Jump to ContentJump to Main Navigation
Failure of EmpireValens and the Roman State in the Fourth Century A.D.$

Noel Lenski

Print publication date: 2003

Print ISBN-13: 9780520233324

Published to California Scholarship Online: March 2012

DOI: 10.1525/california/9780520233324.001.0001

Show Summary Details
Page of

PRINTED FROM CALIFORNIA SCHOLARSHIP ONLINE (www.california.universitypressscholarship.com). (c) Copyright University of California Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in CALSO for personal use (for details see http://california.universitypressscholarship.com/page/535/privacy-policy-and-legal-notice). Subscriber: null; date: 21 June 2018

. Valens and the Eastern Frontier

. Valens and the Eastern Frontier

(p.153) Chapter 4. Valens and the Eastern Frontier
Failure of Empire

Lenski Noel

University of California Press

Abstract and Keywords

This chapter looks at Valens' interactions with the peoples of the eastern frontier: the Persians, the Armenians, the Georgians, the Saracens, and the Isaurians. Although Valens' court historians Eutropius and Festus bolstered his claims to Armenia and upper Mesopotamia with slanted accounts of Roman history, these gains began to disintegrate toward the end of his reign, when multiple military crises forced Valens to withdraw his forces to other hot spots. Iberia and especially Armenia were not faithfully obedient subjects but powerfully independent friends. The relations of Valens with Persia, Armenia and Iberia are examined also. Festus and Eutropius provided Valens with practical guides to Rome's historical claims to the territories of the east. They were engaging in a historical discourse with very real political and military implications. It is shown that Valens was forced to deal with three peoples who shared environmental and cultural similarities; the Maratocupreni, Isaurians, and Saracens.

Keywords:   Valens, Persians, Armenians, Georgians, Saracens, Isaurians, eastern frontier, Eutropius, Festus

Romano-Persian Relations, A.D. 298–363

Valens faced an eastern frontier that offered a confusing tangle of problems. Unlike the lower Danube, where Gothic hegemony confronted him with a single, albeit very dangerous, enemy, the eastern limits of the empire were home to a frustratingly complicated web of interrelated peoples, whose threats and demands had to be balanced one against the next. At no point in his reign was Valens free to let down his guard in the east, for as soon as problems with one group ceased, those with another arose. By investigating all of the conflicts that Valens faced in Persia, Armenia, Iberia, Syria, Isauria and Arabia, we shall see how the eastern emperor struggled to keep pace with unceasing demands on his military and financial resources. Before we begin the investigation, however, it is necessary to look briefly at the social and historical background of the peoples and lands on whom Valens concentrated so many of his resources.

Older textbooks would term much of the terrain with which this chapter deals “the fertile crescent” (see map 5).1 That is to say, much of Valens’s activity on the eastern frontier was concentrated around the territory that stretches in a broad arc from Palestine in the west up through Syria and northern Mesopotamia south of the Taurus Mountains and down again along the base of the Zagros Mountains in the east. The concave side of this “crescent” was roughly outlined by the 200 mm p.a. isohyet, which provided enough precipitation for the dry farming of cereal crops. Beyond this, to the east and south of the isohyet from a Roman perspective, lay desert. Rome had controlled the western end of this crescent since the first century B.C., while (p.154) Persia remained master of the east throughout the Roman period. At the top of the crescent, in northern Mesopotamia, the territories of the two empires met along the natural avenue of communication between west and east. This bottleneck of cultivable territory at the edge of both empires became a bone of bloody contention from Rome’s first arrival in the region until its departure in the wake of the Arab conquests. In this struggle for control of northern Mesopotamia, Rome had since the end of the third century enjoyed the upper hand. When Valens was born in the 330s, it controlled all of northern Mesopotamia up to the Greater Zab. In the year before Valens came to power, however, much of this territory had been ceded, and with it Rome’s position of dominance on its eastern frontier.

Among the major subjects of conflict between Rome and Persia were the territories of Armenia and Iberia, in the rugged sub-Caucasus, just north of the Fertile Crescent. Valens’s own efforts to maintain hegemony over these kingdoms continued a drama that had been unfolding since the first century B.C.2 Pompey had first asserted Rome’s right to crown the Armenian and Iberian kings in 66 B.C., and Rome continued to claim suzerainty in these regions into the fourth century. Even so, its claims had not gone uninterrupted, for they had always been disputed, first by the Parthians and then by the Sassanians, and these had been successful at various periods in imposing their own sovereigns over the kingdoms of the sub-Caucasus. Fuel was added to the fires of discord by the fact that both Armenia and Iberia were inhabited by staunchly independent peoples whose loyalty either side could maintain only with great effort.

Armenia had been ruled since the first century A.D.—with some interruptions—by a monarchical dynasty of Parthian descent called the Arsacids. Arsacid domination was, however, hardly monolithic. Rather, the Armenian throne was locked in a constant struggle to impose its sovereignty on a series of over one hundred powerful territorial princes with origins predating the imposition of central rule. These princes, designated nakharars in Armenian, claimed autonomy in their own regions, which they ruled as the hereditary heads of nobiliar households.3 Because they regarded themselves as virtual equals to the Arsacid kings, their loyalty to the throne was always contingent and always depended heavily on the furtherance of their own interests. Thus, in exchange for cooperation, the Arsacids ceded to the most powerful nakharars control of a wide array of political and military fiefs: the Mamikonean family, for example, dominated the position of commander in (p.155) chief (sparapet),4 the Bagratuni the positions of royal coronant and commander of the cavalry (tagakap and aspet), and the Arkruni that of grand chamberlain (mardpet). Even despite these concessions, indeed, partly because of them, the nakharars guarded their independence fiercely and often broke into open rebellion against their Arsacid rulers.5 The Arsacids were thus under constant compulsion to renegotiate—through force and diplomacy—dominion over their feudal princes.

Because of the inherent weakness of their authority, the Arsacids often turned to the Roman state to help realign nakharar loyalties. Yet the Arsacid kings were themselves willing to grant the empire only limited influence, making their own loyalty to Rome yet another variable in the complex equation of Romano-Armenian relations. Arsacid kings refused to travel in Roman territory without military escorts and reserved the right to decide whether or not to support Roman military endeavors.6 Politically, Arsacid cooperation was contingent upon enticements and sometimes even threats: the retention of royal hostages, the concession of privileges, the payment of tribute, and even the diplomatic exchange of women. In 358, for example, Constantius II offered Olympias, the daughter of a former praetorian prefect, to Arshak II of Armenia in marriage; two years later, still hoping to retain Arshak’s friendship, Constantius summoned the king to Edessa to offer him generous gifts and an exemption from tribute in exchange for a promise of loyalty during Constantius’s planned expedition in Mesopotamia that year.7 Only with such cozening could the Arsacids be won over, for their loyalty was contingent and needed always to be maintained at a price.

Such precautions were essential in a world where the Persian shahanshah constantly lobbied the Armenian kings and nakharars with threats and enticements of his own.8 The Persians regarded Armenia as part of their ancestral territories; indeed, they considered the Armenian throne, which was after all held by a Median dynasty, second in importance only to that of Iran.9 (p.156) This comes as no surprise, given that, in terms of culture, politics, and social structures, Armenia had much more in common with Persia than with Rome.10 It is, therefore, also natural that Armenia often shifted its allegiance to its southern neighbor. In the thirty years prior to Valens’s reign, Armenian kings are known to have allied themselves with the Sassanians temporarily in 312, 337, and again in the late 350s.11 And rebellious nakharars shifted their allegiance to Persia with equal regularity. Armenia was thus constantly in flux from the Roman perspective—indeed, from the Sassanian as well. Its allegiance could never be taken for granted.

Even despite these interruptions, it is fair to say that Rome largely succeeded in commanding Armenian loyalty against the Sassanians through the first two-thirds of the fourth century. As we shall see, much of this stemmed from a reassertion of Roman military supremacy in the region in the last years of the third century. Rome’s success in retaining Armenian loyalty also stemmed at least in part from its greater willingness to grant the Arsacids a freer rein than the Sassanians had allowed when they controlled Armenia in the second half of the third century. Precisely because Armenia was so close to Iranian society culturally and to the Iranian heartland geographically, the Sassanians always expected greater power inside Armenia when they gained the upper hand there. For this reason, Armenia and its kings knew well that they needed to maintain political distance from Iran by cultivating their alliance with Rome. After the conversion of Rome’s empire to Christianity, this meant the conversion of the Armenian monarchy as well. In 314, the Armenian king Trdat (probably IV) the Great (ca. 298–ca. 330) converted from Armenian Zoroastrianism to Christianity and began the slow and painful process of converting his kingdom.12 This he did in alliance with Armenia’s first patriarch, Grigor Lusaworic’ (Gregory the Illuminator), who had been raised as an exile in Cappadocian Caesarea and baptized a Christian. As with all offices in Armenia, that of patriarch promptly became hereditary, and their hold on religious authority, and the wealth and territory that went with it, gave Grigor’s family, the Gregorids, a great deal of influence. It also brought them into frequent conflict with the throne and the nakharars over both social and political issues.13 Christianity thus introduced yet another (p.157) complication into the formula of Romano-Armenian relations. In all, however, Rome managed to take advantage of this religious common ground to strengthen its grip on this proud kingdom.14

Iberia sat directly north of Armenia and was thus frequently at logger-heads with its neighbor over the border territory between the two kingdoms along the river Cyrus (Kur). Iberia was perhaps closer to Rome culturally than Armenia, although there was also much about it that was Persian.15 Even so, Iberia differed from Armenia in that its monarchs succeeded in establishing firmer control over the nobles within their more circumscribed domains.16 As in Armenia, the Persians vied with Rome for authority over the Iberian throne, and in the late third century, they successfully asserted their claim by crowning a Persian dynast named Mirian III.17 Mirian, founder of the Mihranid dynasty, which ruled Iberia into the sixth century, maintained a tight hold on his kingdom down to his death in 361. Nevertheless, although he had been a Persian appointee, after 299, Rome established suzerainty over his kingdom, which would last through Valens’s reign.18 It was during this period of Roman control that Mirian also converted himself and his kingdom to Christianity. Indeed, the strongest cultural bond cementing Romano-Iberian relations in the fourth century was, as with Armenia, Christianity.19

Thus Iberia and especially Armenia were not faithfully obedient subjects but powerfully independent friends. They were not fixed points in the geometry of Rome’s eastern frontier but variables in the calculus of Roman foreign policy. As we shall see, this complicated Valens’s relations with both kingdoms, and particularly Armenia, even in periods when Persia had backed away from conflict.

While Iberia and Armenia posed problems for Rome, Persia posed a definite threat. Much of the tension between Persia and Rome in Valens’s day stemmed from disputes over the control of northern Mesopotamia. In two wars between 195 and 199, Septimius Severus had extended Roman authority deep into the Parthian-controlled territories of northern Mesopotamia, which he formed into the Roman provinces of Osrhoene and Mesopotamia.20 Within two decades of Severus’s death, however, the Parthians had succumbed to a new south Persian dynasty, the Sassanians, and these quickly set about retaking Severus’s gains. Unlike the Parthians, the Sassanians (p.158) succeeded in breaking the power of Persia’s own formidable landed aristocracy to form a centralized and highly organized political machine.21 Much of the newfound vigor they injected into the Persian state was channeled into conquest, particularly the recapture of northern Mesopotamia. Major victories over the Romans enabled Shapur I (r. 240–72), the second Sassanian dynast, to reconquer most of northern Mesopotamia and, after 244, to regain suzerainty over the kingdoms of Armenia and Iberia.22

Third-century Rome faltered in the face of this formidable enemy and only began to regain its balance in the 270s. This process culminated in 298, when Diocletian’s Caesar, Galerius, crushed the forces of the shahanshah Narses (r. 293–303) and captured his royal harem.23 Narses was forced to settle an agreement in 299 under which he turned over control of the remaining territories that Shapur had wrested from Rome.24 Thus the treaty of 299 reextended Rome’s territorial claims on its eastern frontier up to the Tigris, reasserted Roman suzerainty over Armenia and Iberia,25 and limited commercial exchange between Rome and Persia to the city of Nisibis (Nusaybin).26 New to the 299 agreement was the provision that Rome would also gain direct suzerainty over five Armenian satrapies that lay north and east of the Tigris—Ingilene, Sophene, Arzanene, Corduene, and Zabdicene (see map 4).27

These south Armenian principalities, separated from the Armenian midlands by the Taurus,28 had always enjoyed even greater independence vis-à-vis the Armenian throne than the nakharar territories to their north. They were inhabited by a multi-ethnic agglomeration of Syrian and Armenian peoples and were ruled by hereditary satraps, who had frequently claimed total (p.159) independence from the Arsacids in preceding centuries.29 Indeed, the Arsacids had granted two of the satraps in question—those of Arzanene and Corduene—the further distinction of recognition as viceroys (Latin vitaxa; Armenian bdeashkh) of the Arabian and Syrian marches.30 These two marches controlled access over the Taurus into the Armenian heartland and as such constituted the most strategically important territories in the region. By claiming direct suzerainty over these and the three remaining trans-Tigritane satrapies in 299, Rome gained a footing beyond the Tigris in what were now a whole series of tiny client kingdoms.31 Indeed, from 299 on, the satrapies were, like client kingdoms, free from taxes but expected—if not legally obliged—to supply troops for Roman armies, fortifications to protect Roman frontiers,32 and crown gold for Roman coffers.33 The 299 treaty was thus a major coup for Rome, not just reclaiming the Severan frontier up to the Tigris but also extending Roman control over client territories situated beyond it.

Just as Rome’s gains were great, Persia’s loss was devastating, so much so that the Sassanians never accepted the treaty. Not until the 330s, however, was Narses’s grandson, Shapur II, in a position to reassert Persian power in the region. Between 337 and 350, Shapur agitated for control of the Armenian throne, thrice laid siege to Nisibis, defeated a massive Roman army near Singara and captured that stronghold. In the seven years that followed, tensions subsided, because both Shapur and Constantius II were occupied with wars elsewhere in their empires, but in 357, Shapur was again demanding the return of the territories lost by Narses. Having failed to achieve his aim through diplomacy, the shahanshah once again undertook strikes into upper Mesopotamia in 359–60, during which he destroyed the Roman (p.160) fort of Amida, again overcame Singara—since regained by the Romans—and captured and regarrisoned Bezabde as a Persian stronghold.34

A full-scale Roman response to these aggressions came only in 363, when Julian invaded Persia in a lightning strike down the Euphrates.35 He reached the capital of Ctesiphon in less than a month and laid siege to it with spectacular pomp. After his efforts failed, though, Julian’s plans quickly went awry, and his army eventually found itself marching out of Persia along the Tigris under constant attack from the Sassanians. On June 2636—as we saw in the first chapter—the emperor fell back to help fend off an enemy raid, and was struck and killed in the melee. For the fifth time in Roman history, the empire had lost its ruler in the midst of eastern hostilities.37 With no time for ceremony, the army quickly proclaimed Jovian emperor and continued its retreat up the Tigris.

Jovian’s situation grew worse as the Persians pressed his army and supplies dwindled to desperate levels.38 Under these circumstances, Shapur sent an embassy led by his chief commander, Suren, to offer peace.39 Jovian had little choice but to accept. He sent his praetorian prefect Sallustius and the general Arinthaeus to Shapur’s camp, where they negotiated for four days before finally reaching an accord. The length of their negotiations in the midst of such trying circumstances indicates that terms were not simply dictated by Shapur but were actually debated.40 In fact, Shapur’s bargaining position (p.161) was not ironclad: his own supplies must have been limited, since both he and the Romans had destroyed Persian crops and supply depots,41 and he had learned that a Roman reserve force of nearly 15,000 men was poised in Mesopotamia, putting his own men in ever greater danger as Jovian continued north.42

Because there were reasons for concessions on both sides, the treaty that emerged was a compromise: Persia came off the clear winner, but Rome did not lose all that it might have.43 Shapur’s aim in the negotiations was the same it had been since he initiated his attacks on the Roman frontier: he hoped to recover control over Mesopotamia, Armenia, and the trans-Tigritane satrapies ceded by his grandfather in 299.44 Not surprisingly, the terms of the treaty—best preserved in Ammianus and Zosimus45—indicate that he achieved his aim only in part. As in 299, five satrapies were ceded, but not the same five Narses had lost. Galerius had gained Ingilene, Sophene, Arzanene, Corduene, and Zabdicene, but Shapur regained only the last three, that is, only those east of the river Nymphius.46 Together with these, Shapur (p.162) also obtained Moxoene and Rehimene. The latter, it seems, had been subsumed under the larger unit of Corduene in the 299 treaty.47 Although it has been argued that the former was similarly considered a part of the larger satrapy of Arzanene, it is just as likely that this territory had simply not been under discussion at all in the first treaty.48 There was thus a discrepancy between the satrapies ceded in 299 and those retroceded in 363, but it did not, however, matter as much as the fact that the numbers matched: Narses had lost five trans-Tigritane satrapies in 299, and Shapur had won five back in 363.49

Rome also surrendered eastern Mesopotamia, together with its key fortresses of Nisibis, Singara, and Castra Maurorum.50 Nevertheless, it retained control of western Mesopotamia and did not even surrender the inhabitants (p.163) of Nisibis51 and Singara.52 Based on the placement of Rome’s eastern frontier forts in the years that followed, it seems that an effective border was established, which ran north from the Euphrates along the Chabur, then just west of the Mygdonis over the Mons Massius (Tur Abdin) to the Tigris, and finally along the Nymphius (Bohtan Su) into the Taurus Mountains.53 By this reckoning, Rome lost only half of its claims in Mesopotamia.54

The treaty thus created a situation wherein both powers had an equal footing in this strategic region. Although it represented a clear gain for Persia and a devastating loss for Rome, relative to the 299 treaty, it remained a compromise: Rome and Persia each controlled half of upper Mesopotamia and half of the trans-Tigritane satrapies. As such, the treaty created a parity that neither side could easily transgress, and perhaps for this very reason, it proved extremely durable. Aside from its provision on Armenia, which was modified in the two decades following and completely rearranged in 386, the treaty, which had originally been brokered for only thirty years, determined the shape of the Romano-Persian frontier down to 502.55

Regarding Armenia, however, the situation was, as indicated, somewhat different. In listing the terms of the 363 agreement, Ammianus reports very specifically:

(p.164) To these conditions there was added another, which was destructive and impious, that after the completion of these agreements, Arsaces, our steadfast and useful ally, should never, if he asked for it, be given help against the Persians. This was contrived with the double purpose that a man who at the emperor’s order had devastated Chiliocomum might be punished and the opportunity might be left of presently invading Armenia without opposition.56

Ammianus’s phrasing implies that Rome surrendered only the right to defend Armenia’s current king, Arshak II.57 Later in his narrative, though, Ammianus implies that Rome was also forbidden to defend Arshak’s son Pap and to impose him as Arshak’s successor.58 Other sources also point to a broader abandonment of the Armenian throne. Zosimus indicates that the treaty surrendered “most of Armenia”; the Armenian Epic Histories confirm that Rome allowed Persia a free hand in the “Armenian Midlands”; Libanius claims that “all of Armenia” was abandoned; and the Syriac Chronicon miscellaneum speaks of the surrender of “the whole of Armenia together with the regions subject to Armenia itself.”59 Ammianus’s initial testimony thus seems to be contradicted by his later statements and by our remaining sources.

