dukes of eastern frontier.pdf

12
GEOFFREY GREATREX In an influential paper published in 1977, Wolf Liebeschuetz offered a measured assessment of the strength of Roman defences in the East in late antiquity. While marshalling the epigraphic evidence to demonstrate the lack of building activity by the emperor and his officials and the apparent absence of Roman forces in Syria, he also noted the relative infre- quency of invasions and the continued prosperity of northern Syria.' Other scholars since then have been more downbeat, emphasising the impact of the invasions and the deportations which accompanied them, as well as the deterioration of Roman defences;*yet further work in the Limestone Massif of northern Syria continues to point to enduring prosperity, at least until the mid-sixth century.3 The aim of this paper is to contribute to this debate by focussing on administrative changes along the eastern frontier in the fifth and sixth centuries - in particular the proliferation of duces ('dukes'), a development that has so far received little attention." Even if the geographical scope of the power of such officials was eroded through being centred on one particular base, as will be shown, this does not necessarily imply a reduction in their military effectiveness. Too often, it will be argued, scholars have been inclined to draw unduly negative conclusions from a lack of evidence. The eastern frontier in thefifth century At the time of the Notitiu Dignitaturn, i.e. the early fifth century," there were seven duces along the eastern frontier, stationed in Phoenice, Syria (which included Euphratesia), Palestine, Osrhoene, Mesopotamia, Arabia and Armenia (see Figure 9.1): The Notitiu offers a useful list I Liebeschuetz 1977. z E.g. Trombley 1997; Trombley and Watt 2000, xliv-xlvii; Kaegi 1992,47-51. 3 See Tate 1989, 1992, 1996; Foss 1995,220-22 (reviewingTate 1992), 1997,esp. 202; also Ball 2000, 4 A brief paragraph in Whitby 1986,728.Cf. Ravegnani 1988,74; Mazal2001.327. On fourth-century developments, see van Berchem 1952, ch. I. Seeck 1905 is still useful. s Zuckerman 1998a, though still controversial: see Brennan 1998, 35, and Kulikowski 2000,360. 6 Not. Dign. or. 32-38, pp. 67-83 (Dodgeon and Lieu 1991,340-48; Brennan 1998,38-40).The maps used here were kindly prepared by Maurice Clayton. They are based on those to be found in Greatrex 1998, but take into account not only the relevant parts of the Tubinger Atlas des Vorderen Orient (B V 13, 1992, B VI 4, 1984, B VI 5, 1988). but also Talbert 2000 and Hewsen 2001. In the case of the bases in Tzanica, we have preferred the locations indicated in Talbert 2000, which differ somewhat from the Tubinger Atlas and Hewsen 2001. For the provincial boundaries, Jones 1964 and Hewsen 1992 were used. 207-36. 87

Upload: constantius11

Post on 22-Oct-2014

107 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Dukes of eastern frontier.pdf

GEOFFREY GREATREX

In an influential paper published in 1977, Wolf Liebeschuetz offered a measured assessment of the strength of Roman defences in the East in late antiquity. While marshalling the epigraphic evidence to demonstrate the lack of building activity by the emperor and his officials and the apparent absence of Roman forces in Syria, he also noted the relative infre- quency of invasions and the continued prosperity of northern Syria.' Other scholars since then have been more downbeat, emphasising the impact of the invasions and the deportations which accompanied them, as well as the deterioration of Roman defences;* yet further work in the Limestone Massif of northern Syria continues to point to enduring prosperity, at least until the mid-sixth century.3

The aim of this paper is to contribute to this debate by focussing on administrative changes along the eastern frontier in the fifth and sixth centuries - in particular the proliferation of duces ('dukes'), a development that has so far received little attention." Even if the geographical scope of the power of such officials was eroded through being centred on one particular base, as will be shown, this does not necessarily imply a reduction in their military effectiveness. Too often, it will be argued, scholars have been inclined to draw unduly negative conclusions from a lack of evidence.

The eastern frontier in thefifth century

At the time of the Notitiu Dignitaturn, i.e. the early fifth century," there were seven duces along the eastern frontier, stationed in Phoenice, Syria (which included Euphratesia), Palestine, Osrhoene, Mesopotamia, Arabia and Armenia (see Figure 9.1): The Notitiu offers a useful list

I Liebeschuetz 1977.

z E.g. Trombley 1997; Trombley and Watt 2000, xliv-xlvii; Kaegi 1992,47-51.

