building3 may perhaps be linked with the
evidence for Moorish raids affecting the area in the late sixth century and making it necessary to bring Carthago Proconsularis more into line with other provinces through the establishment of a larger number of local garrisons.'*
Carthago Proconsularis was bordered to the south by the province of Byzacium. The list of known fortifications from Byzacium includes twelve built under Justinian, three that are undated and three dated to after the reign of Justinian. The distribution of fortified sites corresponds generally speaking to the areas of most developed Roman-African settlement, that is to say the fertile coastal strip and the
south-eastern edge of the Dorsal, where the Roman townships of the region could benefit from
springs fed by rain-water from the massif. ° The central arid steppe, across which the Byzantine army marched in 548, was apparently devoid of official fortifications, though George of Cyprus mentions the existence of the town of
Madarsuma (M6, 650).6
Communications between the coastal towns was relatively easy, both by land and sea.
Belisarius had followed the coast road from
Caput Vada to Grassa in 533, stopping in towns and camps along the way. The route from
Carthage to Hadrumentum continued to be of strategic importance throughout the sixth
century. It was probably used by Solomon in 534 and 535, by John son of Sisiniolus in 544, by Artabanes in 546 and by John Troglitas in the same year. John Troglitas may have followed the same route again in 547; and, after the defeat of his army by the Moors at Marta, he retreated back along the coastal road as far as Iunci. supplied all the time by his fleet. 7
III.
6-7 99