This ambiguity may not, however, be entirely Ammianus’s fault. Indeed, it may reflect a more general ambiguity in the treaty itself and a willingness on the part of both Persia and Rome to reinterpret this vagueness to their own advantage. As Robin Seager has shown, it is difficult to reconstruct the provisions on Armenia and Iberia from what we learn about the manner in which the treaty was implemented in the years that immediately followed it. In this period, both sides were only too willing to bend the agreement to their advantage, rendering their behavior a poor gauge of the specifics agreed upon. The ability of both sides to manipulate the treaty terms must, however, reflect a certain vagueness on the question of Armenia and Iberia to begin with.60

Ammianus seems right to argue first that Rome was debarred from defending the Armenian king, and later that it was forbidden to crown a successor. For this reason Valens was initially very hesitant to do either, expressly because it might violate the 363 agreement.61 Yet the treaty may only have (p.165) forbidden Roman interference without expressly sanctioning a Persian invasion. This would explain why Ammianus twice refers to Shapur’s attacks on Armenia as violations of the treaty.62 More important, though, once he was militarily able to do so, Valens did occupy Armenia and impose his own nominee there, Arshak’s son Pap.63 Indeed, by 377, Valens was protesting that his new arrangement should stand, since Rome and Persia had agreed that Armenia was to be free of external interference. As we shall see, this new specificity represented a temporary reshaping of the 363 agreement that Valens had coerced out of Persia in 371.64

On Iberia, the surviving sources mention nothing of what was agreed in 363. Themistius, however, states that Iberia sent an embassy to Constantinople in 365/66, apparently with crown gold for Procopius. This would seem to imply that Rome never lost its claim to suzerainty over Iberia.65 Here, it would seem, Shapur was the first to violate the treaty by invading Iberia and replacing its king. Once again, a separate agreement had to be reached to fill the silence of the 363 treaty after Valens had reasserted Rome’s claim to at least the western portion of this kingdom.66

The situation in Armenia and Iberia was thus ambiguous. It is certainly so to us and was apparently ambiguous to Valens and Shapur as well.67 Both exploited this ambiguity to gain as much influence over these kingdoms as possible, even when this meant distorting or contravening specific treaty terms.68 The maneuverings of Valens and Shapur make it difficult to reconstruct the treaty’s terms with respect to Armenia and Iberia, but they gain interest from what they reveal about mutual dissatisfactions with this compromise peace.

Indeed, to contemporary Romans, far from representing an acceptable entente, the 363 treaty was regarded as a shameful disaster. It had denied Rome military superiority in upper Mesopotamia, cost it an extremely productive (p.166) chunk of land,69 and belied Rome’s core political ideology of limitless power. Particularly significant was the loss of Nisibis, the “strongest bulwark of the east,” which had stood up to no fewer than three major Persian sieges in the thirty years preceding the treaty.70 When Jovian arrived there in 363 to surrender the city, his shame was patent: he refused to go within the walls personally, preferring to order the evacuation from a camp he established outside.71 The citizens, who protested that they would gladly defend their city alone, were threatened with death if they did not depart within three days.72 Ammianus and the deacon Ephrem—a citizen of Nisibis himself—both vividly recalled watching Shapur’s legate Bineses flying the Persian flag from the citadel, “so that this standard bearer would declare to the onlookers that the city was slave to the lords of that standard.”73 Shapur would soon transfer 12,000 Persians to Rome’s urbs inexpugnabilis in order to repopulate it as a bulwark of Persian defense.74 The abandonment of Nisibis provoked the violent protests against Jovian in Antioch mentioned in chapter 1 and quickly came to be regarded as the symbolic manifestation of a shameful peace.75

The sentiment was undoubtedly shared by Valens, who must have felt tremendous pressure to overturn the treaty when he occupied the eastern throne in 364. We shall see in what follows that this is precisely what his propagandists advocated, and precisely what Valens attempted to do: by 370, his ministers were directly lobbying for a violation of the treaty; by 371, he had reoccupied Armenia and half of Iberia and garrisoned both with Roman troops; shortly thereafter, he and the Armenians recaptured the satrapies of Arzanene and Corduene; and by 377, he was preparing for a larger invasion of Persia. At every step, then, Valens exploited the pax of 363 as a shelter behind which to hatch his plans. For his part, Shapur had also manipulated the treaty as he struggled to regain those territories that his grandfather had surrendered in 299 that remained in Roman hands. Later, (p.167) he also accused Valens of having infringed its terms. For both Valens and Shapur, the ink of the 363 treaty was thus never truly dry.

Despite the ultimate longevity of this agreement, then, it was hardly regarded as satisfactory in its immediate aftermath.It lasted so long not because it satisfied both sides, but because, as a mutual compromise, it had granted both sides a strong enough position that neither could gain the upper hand needed to overturn it. Only with time did the fragile balance of power it created solidify into lasting peace, and this only after the deaths of both Shapur and Valens. In the years of Valens’s reign, as we shall see, this compromise seemed neither viable nor acceptable. Rather, it left Persia and Rome locked in a border war for almost two decades.

Valens’s Relations with Persia, Armenia, and Iberia

On taking the throne, Valens’s first military ambition was to prevent a Persian attack on the eastern frontier.76 In the summer of 365, he headed east from Constantinople toward Antioch, whence he could have moved quickly to check any threat from Shapur.77 Valens’s resolve was apparently quite firm, given that, even when he learned that the Gothic Tervingi were planning a raid across the Danube that summer, he stopped only long enough to dispatch a pair of mobile units and then continued east.78 Ultimately, he never reached Antioch that summer, nor indeed for the next five years. Only in 370 did Valens finally make good on his initial plans and devote his undivided attention to the eastern frontier. This long delay in what first seemed like urgent plans may reflect the fact that, at least in these first few years of Valens’s reign, the eastern frontier remained more stable than some of our sources indicate.79

Indeed, Valens’s initial move toward the east was probably directed more at a potential than an actual threat.80 After Valens and Valentinian had divided the administrative and military machinery at Naissus and Sirmium in (p.168) midsummer 364, Valens returned to Constantinople, where he remained until the end of July 365.81 He probably had only vague information about Shapur’s activities on the Tigris, nevertheless had he known of urgent trouble, he would not have remained in Constantinople so late into the summer.82 Moreover, even after he undertook his journey eastward, Valens halted once again in Cappadocia to wait out the late summer heat.83 He did not yet know, therefore, of problems demanding immediate attention in 365. With the revolt of Procopius and the Gothic war, Valens continued to be detained west of the Bosporus for the next five years.84 The fact that he undertook the latter extended and arguably unnecessary war in late summer 366 indicates that he had not yet heard of a serious catastrophe in the east by this time either.

In these early years, it seems, Shapur implemented the 363 agreement by simply aligning his newly won territories under his rule, without necessarily invading them.85 He seems to have divided the new acquisitions in upper Mesopotamia between two governors (mobed) directly related to him—Adurad and Zamasp.86 This we know from hagiographic sources, which report that his governors began persecuting Christians inside these former Roman holdings in an effort to force conformity to Persian religious customs.87 Ammianus indicates that he also initiated efforts to win political allegiance by offering enticements or employing force to garner support from Armenian optimates et satrapas.88 The historian’s terminology here is precise, designating the Armenian nakharars more generally (optimates) and specifically the satraps of south Armenia, whose territories had just been ceded to him. The Armenian Epic Histories confirm that power was not simply transferred in these regions but that Shapur had to win them over through guile, threat, and eventually violence against the resistance of their local rulers.89 The same source also reports, however, that the Armenian forces initially held the shahanshah at bay. This would seem to confirm our intimations that major problems had not yet been reported to Valens by 366. In the years immediately (p.169) following the treaty of 363, then, Shapur was preoccupied with the implementation of its provisions in Mesopotamia and the south Armenian principalities. He was not yet able to make serious attacks on Roman territory or even to overrun the Armenian Midlands.

The account of events following 363 given in the Epic Histories is perhaps generally acceptable, but their exaggerated portrayal of Armenian victories over the Persians in these early years masks the more gradual progress Shapur made in gaining control of individual nakharars and satraps. Some of these even included princes beyond the territories to which he had legal claims. Thus he was especially fortunate to win the early defection of Meruzan Arkruni, the lord of Sophene, which technically remained under Roman suzerainty by the terms of the 363 treaty.90 With this defection, it seems, Shapur obtained the toehold he needed to begin laying waste to Rome’s remaining client satrapies. Probably in 367, while Valens was busy with his first Gothic war and King Arshak stood guard against a Persian assault on Armenia’s southeastern border, Shapur launched a devastating attack on the western flank of Armenia. He penetrated as far north as Ani (Kemah) in Akilisene, opposite Roman Satala (Kelkit) along the Euphrates. This fortress he plundered, capturing among other treasures the bones of the Arshakuni kings, which had traditionally been entombed there. His desecration seems to have been a gesture symbolic of what he planned for Arshak and his house.91 In this bold drive, Shapur had further overstepped the 363 agreement with his invasion of the Roman-controlled satrapies of Ingilene and the two Sophenes.

Shapur’s drive through Rome’s client principalities was probably meant less to provoke Rome than to surprise the Armenian forces under Arshak, who had been prepared for attacks at the opposite end of his kingdom. Indeed, only after they had already heard of Shapur’s raid in the west did Arshak and his sparapet Vasak Mamikonean hasten to move their forces from the Median border to the Armenian Midlands to forestall further encroachment. Although they successfully prevented an invasion of the central territory of Ayrarat, Shapur’s successes had been undeniable. In their wake, Arshak could no longer prevent most of his nakharars—who had grown disgusted with what had in effect been a state of active hostility with Persia since 337—from defecting to Shapur.92 By deserting their king, the nakharars, who constituted the backbone of the Armenian cavalry, deprived Arshak of (p.170) any hope of further resistance. In the face of impossible odds, Arshak was himself persuaded to go over to Shapur in 367.

The Epic Histories tell the tale of what ensued with characteristically legendary coloring. After fêting Arshak in traditional Sassanian fashion, Shapur is said to have set a test to prove the king’s newfound loyalty. He prepared a tent half of whose floor was covered with Persian soil and half with Armenian. When Arshak was led to the Persian soil, he protested his loyalty to Shapur, but once he returned to the Armenian, he slandered Shapur and promised to attack him. Apocryphal though the story may be, it demonstrates marvelously the underlying principles of Perso-Armenian relations. Indeed, the test constitutes a literary crystallization of what past experience had already shown to Shapur: Arshak’s promises of alliance were not to be trusted. In light of this knowledge, Shapur had Arshak blinded and thrown into his Prison of Oblivion, where he kept him into the 370s.93

After suppressing the Armenian monarchy, Shapur sent a Persian invasion force into Armenia and ordered it to besiege the fortress of Artogerassa. There Arshak’s wife, Paranjem, and son, Pap, were holed up with the Armenian treasure, defended by a troop of azats—nobles of a lower grade than the nakharars.94 Ammianus tells us that Shapur’s invasion force was led by the Armenian defectors Cylaces and Artabanes, and that Shapur had entrusted Armenia to these two.95 Based on this testimony, R. C. Blockley has argued that Shapur intended to replace the traditional Arsacid monarchy with a non-Arsacid but still Armenian nakharar dyarchy.96 The Epic Histories indicate that the situation was considerably more complex.97 These say that Shapur commanded two officials with Persian noble titles, Zik and Karen, to besiege Artogerassa and rule Armenia. They also speak of two Armenian nakharars, Meruzan Arkruni and Vahan Mamikonean, in leadership positions under Shapur’s suzerainty.98 A careful comparison reveals that the Epic Histories and Ammianus have intersections, but that any attempts to recover Shapur’s administrative scheme in Armenia based on the two is futile. The sources name up (p.171) to five figures with leadership roles: Zik and Karen, Meruzan Arkruni, Cylaces (who is probably to be identified with a figure the Epic Histories name Glak), and Artabanes (who may be identifiable with the Epic Histories’ Vahan Mamikonean). Given these confusions, we can say only that Shapur combined Persian administrators (marzban) with traditional Armenian aristocrats (nakharars) in a system that completely suppressed the Arsacid house. This system never had time to solidify, because the nakharars whom Shapur set up in positions of power—Cylaces/Glak and Artabanes/Vahan—did not remain faithful to him but played all sides against one another to their own advantage. Armenia was in flux, and Shapur was only beginning to experiment with administrative structures he thought he could control.

In the same year in which Shapur ordered the siege of Artogerassa, 367, he also seized Iberia, once again in contravention of the 363 treaty. He deposed its king, Sauromaces, the son of Rome’s ally Mirian III, and installed Sauromaces’ cousin Aspacures in his place.99 Meanwhile, the siege of Armenian Artogerassa continued through the winter of 367/68. In midwinter, Cylaces was able to initiate negotiations for the surrender of the fortress, during the course of which Paranjem, Arshak’s wife, appealed to Cylaces and Artabanes in the name of her husband. In typical fashion, the two Armenian defectors defected back to the Armenian throne and engineered a treacherous foray against the Persian forces they had been leading. In the event, the Persians were badly beaten, and Pap escaped to make his way to Valens, who was then wintering at Marcianople. Pap’s arrival was noted by Themistius, who mentioned the young prince in an oration delivered to Valens on March 28, 368.100 Valens eventually sent Pap back to Neocaesarea (Niksar) in Pontus Pole-moniacus, three hundred kilometers from the Armenian border. There the young king seems to have spent the rest of 368 enjoying “liberal support and education” while Valens prepared for a final year of campaigning against the Goths.101 Meanwhile, the siege of Artogerassa continued, and the Epic Histories inform us that Pap, in “the land of the Greeks,” was in communication with his mother inside the fortress, whom he encouraged to await his rescue.102

Unfortunately for Paranjem, Valens was willing to intervene and reimpose (p.172) Pap only in 369. The emperor had received legates from Cylaces and Artabanes requesting that Pap be returned as their king, along with a military force.103 Ammianus’s wording implies that Cylaces and Artabanes were not with Pap when they made their request, but inside Armenia. If their identification with the Epic Histories’ Glak and Vahan is correct, they did indeed remain in Armenia, acting nominally under Shapur’s direction during the siege of Artogerassa. Their communications with Valens show, however, that they were also working clandestinely for an Arsacid restoration.104 In the summer of 369, Valens responded timidly to their request ordering a Roman general, the comes Terentius, to escort Pap back into Armenia, but forbidding him to confer royal honors on the prince, lest this be judged a violation of the 363 treaty.105 When Shapur learned of Pap’s restoration, he became so incensed that he invaded Armenia personally, apparently in the autumn of 369. In the wake of this invasion, Pap was again forced into hiding somewhere in the southwestern Caucasus at the Roman frontier with Lazica. There he remained for the next five months, that is, until the spring of 370.106 Rather than hunt down Pap, Shapur laid waste to Armenia and concentrated his forces against the long-besieged Artogerassa. The fortress finally fell, after more than two years, in the winter of 369/70. The Persian forces captured the royal treasure and Arshak’s wife, Paranjem, whom they brutally raped until she was dead.107 They also captured and destroyed the Arsacid capital of Artaxata and a number of other urban foundations, former strongholds of Arsacid influence.108 When Shapur returned to Persia after these victories, he left behind ostikans—overseers—charged with holding the fortresses he had built or captured.109 He also began systematically persecuting Christians, forcing apostasy to Mazdaism, closing churches, and consecrating fire temples.110 He thus tightened (p.173) his grip on Armenia after his attempts to combine Persian and Armenian administrative structures had failed. Both politically and culturally, Armenia was to become Persian.

By late summer 369, Valens had finally brought his protracted Gothic War to a close, not least because of the desperate state of affairs in the east.111 He sent his magister peditum Arinthaeus to negotiate peace with the Goths in the late summer of 369 and then ordered him straight to the eastern frontier with a full military command.112 It probably was not until the following spring that Arinthaeus entered Armenia with his army and restored Pap for the second time.113

Arinthaeus’s arrival came none too soon, given that Shapur had contacted Pap, apparently still in hiding, and persuaded him to come over to the Persians. Under Shapur’s influence, Pap had murdered the duplicitous Cylaces and Artabanes and sent their heads to the shahanshah as a sign of loyalty.114 When Arinthaeus reached Armenia, he halted this rapprochement, and his presence in Armenia through the summer headed off a second Persian invasion. Shapur had to content himself with an embassy to Valens protesting this obvious violation of the 363 treaty.115 After this first Persian embassy, Valens also sent his general Terentius with twelve legions—about 12,000 men—to restore Sauromaces in Iberia. When this force reached the river Cyrus, Terentius and Sauromaces worked out an agreement with Sauromaces’ cousin Aspacures to divide the kingdom in two, with the river as the boundary. Aspacures indicated that he had considered defecting to Rome, but feared for the life of his son, who was by then a Persian hostage. The agreement was later approved by Valens, although not by Shapur, who regarded it as grounds for war. Even so, Valens did not hesitate to refer to this settlement later as the legal basis for his occupation of western Iberia.116 It was probably at this point that Valens also put two garrison forts on the Georgian coast and fortified the major passes of the Caucasus. This action may be at the heart of a confused passage in John Lydus to the effect that the Persians and Romans settled a quarrel over the terms of the 363 treaty by mutually agreeing to fortify the Caucasus pass(es) against the Huns.117

(p.174) Valens had thus finally met Shapur’s earlier violations of the 363 treaty with violations of his own. He had regained part of Iberia—to which Rome probably still had claims—and all of Armenia—to which it did not. He did so, not in the belief that he was still operating within the provisions of the Dura agreement, but because he now had the military might to enforce his occupation. His Gothic war had ended, he had already been able to send advance comitatensian forces eastward, and, as we shall see, his propagandists Festus and Eutropius had already begun openly advertising Valens’s intention of breaking the treaty. In the spring of 370, after a brief stay in Constantinople, he raced east himself and reached Antioch with his army by late April.118 Soon afterward, he moved out to Hierapolis, from where he could coordinate frontier operations, and for most of the following eight years, he operated on a similar pattern, wintering in Antioch and traveling east for the summer to coordinate the army from Hierapolis, Edessa, and even Caesarea.119 As long as he remained near the eastern frontier with his massive army, Rome retained control of the Armenian kingdom. However, although Valens continued to take measures to avoid open warfare, he stood in clear violation of the 363 accord.

Naturally, Valens’s recapture of Armenia and his dispositions in Iberia did not sit well with Shapur, who declared the peace void in the winter of 370 and promised an invasion the following year.120 The shahanshah began collecting an army and assembling support from his allies,121 and, as he had promised, in the spring of 371, he sent an invasion force into Armenia. Valens had dispatched his generals Traianus and Vadomarius to meet (p.175) the Persians, although he ordered them to defend rather than attack, apparently in hopes of breaking the treaty without breaking the peace. The Persians met Traianus and Vadomarius at Bagawan, a valley at the foot of Mount Npat, near the source of the Arsanias River.122 The Roman troops at first resisted combat, but they were eventually forced to respond to the Persian cavalry attacks and came off victorious.123 The Epic Histories paint a picture that gives considerable credit for the victory to Pap’s sparapet, Mushel Mamikonean, although they admit that Pap personally played no role but watched the battle from Mount Npat. It is impossible to determine the relative contributions of Armenians and Romans, although it is certain that the Armenians would have been hard-pressed to repel the Persians without Roman help. In the aftermath of the battle, several other engagements were fought, which initiated the process of reclaiming additional Armenian territories.124 According to the Epic Histories, these included the satrapies of Arzanene and Corduene, which had been ceded to Persia in 363.125 With these acquisitions and the capture of Armenia, Valens had broken the treaty on several counts.