3 See Tate 1989, 1992, 1996; Foss 1995,220-22 (reviewing Tate 1992), 1997, esp. 202; also Ball 2000,

4 A brief paragraph in Whitby 1986,728. Cf. Ravegnani 1988,74; Mazal2001.327. On fourth-century developments, see van Berchem 1952, ch. I . Seeck 1905 is still useful.

s Zuckerman 1998a, though still controversial: see Brennan 1998, 35, and Kulikowski 2000,360.

6 Not. Dign. or. 32-38, pp. 67-83 (Dodgeon and Lieu 1991,340-48; Brennan 1998,38-40). The maps used here were kindly prepared by Maurice Clayton. They are based on those to be found in Greatrex 1998, but take into account not only the relevant parts of the Tubinger Atlas des Vorderen Orient (B V 13, 1992, B VI 4, 1984, B VI 5 , 1988). but also Talbert 2000 and Hewsen 2001. In the case of the bases in Tzanica, we have preferred the locations indicated in Talbert 2000, which differ somewhat from the Tubinger Atlas and Hewsen 2001. For the provincial boundaries, Jones 1964 and Hewsen 1992 were used.

207-36.

87

Page 2: Dukes of eastern frontier.pdf

88 WOLF LIEBESCHUETZ REFLECTED

AO.rtO0

0 100 200 300kKm

Figure 9.1. Dukes of the eastern frontier, c. 400

Page 3: Dukes of eastern frontier.pdf

GEOFFREY GREATREX: DUKES OF THE EASTERN FRONTIER 89

of the forts within the province (or provinces) under the control of each dux, as well as of the units based there. It unfortunately does not make clear, however, where the dux himself was based. It is usually assumed that he would reside in the metropolis of the province to which he was assigned, e.g. at Edessa in the case of O~rhoene.~ Yet this may not always have been the case: as Wolf Liebeschuetz noted already in 1972, the Martyrdom of Sergius and Bacchus clearly places the seat of the dux of Syria and Euphratesia at Barbalissus rather than Hierapolis.x Similar is the case of Mesopotamia, where the seat ofthe governor will have been at Amida, the metropolis, while the dux was stationed at Constantia (and later at Dara)."

The peace that prevailed along the eastern frontier for much of the fifth century means that evidence concerning the evolution of the command structure is sparse. We hear of a dux in his praetorium at Constantia around the year 450."'Some twenty years later a law of Leo con- cerning enrolments in government service offers a list of ducates along the eastern frontier: it mentions duces of Palestine, Mesopotamia, the 'new limes' of Phoenice, Osrhoene, Syria, Euphratesia, Arabia, of both Armenias and of both provinces of Pontus. ' I Clearly there had been significant changes since the time of the Notiria. The new command in Pontus is important, since it prefigures the further additions made by Justinian in the sixth century. It may be supposed that the seat of this new dux lay at Trapezus, hitherto the residence of the prefect of the first Pontic legion.'* To the south, measures were taken probably in response to the withdrawal of Roman posts along the Strata Diocletiana. The area south of the Euphrates, sometimes referred to as the barbaricum, lay increasingly exposed to sudden raids by nomads." No doubt in order to check these incursions, separate ducates were created for Euphratesia - centred probably on Barbalissus - and for Syria. The dux of the 'new limes' of Phoenice was based at Emesa until the reign of J~st inian. '~

7 Cf. Dillemann 1962, 107; Segal 1970, 117.

8 Pussio Antiquior SS. Sergii ef Bucchi (ed. van den Gheyn 1895), 385, cited by Liebeschuetz 1972, 115 n. 9.

9 See Dillemann 1962, 107; Lauffray 1983,31. This despite the fact that the Notitiu, whose geography is somewhat erratic here, actually places Constantia in Osrhoene, as does Hierocles, Synekdemos (ed. and comm. Honigmann 1939), 714.2. For a full discussion of the province to which Constantia belonged see Burgess 1999,280-82. For 'Constantia' not 'Constantina' see Greatrex 1998, 101 n. 82; Burgess 1999,277-78. On Fig. 9.1 Emesa is indicated as the site of the dux of Phoenice, but it is possible that he was stationed at Palmyra.