At the end of the summer, a truce was settled, leaving Shapur to return to Ctesiphon and Valens to Antioch. It is by no means clear that Valens or Shapur regarded this agreement as a long-term accord in 371: it was hardly a treaty, and Ammianus states explicitly that both sides remained at odds even in its aftermath.126 As it turned out, though, this temporary settlement had the effect of sanctioning Roman control of Armenia for the rest of Valens’s reign. The sixth-century Greek historian John Malalas confirms this:

Valens did not fight himself, but marched out and made a peace treaty [. Valens and the Eastern Frontier].…The Persians had come and sued for peace. Thus, having arrived in Antioch in Syria with the greater part of the military forces on November 10 in the fourteenth indiction, Valens lingered there to conclude the peace treaty [. Valens and the Eastern Frontier] with the Persians. He negotiated a treaty for seven years, with the Persians suing for peace and ceding half of Nisibis [. Valens and the Eastern Frontier . Valens and the Eastern Frontier].127

(p.176) Some of the details had become hazy by Malalas’s day, but much of this information corresponds with what we know from other sources. Malalas’s notice that the Persians asked for terms seems logical, since it was they who had suffered setbacks. What Malalas means by the Persian return of “half of Nisibis” is vague but interpretable. It would have been difficult to relinquish half of a city, but Malalas may be implying that the Persians had given up half of the Nisibis accord, which, in effect, they had. The notice that the “peace agreement” lasted seven years corresponds precisely with Ammianus’s account, which portrays no open conflict on the frontier for the seven years between 371 and 378. In fact, Ammianus’s notice on events from early 372 directly confirms that peace with Persia had begun that winter.128

Other sources tell the same story. Socrates indicates that peace prevailed in 375, when Valens instituted severe measures against recalcitrant Nicenes, and Themistius confirms peace with Persia in the same year.129 This “peace” is best attested in Themistius’s eleventh oration (March 373): “[Valens is] sated with victory and even willing to put aside [war], but he is insatiable for the logos of philosophy. He more gladly concedes leisure to the soldier than silence to the philosopher.”130 In the same speech, Themistius notes that Valens had armies in the Caucasus, Iberia, Albania, and Armenia.131 The emperor had thus regained much of what the empire had lost in 363 and now held it with Roman troops under the sanction of a truce. Only the Armenian tradition informs us as to why Shapur allowed this. The Epic Histories report that the Kushans—a people on the northeastern edge of the Sassanian empire—occupied the shahanshah with war after his loss at Bagawan.132 The notice is credible, given that Ammianus himself reports how these same people had kept Shapur from attacking the Roman frontier in the early 350s.133 Whether Shapur was already aware that the Kushans had revolted when he sanctioned the truce of 371 is unclear. Ultimately, however, this conflict enforced a “peace” that allowed Valens to hold Armenia for seven years, in direct contravention of the 363 agreement.

(p.177) While peace prevailed with Persia, the situation inside Armenia began to crumble. As we have seen, in the period after the victory at Bagawan, Pap’s sparapet Mushel Mamikonean regained control over many of the nakharars who had defected to Shapur in the 360s.134 This was probably largely an Armenian concern. Themistius says that a Roman army remained stationed in Armenia, but the Epic Histories are probably right in saying that it was left to the Arsacid king and his sparapet to realign the nakharars under their authority. This is apparently what Themistius was referring to in 373 when he said of Valens:

For, without touching his sword, but using only the clemency of his spirit, he won back no small section of the neighboring barbarians who are normally hard to persuade and control, and he bound peoples more untrustworthy than the ancient Thessalians. So that, although they still quarrel with one another, they cooperate with and are in spirit with the Romans. Toward each other they follow their usual nature, but toward the emperor, they follow necessity. They have been defeated neither by spears nor by bows and slings but by the open patience in which they exist—a fact surprising to the listener.135

If we are to believe Themistius, Valens’s clemency had achieved a realignment of the nakharars and thus benefited Rome. The Armenian sources, by contrast, reveal that the nakharars had in fact been won back by the Arsacid sword, but with only limited success.

It seems safest to assume that the situation inside Armenia was far from settled. The young Pap was probably struggling to hold together a kingdom that Shapur had only just dismantled. His tenuous grip on power led him to measures that were less than judicious. Unfortunately, we have no firm chronological indicators for the dates of most of Pap’s activities. Pap had the powerful Armenian patriarch—and close Roman contact—Nerses poisoned at some point, perhaps late in 371.136 After doing so, the young king nominated a certain Yusik as a replacement and sent him for consecration to Caesarea, as was the custom for Armenian patriarchs. The Epic Histories tell us that the bishop of Caesarea—who would have been Basil—refused the nominee. In the meantime, Pap began systematically dismantling the religious (p.178) foundations that Nerses had established.137 Valens responded by having his general Terentius invest Basil with the authority to resolve the dispute over the ecclesiastical succession. A number of Basil’s letters regarding the incident survive.138 Apparently, Basil never succeeded in appointing an acceptable successor, in part because of Pap, in part because of the intrigues of Basil’s rivals Theodotus and Anthimus. This failure meant that, from the death of Nerses on, the Roman see of Caesarea ceased to claim its traditional authority over the consecration of Armenia’s patriarch.139 This loss of ecclesiastical control from inside the empire would certainly have disturbed both Terentius and Valens.

Insofar as Terentius was charged with the resolution of the episcopal problem, we might assume that he was the commander of the Roman army stationed in Armenia at this time. This would explain why he is one of only two Roman generals whose names survive in the Armenian sources for the period.140 It would also explain why it was Terentius who first began writing to Valens to complain of Pap’s activities. Terentius had allied himself with a number of the nakharar lords (gentilibus) whom, as Themistius indicates, Pap had disaffected through his heavy-handed tactics.141 Terentius complained of this to Valens and reminded the emperor that, in 370, Pap had murdered Cylaces and Artabanes and nearly defected to Shapur.142 Terentius was rigorously Nicene and probably did not take kindly to Pap’s murder of the Nicene patriarch Nerses or to Pap’s refusal to cooperate with Terentius’s personal friend Basil. Because Terentius was a staunch Christian, it is natural that Ammianus chose to play up his morose carping, but some weight must be given to the credibility of his complaints. According to the Epic Histories, Pap’s behavior was even more outrageous than Ammianus indicates: he began openly courting Persia and sent to Valens demanding control of Caesarea and ten other Roman cities, including Edessa, which he claimed were Arsacid foundations.143

(p.179) In response to this insubordination, Valens decided to arrest and execute the brash young Armenian king. He invited Pap to a meeting in Tarsus. Pap came with a 300-man mounted escort but, after arriving, became anxious when Valens would not receive him in audience. According to Ammianus, Pap learned that Terentius had secretly recommended that he be replaced, so as not to disaffect other nakharars or run the risk of Pap defecting to Persia. On learning of Terentius’s plan, Pap made a break for Armenia. In his race to escape, he threatened the local governor in Tarsus with death and attacked a legion that was sent to pursue him, thus making himself an open enemy of Rome.144 The young king continued east to the Euphrates, which he reached within two days. While he and his men struggled across, Count Danielus and the tribune Barzimeres, sent by Valens in pursuit with a force of Scutarii, outflanked Pap and crossed the river first.145 They set up roadblocks on the two routes Pap might take east, but did not manage to capture the fugitive king. A traveler who had seen the roadblocks helped Pap escape along an intermediary path between them. Valens’s assassination attempt had thus failed, and he knew well that he stood in grave danger of losing Pap’s loyalty and with it the control of Armenia.

Under considerable pressure to explain the debacle, Danielus and Barzimeres claimed that Pap had used magical powers to disguise his party and mask them in a cloud of darkness, under which they had escaped. The charge fits well with claims in the Epic Histories that Pap was reputed since birth to have been possessed of demons (dews), who dictated his erratic behavior. Daniel and Barzimeres seem merely to have capitalized on this reputation to save themselves from charges of incompetence. Given that at least Barzimeres kept his office, the story was apparently accepted.146 More important, (p.180) however, its significance stretches beyond this single incident. The stories of Pap’s demonic possession and magical powers bespeak a strong undercurrent of pagan sentiment in the young king. Taken with his murder of Nerses and his resistance to Roman intervention in the appointment of a patriarchal successor, these stories help confirm that Armenia’s peculiar version of Zoroastrianism was alive and well in the later fourth century, despite its nominal conversion to Christianity two generations earlier.147 As noted at the beginning of this chapter, religion was only one of many battlegrounds on which Persia and Rome competed for Armenian loyalty.

Pap’s escape probably occurred in 373 or 374. Upon his return to Armenia, he was received, according to Ammianus, “with the greatest joy by his subjects.”148 Ammianus also tells us that, despite the offenses he had endured, Pap remained loyal to Valens. Whether we can trust Ammianus, who favored Pap just as much as he hated Valens and his generals, remains in doubt. The Epic Histories make Pap’s position in Armenia look much less stable and his loyalty to Valens much less firm. At any rate, Valens was aware that amends had to be made and new plans laid to avert the loss of Armenia. He thus dismissed Count Terentius, who had engineered the bungled assassination plot, and turned over the army in Armenia to Traianus, with instructions that he should win his way back into Pap’s confidence.149 At the same time, Valens sent a secret missive instructing Traianus to have Pap killed as soon as he could.

Traianus convinced Pap that he had nothing to fear from Valens by showing him letters that indicated Valens’s favor. With this ploy, he won his way into the king’s feasts, important occasions for the social cohesion of the Armenian aristocracy. When Traianus invited Pap to a banquet of his own, Pap came with little reluctance, indicating that the general had indeed succeeded in gaining Pap’s confidence. This was no doubt because Traianus played his role well, as the circumstances of the feast would seem to confirm: the general organized the meal according to Armenian principles, at midday, with musical entertainment on strings and winds, and with the honored seat being reserved for the king himself.150 These amenities lured Pap into complacency, (p.181) making it easier for a barbarian guard to go in at a signal and murder him.151 The Quadic king Gabinius had been lured to his death in this way by Valentinian’s general Marcellianus, and Valens’s general Lupicinus would attempt similar tactics against the Gothic leaders Alavivus and Fritigern in 377. Ammianus draws an explicit parallel between the treacherous murders of Gabinius and Pap. To Valens, as to most of his contemporaries, such acts no doubt seemed as justified as they were expedient.152

Even so, the Epic Histories tell us, the Armenians were furious. However, as Valens had probably calculated, their rage would have been tempered by resignation to the reality of the current situation. The presence of a Roman army in Armenian territory probably left the Armenian nobility with little option but to accept Pap’s murder without resistance. They also accepted the Roman imposition of a replacement for Pap, an Arsacid prince named Varazdat.153 According to the Epic Histories, Varazdat, still a boy, ruled under the regency of the sparapet Mushel Mamikonean.154 The regency of a Mamikonean, a notoriously pro-Roman family, helps explain Armenia’s continued loyalty to Valens in this trying period. Moreover, the nakharars accepted what had happened because they still believed that they were best served by remaining Rome’s ally. In the Epic Histories, they are reported as saying: “We cannot become servants of the heathen Persians or be hostile to the king of the Greeks. Neither can we carry on hostilities with both of them. We cannot maintain ourselves without the support of one of them.”155 The Armenians knew they had to play one side against the other. Fortunately for Valens, despite his oppressive manipulations, Christianity proved a strong enough bond to decide the nakharars in his favor. Here again, religion constituted a key element in mediating the struggle for Armenian loyalty.

Pap’s murder occurred in 375. Naturally, it triggered a reaction from Shapur, who had been courting the king in hopes that he would defect to Persia. With Pap’s death and replacement by a Roman nominee under a pro-Roman regent, the shahanshah had to find other avenues back into Armenia. In the autumn of 375, Shapur reopened negotiations with Valens. He sent a legate demanding that Valens either evacuate Armenia, the “perpetual (p.182) source of troubles,” or withdraw from the part of Iberia where Roman troops now supported Sauromaces.156 Shapur, who had apparently ended his war with the Kushans, was thus demanding the return of half of what Valens had gained in 370 and 371. Valens responded that he could not renege on agreements made with the consent of both sides.157 A brief lacuna in Ammianus’s text mars an exact understanding of what followed, but it is clear that Shapur responded with a second embassy, this time in midwinter 375/76. Here, the shahanshah claimed that the discord could not be rooted out unless the original negotiators of the 363 treaty were present, an impossibility, given that some of them were already dead. This statement seems to add weight to the assumption that the spirit of the treaty and perhaps even its text were not entirely clear.158 Shapur was attempting to exploit the chinks in the treaty through which Valens had wormed his way back into Armenia to root him out again.

The following summer, 376, Valens sent an embassy of his own under Victor, the magister equitum, and Urbicius, the dux of Mesopotamia. These told Shapur that his demands for Armenia were unjust, given that its inhabitants had been granted the right to live according to their own decisions.159 They also informed Shapur that unless the Roman troops assigned to protect Sauromaces in western Iberia were allowed to return to Iberia unhindered, as had been agreed (ut dispositum est), Shapur would be compelled to do what he had refrained from doing on his own. This last clause was a threat to force Shapur into the war he had hoped to avoid through diplomacy. Valens thus raised the stakes by promising to reopen hostilities after five years of peace. As Ammianus notes, Valens’s confidence was boosted because he could now chose from an array of options rather than being compelled to diplomaticy (p.183) tergiversation.160 He had apparently learned that he would soon be able to fill the ranks of his army with auxiliary recruits from the Goths, whom he now permitted to settle in Thrace.161

Ammianus tells us that Valens based his firm stance in both Armenia and Iberia on previous agreements.162 Since Valens was plainly not referring to the 363 treaty, under which Rome had relinquished Armenia—although not Iberia—he must have issued his ultimatum on the strength of later accords: that reached about Iberia at the river Cyrus in 370, which, as Ammianus indicates, allowed the Romans to send troops to protect Sauromaces each year, and the truce reached at Bagawan in 371, which must have granted the nakharars the right to choose their own form of rule.163 This latter agreement would clearly have favored Rome, since the nakharars were generally inclined to maintain the Arsacid monarchy, which had for so long leaned—however tenuously—in Rome’s favor. Victor and Urbicius had thus presented Shapur with a hard-line stance based on what they could claim were Rome’s sanctioned rights. Shapur either had to accept the “previously agreed” Roman presence in Iberia and Armenia or go to war.

Fortunately for Shapur, the two legates had committed a blunder. During their diplomatic journey, they had improperly accepted offers from two regiones exiguae to be accepted under Roman suzerainty.164 This diplomatic infraction offered Shapur a new bargaining chip with which to revive negotiations. Later in 376, he sent a third embassy under the Suren offering the two territories illegally accepted by Victor and Urbicius in exchange for concessions. (p.184) Valens received the Suren graciously but sent him back with the same message: the Romans were unwilling to negotiate and would launch a tripartite invasion of Persia the following spring, 377. For this purpose, Valens was already mustering a huge army and assembling auxiliaries from the Goths who had entered Thrace that autumn.165 Given that this was the first time that Valens had had the strategic potential to seize back the Mesopotamian territory lost in 363, and given that he jumped at the opportunity, we can assume that he had been intent on invading Persia and restoring Roman territory for some time. He now had both the excuse and the military means to do so.

Shapur responded to Valens’s rebuff by ordering the Suren to seize back the two territories offered to Victor and Urbicius and to harass the troops who would be sent into Iberia in the spring of 377. As Ammianus indicates, Valens never had the chance to respond.166 News of the Gothic revolt in early 377 reached Valens sometime that spring, and he was forced to send a second embassy to Shapur, this time asking for a settlement rather than war.167 Once again, Victor conducted the negotiations, whose outcome is not reported. While he did so, Valens withdrew the units he had stationed in Armenia and sent them west to face the Goths in early 377. Later that summer, they were defeated in two engagements in Scythia Minor.168 The defeat of this army would necessitate Valens’s own journey to Thrace the following spring. He arrived at Constantinople on May 30, 378,169 and was killed in battle at Adrianople that August.

As we might imagine, the situation in Armenia collapsed after the withdrawal of the Roman forces in 377. The Roman nominee Varazdat, who was young and inexperienced, had been governing under the regency of Mushel Mamikonean, who had fought with the Romans at Bagawan and maintained strong Roman connections. So strong were Mushel’s ties that Varazdat came to suspect that he himself stood in danger of suffering the same fate as his predecessor. He thus eventually had Mushel killed, probably some time after the withdrawal of Roman forces.170 The power vacuum created by the (p.185) death of this sparapet was quickly filled by another Mamikonean, Manuel. Manuel had served under Shapur in the recent Kushan war, and his loyalty to Persia proved strong after he assumed the sparapetut’iwn. He eventually quarreled with Varazdat and expelled him from Armenia. While Varazdat sought refuge with Rome, Manuel began courting Persia. Together with Varazdat’s wife, Zarmandukht, and his son, Arshak III, Manuel established regency over what remained of the Arsacid dynasty and allied that dynasty to Persia.171 Shapur then sent the Suren with a 10,000-man army to garrison Armenia, much as Valens had been doing up to 377. Eventually, however, Manuel revolted even against Persia and defended Armenian independence from both powers down through the early 380s.172

Armenian independence was possible in a political climate where Rome was preoccupied with the Goths and the death of Shapur II in 379 and his heir Ardashir II four years later left the Sassanian house weak. Only after Manuel’s death was a settlement reached between Shapur III and Theodosius in 386 by which “Armenia was divided and split into two parts like a wornout garment.”173 Rome retained only the western fifth of the country, while Persia controlled the eastern four-fifths. After the death of Arshak III (387), Rome abolished the Arsacid monarchy and established the province of Armenia.174 Forty years later, the Sassanids followed suit, abolishing their branch of the Arsacid monarchy in “Persarmenia” by A.D. 428. Valens was thus the last Roman emperor to regain suzerainty over an independent Armenian kingdom. Although his settlement was always tenuous and crumbled immediately after he withdrew his troops, he had restored Armenia as a Roman client state for the last time in history.

Festus and Eutropius: Historical Propaganda and the Eastern Frontier

As his first Gothic war dragged on into 369, Valens must have grown increasingly agitated at the state of affairs on the eastern frontier. Shapur had occupied most of the former Roman territories and fortresses in Mesopotamia; he had deposed the pro-Roman kings of Armenia and Iberia; he had imposed a new king in Iberia, dismantled the Arsacid kingship in Armenia, (p.186) and begun eradicating Armenian Christianity. Shapur could do most of this under the authority of Jovian’s 363 treaty, and even what the treaty did not officially sanction, it tacitly allowed through its ambiguities. No Roman was under any illusion but that the 363 agreement had seriously compromised Roman power in the east and weakened Roman claims to world dominion. Once Valens brought his Gothic war to a close in late summer 369, he began making preparations for a strategic response to this new shift. Along with mustering troops and arranging supplies, Valens took care to establish the ideological framework for his challenge to Shapur’s new claims. This he did with the help of two successive magistri memoriae, Eutropius and Festus, whom he set to the task of composing separate abbreviated accounts of Roman history. A close examination of their breviaria gives some idea of how Valens hoped to use the power of historical discourse to his advantage as he undertook his attack on Shapur’s position.

Fortunately, we are able to date both breviaria with great precision. Since Eutropius accords Valens the epithet Gothicus Maximus in his preface, he must have finished his work after Valens claimed that title in late summer 369.175 In the same preface, Eutropius refers to himself as magister memoriae, a post that he ceased to hold in early 370. He must then have written shortly after the close of the first Gothic war.176 Festus succeeded Eutropius as magister memoriae in the period after Valens’s arrival in Antioch in spring 370.177 One of the oldest and best manuscripts of his breviarium indicates that he too was serving in this office when he dedicated his work. Since he held the post down to 372, he cannot have written later than this date,178 and Mommsen noticed that it is possible to offer even greater precision. The provincial lists cataloged in Festus omit the British province of Valentia, newly created and named after Valens himself in late 369. Festus must then have written before news of this had reached the east.179 If so, both breviaria can be dated around the period when Valens was preparing to confront Shapur directly in early 370.

(p.187) Abbreviated histories were very much a fourth-century vogue. They appeared sometimes as epitomes of earlier histories, sometimes as abbreviated accounts of a given period, sometimes as short biographies or lists of exempla.180 At least two other epitomes dedicated to fourth-century emperors survive, and one of these bears striking resemblances to the breviaria of Eutropius and Festus. The concise biography of Alexander and Trajan known to us under the misleading title Itinerarium Alexandri was dedicated to Constantius II. It was written for the eastern Augustus in the early part of his reign, probably in 340, by the well-connected courtier Flavius Polemius.181 Constantius was preparing a Persian invasion for that year, and Polemius openly states that his Itinerarium was designed to serve as an incitement to Constantius as he set out.182 It survives in a single manuscript, which breaks off before the narrative on Trajan has begun,183 but the extant text, with its hortatory introduction, its information on strategy and battle sites, and its choice of Alexander and Trajan as exempla of successful invaders of Persia, leaves no doubt about the intentions of its author. He was using history and historicizing rhetoric to influence the events of his own day.

Festus and Eutropius did much the same, although they differ from one another in the degree of emphasis they place on the east. Eutropius devotes great attention to the importance of military activity and territorial expansion throughout Roman history. He is especially careful to catalog affairs on the eastern frontier, but his decision to treat periods even before Roman expansion eastward makes it more difficult to identify a geographical center of gravity for his work. Perhaps for this reason, within months after Eutropius had completed his breviarium, Valens set Festus to compose a breviarium de breviario, which kept a much narrower focus on the east.184 Festus organized his history geographically rather than chronologically. Unlike Eutropius, he did not begin “from the foundation of the city” and was thus able to gloss over most republican history before Rome’s first encounters with Parthia. (p.188) He divided his narrative into three parts, set off by programmatic lemmata. These begin with a survey of Rome’s acquisition of provincial territories west of the Hellespont (4–9) and then move, with much greater emphasis, to the acquisition of eastern territories (10–14). This second part is introduced: “In obedience to your command, I shall now explicate the eastern territories and the entire Orient and the provinces placed under the neighboring sun in order that the zeal displayed by your clemency for the propagation of those same may be incited all the more” (10). A similarly revealing programmatic passage introduces the third and longest part (15–29), which enumerates the conflicts between Rome and Persia down to Festus’s own day.185 Thus more than two-thirds of Festus’s breviarium is devoted to the eastern empire, most of it to Romano-Persian military history. Lest there be any doubt as to his purpose in assembling this information, Festus closes with a valedictory wishing Valens a Persian victory to match his recent Gothic success.186 In both breviaria, then, and particularly in that of Festus, the purpose is beyond doubt: to employ historical discourse as a practical instrument to prepare for and justify a war against Persia.187

On the simplest level, this implied cataloging the essentials of warfare. Thus, Festus and Eutropius, like Polemius before them, offer statistics on troop strength and casualty figures in previous engagements.188 Much more important than practical concerns, however, were ideological issues. Late Romans were obsessed with the growth of their empire, both as a historical fact and as a contemporary desideratum. Imperial panegyrics gush with the rhetoric of conquest, forever encouraging emperors to extend the limites or praising them for doing so.189 Festus and Eutropius were also active participants in this discourse. With Festus, this is obvious in the schematic outline he offers of the growth of the empire. He prefaces this with an overview of “how much Rome expanded [quantum Roma profecerit]” under each of three “types of rule”: kings, consuls, and emperors (3). He then proceeds to offer (p.189) region-by-region summaries describing, “in what order the Roman state acquired its individual provinces.”190 These summaries culminate in precisely the territories in Armenia and Mesopotamia whose control Valens wished to dispute.

In the second half of his narrative, Festus follows a similarly schematic pattern, in which Roman territorial compromises are always followed by new conquests. Nero loses Armenia to the Persians, but Trajan regains it and forms new provinces (20). Hadrian abandons Trajan’s eastern provinces, but Severus creates them anew (20–21). Gallienus allows Mesopotamia to lapse, but Galerius and Diocletian regain it (23, 25). This cycle of loss and reconquest concludes with Jovian’s loss of Nisibis and Mesopotamia (29). By the pattern of history established in Festus, it fell to Valens to restore these territories.191 Themistius confirms that the emperor intended to do so in an oration of 368, where he names the reacquisition of Mesopotamia at the head of a wish list that includes the Gothic and Alamannic victories also anticipated by Valens and his brother.192

Eutropius, although less schematic, was equally concerned with Rome’s expansion. His early chapters chart Rome’s slow early growth as it conquered peoples twelve, then sixteen, then eighteen miles from the city.193 With the advent of republican magistrates, Eutropius notes an increase in the pace of expansion, which continued down through Augustus (2.1). This phenomenon was particularly remarkable under Julius Caesar, who added a region that extended “in circumference up to thirty-two hundred miles” (6.17), and Augustus, who gained control of a previously unprecedented number of territories and peoples (7.9–10). Augustus’s acquisitions were matched only by those of Trajan, whom Eutropius ranks above all emperors: “He extended far and wide the boundaries of the Roman Empire, which, after Augustus, had been defended rather than honorably enlarged.”194 With his Dacian (p.190) province alone, Trajan added a region whose circumference was a thousand miles (8.2). Little wonder that he and Augustus were, for Eutropius, the touchstones of imperial felicitas.195 They better than anyone fulfilled the most important requisite of imperial rule, expansionism.

By contrast, those rulers who failed to acquire territories, or, worse yet, those who actually lost them, were judged harshly. Tiberius was criticized for failing to campaign personally, Nero for nearly losing Britain and suffering setbacks in Armenia, and Gallienus for allowing Dacia to lapse.196 Worst of all, Hadrian actually chose to abandon Trajan’s new provinces of Assyria, Mesopotamia, and Armenia and wanted to desert Dacia as well, out of envy for Trajan (8.6). This shameful retreat was matched only by that engineered under Valens’s predecessor: “[Jovian] made what was, in fact, a necessary but shameful peace [pax ignobilis] with Shapur, for he was punished territorially and surrendered a certain portion of the Roman Empire. Before his time, this had never happened in practically one thousand and eighteen years since the Roman Empire had been founded.”197 The claim that Jovian’s withdrawal was unique in Roman history is contradictory, given that Eutropius has just described Hadrian’s abandonment of the same territories and Aurelian’s official renunciation of Dacia.198 Precisely because of its tenuousness, however, Eutropius’s claim draws attention to itself as a crucial element in the rhetoric of his breviarium. In fact, the claim probably represented the official position of Valens’s administration, which is also why Festus repeated it in his own breviarium, and Ammianus and Eunapius later took it up as well.199 Eutropius’s “official” version thus became normative for viewing the abandonment of Nisibis and the five trans-Tigritane territories. Placed under his spotlight, the event glared back as an egregiously (p.191) disgraceful blemish in Roman history: the treaty was unique and therefore unacceptable.

Hand in hand with the ideology of expansionism goes the ideal of triumphal rulership. Both Festus and particularly Eutropius were obsessed with this as well.200 Given that military victories constitute high points in history, they feature in all breviaria, but Eutropius lists no fewer than sixty-one. Nor was he content simply to relate victories in battle. Rather, he carefully catalogs actual triumphal processions,201 making special note of years with multiple triumphs, triumphs over new peoples, or significant features in triumphal processions.202 This obsession partly reflects the fact that Valens had just celebrated a Gothic triumph of his own, but it also betrays an assumption that successful rule can be equated with success in battle.

Nor was military success anywhere more glorious than in the east. Eutropius noted this when he called attention to the simultaneous triumphs of Marcus and Lucius Lucullus over Thrace and Pontus/Armenia respectively. The latter, he says, won greater glory for having conquered such powerful kingdoms (6.10). Indeed, all emperors must have regarded an eastern victory as the apex of success; Alexander the Great had paved a path to glory that every emperor yearned to follow. Some actually did so in a self-conscious, at times even maniacal, fashion. Caracalla was fortunately murdered before he could finish acting out his Alexander fantasy, and Julian’s more sober, although equally dangerous, Persian invasion was similarly tinged with imitatio Alexandri.203 Even more temperate sorts found the example of Alexander a powerful stimulus. With his Itinerarium Alexandri, Polemius kindled Constantius’s ambitions with direct comparisons between Alexander and his dedicatee and directly encouraged the emperor to subjugate all of Persia.204 So too, Festus and Eutropius used the examples of (p.192) those successful in the east to incite Valens. Festus offers a catalog of the campaigns of succeeding commanders and emperors in which he judges leaders based almost exclusively on their performance in the Orient.205 Similarly, Eutropius carefully lists all of the major figures who fought on the eastern frontier,206 laying particular emphasis on Trajan’s successes, to which he devotes an unparalleled number of chapters.207 Here, Eutropius, like the author of the Itinerarium Alexandri, makes a conscious effort to draw attention to the similarity between his patron and his great imperial predecessor. By setting Trajan’s Dacian campaigns “in those territories which the Taifali, Victoali and Tervingi now hold,” Eutropius emphasizes how Valens’s recent victory in the former Dacia mirrored Trajan’s own successes there.208 By extension, it implies the expectation of similar successes in the east to follow those of Trajan.

Most important, Festus and Eutropius provided Valens with practical guides to Rome’s historical claims to the territories of the east. With the rise of the Sassanians in the early third century, the Romans came to face an empire that based its aggressive stance toward Rome at least partly on the privileges of the past. From the beginning, Ardashir I seems to have claimed links with the Achaemenid Persians, whose empire had once stretched as far west (p.193) as the Aegean and even up to the river Strymon in Macedonia.209 Either he or his son, Shapur I, also began to argue that their Achaemenid background justified claims to these “ancestral” territories.210 A letter from Shapur II to Constantius that Ammianus preserves confirms that the claim remained very much a bone of diplomatic contention in Valens’s own day.211 Indeed, it probably colored the demands of Shapur II when he oversaw the rearrangement of the eastern frontier in 363. In the face of this historically determined justification for Sassanian aggression, Valens must certainly have understood the importance of historical discourse for his own ambitions.

Here Festus in particular would have been useful in helping Valens make his case. We have seen that his account offers a schematic outline of Rome’s eastern acquisitions. This overview (10–14) emphasizes the legal rights Rome had obtained through inheritance, conquest or the requests of indigenous peoples for Roman annexation.212 Armed with this information, Valens could promote the empire’s legal rights to territories that Shapur argued should be his. Of course, the ideal and the reality of Shapur’s claims were far from compatible. Although his letter to Constantius promotes an ancestral right to all of Asia, he had little practical expectation of actually winning this land. Thus, despite its rhetoric, Shapur’s letter, in fact, demands only the return of the territories on the Sassanian border surrendered by Persia under his grandfather in 299.213

These territories, upper Mesopotamia and Armenia, are treated in particular detail in Festus’s fourteenth chapter. To summarize his summary: Lucullus first conquered Armenia and the city of Nisibis,214 although he soon lost both; Armenia was later won back by Pompey, along with the territories of Syria and Phoenice. Trajan won Armenia, Mesopotamia, and Assyria and first extended the “eastern frontier beyond the banks of the Tigris [limes oreintalis supra ripas Tigridis]”; although these were ceded by Hadrian, Mesopotamia at least remained an object of dispute; it was won and lost again four times until Diocletian (read Galerius) defeated the Persians, established suzerainty (p.194) over the five trans-Tigritane territories, and restored the “frontier beyond the banks of the Tigris [limes supra ripas Tigridis]”; this peace had lasted down to the time of Constantius II.215 To drive home the point, Festus develops this synopsis even more fully in the second half of his history. Such an overview must have been invaluable to an emperor facing a Persian dynasty that pressed its own territorial ambitions under the guise of historical right. By providing Valens with a blueprint of Rome’s historical claims to upper Mesopotamia, Festus was justifying and reinforcing those claims.

Both Eutropius and Festus also note the origin of Rome’s right to crown the client kings of Armenia, Iberia, Panticapaeum, and the Albani and Arabs. The knowledge that Rome’s suzerainty over these client kingdoms stretched back to Pompey must also have been useful to Valens, who eventually chose to transgress the 363 treaty in these kingdoms of the sub-Caucasus.216 By the same token, Festus and Eutropius make no mention of Philip the Arab’s treaty of 244, which had ceded the right to crown the Armenian king to Persia. Other sources confirm Rome’s loss of Armenian suzerainty in the years after 244, but Festus and Eutropius omit the treaty altogether.217 If their omission was deliberate, it must have represented an attempt to present Valens with a stronger “historical” claim than history actually allowed.218

Eutropius goes even further in shaping history to meet the exigencies of Valens’s frontier policy. Not only was he obsessed with triumphs, but he offers a good deal of information on treaties as well. He mentions a total of fourteen treaty negotiations, often with explicit details of terms and obligations.219 Once again, this was certainly conditioned in part by Valens’s recent treaty with the Goths, for which comparanda might have been useful. Nevertheless, in his presentation, Eutropius gave special emphasis to Rome’s stubborn insistence on obtaining favorable terms and its staunch refusal to (p.195) ratify treaties that were unacceptable. Thus the Romans had refused Pyrrhus’s first peace offer in 280 B.C. because he would not evacuate Italy, and Scipio went to war with Hannibal in 202 B.C. when the latter balked at the peace terms offered by Rome.220 After the disaster at the Caudine Forks in 321 B.C., the Senate had refused to ratify the treaty terms agreed to by T. Veturius and Sp. Postumius, choosing instead to take the war back to the Samnites.221 When Calpurnius Bestia made an outrageous treaty (pax flagitiosissima) with the Numidians in 111 B.C., the Senate had rejected it outright and waged war on Jugurtha. So, too, after the Numantines had defeated Q. Pompeius and forced a shameful treaty on him (pax ignobilis) and later worsted C. Hostilius Mancinus and imposed the same on him (pax infamis), the Senate had once again refused to ratify their agreements and made war on Numantia in 133 B.C.222 It is precisely these exempla that Eutropius later cites when he condemns the latest pax ignobilis, Jovian’s 363 treaty.223 He thus sets up a scenario in which it is acceptable to refuse obedience to treaty terms when these are reprehensible. By his logic, not only would Valens be justified in refusing to abide by Jovian’s treaty; in keeping with the precedents of Roman history, he was obliged to break it.

Eutropius and Festus were thus engaging in a historical discourse with very real political and military implications. Writing for Valens in precisely the period when his policy was shifting from passivity to aggression, both helped to promote that policy with the rhetoric of aggression, expansion, and real-politik. Both concentrated heavily on the eastern frontier, whose history they schematized as an ongoing struggle by Rome to retain its just claims to territorial control established in 66 B.C. Both vaunted the glories of imperial expansion and derided past rulers who had allowed subject territories to lapse. And both—but especially Eutropius—emphasized that integral to Rome’s history was a stubborn refusal to accept any treaty that—because of its concessions—was a priori unacceptable. When Valens undertook his war against Shapur in 370, then, he had in his arsenal a powerful weapon in the (p.196) ongoing ideological struggle between Rome and Persia over the rights to Armenia and upper Mesopotamia.

Bandits and Barbarians of the East: Valens and the Maratocupreni, Isaurians, and Saracens

While he was defending what remained of Roman Mesopotamia, while he was upholding Roman interests in Armenia and Iberia, while he was keeping an eye on the Danube frontier and guarding against the omnipresent threat of usurpation, Valens was faced with a number of separate and equally dangerous security issues in the east. Some of these involved peoples residing well within what we conceive of as the established “boundaries” of Roman rule; others, peoples intertwined with its fringes, neither firmly within Roman control nor completely outside it. If we are fully to understand Valens’s policies on the eastern frontier, we must examine the history and causes of these conflicts between the imperial authorities and the various groups that inhabited Syria, Isauria, and Arabia.

In Roman Syria, Valens faced a security problem with a group of people who called themselves Maratocupreni and dwelt in a homonymous village near Apamea. The name Maratocupreni, from the Syriac m’arta (cave), implies that the bandits used the cavernous hillsides north of Apamea to shelter themselves from reprisals by civic and imperial authorities.224 Ammianus, our best source for their activities, describes just one of their schemes: they formed a band, disguised themselves as an imperial official (rationalis) and his retinue, and plundered the house of a wealthy Syrian under the pretense that he had been proscribed. With this and other such plots, the Maratocupreni amassed considerable wealth, which, Ammianus says, they used to establish luxurious homes and settle their families.225

Ammianus’s placement of the notice in his narrative indicates that these raids became particularly acute around the time of the revolt in Britain between 367 and 369. Although scholars have been reluctant to assign a firmer date to the apex of their banditry, a reference in Themistius may confirm that news of an increase in troubles reached Valens in early 368.226 In light of this news, Valens apparently felt that the Maratocupreni had become threatening enough to require imperial intervention. Ammianus reports that they were exterminated by an imperialis motus, probably the movement of (p.197) Valens’s army into Syria in early 370. Libanius later described the executions in graphic terms: mothers with babes in arms thrust into the flames.227 Valens apparently wanted to eliminate the problem root and branch.

Such tares are not, however, so easily extirpated. Prior to this incident, the urban centers of Syria had been unable to check the bandits on their own. Indeed, banditry had regularly presented problems in the region, even long before the arrival of Roman power in the first century B.C.228 The rocky terrain of the limestone massif north of Apamea was ideally suited to sheltering bandits, and locals regarded raiding as an acceptable occupation. Not surprisingly, then, such raids crop up again in the fifth-century sources. They could no more be ignored than the problems with Persia, since, although smaller in scale, they were often more immediate in their impact on provincials.