1 0 Segal 1970, 103. Cf. Flemming 1917,82.25/83.36, on the dux Florus, with PLRE 2, Florus 1 . The relevant section of the Acts of the Council of Ephesus of 449 is translated also in Honigmann 1944.

I I CJ 12.59(60).5, undated. See Jones 1964,224,609. Trombley and Watt 2000.50 n. 244 suggest a date of c. 470. Cf. Garsoyan 1998, 249. The Pussio of Eustratius, Auxentius, Eugenius, Mardarius and Orestes, PC 1 16,470, ch. 2-3, dated by Zuckerman 1998b, 124 n. 53, to the fifth century, places a dux named Lysias in Satala and refers to his command over the limitunei; no other source, however, refers to a dux at Satala.

1 2 Not. Dign. or. 38.16, p. 84.

1 3 Greatrex and Lieu 2002,238; Key Fowden 1999,65-66. Cf. however, Konrad 1999,408-10, for a greater Roman military presence, at least close to the Euphrates.

14 On Barbalissus see above n. 8. The location of the headquarters of the dux of Syria is difficult to determine. The quarters of units listed in the Notitiu lie well to the east of the major cities of Syria. In

Page 4: Dukes of eastern frontier.pdf

90 WOLF LIEBESCHUETZ REFLECTED

The eastern frontier in the early sixth century

The war that broke out in 502 offers a glimpse of the state of the Roman military hierarchy at this time (see Figure 9.2). By far our best source is the account attributed to Joshua the Stylite which frequently makes mention of duces. Specifically, it refers to Olympius, dux of Tella (i.e. Constantia) ($51), Eugenius, dux of Melitene ($51), Timostratus, dux of Callinicum ($$64,69), as well as (briefly) to Gainas, dux of Arabia ($75) and to a dux Romanus ($92).15 The first commanders mentioned must be the duces of Mesopotamia, of both Armenias, and of Osrhoene respectively. We may note in passing that all three duces acquitted themselves well, which implies that they were able to draw on adequate forces. Among these will have been units of comitatenses, which, by a law of Anastasius of 492, were placed under the command of the dux in whose province they were stationed.16 But Pseudo-Joshua’s evidence may be pressed further. His tendency to refer to a dux as of a city, rather than of a province, is significant. Given the high level of accuracy of his account and his frequent use of official terms, it may well reflect practice at the time.” It is likely therefore that already by 500 the post of dux was beginning to detach itself from a particular province and to associate itself rather with a particular base, where, presumably, ducal forces were concentrated.

The eastern frontier under Justinian

We come now to the situation during the reign of Justinian. Our prime source concerning the command structure is Procopius’ De aedificiis, supplemented by the emperor’s laws on Armenia and a few notices in Malalas and elsewhere. The increasing importance of the northern sector of the frontier is immediately apparent. In 528 the emperor not only created a new high command for this sector, henceforth under the control of the magister militum per Armeniam et Pontum Polemuniacum et gentes, but placed under him several new duces. He was also careful to recruit new soldiers for the magister militum, as well as transferring to him seasoned units from other armies.18 In the former Armenian satrapies, now abolished, duces were stationed at Martyropolis and CitharizBn, both important strongpoints guarding potential invasion routes. In the De aedificiis, Procopius indicates that both duces received substantial garrison forces.” Another dux was positioned at Artaleson, to the north; he too was endowed

the sixth century, Chalcis appears to have been a major base; it may have already taken on this role by the late fifth century, cf. n. 29 below. On Emesa see Stein 1949,289.

I 5 See Trombley and Watt 2000, ad loc. for detailed commentary; also Luther 1997. Whether Callinicum was the normal base of the dux of Osrhoene may be doubted; Timostratus may have been sent there by the magister militum Ambindus, who had himself chosen Edessa as his headquarters (contra Greatrex 1998.3 1 and n. 96). On Melitene as the base of the dux of Armenia see Garsoran 1998,248.