A more profound security threat came from a bandit people who were much better organized and far more destructive, the Isaurians.229 The territory of Isauria, on the southern coast of Anatolia, was home to an indigenous culture dating to Hittite times that remained remarkably resistant to assimilation by larger empires.230 Here, as in Syria, the presence of mountains, in this case the precipitous peaks of the Taurus, provided refuge from central authorities. Indeed, the terrain of the Isaurian hinterland was so rugged that it afforded its inhabitants refuge equivalent to a giant natural fortress.231 Even so, literary and archaeological evidence indicates that the problem of Isaurian banditry had been brought under heel in the first two centuries A.D.232 During this period, the Isaurian elite appears to have become largely sedentarized and to have assimilated itself to the provincial elites in neighboring regions. Unfortunately for Rome, this calm was uncommon for the region and ultimately short-lived. Following Shapur I’s invasion of the eastern empire in 260, an invasion that carried a Persian division into the Isaurian Taurus, the problem seems to have flared up with renewed vigor.233 The wealth and social organization afforded by Roman rule in the region had transformed Isauria’s elite into a much more powerful force. (p.198) Thus, by the fourth century, Isaurian troubles had grown to such an extent that the Romans no longer faced isolated, small-scale raids, but an organized, perhaps confederated enemy in what amounted to open warfare.234

Thanks again to Ammianus, we have particularly good evidence for Isaurian activities in the mid fourth century.235 Already in the second chapter of his extant narrative, Ammianus reports a major uprising in 353 that affected territory stretching into Lycia and Pamphylia in the east and Lycaonia in the north.236 In this revolt, the Isaurian raiders were eventually able to lay siege to the provincial capital of Seleucia, which, like other urban centers surrounding the hinterland, remained in Roman control.237 Despite resistance from civic and limitanean garrisons stationed in the region, the highland rebels were only forced back into the mountains with the arrival of imperial field troops.238 In 359, they broke into revolt once again and were only placated after a settlement was brokered by the inhabitants of Germanicopolis, a small city in the hinterland that apparently served as a nerve center for bandit coordination.239

Finally, Ammianus tells us that in the first years of Valens’s reign, a third large-scale revolt broke out, which was able to gain momentum because Valens failed to send imperial troops to quash it in its early stages.240 Eunapius reports that the situation was bad enough that Musonius, the vicarius Asiae, felt compelled to lead a band of lightly armed civic police (diogmitae) against the raiders.241 These were, however, hardly prepared to contend with fighters as experienced as the Isaurians and were easily overwhelmed and slaughtered. Only later, when Valens sent comitatensian units into the region, were the bandits once again driven back into the highlands and forced to negotiate a peace. Shortly after this, the Isaurian governor constructed a fortified harbor at Corasium, about ten kilometers north of the mouth of the Calycadnus, apparently in anticipation of future troubles.242 At the same time, Cilicia’s governor dedicated three statues to Valentinian, Valens, and Gratian in Tarsus; similar dedicatory bases are found in Iconium and Pisidian (p.199) Antioch.243 All were apparently offered in thanksgiving for the emperors’ aid in suppressing the Isaurians, a good measure of how wide an area was affected. A number of indications date this last revolt to 367–68, which would explain Valens’s tardiness in sending forces, because he was engaged with the Goths north of the Danube in 367 and would have been better able to spare troops when floods prevented his planned Danube crossing in 368.244

The eastern emperor could thus not ignore this major security issue festering in the southern reaches of Roman Anatolia. When the Isaurians attacked, local limitanean units—of which there were at least three—and civic police forces were unable to squelch their raids.245 The provincials surrounding the Isaurian hinterland were thus forced to rely on imperial intervention to put down these uprisings. Moreover, once the Isaurians had been driven back into the highlands, even the imperial authorities were unable to eliminate these raiders but were forced to settle peace accords with them as if they were a foreign people.246 Banditry had turned to war on such a scale that it transformed a part of southern Anatolia into a militarized zone that regularly diverted imperial troops and attention.

Ammianus mentions no more Isaurian uprisings after this 367–68 revolt. Zosimus, however, describes an uprising around 375, and Eunapius’s fragments seem to confirm problems at this time.247 While some have questioned the validity of these notices, independent testimony for a revolt in the second half of this year survives in some previously unnoticed passages from the letters of Basil of Caesarea.248 Taken together, this evidence confirms that the 375 event was again more like war than simple raiding. The Isaurians apparently reached as far east as Lycia and as far north as central Anatolia; there they actually cut off communication routes between east and west during the winter of early 375.249 Valens was once again compelled to send imperial mobile units, apparently under the direction of the general Saturninus, to combat the Isaurians from both the northern and southern approaches to the Taurus.250

(p.200) As in the earlier cases, the arrival of these imperial forces seems to have settled the 375 uprising quickly, but Valens’s men apparently felt compelled to remain in the region to maintain the peace and oversee the rebuilding of the road into Iconium, the northern base of operations against the Taurus.251 The presence of these units in Isauria in precisely the year when the Persian frontier was beginning to become volatile makes Ammianus’s silence about this uprising all the more puzzling. Valens was confronted with a problem that forced him to commit comitatenses who might otherwise have been used profitably on the Persian frontier. Since they were retained in the region down to 377, the Isaurian problem also affected troop availability when the time came to resettle the Goths in Thrace in 376.252 This obviously sheds new light on Ammianus’s testimony about the shortage of manpower to oversee the Gothic immigration. The internal security threat posed by Isauria had grown so great that it could distract Valens from his efforts against foreign powers, first Persia and then the Goths.

Similar difficulties faced Valens in yet another marginal region, the Arabian Desert.253 Desert territory abutted the Roman provinces of Syria, Arabia, and Palestine on the east and south. Even so, eastern Syria and much of Rome’s province of Arabia lay within the 200 mm p.a. isohyet, which allows for dry farming and thus settled agriculture. This permitted the Romans to continue the process of sedentarizing and urbanizing the Arab peoples, which was already well advanced when they arrived. Even just beyond the isohyet, Arab peoples regularly settled in oasis communities, the most famous of which was Palmyra, a powerful ally of Rome by the second century A.D. Because of its strategic location in the eastern desert of Syria, Palmyra was able to control not just trade routes and military operations against Persia but nomadic Arab groups as well.254

Indeed, despite the remarkable degree of sedentarization around the edges of the desert, pastoral transhumance remained a common way of life in the region. For centuries, pastoral nomads, the Bedouin, had been moving their herds of sheep, goats, and camels between the cultivable edges of Syria and Arabia and the deserts east and south of these. Not surprisingly, after the Romans arrived, the nomads continued to migrate in and out of provincial territory and to maintain contact with the sedentarized communities (p.201) whose territory they passed through.255 Some degree of symbiosis between nomads and sedentarists is documented in the Roman period, but tensions always remained a feature of Romano-Arab relations.256 After the sedentary stronghold of Palmyra was suppressed in 273, these tensions intensified as new concentrations of power coalesced around the Bedouin. Not surprisingly, at precisely this time, a new word for these nomads first comes into common usage in the Greek and Latin sources, Saracen.257 A recent and credible hypothesis locates the root of this word in the Arabic sirkat, meaning precisely “confederation.”258 It would seem that, with the power vacuum created by the reduction of Palmyra, the Bedouin began to unite in broader associations that came to challenge Roman authority on the desert fringe.259

At the same time that the name “Saracen” comes into common currency, we also begin to see more powerful leadership structures emerge among the nomads. Bedouin tribes were ruled by sheikhs, some with thousands of followers.260 From the late third century, however, we find evidence that these sheikhs began to unite under more powerful rulers who styled themselves kings—Arabic malik—and claimed hegemony over broader swaths of territory and broader groupings of tribes. Although fiercely guarding their own independence, these kings often maintained treaty relations with Rome or Persia. Our best evidence for such chiefs comes from an inscription found at Namara that records the burial in 328 of Imru al Qays, son of Amr ibn Adi.261 We know from the Arab writer al-Tabari (838?–923) that Amr ibn Adi headed a confederation called Lakhm, which was centered around Hira in southern Persia and had been allied with the Sassanians in the third century.262 The Namara inscription indicates, however, that by the early fourth century, the Lakhmids had moved into Roman territory and shifted their allegiance to the empire. In the inscription, Imru al-Qays boasts of being “the (p.202) king of the Arabs,” a claim he substantiates with reports of military successes stretching from Syria deep into the Hejaz. He also reports that he “dealt gently with the nobles of the tribes, and appointed them viceroys and they became phylarchs for the Romans.” The inscription thus provides evidence of central authority established under a dynasty that initiated allied relations with the Roman Empire and conducted wide-ranging military operations.

Nor were Imru al-Qays and his Lakmids unique. Ammianus refers to a malik whom he calls Podosacis phylarchus Assanitarum. The Assanitae are almost certainly the Ghassanids, the Saracen group most closely allied to Byzantium in the sixth century. In the fourth century, by contrast, Podosacis and his Ghassanids were friends of Persia. Indeed, Ammianus tells how they had earlier made many raids into Roman territory and how they harassed Julian’s army during its retreat from Persia in 363.263 In addition to Lakhm and Ghassan, confederations of Tanukh and Salih, both allied to the Romans, are reported in other sources for the period. Considerable efforts have been made to pinpoint the geographical locus of these groups and something of their political history, yet only the barest outlines have emerged.264 What is clear, is that the Saracen nomads had coalesced into powerful confederations under strong leaders, some allied with Rome and others with Persia. Even despite their alliances with the major powers, these groups always maintained sufficient independence to pose a threat to either empire when they chose.

The threat posed by the Saracens is clear from a number of sources. Already in the late third century, a Latin panegyrist briefly alludes to a Saracen uprising around Syria that occurred in 290 and was large enough to demand the intervention of the emperor Diocletian.265 In the years that followed, Diocletian restructured the old Arabian province by shrinking it to a more manageable size and reassigning much of its southern territory—Sinai and the Negev—to the province of Palestina III.266 He also began construction of the massive defensive system referred to in contemporary sources as the Strata Diocletiana.267 This fortified highway stretched from around Callinicum on the Euphrates to Azraq in the south, that is, it followed the 200 mm p.a. isohyet that marked the eastern limit of Roman territorial control along the desert. Considerable recent debate has been focused on the nature and purpose of this defensive system, but it remains reasonable to assume that it was (p.203) aimed at hampering Saracen attacks. The construction of defensive structures across the desert frontier throughout the early fourth century indicates that further conflicts continued to occur.268 Even where slight textual evidence of large-scale warfare can be found, then, other literary and epigraphic sources confirm a constant threat from smaller-scale raiding.269

Nor is direct testimony of broader conflict in the early fourth century entirely lacking, although the same evidence often indicates cooperation as well. Julian’s First Oration, for example, reports serious problems around Syria ca. 337.270 Later in the same speech, however, Julian indicates that Constantius had engaged the same Saracens in diplomacy and eventually employed them against the Persians, probably in 338.271 By 354, the Arabs were once again attacking Roman territory, as Ammianus reports.272 Yet in 363, Julian was met near Callinicum by a group of Saracens who offered him crown gold and joined his expeditionary forces as auxiliaries during his Persian campaign.273 When, in the course of that campaign, Julian refused their demands for tribute, these Arab allies broke away from his army and hampered its retreat up the Tigris. Indeed, it is likely that these angry Saracens were actually responsible for killing Julian.274 The Saracen tribes neighboring Roman territory thus acted alternately as friend and foe. They were neither entirely outside nor entirely inside Roman territory and were thus neither entirely outside nor inside the sphere of Roman authority. Instead, they constituted a powerful autonomous group allied to Rome in a tenuous relationship that at times benefited the empire and at others hurt it gravely.275

Most have assumed that in the immediate aftermath of Julian’s failed Persian expedition, Romano-Saracen relations quickly simmered down. These have failed to take account of two letters from Libanius to Ulpianus, governor (p.204) of Arabia in 364. In one of these, Libanius expresses surprise that Ulpianus has time to write on top of dealing with provincials and battling against barbarians.276 Although the reference is vague, it probably alludes to continued tension with the Saracens in the year after they had been snubbed by and perhaps murdered Julian. Even so, at some point early in his reign, Valens must have reestablished concord. By 375, Themistius could visit the eastern frontier and report that the Scenitae—Themistius preserved the older name for the Bedouin—were at peace.277

Peace, of course, never meant a cessation of tensions. Many building inscriptions record the construction of forts in the region under Valens: one marks a tower at Khirbet es Samra (ca. 367/75), a second, the rebuilding of the castellum on the Strata Diocletiana at Deir el Keif (ca. 368/71), a third, an unspecified military installation at Dibin east of the Dead Sea (ca. 368), and a fourth, fifth, and sixth (ca. 368 and 371), the erection of burgi in the area around Umm el Jimal.278 This intensive activity indicates that Valens felt the need to continue fortifying Rome’s desert fringe against Saracen attacks. The Notitia dignitatum records the establishment of two units in the province of Palestina under Valens and two more in Arabia, further confirming Valens’s defensive efforts.279 Given that the new structures and garrisons were part of Valens’s and Valentinian’s larger program of frontier defense, they need not imply a response to specific conflicts.280 Rather, they were representative of the ongoing effort to control smaller nomadic raids and protect settled populations that abutted on and even shared Saracen territory.

At some point toward the end of Valens’s reign, ecclesiastical sources tell us, the king281 of Rome’s Arab allies282 died, leaving only his wife, Mavia, to rule as queen.283 In the aftermath of this succession crisis, a major conflict arose, which led Mavia into open war with the Romans.284 Unfortunately, as (p.205) with the Isaurian revolt of 375, Ammianus reports nothing on Mavia. Our sources for the event are entirely Christian and are consequently preoccupied with affairs related to Christianity.285 Although we cannot deny the religious implications of this uprising, we must also be careful to look beyond them for nonreligious elements that our sources were less interested in treating.

The extent of Mavia’s revolt is reported in Rufinus: “Mavia began to disturb the towns and cities of the Palestinian and Arabian frontier and at the same time to lay waste to neighboring provinces.”286 Sozomen further clarifies: “[T]hey plundered the cities of the Phoenicians and Palestinians as far as the regions of Egypt lying left of those who sail toward the source of the Nile, which are generally named Arabia.”287 The raids thus reached a vast territory stretching from the province of Phoenice in the north, through Arabia and Palestine all the way deep into the Sinai. What appears to be a contemporary account of the events, the Relatio Ammonii, confirms that the attacks reached as far south as Mount Sinai itself.288 The Relatio records the martyrdom of forty monks who were killed there by the Saracens at some point during the exile of Bishop Peter of Alexandria,289 that is, between 373 and 378. The Saracens had risen up, reports the Relatio, after the death of their phylarch,290 almost certainly the husband of Mavia. These Saracens beset the anchorite communities around the holy mountain and killed all but a handful of monks, who held out in a tower until a firestorm scared away the raiders. The uprising was thus massive and affected a broad swath of the southeastern frontier from Phoenicia deep into the Sinai.

Mavia’s uprising was also apparently quite well organized and dangerous. (p.206) Sozomen, who offers the most detail on the events, reports that the limitanean troops stationed in the region could not handle it by themselves. The comes (?) of Palestine and Arabia thus had to call in the magister equitum et peditum per Orientem for assistance.291 Given that Julius is attested in this latter office in 371 and again in late 378, it is likely that he is the officer to whom Sozomen refers.292 According to Sozomen, the magister equitum et peditum attempted to meet the Saracens in open battle but was nearly overwhelmed. Indeed, he escaped the fray only with the aid of the commander of Palestine and Arabia, whom he had earlier ridiculed for his inability to suppress this female adversary. The defeat was so noteworthy that, according to Sozomen, the locals continued to talk of it and the Saracens to memorialize it in song into the mid fifth century.293 Mavia’s revolt was thus a major military event. It must have involved a broad Arab confederation and certainly occupied a large number of imperial units.

Indeed, if we are to believe our Christian sources, it was ultimately the Romans who felt compelled to call for peace when their efforts to defeat the Saracens by force had failed. Mavia granted this only on the condition that the Saracens receive as bishop a local holy man named Moses, who was to be consecrated by Roman authorities.294 When Moses himself refused appointment at the hands of the Arian patriarch Lucius of Alexandria, he was taken into the desert to receive his office from bishops whom Valens had exiled. To further cement the treaty, the general Victor, himself a barbarian, took in marriage the daughter of Mavia.295 The success of this arrangement is evident from the events that followed. In the aftermath of the peace, the Saracens willingly sent auxiliary troops with Valens to Thrace to fight against the Goths in 378.

Saracen participation in Valens’s forces in 378 is confirmed not just by our ecclesiastical sources, but also by Eunapius, Zosimus, and even Ammianus, who, as noted, mentions nothing of the revolt itself.296 Themistius too mentions Arab participation in the imperial army in Thrace as late as 382.297 This would seem to indicate that peace prevailed among the Saracens in (p.207) the early 380s. That the Saracens of the Sinai were quiescent in this period is also confirmed by the testimony of the pilgrim Egeria, who was able to travel along the Sinai desert coastal road unmolested in 382, albeit with military escorts.298 By 389, however, we learn from Pacatus’s panegyric that Saracen resistance had once again erupted in Arabia in the recent past. Based on some references in Libanius, Irfan Shahid has dated this revolt to 383—the year after the Gothic war in Thrace ended.299 One cannot help but wonder if the continued retention of the Saracen auxiliaries in Thrace between 378 and 382 played a role in provoking problems once these auxiliaries were returned.

Indeed, the issue of supplying auxiliary manpower to the Roman armies may have been at the root of Mavia’s uprising as well. Given the absence of explicit testimony, most have avoided attempts to pinpoint a cause for Mavia’s revolt. Following the ecclesiastical sources, however, Shahid argues at length that Mavia’s Saracens had long been Christian, and that these broke into revolt because Valens attempted to force Arianism on them by having their choice for bishop, Moses, consecrated by the Arian patriarch Lucius.300 This analysis has several flaws. First, there is no evidence for widespread conversion among the Saracens by this date.301 Indeed, the ecclesiastical historians trumpet Mavia’s tribesmen precisely because they were supposedly the first Saracens to become Christian. Although we do know of earlier instances of conversion by smaller groups of nomadic Arabs, we also know of many more examples from considerably later.302 Moreover, the same sources on which Shahid bases his case make it quite clear that Mavia’s Saracens converted after Moses’ consecration, at the end of the revolt.303 It is highly unlikely, then, that Nicene Christian piety led Mavia to rise up in the first place. Indeed, the treaty with Mavia that worked out the terms of conversion seems to have stipulated that Moses be consecrated under Roman authority; in the circumstances, this meant by an Arian bishop. It was Moses’ own choice, not Mavia’s, to refuse consecration by Lucius, and, according to the ecclesiastical sources, Moses was disturbed not by Lucius’s Christological stance but by the violence the Alexandrian bishop had perpetrated against fellow Christians.304 (p.208) Mavia herself was probably even more indifferent to questions of doctrine. Rather, she favored Moses because he was himself of Saracen birth and because he had had earlier associations with her tribesmen, near whom he lived.305 Religion was thus not so much a cause for Mavia’s revolt as an element in its settlement.