16 CJ 12.35.18 (of492), cf. Greatrex 1998.31.

17 See Trombley and Watt 2000, xxx-xxxiv, with Howard-Johnston 1995, 166 n. 13, and further below.

1 8 Adontz 1970, 107-12; Zuckerman 1998b, 126-27; Garsolan 1998.24546; Greatrex and Lieu 2002, 83. 19 Procop. Aed. 111.2.1-14 on Martyropolis, without mention of troops, although those installed at nearby PheisSn (3.1-6) were no doubt placed under the command of its dux; cf. Zuckerman 2002, 17 I. Cf. Adontz 1970, 107; Whitby 1984, 181 (on Martyropolis); Howard-Johnston 1989 (on CitharizSn).

Page 5: Dukes of eastern frontier.pdf

GEOFFREY GREATREX: DUKES OF THE EASTERN FRONTIER 91

30!fKm 0 100 200

Figure 9.2. Dukes of the eastern frontier, c. 500.

Page 6: Dukes of eastern frontier.pdf

92 WOLF LIEBESCHUETZ REFLECTED

with adequate forces according to Procopius. The mugister militum had his headquarters at Theodosiopolis, where his troops could man the city defences, soon to be strengthened by the emperor.’”Further north, in the newly pacified region of the warlike Tzani, three more duces were established, based at HoronBn, SchamalinichBn and TzanzacBn; Procopius does not refer in their case to the assignment of forces, but it seems safe to assume that they were equipped with sufticent troops to quell the uprisings which inevitably broke out.”

Around the same time as he introduced these new posts in Armenia, Justinian also reorganised Roman defences further south. Hitherto the dux of Phoenice Libanensis had been stationed at Emesa, but in response to increasingly daring raids by the Lakhmid ruler, al- Mundhir, the emperor added a dux to the province and stationed the two duces at Damascus and Palmyra, the latter being much closer to the area affected by raiding.” Malalas, moreover, specifically states that a numerus of regular troops was assigned to the dux at Palmyra, along with the limitunei.” In Euphratesia, Procopius describes the strengthening of fortitications at Zenobia, where a commander (urchbn) was installed with regular troops and a garrison (presumably of limitunei). It is highly likely that he is referring to a dux.24 This need not imply the absence of a dux further west in the province, at Hierapolis (the metropolis) or elsewhere: during Khusro’s invasion of 540, Bouzes, supreme commander of the frontier region, assembled his forces at Hierapolis.2s In Osrhoene, Procopius, again in the context of the strengthening of fortifications, describes the appointment of a dux to take command of Circesium, at the confluence of the Khabur and Euphrates. There is no reason to suppose that the dux based in Edessa was transferred here: northern Osrhoene and Mesopotamia required more vigilant defence than any other sector of the frontier. Rather, just as in the case of Phoenice and, probably, Euphratesia, Justinian sought to shore up defences by the addition of a dux.26 In Mesopotamia, on the other hand, it appears that there continued to be only one dux, based initially at Constantia, then at Dara. Under the terms of the Eternal Peace of 532, the base

20 Procop. Aed. 111.3.13-14 on ArtalesBn; 111.5.1-12 on Theodosiopolis. Cf. Stein 1949,290; Adontz

21 Aed. 111.6, esp. 17,26. The first major revolt of the Tzani broke out in the 550s: Stein 1949.5 16.

22 Joh. Mal. 18.26 with Stein 1949, 289; Shahid 1995, 172-73; Greatrex and Lieu 2002, 84-86. Treadgold 1995,97 n. 35 seems to doubt the existence of two duces in the province, although both Procop. Bell. (Pers.) 11.16. I7 and Malalas clearly refer to a dual command. Justinian’s Fourrh Edict (probably dating from 535 or 536) unhelpfully refers to both a dux and duces (Just. Ed. IV.2).

23 Joh. Mal. 18.2. Cf. Greatrex and Lieu 2002,85.

24 Aed. 11.8. I 1. The commander is said to be an &pxov o r p a r t o r r ~ ~ h ~ a t a h 6 y o v , i.e. a commander of military units. At 111.3.14 Procopius refers to the placement of ~ t p a t ~ o r t l t o i Katdlhoyot at ArtalesBn under the command of an &pxov, which he then explains as being the equivalent of a dux. Cf. 111.6.17 (HoronBn); 6.26 (Schamalinichh and TzanzacBn); 11.6.9 (Circesium). Note also the term &pxov being applied to duces in Bell. (Pers.) 11.8.2, 18.16,24.14.

25 Procop. Bell. (Pers.) 11.6.1-2.

26 Procop. Aed. 11.6.1-1 1. Cf. Stein 1949,289 and n. 4. Whitby 1986,727, places the change early in Justinian’s reign, but see contra Greatrex 1998, 196 n. 1 I . In addition, Elias, V. /oh. Ep. Tell. 39/42, refers to the pruetorium of a dux at Resaina (i.e. Theodosiopolis in Osrhoene) in the early sixth century. This hagiography is a detailed, reliable and contemporary source and deserves credence, cf. Palmer 1990, 24. Thus we should probably add a further dux to the two already noted in Osrhoene.

1970, 1 18-25.

Page 7: Dukes of eastern frontier.pdf

GEOFFREY GREATREX: DUKES OF THE EASTERN FRONTIER 93

was moved back to Constantia at the Persians’ req~est.~’ Although Justinian repaired the defences of Amida, the metropolis of the province which had fallen to the Persians after a three-month siege in January 503, nothing is heard of the presence of a dux. It is sometimes claimed that no Roman soldiers at all were present in the city when the Persians arrived before i t in Octobcr 502, but the vigorous defence described in all the sources, involving artillery, counter-mines and other devices cannot have been exclusively the work of civilians. Given the irnportancc of the city and the frequency with which it was attacked by the Persians, it cannot bc ruled out that a dux was positioned here too, even if no rcference to one can be found in the sourccs.” In Arabia, aduxcontinued to be based at Bostra, while in Palestine the dux still had his headquarters at Caesarea. The case of Syria is less clear, as has already been noted, but the well-attested construction work carried out at Chalcis in 550 points to the importance of the city in the defence of the region, and it is highly probable that it was the base of the dux.’9

From ‘dux provinciae ’ to ‘dux civitatis ’

It is time to take stock of the situation during the reign of Justinian (see fig. 9.3).3” Duces were certainly stationed at TzanzacBn, Schamalinichh, HoronBn, ArtalesBn, CitharizBn, Martyropolis, ConstantialDara, Edessa, Zenobia, Circesium, Palmyra, Damascus, Bostra and Caesarea.” Wherever he was based, there was certainly a dux in Syria. This makes a total of at least fifteen duces by the 530s. We must now consider the impact of this growth on the defence of the frontier and on the office of dux itself. It is clear that a transformation had taken place since the time of the Notitiu. By the reign of Justinian, it was no longer the norm to have just one dux assigned to a province throughout which were scattered detachments at his command, as described in the Notitia. As Constantin Zuckerman has argued in the case of Africa after Justinian’s reconquest, a dux was now associated with a single strongpoint (even if his command extended to the surrounding region and whatever forces were stationed there). Thus, for instance, the dux of Byzacena was based at Hadr~metum.~~ But we may go further than Zuckerman in our discussion of the eastern frontier. With the increase in the number of

27 Procop. Bell. (Pers.) 1.22.3, with Greatrex 1998,31,213, 216.

28 On the siege of Amida see Greatrex 1998, 83-94, and now DebiC 2003; Greatrex and Lieu 2002, 63-67. In both places 1 accepted the statement of Procop. Bell. (Pers.) 1.7.4 - that no soldiers were present - at face value. But see Josh. Styl., $050, 53, on the sophisticated devices of the defenders, with Trombley and Watt 2000, 54 n. 260, linking them to the presence of soldiers.

29 See Greatrex and Lieu 2002,243 for the building work with Fourdrin 1994. Procop. Bell. (Pers.) 11.12. I refers to an &pxov r3v arpar~or3v there in 540, who had to be hidden from the Persians by the city’s inhabitants. Cf. Trombley 1997, 184, who argues that Chalcis was the headquarters of the dux of Syria 1. On Palestine see Mayerson 1988.

30 Fig. 9.3 takes into account the provincial changes implemented in Armenia in 535: Adontz 1970, ch. 7; Zuckerman 1998b, 126; Hewsen 2001,85-86 and map 65.

3 1 Note also that the comes of Armenia Ill acquired military as well as civilian functions. Accordingly his base at Melitene (Adontz 1970, 140) is indicated as a ducal base on the map here (fig. 9.3).