I would contend, rather, that the major factor in provoking the revolt was Valens’s own insistence that the Saracens send military auxiliaries with him to fight the Goths in Thrace. Here the date of Mavia’s revolt is crucial. Although there has been much speculation, Glen Bowersock has rightly noted that the most reliable chronological indicator for the revolt can be found in Socrates and Sozomen, who agree that it occurred when Valens attempted to leave Antioch for Thrace to fight the Goths.306 Bowersock assumes this means the spring of 378, since Valens’s is known to have journeyed from Antioch to the west beginning around April of that year. It seems more likely, however, that the revolt began the previous winter. The conflict outlined in Sozomen must have occupied considerably more time than Bowersock allots: it began with widespread Saracen attacks all across the southeastern frontier,307 led to two separate campaigns, one by local and the second by imperial troops, eventually demanded a diplomatic settlement, which entailed a journey to the desert to arrest Moses,308 a second journey with Moses to Alexandria, and a third journey back into the desert for Moses’ consecration, and was crowned by the marriage of Victor to Mavia’s daughter. The beginnings of this sequence must have stretched back into 377, which is precisely the year when Theophanes dates the event.309 Moreover, the Relatio (p.209) Ammonii dates the massacre of the monks of Sinai to December 28.310 The revolt, which was probably already in full swing by the time the monasteries of the Sinai were attacked, must then have begun sometime in the winter of late 377.311

A date in the winter of 377/78 would be entirely in keeping with the theory that Valens provoked the uprising with demands for Saracen auxiliaries to fight in his Gothic war. Already in the summer of 377, the emperor had learned of the defeat of his advance forces in Scythia and had sent Victor to strike a peace with Persia so that he could mobilize his own army. After Victor’s return, Valens probably made his first effort to move west, late in 377. As he prepared to leave, he no doubt attempted to draw on the Saracen federates, much as Constantius had done in 338 and Julian in 363. This latter long-range expedition had been extremely unsatisfactory for both sides. The anxiety of a new expedition, this time to the unfamiliar territory of Thrace, apparently provoked an uprising among a people already in turmoil after their king had died without an heir. It would be logical that Valens made his first efforts to mobilize in late 377, since he probably hoped to deal with the Goths as quickly as possible. But because of this Saracen revolt, his efforts were delayed. Indeed, as we shall see, he had probably intended a strike against the Goths much sooner than he was eventually able to muster one. Had he been able to arrive earlier, he might have hit the Goths—battered in their first major engagement in summer 377—with a blow hard enough to force an end to their uprising. Instead, he was tied down with the Saracens until the following spring.312 The net result was the reconstitution of an expanded Gothic force and the consequent disaster at Adrianople.

We have thus seen that Valens was forced to deal with three peoples who shared environmental and cultural similarities: the Maratocupreni, Isaurians, and Saracens. All lived in marginal terrain that afforded them protection and isolation from Roman authority; all had inhabited that terrain since the arrival of Roman power in their territory; and all operated both inside and outside spheres of Roman influence. The tensions stirred up by the clash between Roman and indigenous elements in these marginal areas created an ongoing cycle of violence. Romans met indigenous violence with violence (p.210) of their own in a deadlock where neither side was capable of gaining absolute dominance. When these bandit groups were small enough and weak enough to be crushed by imperial forces, every effort was made to crush them, as with the Maratocupreni. Yet, although this response eliminated one of the organized groups disturbing Syria, it hardly ended the problem permanently, since the Syrian massif quickly sprouted new bandits where the old had been pruned back. In the case of Isauria, the fourth century witnessed the “frontierization” of an entire region. There, Valens confronted an indigenous population that was already so powerful that it could hardly be crushed, although it could be driven back into the highlands by imperial forces. By the fifth century, the Isaurians had gained enough independence to supply the empire with auxiliaries like other foreign peoples.313 Yet even in the fourth century, they were already threatening enough to derail military operations in other regions by draining off troop strength when they chose to rise. In Valens’s reign, this menace affected both his ability to deal with Persia and his efforts to resettle the Goths in Thrace. The Saracens, too, posed a threat that occupied Roman military authorities on an ongoing basis. Already in the early fourth century, they had formed fully articulated confederations, which Rome treated as autonomous federate groups. As such they were often tapped for Rome’s military operations, which allowed them to increase their power and thus to grow increasingly independent of Roman authority. The force they could unleash was great enough to cripple Valens’s army in the face of the Gothic revolt and help delay his arrival in Thrace until the Gothic problem had spiraled beyond his control.


(1) . For the topography and historical geography of the region, see Oates 1968, 1–18; Frye 1984, 1–20; Kennedy and Riley 1990, 24–46.

(2) . On Romano-Armenian relations in the first three centuries, see Chaumont 1969; 1976; Redgate 1998, 65–113. On Romano-Iberian relations, see Braund 1994.

(3) . On nakharars, see Toumanoff 1963, 112–41, 192–252; Adontz 1970, passim, esp. 183–371; cf. Hewsen 1996; Redgate 1998, 97–102.

(4) . On the Sparaptetut’iwn, see Bedrosian 1983.

(5) . E.g., BP 3.8–9, 4.23, 5.8–19; MX 3.4–6.

(6) . Tac. Ann. 2.56: “Ambigua gens ea antiquitus hominum ingeniis et situ terrarum…maximisque imperiis interiecti et saepius discordes sunt, adversus Romanos odio et in Parthum invidia.” For escorts, see AM 20.11.3, 27.12.9, 30.1.5; Proc. Bella 1.5.16. On military support, see AM 23.2.2; Zos. 2.51.4; Lib. Or. 18.215; Soz. 6.1.2–3.

(7) . On gifts and exemption from tribute, see AM 20.11.1–2; CTh 11.1.1; Proc. Bella 2.3.35. On marriage, see AM 20.11.3; BP 4.15; MX 3.21; Vita Nerses 8 p. 29; cf. Garsoïan 1969; 1989, 285–86 n. 40. Similar enticements were offered to both the Armenian and Iberian kings in 361, AM 21.6.8. On hostages, see BP 4.5, 11, 15, 55. For more on the client king’s position and role at the frontier, see Braund 1984, 91–107.

(8) . AM 20.11.2, 27.12.3, 14, 16, 30.2.1; BP 4.16, 19, 21, 5.37–38; Proc. Bella 1.5.14–16.

(9) . Garsoïan 1976, 6–8 esp. n. 17 and ead. 1980.

(10) . Asdourian 1911, 86–87; Garsoïan 1976; Redgate 1998, 81–83, 104–12.

(11) . For 312, see Malalas 12.46–47; cf. Eus. HE 9.8.2–4. For 337, see Jul. Or. 1.18d–19a, 20a–21a; cf. BP 4.20. For the 350s, see BP 4.11, 16, 20; MX 3.25; Vita Nerses 8 p. 31; cf. Garsoïan 1969. None of these sources is undisputed. For recent bibliography, see Garsoïan 1989, 290 n. 8; Dodgeon and Lieu 1991, 380 n. 22, 383 n. 2.

(12) . On Trdat’s conversion, see Kettenhofen 1995, 92–104, 163; cf. Redgate 1998, 113–19. On Armenian Zoroastrianism, see Russell 1987, 173–211. Early Armenian Christianity retained many Zoroastrian elements, and fourth-century nakharars often apostasized; cf. Russell 1987, 126–30.

(13) . Redgate 1998, 119–38.

(14) . Garsoïan 1967 and ead. 1969.

(15) . Braund 1994, 253–59.

(16) . Toumanoff 1963, 84–103, 141–44.

(17) . RGDS ll. 2–3; Braund 1994, 239, 245.

(18) . Petrus Patricius fr. 14 (FHG 4.189). For the survival of Roman suzerainty through 361, see AM 21.6.7–8, and see n. 65 for the period of Valens’s reign.

(19) . Thelamon 1981, 85–91; Braund 1994, 239, 246–61.

(20) . Oates 1968, 73–92; Birley 1972, 115–7, 129–35; Millar 1993, 121–41.

(21) . For the growing consensus on the early development of strong centralized state structures, see Adams 1981, 200–214, 246–47; Gnoli 1989, 129–74; Gyselen 1989; Lee 1993, 15–20. The classic study on Sassanian Persia remains Christensen 1944, to which Frye 1984, 287–339, adds useful new material. On Romano-Sassanian relations, see now Dodgeon and Lieu 1991 and Blockley 1992.

(22) . On Shapur’s campaigns, see Kettenhofen 1982; Millar 1993, 159–73. Sources at Lieu and Dodgeon 1991, 49–67. On Armenia, see n. 85. On Iberia, see Braund 1994, 240–44.

(23) . Winter 1988, 130–45; Millar 1993, 177–79. Sources at Dodgeon and Lieu 1991, 125–31.

(24) . The treaty terms are reported at Petrus Patricius fr. 14 (FHG 4.189). On the date, see Barnes 1976a, 182–86. For the extensive bibliography on the treaty, see Winter 1988, 152–215, and id. 1989, to which add Blockley 1992, 5–7.

(25) . Rome had already restored an Armenian king, whose son, Trdat III (r. 287–98), regained much of Armenia from the Sassanians; see Agathangelos 1.46–47; MX 2.82, with Redgate 1998, 94–95; cf. Chaumont 1969, 93–111.

(26) . Lieu 1986, 491–95; Rugullis 1992, 105–15; and Lee 1993, 62–64 argue that this last measure may have been designed to limit cross-border espionage.

(27) . For the locations and history of these territories, see Dillemann 1962, 110–12, 116–26; Winter 1989, 563 nn. 47; Dodgeon and Lieu 1991, 377 n. 48, and above all ASX v.22.ii–iii, v–vi, with Hewsen 1992, 153–62, 168–76.

(28) . AM 18.9.2.

(29) . Roman and Greek sources refer to these territories as gentes or . Valens and the Eastern Frontier, and to their leaders as satrapae; cf. Festus 14; Eutr. 8.3; Zos. 3.31.1; AM 18.9.2; CTh 12.13.6; CJ 1.29.5; Nov. Just. 31; Proc. Aed. 3.1.17. On the status of these satrapies, see Güterbock 1900; Toumanoff 1963, 131–37, 170–75; Adontz 1970, 25–37; cf. Matthews 1989, 51–57.

(30) . There were a total of four of these viceroys, who controlled the Arabian, Syrian, Median, and Iberian marches (Asorestan, Aruastan, Nor Sirakan and Moskhia). R. H. Hewsen (1988–89; 1990–91) overturns the widely held theory, first introduced by Markwart 1901, 165–79, that Petrus Patricius’s groupings of “Ingilene along with Sophene” and “Arzanene with the Corduenes and Zabdicene” (fr. 14 = FHG 4.189) correspond to the Syrian and Arabian Marches.

(31) . AM 21.6.7 speaks of “Transtigritani reges et satrapae”; cf. Proc. Aed. 3.1.18–23. See Güterbock 1900, 34–39, and Toumanoff 1963, 133–35, 154–92 (but cf. also 170–71, 176), on the legal and political status of the satraps of these regions, who continued to cooperate with the Arsacids even while under Roman dominion between 298 and 363.

(32) . On troops, see Güterbock 1900, 36–37. Pace Dillemann 1962, 218–20, Rome still claimed fifteen fortresses beyond the Tigris in 363 (cf. AM 25.7.9; Zos. 3.31.1). Archeological remains of at least two—at Tilli and Hiksan—have been discovered (cf. Mitford 1986; Light-foot 1986; Algaze 1989, 244 with figs. 12, 14, 33). These were probably garrisoned by native, not Roman forces.

(33) . CTh 12.1.6 (a. 387). See more on crown gold at p. 289.

(34) . Sources in Dodgeon and Lieu 1991, 164–230. Analysis at Warmington 1977; Frye 1984, 308–16; Blockley 1988; 1989; 1992, 12–24; Matthews 1989, 41–66; Portmann 1989; Seager 1997, 253–62.

(35) . Sources on Julian’s Persian campaign in Dodgeon and Lieu 1991, 231–74. Analysis at Bowersock 1978, 106–16; Matthews 1989, 130–79; Blockley 1991, 24–30; Barnes 1998, 162–65. The Sassanians were said by Libanius (Or. 18.305; 24.18–19) to have depicted Julian as a lightning bolt.

(36) . On the date, see p. 14.

(37) . Rome also lost Caracalla, Gordian III, Valerian, and Carus.

(38) . On the dearth of supplies, see AM 25.7.7, 8.1, 15; Zos. 3.30.5, 33.1; Lib. Or. 18.278, 280; Festus 29; Eutr. 10.17; Greg. Naz. Or. 5.12; Philostorg. 8.1; August. De civ. Dei 4.29; Pass. Artemii 70; Soc. 3.22.5–9; Soz. 6.3.2; Ruf. HE 11.1; Zon. 13.4; Joh. Chrys. De S. Babyla 22 (PG 50.569). The supply problems continued when Jovian reached Antioch; cf. Theod. HE 4.4.1–2. On the effects of Shapur’s attacks, see Eutr. 10.17; cf. Lib. Or. 18.264–67.

(39) . The sources agree that Shapur initiated negotiations; cf. Festus 29; AM 25.7.5; Zos. 3.31.1; Lib. Or. 18.277; Malalas 13.27 (p. 336); Chron. pasch., p. 553; Theod. HE 4.2.2; Ruf. HE 11.1; Tabari 842–3 (p. 62 Bosworth).

(40) . AM 25.7.7–8; cf. Malalas 13.27 (p. 336); Chron. pasch., p. 553; Lib. Or. 18.277–79. Blockley 1984, 30, 34, 37, argues that Jovian’s legates were inexperienced negotiators and that the negotiations were rushed, leaving the treaty ambiguous. Lee 1993, 40–48, however, demonstrates that professional negotiators did not exist in the ancient world: Sallustius and Arinthaeus, both experienced officers, were typical and quite appropriate diplomatic liaisons. Moreover, as Ammianus and Libanius indicate, four days was actually quite a long time, given that Jovian and Shapur were camped in close proximity, eliminating the need for lengthy embassies. The treaty was indeed ambiguous, but more because it was a compromise than because it was poorly constructed.

(41) . Lib. Or. 18.222–23, 232, 250; Tabari 843 (p. 62 Bosworth).

(42) . AM 25.7.2. Cf. Lib. Or. 18.214; AM 23.3.5, 25.8.7, 16; Zos. 3.12.5, 4.4.2; Malalas 13.21, 27; Soz. 6.1.2. For debate on the exact numbers of men involved, see Hoffmann 1969–70, 1: 304–8.

(43) . For the treaty as a successful compromise, see Honigmann 1935, 5; Dillemann 1962, 223–24; Wirth 1984, 364–69; Blockley 1984, 28; Isaac 1998, 442. Even so, the contemporary Themistius (Or. 5.66a–c [a. 364]) certainly overstates the mutual goodwill of Persians and Romans in 363, as a later oration proves (cf. Or. 5.65a with 9.124a). For a similarly rosy picture, see Greg. Naz. Or. 5.15; Theod. HE 4.2.2; Malalas 13.27 (p. 336); Ruf. HE 11.1.

(44) . AM 17.5.6, 25.7.9; cf. 26.4.6; Lib. Ep. 331; Zon. 13.9. Note the legal language Ammianus uses to describe Shapur’s recapture of Armenia, AM 27.12.1: “iniectabat Armeniae manum ut eam…dicioni iungeret suae”; cf. 26.4.6. Manus iniectio was the civil law term for the personal execution of a properly adjudicated debt.

(45) . AM 25.7.9–14; Zos. 3.31.1–2. Various pieces of the agreement are also reported at Them. Or. 5.66a–c; Philostorg. 8.1; Pass. Artemii 70; August. De civ. Dei 4.29; Chronicon miscellaneum, p. 134; Eutr. 10.17; Festus 29; Lib. Or. 1.134, 18.278, 24.9; Eunap. Hist. fr. 29.1 (Blockley) = Suid. I 401; Greg. Naz. Or. 5.15; Joh. Chrys. De S. Babyla 22 (PG 50.569–70); Soc. 3.22.6–7; Soz. 6.3.2; Theod. HE 4.2.2–3; Ruf. HE 11.1; Oros. 7.31.1; Malalas 13.27 (p. 336); Theoph. a.m 5856; Zon. 13.14.4–6; Jer. Chron. s.a. 363; Chron. pasch., p. 554; Tabari 843 (p. 62 Bosworth); BP 4.21; Agathias 4.25.7; Joh. Ant. fr. 181; Jord. Rom. 306; Josh. Styl. Chron. vii–viii. Julian’s death and the 363 treaty may be represented in a rock relief at Taq-i Bustan; see Azarpay 1982. Previous bibliography on the treaty is summarized at Blockley 1984, to which add Blockley 1987 and Seager 1996.

(46) . Ammianus makes it clear that Shapur’s rhetoric of reconquest did not jibe with reality; cf. AM 25.7.9: “Petebat autem rex obstinatius, ut ipse aiebat, sua dudum a Maximiano erepta, ut docebat autem negotium, pro redemptione nostra quinque regiones Transtigritanas.” Cf. Blockley 1984, 29; Dillemann 1962, 219–20. Dillemann 1962, 217–20, rightly argues that much of what Shapur gained de iure in 363 he had already retaken de facto; cf. Winter 1988, 176–78; 1989, 558–59. In the case of these satrapies, this was as much due to the shifting loy-alties of their princes as to military or political successes on Shapur’s part. Thus, for example, the satrap of Corduene Jovinianus is said to have been under Persian suzerainty in 359 (AM 18.6.20), but his father must have been a Roman client, because he had sent Jovinianus as a hostage to Syria. By 363, Ammianus (25.7.8; cf. 24.8.5) indicates that Corduene had turned back to Rome, and the “Passion of Mar Sabha” (Hoffmann 1880, 23) confirms this. Corduene was thus favorable to Rome in the 330s, to Persia in 359, and to Rome again in 363. The satrapy of Sophanene was equally quick to shift allegiance in 502 (cf. Proc. Aed. 3.2.1–10). Under such circumstances of constant flux, it is difficult to say that either Rome or Persia had firm control; it is at least true, however, that Rome made every effort to keep alive its claims up to 363, as indicated at AM 19.9.2, 21.6.7, 22.7.10, 25.9.3, 12; cf. Festus 14, 25; Ephrem CJ 2.19–20, 4.15.

(47) . Blockley 1984, 41 n. 19; Winter 1988, 174 n. 2; 1989, 556 n. 10. On the complex divisions of the territory of Corduene (Arm., Korduk’), see Hewsen 1992, 170–76. See also ibid., 344 for, the suggestion that the elusive territory of Rehimene should be localized in the Zarva River valley.

(48) . Zos. 3.31.1, who explicitly names only the regions that Rome relinquished—as opposed to those which Persia gained—omits Moxoene. This may indicate that Moxoene was never claimed by Rome in 299. It was, after all, some distance north and east of the Tigris (Hewsen 1992, 168–69) and thus quite remote from Roman territory. Contrast Winter 1988, 174 n. 1; 1989, 556 with n. 9, summarizing earlier opinions, and yet another reconstruction at Hewsen 1990–91, 161–63.