32 Zuckerman 2002, 172. In 54617, Corippus implies the existence of two duces in Byzacena, cf. PLRE 3, Anonymi 41 (not noted by Zuckerman). Note, however, the criticisms of Duval 2002.47, who is rightly sceptical about some of the inferences drawn by Zuckerman.

Page 8: Dukes of eastern frontier.pdf

94 WOLF LIEBESCHUETZ REFLECTED

3004Km 0 100 200

Figure 9.3. Dukes of the eastern frontier, c. 540.

Page 9: Dukes of eastern frontier.pdf

GEOFFREY GREATREX: DUKES OF THE EASTERN FRONTIER 95

duces, the whole notion of a dux of Osrhoene, Armenia or Phoenice becomes obsolete. Although modern scholars continue to apply these terms to officials down to the end of the sixth century, it is doubtful whether they are still appropriate. Pseudo-Joshua tends to associate duces with a particular city, and he is not alone: Malalas, for instance, refers to a former dux of Damascus.” Procopius, it is true, refers to the dux of Mesopotamia and to duces of Phoenice Libancnsis, but never to any others.’“ When we turn to sourccs from later in the ccntury, there is little tracc of duces of particular provinces: Theophylact Simocatta refers, I‘or instance, to Eilifrcdas as the archBiz of Emcsa, i.c., as onc of thc duces in Phoenice, whose hasc had by then returned to this city.” Marinus is described in 591 as hcing the commander of thc troops stationed at Chalcis, as is Juventinus in 572. Both are therefore identified as being duces of Syria in PLRE 3.16

Dukes and garrisons

The officc of dux was thus undergoing a transformation in the period. With the multiplication of duces and their association with specific garrison-fortresses came an inevitable reduction in their relative importance. The Strategikon attributed to the Emperor Maurice confirms their lower status: a dux is a commander of a moiru, an army unit that consists of several smaller units, each of a few hundred men. This is in sharp contrast to Procopius’ duces, each of whom, commanding a few numeri or katalogoi, alongside the local lirnitunei, controlled several thousand.” The temptation is to infer an inevitable debasement in the quality of frontier defence, now entrusted to a dozen or so duces and the meagre forces at their disposal. Such is the conclusion of Zuckerman concerning North Africa: outside the fortresses in which the duces resided the provinces were almost bereft of defences. To be sure, great effort was expended in upgrading fortifications and defensive installations, as Procopius details in the De aedijiciis, but, Zuckerman argues, these places were merely centres of refuge for the population. He makes precisely the same point for the eastern frontier, stressing that Procopius makes no mention of garrisons being placed in Satala, Sebastea, Nicopolis or

33 Joh. Mal. 18.26.

Procop. Bell. (Pers.) 1.22.3, I I . 18. I6 (Mesopotamia); cf. 11.8.2, 16. I7 (Phoenice). His referencc to Isaac, 11.24.14, is less clear: he is described as the commander of forces ‘here’, presumably at Cithariz6n.

35 Theoph. Sim. 11.3. I (p. 73.16); cf. PLRE 3, Eiliphredas. The date is 586. CIzrort. 1234, 21 1/166, cf. PLRE 3, Anonymus 43, refers to a dux at Emesa in 58 1 . See also Sartre 1982, I 12, for duces attested in Arabia in the 580s.

36 PLRE 3, luventinus, Marinus 5 (Theoph. Sim. V. 1 1.4 on Marinus; Joh. Epiph. 3 on Juventinus, tr. in Greatrex and Lieu 2002, 142). Evagrius, it is true, refers to the commander of forces in Phoenice Libanensis, Germanus (Evagr. HE V1.5, p. 225.3), cf. PLRE 3, Germanus 6. The absence of any reference to either B p ~ o v r e ~ or duces in Agathias is striking; it is likely, as Stein I949,8 14- 15, suggests (in the case of Theodore the Tzan), that he uses the term taSiapxoG (‘taxiarch’) as an equivalent.

37 Maur. Srrclr. (ed. Dennis 198 I , trans. idem 1988), 1.3-4, cf. Kazhdan, ODB, 659. Sfrrtr. I .4.12-13 puts an upper limit on the size of a rnoiru (of cavalry) at 3000.