(49) . Festus 14, writing seven years after the 363 treaty, confirms that contemporary rhetoric emphasized the exchange of “quinque gentes trans Tigridem constitutas” as central to the treaties; cf. AM 25.7.9 “quinque regiones Transtigritanas.” On this point, see Adontz 1970, 36; Blockley 1984, 31–32. Hübschmann 1904, 220 n. 3; Chaumont 1969, 123, and Winter 1988, 174–76; 1989, 557, argue that nine territories were ceded in 299 and only the five retroceded. This assumes too strict a correlation between the 299 and 363 treaties and disregards the fact that the territorial units ceded in 299 subsumed smaller territories. This was true not only of Corduene (see n. 47) but of “Sophene,” which technically included Greater Sophene (or Sophanene), Ingilene, Anzitene, and Lesser Sophene, all of which were later treated by Rome as independent satrapies; see ASX v.22.ii, with Hewsen 1992, 154–55; cf. Adontz 1970, 27–35.

(50) . Shapur had already made significant strides in regaining the fortresses of Upper Mesopotamia; see AM 20.7.1–16, with Dodgeon and Lieu 1991, 214 (Bezabde); AM 18.10.1–2 (Busan and Reman); AM 19.6.1–2 (Ziata); cf. AM 18.6.3–5 (Amudis).

(51) . On the preservation of the Nisibenes, see AM 25.7.11; Ephrem CJ 2.26; Theoph. a.m. 5865; Zon. 13.14; Malalas 13.27 (p. 336); Chron. pasch., p. 554. Most were transferred to Amida, which Valens rebuilt; see Zos. 3.34.1; BP 3.10; Historia S. Ephraemi 10 col. 26; cf. Chrysos 1976, 27 n.1. Under other circumstances, Shapur would have been happy to transfer the inhabitants of Nisibis to Persian territory, a fate that befell the inhabitants of the fifteen fortresses in the trans-Tigritane satrapies (Zos. 3.31.1; cf. Lieu 1986, 476–87, 495).

(52) . Singara had been captured in 360 (AM 20.6.1–8) and its inhabitants led off into slavery. Thus, Ammianus’s notice (25.7.11; cf. Zon. 13.14) that Jovian was able to rescue the inhabitants of Singara from Persian captivity seems puzzling. Given this explicit testimony, it is likely that the force Julian sent to Mesopotamia in 363 had regarrisoned the city. Ephrem CJ 2.15 says that Julian planned to rebuild Singara (cf. AM 23.5.18), and evidence that he did so may be found in the Notitia dignitatum. AM 20.6.7–8 tells us that Singara was defended by the legio I Flavia and legio I Parthica in 360 when Shapur overcame the fortress and captured these. The Not. dign. or. 36.29 registers a legio I Parthica Nisibena at Constantina (Tella). Julian’s Mesopotamian force may have reconstituted the legio I Parthica at Singara from inhabitants of Nisibis. Lightfoot 1988, 108–9, offers a different solution to this conundrum. On Singara, see Kennedy and Riley 1990, 125–31.

(53) . Honigmann 1935, 3–16; Sinclair 1987–89, 3: 367; Blockley 1984, 35; 1992, 27. Cf. Dillemann 1962, 223–40, with fig. 31.

(54) . Festus 29: “ut Nisibis et pars Mesopotamiae traderetur”; Jer. Chron. s.a. 363: “Nisibin et magnam Mesopotamiae partem Sapori regi tradidit”; cf. Oros. 7.31.2; Jord. Rom. 306; August. De civ. Dei 4.29; Proc. Bella 1.17.25.

(55) . On the thirty-year term, see Lib. Or. 24.20; AM 25.7.14; Zos. 3.31.1; Philostorg. 8.1; Cedrenus, p. 539. Them. Or. 5.66a–c, 69b, seems to have had some sense of the value of this compromise already in early 364. For the Romano-Persian wars of the fifth and sixth century, see Greatrex 1998.

(56) . AM 25.7.12.

(57) . It has been taken thus by Chrysos 1976, 33–35; cf. Blockley 1987, 225.

(58) . AM 27.12.10, 29.1.3; cf. 27.12.15, 18.

(59) . Zos. 2.31.2: . Valens and the Eastern Frontier; Lib. Or. 24.9: . Valens and the Eastern Frontier; cf. 1.134; Chronicon miscellaneum, p. 134; BP 4.21: “I give you Mcbin [Nisibis] and I am withdrawing from the Armenian Midlands. If you are able to attack and subject them, I shall not support them.” “Armenian Midlands” (Mijnasxarh Hayoc’) refers to the central region of Greater Armenia, the core of Arsacid power (cf. Garsoïan 1989, 481). Following the German translation of M. Lauer, Blockley 1987, 224, and Chrysos 1976, 33, have mistranslated the phrase as “half of Armenia,” with serious consequences for their arguments.

(60) . Seager 1996, clarifying Blockley 1984, 36–38; 1987; Chrysos 1976.

(61) . AM 27.12.10, 29.1.3.

(62) . AM 26.4.6, 27.12.1, 11

(63) . AM 27.12.13, 29.1.2–3. Blockley 1987, 225, argues that the Roman advance up to Bagawan by 371 proves that the Romans were allowed to garrison Armenia up to Bagawan, obvious petitio principii. The Romans occupied the territory by force, not necessarily by right.

(64) . AM 30.2.4. On the rearrangement of treaty terms in 371, see pp. 175–76.

(65) . Or. 7.92a. It matters little whether the ambassadors were genuine or imposters; Themistius’s intention is clearly to demonstrate support from the provinces and client states known to have been under Roman jurisdiction. Christensen 1944, 238, and Toumanoff 1963, 150 n. 5, 360, 460–61, argue that Shapur did gain claims to Iberia in 363. Contrast Chrysos 1976, 45–8. None noticed the reference in Themistius.

(66) . On Shapur’s invasion, see AM 27.12.4. On Valens’s restoration, see AM 27.12.16–17. Cf. Braund 1994, 260.

(67) . This seems to be at the heart of Shapur’s statement (AM 30.2.3) that the disputes over the treaty could not be resolved in the absence of those who had actually struck it in 363.

(68) . On the exploitation of ambiguities, see Köhler 1925, 76–77; Blockley 1984, 36; 1987; 1992, 34–36.

(69) . Matthews 1989, 48–51.

(70) . AM 25.8.14: “orientis firmissimum claustrum.” On the importance of Nisibis in the eastern command structure, see AM 14.9.1, 18.6.8, 20.6.1. Cf. Ephrem CJ 2.18, 25–26, 3.3; Joh. Chrys. De S. Babyla 22 (PG 50.569–70). See also Dilleman 1962, 221–23; Lightfoot 1988; Pollard 2000, 71–81, 286–87.

(71) . AM 25.8.17; Zos. 3.33.2; Malalas 13.27 (p. 336); Chron. pasch., p. 554.

(72) . AM 25.8.13–9.6; cf. Malalas 13.27 (p. 336); Historia S. Ephraemi 8 col. 23.

(73) . CJ 3.1; cf. AM 25.9.1. Malalas 13.27 (p. 336) and Chron. pasch., p. 554, name the Persian legate Junius. Bineses may be the Persian commander Vin who, according to BP 4.26, invaded Armenia in the mid 360s.

(74) . Tabari 843 (pp. 62–63 Bosworth). Jul. Or. 1.27a–b suggests that Shapur planned to re-populate recaptured Roman cities as early as 350.

(75) . For the protests in Antioch, see p. 18. For the subsequent reaction to the loss of Nisibis, see Turcan 1966.

(76) . For secondary studies of Valens’s relations with Persia, Armenia, and Iberia, see Baynes 1910, 636–42; Asdourian 1911, 155–62; Seeck, Untergang, 58–69; Köhler 1925, 70–94; Solari 1932c; Nagl, “Val.,” 2113–17; Stein 1959, 186–87, 205–6; Piganiol 1972, 175–77; Blockley 1975, 62–72; 1987; 1992, 30–39; Demandt 1989, 118–19; Redgate 1998, 132–39; Greatrex 2000, 35–41.

(77) . AM 26.6.11; Zos. 4.4.1, 13.1–2; Soc. 4.2.4; Philostorg. 9.5. For the chronology of all that follows, see Lenski 1995b, 499–522.

(78) . AM 26.6.11–12; see p. 116.

(79) . Zos. 4.4.1 indicates that the Persians had initiated attacks on . Valens and the Eastern Frontier in the first year of Valens’s reign. Ammianus (26.4.6) also implies that the Persians invaded in the year after Julian’s death, although this rhetorical passage is not to be trusted for purposes of chronology (see pp. 141–42). See Blockley 1992, 189 n. 25; cf. Blockley 1987, 224–25.

(80) . Soc. 4.2.4–5 and Soz. 6.7.10 indicate that Valens suspected the Persians but that they remained quiet.

(81) . Seeck, Regesten, 225, with 33, 37. Cf. AM 26.6.11: “consumpta hieme.”

(82) . Lib. Or. 12.73 indicates that Valens would have been seventy days’ journey from the eastern frontier in Constantinople. On the problem of intelligence at the eastern frontier, see Lee 1993, 109–28, who demonstrates that large invasions were generally learned of in advance even if tactical movements were difficult to determine.

(83) . See p. 77.

(84) . Zos. 4.10.1: . Valens and the Eastern Frontier.

(85) . Similarly, it took Shapur I eight years after winning suzerainty over Armenia at the battle of Misik to invade the region and depose its Arsacid ruler; cf. Chaumont 1969, 49–66; 1976, 167–76; Kettenhofen 1982, 38–43.

(86) . Hoffmann 1880, 23–24 (“Passion of Mar Sabha”).

(87) . BHO 174–76, 1031, with Fiey 1977, 166–70; cf. Historia S. Ephraemi 10 col. 26.

(88) . AM 27.12.2; cf. 27.12.12.

(89) . BP 4.21–43, 45–49. Garsoïan rightly cautions against assigning too much credence to BP on these Armenian resisters (1989, 295 n. 1).

(90) . BP 4.23; cf. MX 3.29, 35–37. On Meruzan as satrap of Sophene, see Garsoïan 1989, 391–92.

(91) . BP 4.24; cf. MX 3.26, 28; cf. Garsoïan 1980, 43–45. BP 17–18 confirms that Shapur continued to retain the loyalty of Ingilene, Greater Sophene, and Anzitene until after 371.

(92) . BP 4.50–51. BP speaks of a 34-year war with Persia at 4.50, but of a 30-year war at 4.20, 54. Proc. Bella 1.5.10, whose source was BP, speaks of a 32-year war. The terminus post quem is clearly the rise of hostilities at the end of Constantine’s reign (see Garsoïan 1989, 291 n. 19).

(93) . BP 4.52–54; cf. AM 27.12.3; Proc. Bella 1.5.13–29; Vita Nerses 10 p. 33. MX 3.34–35 locates Arshak’s capture after Valens’s death in his narrative. AM 27.12.3 implies that Arshak was murdered shortly after his capture, but BP 5.7 (cf. MX 3.35; Vita Nerses 10, p. 33; Proc. Bella 1.5.30–40) relates that Arshak committed suicide with the help of Drastamat after the K’usan war of ca. 372/75. The latter version, although embellished in the Epic Histories, seems preferable; one can imagine the value of keeping Arshak alive, especially if the treaty of 363 specifically granted Shapur control over him (see p. 164).

(94) . AM 27.12.5 with BP 4.55, which gives the figure of 11,000 men; cf. MX 3.36. On the status of the azat, see Christensen 1944, 111–13; Toumanoff 1963, 124–27; Adontz 1970, 213. Arshak may have rebuilt Artogerassa during his reign; cf. BP 4.19, with Garsoïan 1989, 289 n. 6.

(95) . AM 27.12.5.

(96) . Blockley 1987; cf. Baynes 1910, 638 n. 71.

(97) . See app. B.

(98) . BP 4.55, 58–59, 5.1.

(99) . AM 27.12.4: “eiusdem…gentis”; cf. 27.12.16: “consobrinus.”

(100) . Or. 8.116c. Seeck, Untergang, 448 n. 26, associates this reference with Pap. Hoffmann 1978, 314–16 (cf. Heather and Matthews 1991, 23, 31. n. 5) believes rather that it refers to the Iberian prince Bacurius. I favor Seeck’s reading, but either interpretation supports an attack on the sub-Caucasus in this year and the flight of the Armenian and Iberian dynasts.

(101) . AM 27.12.5–9. BP 4.15 claims that when Pap reached puberty, he was sent to the “land of the Greeks”; 4.55 says that Pap was with the “king of the Greeks” when Artogerassa was first besieged (cf. MX 3.29). BP clearly preserves the memory of Pap’s presence at Valens’s court, but he seems to have elevated his status there by converting him from a refugee to a hostage. On the status of royal hostages and refugees, see Braund 1984, 9–21, 165–80.

(102) . BP 4.55; cf. MX 3.25.

(103) . AM 27.12.9. BP 4.55, 5.1, says that the sparapet Mushel Mamikonean negotiated directly with Valens on behalf of Pap. More likely Ammianus is correct to say that legates were used.

(104) . AM 27.12.8 implies that Pap escaped without Cylaces and Artabanes, with whom he is next associated only after returning to Armenia at AM 27.12.14. Vahan clearly remained in Armenia until his death at BP 4.59. Glak and his anonymous alter ego in the Epic Histories apparently did as well (cf. BP 4.55, 5.3).

(105) . AM 27.12.10. Here MX 3.36–37 is to be preferred to BP 5.1, who fuses the restoration of Pap by Terentius (Terent) with that of Arinthaeus (Ade/Adde); cf. Markwart 1896, 215; Blockley 1975, 187–88; Garsoïan 1989, 307 n. 4.

(106) . AM 27.12.11.

(107) . AM 27.12.12; BP 4.55–58. BP indicates that Shapur was not present for the capture of Artogerassa but only invaded Armenia personally after its fall. BP also mentions a plague that devastated the garrison of Artogerassa. MX 3.35 says that the garrison surrendered.

(108) . AM 27.12.12; BP 4.55, 58.

(109) . BP 4.58.

(110) . BP 4.55–57, 59, 5.1. MX 3.36 says that Meruzan destroyed all written texts in an effort to wipe out Christianity. On Shapur’s Christian persecutions more broadly, see Decret 1979, 135–48. On fire temples, see Christensen 1944, 160–62.

(111) . AM 27.5.7; Zos. 4.11.4; cf. p. 133.

(112) . AM 27.5.9.

(113) . AM 27.12.13; MX 3.37. Arinthaeus probably took the road between Amaseia and Satala, a stretch over which three milestones of Valens have been found, indicating that repair work was done to facilitate the transport of troops and supplies; cf. RRMAM nos. 59, 63, 933.

(114) . AM 27.12.14. Cf. p. 382.

(115) . AM 27.12.15.

(116) . AM 27.12.16–17, 30.2.4. Cf. Braund 1994, 260–61.

(117) . On Valens’s Georgian garrisons and fortification of the Caucausus, see app. A. Joh. Lyd. De mag. 3.52 clearly has some chronological confusions, which led Blockley 1985, 63–67, and Chrysos 1976, 30 n. 3 to reject it as anachronistic. This is unnecessary. The passage clearly refers to an agreement reached shortly after 363 because of ongoing disputes over the construction of a fort (. Valens and the Eastern Frontier) to guard the pass (. Valens and the Eastern Frontier) through the Caucasus. There were in fact two main Caucasus passes, the Dariel in the west and the Derbend in the east. A division of Iberia at the Kur would have left Rome in command of one and Persia of the other. Once Persia assumed control of all defenses in the region, it began demanding tribute from Constantinople to fulfill the Roman half of the agreement.

(118) . Soc. 4.14.1; Soz. 6.13.1; Zos. 4.11.4.

(119) . This is clearly the meaning of Zos. 4.13.2; cf. Barnes 1998, 248 n. 4. Valens is attested at Hierapolis in the summers of 370, 373 and 377; see Seeck, Regesten, 241, 245, 249; Barnes 1998, 252–53; Greatrex 2000, 37–38. To these add P Lips. 34, 35. This is also how Constantius operated in the 340s; see Lib. Or. 18.207; cf. Or. 15.17; AM 22.23.2, 7: “Hierapolim solitis itineraribus venit” (of Julian). For Valens at Edessa, see Soc. 4.18.2–10; Soz. 6.18.2–7; Theod. HE 4.17.1–4; Ruf. HE 11.5. Valens’s visit to Edessa may be dated to 373 at Chron. Edes. no. 31 (a. 684 = A.D. 373). For Valens in Caesarea, see pp. 252–54. Here again, Constantius had used Caesarea as a base for conducting affairs with Armenia; cf. AM 20.9.1, 11.1–4. For Valens’s winter camp in Antioch, see Theod. HE 4.26.1; Theod. HR 8.5.

(120) . AM 27.12.18: “velut obseratis amicitiae foribus.”

(121) . AM 27.12.18, 29.1.1. BP 5.4 and Vita Nerses 11 p. 34–35 confirm that Shapur’s army included numerous allies, not least the Hun king Urnayr; cf. Baynes 1910, 640. On the use of auxiliaries by the Sassanians, see Christensen 1944, 209–10.

(122) . On the location, see ASX v.22.xiv with Hewsen 1992, 215–16 n. 287; cf. Kettenhofen 1989, 69. Bagawan was just across the Arsanias River from the former Persian camp at Zarehawan; cf. BP 4.55, 58.

(123) . AM 29.1.2–3; cf. BP 5.4, which continues to name “Terent” as the “Greek” commander rather than Traianus; cf. 5.5–6. MX 3.37 and Vita Nerses 11, p. 34, confuse the battle of Bagawan with the later battle of Jirov.

(124) . AM 29.1.4; BP 5.5, 8–19; MX 3.37. In the early part of 371, Valens had been in Constantinople. He only moved to the eastern frontier late in the summer; cf. p. 280.

(125) . BP 5.10, 16. Cf. Adontz-Garsoan 1970, 176; Toumanoff 1963, 181; Hewsen 1992, 158.

(126) . AM 29.1.4: “pactis indutiis ex consensu, aestateque consumpta, partium discessere ductores etiam tum discordes.”

(127) . Malalas 13.30 (p. 338).

(128) . AM 29.2.21: “parthico fragore cessante”; cf. 29.1.4: “securus interim hostium externorum”; 30.4.1: “in eois partibus alto externorum silentio.”

(129) . Soc. 4.32.1; Them. Or. 13.166c, delivered in 376 but referring to 375; cf. Vanderspoel 1995, 179–84, 251.

(130) . Or. 11.144a; cf. 148d–149a; on the date, see Vanderspoel 1995, 177, 251.

(131) . Or. 11.149b.

(132) . BP 5.7. At BP 5.37, Shapur was still fighting the Kushan war after Pap had been murdered. Proc. Bella 1.5.30–40 and MX 3.29, both of which drew on the Epic Histories, report the same war. On this Kushan war, see Markwart 1901, 50–52; Seeck, Regesten, 65; Garsoïan 1989, 313 n. 3. On the Kushans more broadly, see Frye 1984, 249–69.

(133) . AM 16.9.3–4, 17.5.1; cf. 14.3.1, 15.13.4, 18.4.1, 6.22, 19.1.7, 10. See also Matthews 1989, 62, with 488 n. 26, on the Chionitae, a Hunnic people, who also occupied Shapur at this time.

(134) . See n. 125. BP 5.8–19. Not at MX 3.8, probably reflecting MX’s anti-Mamikonean bias. On the extent of the territories regained, see Hewsen 1965, 336–37; Toumanoff 1963, 458–60 n. 98; Adontz 1970, 178–80.