Page 10: Dukes of eastern frontier.pdf

96 WOLF LIEBESCHUETZ REFLECTED

Melitene, for instance, notwithstanding the building work undertaken the~e. '~ A closer examination of the evidence, however, suggests different conclusions.

At the heart of the matter is the issue of evidence. If Procopius fails to mention the presence ofa garrison, can we legitimately infer its absence? What conclusions can be drawn from the lack of any epigraphic tracc of the Roman army? Argumentu ex silentio are always weak. It must be noted that Procopius mentions no garrison at Constantia, at Amida or at Edcssa; in the casc of Dara, he only alludes in passing to the construction of military barracks."' YCL no onc has argued that there were no regular units stationed at these important cities, especially as they were for the most part the seat of adux. Procopius' readers would need no rcminding that thcrc were Roman forces positioned in such places. On the other hand, since there had been no conzit- aterzses at Zenobia or Circesium hitherto, Procopius stresses the installation of a commander and his nunzeri there?" Likewise, there had probably been few, if any, forces garrisoning Thannuris, a post that under Justinian received a sizeable garrison, according to Proc0pius.J' Evidently not only were forces present at the base of the dux, but detachments continued to serve in at least some other strongpoints: in addition to Thannuris, Procopius mentions soldiers in Sergiopolis, PheisBn, in Tzanica generally and SisilissBn in par t ic~lar .~~ Comparison with the African case discussed by Zuckerman is instructive, for there too, as Noel Duval notes, Procopius specifically refers to the installation of gamsons elsewhere than the ducal command post, as, for instance, in the Aurks.4' Thus Zuckerman's supposition that no forces whatever were to be found in cities such as Satala and Melitene appears implausible; for although Theophylact Simocatta indeed refers to the latter as 'defenceless' in 576, this is almost certainly because its garrison had been called away to participate in the engagement with Khusro's army."

As far as epigraphic evidence is concerned, we must bear in mind the sparseness of the material for the whole eastern frontier from Euphratesia northwards. We do find occasional references to a dux or to military installations, but no direct evidence of the presence of army units. This is not unlike the situation in North AfricaA5 For the most part, therefore, we are

38 Zuckerman 1998b, 225-27; idem 2002, 171-72. A similarly bleak view may be found in Lauffray

39 Procop. Aed. 11.5.2-1 I , 3.27,7.1-16, 11.3.26 (barracks at Dara).

40 Aed. 11.6.9, 8.1 1.

41 Aed. 11.6.16. Procopius similarly emphasises the garrisons installed in Armenia and Tzanica, Aed. 111.1.28, 3.7, 3.14, 5.12, 6.17. No troops are specifically mentioned in connection with the duces at SchamalinichBn and TzanzacBn (6.26), but I suspect they are covered by Procopius's reference at 6.22 to a garrison installed at SissilisBn, 'as in all the other places'. Zuckerman's insistence (2002, 172) on limiting the duces to one base, with perhaps one additional fort at his command, is thus far too restrictive.

42 Aed. 11.9.8,II1.2.3,6.I3,6.22. In the Wars we hear of soldiers in Beroea in 540 (Bell. [Pers.] 11.7.6, 7.12.7.37).

43 Aed. V1.7.8; cf. 7.1 with the remarks of Duval2002,47-48.

44 Zuckerman 1998b. 125 n. 57 on Melitene, citing Theoph. Sim. 111.14.1 I .

4s Cf. Greatrex and Lieu 2002,242-43, noting, for example, the inscription in Mango and Mango 199 I , 465-68 (cf. PLRE 3, Thomas 1 I ; Feissel 2OOO.97-98). a dux at Constantia c. 542. On North Africa, Duval2002,47.

1983, 33-37.

Page 11: Dukes of eastern frontier.pdf

GEOFFREY GREATREX: DUKES OF THE EASTERN FRONTIER 97

obliged to rely on references in Procopius and other sources. As we have seen, we can infer from this that not only were the bases of duces supplied with powerful garrisons but so also were certain other strongpoints. Procopius, we must remember, was fully aware of the distinction between a garrisoned post and a place of refuge: he clearly describes the latter at Aed. 11.4. I8 (at BasileBn, north of Amida).4 It is possible to bring one further element to bear on thc issue - the nature of the frontier fortifications themselves. As Duval points out in his response to Zuckerman, we must take into account the design of the forts. If, for instance, they contain platforms intended to support artillery, then it is reasonable to infer that they were built to house trained garrisons!' While such an analysis is almost impossible in the northern sector of the frontier, where the forts named by Procopius have often not been identified with any certainty, there is more scope for an assessment in Mesopotamia and Syria. The formidable remains of Zenobia, which date from the reign of Justinian, were clearly destined to be occupied by a powerful garrison, as Procopius asserts. Artillery could certainly have been employed on the numerous towers, and the presence of arrow slits in the towers increases the likelihood that these defences were constructed with archers in mind!x A similar case could be made for Thannuris, Amida, and Serg i~pol i s .~~

In his article of 1977 Liebeschuetz proposed that 'many fortifications had no soldiers attached to them and that Procopius fails to mention a garrison in connection with the great majority of fortifications in the De aedificiis because they were, in fact, not garrisoned.''" No doubt the vast majority of the forts listed by Procopius in the Balkans, for instance, were not manned. but merely offered places of refuge," yet, as I hope I have shown, this need hardly call into question the presence of soldiers at important cities on the far eastern frontier, such as Amida, Melitene and Satala. It is in northern Syria, where the epigraphic evidence is far more abundant, that the absence of any record of Roman soldiers is significant, as Liebeschuetz demonstrated. Inscriptions show that defensive works were often constructed at the initiative of a local magnate. Towers and other somewhat rudimentary defensive works dot the landscape, but Procopius barely discusses the area (Aed. 11.1 1). While cities such as Antioch, Chalcis and Beroea housed garrisons, the region remained vulnerable to Arab raids and Persian invaders, but - which is crucial - only after they had broken through or bypassed the ducal bases close to the frontier.'* Our conclusion must be that, despite the steady

46 Cf. Dillemann 1962,225 and fig. 3 I .

47 Duval 2002,48.

48 Cf. Kennedy and Riley 1988, 1 17-1 8, for a brief overview, Lauffray 1983, 147-49 for more detail.

49 The first and last of these are rightly categorised by Kennedy and Riley 1988, I 16-2 I , as fortress cities. On the fortifications of Amida see Trombley and Watt 2000, 120; their sheer extent would require an adequate garrison. See also Konrad 1999, 408-9 and Ulbert 2000, 145, on soldiers at Sergiopolis. The case of Sura (cf. Kennedy and Riley 1988, 115-16). is more ambiguous, since the fortifications there were scaled back by Justinian, cf. Konrad 1999, 398-99, Ulbert 2000, 146-47.

so Liebeschuetz 1977,494, cited with approval by Zuckerman 1998b. 128 n. 66.

SI Aed. IV. I 1, cf., for example, Whitby 1988,72.

52 Liebeschuetz 1977,490-94. Cf. Trombley 1997, esp. 161-62, Gatier 2001, esp. 105; Greatrex and Lieu 2002,242-44. See also the references above at n. 3.

Page 12: Dukes of eastern frontier.pdf

98 WOLF LIEBESCHUETZ REFLECTED

reduction in the number of forts occupied by Roman troops, Roman defenccs in the sixth century wcre far from being as sparse or as ineffectual as some have claimed, thanks probably in part to the administrative restructuring and concentration of forces begun in the previous ccntury. The conclusion of Liebeschuetz is still apt: ‘The military effort had been reduced, but this did [not] mean that the Syrian provinces were no longer defended adequately ... Looking at thc ccntury as a whole, nomads and Pcrsians wcre kept out.’

Postscript (Mriy 2007)

W. Wright, Cutulogue of Syriac Munuscripts itr the British Museuni, vol. 2 (London I87 1 ), p. 953, reports a rcferencc to a ‘Theodosius doukas’ of Callinicum, to whom two letters of Paul bar Arab are addressed (in MS Add. 12,155 = DCCLVII, l‘o1.259b). It seems likely that ‘doukas’ is a distorted rcndering of the title dux and that we have here further confirmation of the stationing of a dux at Callinicum, probably in the sixth century (see above p. 90), although neither of the individuals is elsewhere attested.