(135) . Or. 11.149d–150a; cf. 13.166c.

(136) . BP 5.23–24; Vita Nerses 12–14, pp. 36–42; MX 3.38. Inasmuch as he was present at the battle of Bagawan (BP 5.4), Nerses’s murder must have occurred after summer 371; most have assumed he was poisoned shortly before Valens’s visit to Caesarea on Jan. 6, 372 (Giet 1943, 363–65; Hauschild 1973–93, 1, at Ep. 99 n. 15 and Ep. 120 n. 74). May 1973, 54, 58, dates the murder to 373 (cf. Seeck, Untergang, 59–63). Pap was not the first Arsacid king to murder a Gregorid patriarch; see BP 3.12, 14.

(137) . BP 5.29–31. MX 3.39 names Pap’s nominee for successor as Shahak of Manazkert; cf. Garsoïan 1989, 322 n. 1, 432.

(138) . Basil Ep. 99, 102, 120–122. Basil names Pap’s replacement “Faustus,” perhaps a Latinizing version of the Armenian “Yusik” (Hopeful One). Garsoïan 1983, followed by Rousseau 1994, 281–82, argues that Basil was assigned to appoint a metropolitan for the province of Armenia Minor rather than a patriarch for the kingdom of Armenia, but her argument must be rejected on several counts; see Lenski 1996, 441–42.

(139) . On Caesarea’s role in consecrating Armenian patriarchs, see BP 3.12, 16, 17, 4.4, 5.29; Vita Nerses 3, p. 24; Basil Ep. 122.

(140) . BP 5.1, 4–6, 32; MX 3.36, 37, 39; cf. Seeck, Untergang, 63, with 450 n. 25. Theod. HE 4.32.1 also implies that Terentius was commander in Armenia.

(141) . Or. 11.149d–150a.

(142) . AM 30.1.2–3.

(143) . BP 5.32; cf. MX 3.39. AM 30.2.1 confirms that Shapur had been negotiating with Pap. C. F. Lehmann-Haupt argued that a very fragmentary inscription he discovered at Martyropolis (Farkin) confirmed Pap’s demands for former Arsacid territory (Lehmann-Haupt 1908; 1910, 408–18). Markwart 1930, 134–59, expressed reservations, and Mango 1985, 95–104, has since demonstrated that the text is best placed in a late sixth-century context. Garsoïan doubts BP’s veracity with regard to Pap’s demand for the Roman cities because of a lack of outside corroboration (1989, 324–25 n. 3). Nevertheless, other sources attest Armenian claims to Edessa/Urha (Markwart 1901, 160), and Pap’s execution by Valens implies that his transgressions were serious. For the charges against Pap, see Baynes 1910, 640; Nagl, “Val.,” 2116; Blockley 1992, 35.

(144) . AM 30.1.1–7.

(145) . AM 30.1.8–12. To judge by their names, Danielus and Barzimeres were of Ibero-Armenian origin, and Ammianus makes it clear that their familiarity with the terrain helped them outmaneuver Pap. On Danielus’s (south Armenian?) and Barzimeres’s (Iberian?) origins, see Markwart 1930, 139–40 nn. 23. Hoffmann 1969–70, 1: 240–41, makes a strong case that Danielus and Barzimeres led the Tertii Sagitarii Valentis and Sagitarii Domnici, auxiliary archery units newly formed under Valens.

(146) . AM 31.1.16. On Pap’s possession by demons, see BP 4.44, 5.22; Vita Nerses 12, p. 36. See also Markwart 1930, 135–48; Russell 1987, 341–42. Garsoïan 1967, 311–13, believes that Pap’s supposed “possession” is code for his Arian heresy, but the fact that his demonic powers are noted in both Armenian and Roman traditions points to a more straightforward interpretation. Barzimeres remained a tribune of the scutarii in 378 (AM 31.9.9).

(147) . Russell 1987, passim, esp. 437–80.

(148) . AM 30.1.15.

(149) . As noted on p. 178 above, Terentius seems to have been commander-in-chief in Armenia after the battle of Bagawan; cf. AM 30.1.2–4; Basil Ep. 99. A letter of Basil’s (Ep. 214) plausibly dated to 375 (Loofs 1898, 21) indicates that Terentius had retired or been relieved from his command and was residing in Antioch. After Pap’s return to Armenia, Traianus is attested as commander; see AM 30.1.18; cf. 31.7.1–2.

(150) . On feasting in Armenian society, see Garsoïan 1976, 27–30; 1980, 46–63; cf. BP 4.2, 16, 54.

(151) . AM 30.1.18–21; BP 5.32. MX 3.39 offers a completely different account; cf. Vita Nerses 15, p. 42. The Epic Histories are remarkably close to Ammianus. Although they do mistakenly attribute the murder to Terent rather than Traianus, they alone give us the name of the place where Pap was killed, “Xu in the district of Bagrawand,” the same place where the Roman army retained its camp; cf. BP 5.1, 4.

(152) . On Gabinius, see AM 29.6.5. On the Goths, see p. 328. For Ammianus’s parallel between the two murders, see AM 30.1.22–23. At 31.1.3, Ammianus contends that the murder of Pap haunted Valens before the disaster at Adrianople.

(153) . BP 5.32–34, 37. Cf. MX 3.40. AM mentions nothing of Pap’s successor.

(154) . BP 5.34–35.

(155) . BP 5.33.

(156) . AM 30.2.2: “Arrace legato ad principem misso, perpetuam aerumnarum causam deseri penitus suadebat Armeniam: si id displicuisset, aliud poscens, ut Hiberiae divisione cessante, remotisque inde partis Romanae praesidiis, Aspacures solus regnare permitteretur, quem ipse praefecerat genti.” Deseri is a conjecture for deleri. Chrysos 1976, 37–38, and Blockley 1987, 226, make the case for the MS reading. Both attach great significance to the proposal to “destroy Armenia,” which they take to mean that Shapur intended to eliminate the Arsacid crown and divide Armenia (cf. Blockley 1992, 35–36; Seager 1996, 281–82). This reading is unsatisfactory, first because it fails to recognize the parallel nature of Shapur’s proposals—either withdraw from Armenia or from Iberia—and second because it incorrectly assumes that Ammianus understood “Armenia” to mean the “Armenian crown” or “Armenian government.” Ammianus invariably uses “Armenia” to refer only to the territory or people of Armenia (cf. Viansino 1985, 133). Shapur did hope to destroy the Arsacid monarchy, and had already attempted to do so, but deseri must be read, since “deleri penitus…Armeniam” makes no sense, at least in a pre-atomic age.

(157) . AM 30.2.3: “placitis ex consensu firmatis”; cf. 27.12.17–18, 29.1.4.

(158) . AM 30.2.3. Blockley 1987, believes this dispute indicates that some of the terms were agreed to orally, a possible, although not necessary, interpretation of the passage. On foreign policy records, see Lee 1993, 33–40.

(159) . AM 30.2.4: “ad arbitrium suum vivere cultoribus eius permissis.”

(160) . AM 30.2.4: “ingravescente post haec altius cura, imperator eligere consilia quam invenire sufficiens.…” Given the boldness of the demands that follow this phrase, it only makes sense when translated as “though concerns grew worse after this, the emperor, being in a position to make a choice of plans rather than invent them.…”

(161) . See pp. 318–19.

(162) . AM 30.2.3–4, cited in n. 157.

(163) . For Iberia, see AM 27.12.17–18: “dividi placuit Hiberiae regnum”; cf. 30.2.4: “ni Sauromaci praesidia militum impertita principio sequentis anni (ut dispositum est) inpraepedita reverterint.” For Armenia, see AM 29.1.4: “pactis indutiis ex consensu.” See p. 176 for further evidence that the truce after Bagawan was treated as a separate agreement. Seager 1996, 282, is thus wrong to argue that at AM 30.2.3 “there is no good ground for supposing that a reference to anything other than the peace of 363 is intended.”

(164) . AM 30.2.5 (cf. Theod. HR 8.4, with Greatrex 2000, 40 n. 27). Chrysos 1976 is surely wrong to regard this as the occasion for the division of Armenia commonly dated to 386 (cf. Greatrex 2000, 39). Adontz 1970, 36–37, argues more plausibly that the two territories were rather the satrapies of Asthianene and Belabitene, which later show up among the satrapies under Roman suzerainty (cf. CJ 1.29.5; Nov. Just. 31.1.3). These, together with the principalities of Sophene, Sophanene, and Ingilene-Anzitene were later referred to collectively as the “five satrapies” (see Proc. Aed. 3.1.17: . Valens and the Eastern Frontier; cf. Toumanoff 1963, 131–33; Hewsen 1992, 18). If Valens was indeed the first to gain suzerainty over Asthianene and Belabitene, he may have been looking for the rhetorical kudos to be gained by having reestablished “five transTigritane principalities”, the same number won in the treaty of 299.

(165) . AM 30.2.5–6; cf. AM 31.6.1–2. Septimius Severus’s invasion of Mesopotamia in 195 had been in three divisions, as was Severus Alexander’s Parthian expedition in 232 (Dio Cass. 75.2.3; Herodian 6.5.1–2).

(166) . AM 30.2.7–8.

(167) . AM 31.7.1. Cf. Eunap. Hist. fr. 42 (Blockley): . Valens and the Eastern Frontier . Valens and the Eastern Frontier; Oros. 7.34.6: “Persae qui…nunc etiam Valente in fugam acto recentissimae victoriae satietatem cruda insultatione ructabant.”

(168) . AM 31.7.1–2, 5–16, 8.9–10. The date of the Gothic revolt is confirmed at Cons. Const. s.a. 377; Jer. Chron. s.a. 377; Prosper Chron. 1160. Cons. Const. s.a. 377 also confirms the date of the first dispatch of Roman troops against the Goths in this year.

(169) . Cons. Const. s.a. 378; Soc. 4.38.1.

(170) . BP 5.35–36.

(171) . BP 5.37. As we would expect, MX 3.41 does not mention the regency of Manuel Mamikonean; cf. Thomson 1978, 303 n. 1. Oros. 7.34.8 implies that Valens suffered a defeat at the hands of Persia, but his passage is too rhetorical to be taken seriously.

(172) . BP 5.38–39. BP 5.42 says that Armenian independence lasted for seven years, perhaps from Shapur II’s death in 379 until the division of Armenia in 386; cf. MX 3.42.

(173) . Lazar P’arpec’i 1 [1]–8 [11] with reference to Matt. 9.16. Cf. BP 6.1. See also Seeck, Untergang, 69; Blockley 1987, 229–32; 1992, 42–45; Greatrex 2000; contrast Chrysos 1976, 37–40, who dates the division to 378.

(174) . Proc. Bella 2.3.25; Aed. 3.1.14–15; MX 3.46.

(175) . Eutr. praef. To judge from his detailed knowledge of Thrace and the Gothic confederation (Brev. 8.2, 9.15), Eutropius was with Valens during the Gothic war; cf. Bird 1986, 12; 1993, xiii. On the career of Eutropius, see PLRE I, Eutropius 2; cf. Herzog-Schmidt 1989, 202–3, with recent bibliography, esp. Bonamente 1977, on the authenticity of the dedicatory preface, and Bird 1993, vii–xviii.

(176) . Bird 1993, xiii–xiv, suggests that the Breviarium was composed for Valens in celebration of a triumph over the Goths in Constantinople in winter 369/70 . This seems likely given Eutropius’s obsession with triumphs (see p. 000).

(177) . On Festus’s identity and career, see PLRE I, Festus 2; cf. Eadie 1967, 1–9; Herzog-Schmidt 1989, 207–8, with bibliography, esp. the caveats at Boer 1972, 178–83.

(178) . MS B: “Incipit breviarium Festi vc magistri memoriae.” Baldwin’s 1978, 197–99, caveat that only this MS refers to Festus as magister memoriae was already sufficiently anticipated at Eadie 1967, 6. On the dedication to Valens rather than Valentinian, see Eadie 1967, 3, 22–23.

(179) . Mommsen 1862, 587, accepted at Eadie 1967, 1. Contrast Boer 1972, 198, who (oddly) believes Festus omitted Valentia because he was assembling his provincial list from memory. Tomlin 1974 tried to redate Theodosius’s renaming of Valentia to 368, but Salway 1981, 382, still accepts 369.

(180) . Momigliano 1963; Eadie 1967, 10–20; Herzog-Schmidt 1989, 173–75; Chauvot 1998, 207–19.

(181) . It. Alex. 1, 4. On the date, see Barnes 1985a, 135. On the authorship, see Lane Fox 1997, 240–47. I use the critical edition of Tabacco 1992, 112–51, for sections 1–23 and the editio princeps of Mai 1817 for the remainder. For bibliography, see Herzog-Schmidt 1989, 214–15 and Lane Fox 1997, 239 n. 1.

(182) . It. Alex. 3: “scilicet ut incentivum virtutibus tuis.”

(183) . Milan, Ambros. P 49 sup., fols. 54v–64v.

(184) . The incipit: “Breviarium Festi de breviario rerum gestarum populi romani” is found in MSS E B P. Momigliano 1963, 85–86, had suggested—perhaps half in jest—that Valens felt Eutropius’s brief history remained too long and commissioned Festus to abbreviate it further. To the similar conclusion of Eadie 1967, 14–16, see the review of Cameron 1969. On this question, see also Boer 1972, 173–6; Bird 1986, 18–20; 1993, xxiv.

(185) . Festus 15: “Scio nunc, inclyte princeps, quo tua pergat intentio. Requiris profecto, quotiens Babyloniae ac Romanorum arma conlata sint et quibus vicibus sagittis pila contenderint.”

(186) . Festus 30: “ut ad hanc ingentem de Gothis etiam Babyloniae tibi palma pacis accedat.”

(187) . Both Festus and Eutropius openly state that they were writing at Valens’s request (Eutr. praef.: “ex voluntate manseutudinis tuae.” Festus 1: “clementia tua praecepit; parebo libens praecepto”; cf. 10, 15, 30). Previous scholars have placed too much emphasis on the personal role of these authors in actively directing Valens’s eastern policy (see, e.g., Peachin 1985, 159; Bird 1986; 1993, xx). In fact, Valens must have known what his stance toward Persia would be before he ever set his magistri memoriae to work. Their breviaria are not reflections of their personal opinions, but rather historical justifications for Valens’s intended eastern campaigns.

(188) . Festus 6, 12, 16, 25. Eutr. 2.5, 6, 9, 13, 14, 20, 21, 24, 27, 28, 3.8, 13, 16, 20, 4.4, 6, 5.1, 6, 6.8, 12, 20, 9.22. It. Alex. 17, 56, 59, 80–81, 95–96. Cf. Boer 1972, 142–45.

(189) . See pp. 143–45. On the rhetoric of expansion earlier in the empire, see Campbell 1984, 394–401.

(190) . Festus 4: “quo ordine autem singulas provincias Romana res publica adsecuta sit.” Sicily and Africa (4), Spain (5), Gaul and Britain (6), Illyria and Greece (7–8), Thrace (9), Asia Minor (10), Anatolia (11), Syria and Palestine (12), Egypt and Cyrenaica (13), Armenia and Mesopotamia (14). On the provincial lists that conclude each section, see Eadie 1967, 154–71; Boer 1972, 197–99; Peachin 1985, 159–60.

(191) . Eadie 1967, 153, offers the odd interpretation that Festus concluded his narrative with the 363 treaty because “there simply was nothing further to report.” Bird 1986, 19, is surely right that Festus’s aim was to encourage Valens to follow the examples of Trajan, Diocletian, and Julian by re-conquering this land.

(192) . Them. Or. 8.114c; cf. 11.148d–149a.

(193) . Eutr. 1.4, 5, 8. For further examples of Eutropius’s concern with the distance of battles and conquests from Rome, see 1.15, 17, 19, 20, 2.5, 8, 12, 3.14. See also Boer 1972, 120–24.

(194) . Eutr. 8.2. Cf. Festus 20: “Traianus, qui post Augustum Romanae rei publicae movit lacertos, Armeniam recepit a Parthis.” This metaphor is taken from Florus praef. 8: “[populus Romanus] sub Traiano principe movit lacertos,” where it fits into a larger analogy between the human body and the empire.

(195) . As Eutropius 8.5 notes, later emperors were praised as “Felicior Augusto, melior Traiano!” Cf. Aurelius Victor’s relative lack of interest in territorial expansion, which he mentions only five times in De Caesaribus (1.2, 2.3, 13.3, 20.15–16, 39.37) and the same relative lack of interest at Epit. 1.7, 2.9, 5.4, 9.13, 11.2, 48.5. See Bird 1984, 41–59, on Aurelius Victor’s general lack of interest in militaria.

(196) . On Tiberius, see Eutr. 7.11. On Nero, see Eutr. 7.14; cf. Festus 20: “amisit Armeniam.” On Gallienus, see Eutr. 9.8; Festus 23. On the shamefulness of abandoning territory, see AM 25.9.3; cf. Potter 1990, 12–13.

(197) . Eutr. 10.17.

(198) . Eutr. 8.6, 9.15. On Hadrian, see Festus 20. On Dacia, see Festus 8, with Boer 1972, 201–3. Eutropius certainly accompanied Julian’s expedition and must have been with the army when Nisibis was evacuated (Eutr. 10.16; cf. Bird 1993, x–xi; xxii). This helps explain his dismay at the treaty, but not his mendacity about its uniqueness.

(199) . Festus 29: “condicionibus (quod numquam antea accidit) dispendiosis Romanae rei publicae inpositis.” AM 25.9.9–11. Cf. Bird 1993, 163 n. 367; Matthews 1989, 187. Zosimus (3.32.1–6) must derive from Eunapius, pace Paschoud 2.1: 221–23 n. 93 and 1975, 184–206. Similar rhetoric on the unjust loss of territory in 363 can be found at Them. Or. 8.119c.

(200) . Boer 1972, 147, 159, 164–66, 203–4. For triumphs in Festus, see 21, 22, 25. On triumphal rulership in late antiquity, see McCormick 1986; Straub 1939, 7–75.

(201) . Eutr. 1.6, 11, 20, 2.1 (three), 2, 5, 9, 14 (two), 16, 17, 18, 19, 20, 24, 3.2, 3, 4, 5, 6, 23, 4.2, 4, 5, 8, 14 (three), 19 (two), 22, 23, 25 (two), 27 (two), 5.9 (two), 6.2, 3, 5 (four), 10 (two), 15 (three), 7.5, 13, 20, 23 (three), 8.9, 13, 9.27. Cf. Florus, whose much fuller narrative uses triumphus and its cognates much more sparingly and much more loosely as an equivalent to “victory in battle.” He uses the word to refer to triumphal processions only nine times (1.13.26, 18.9–10, 28.12, 30.5, 34.17, 36.17, 38.10, 2.10.9, 13.88–89); cf. Fele 1975, 685–86.

(202) . For multiple triumphs, see Eutr. 2.1, 4.14, 19, 25, 27, 5.9, 6.5, 10, 15. For triumphal firsts, see 1.6, 2.9, 14, 3.4, 7.5. For special features, see 3.23, 4.4, 8. He also catalogs generals who won important triumphal titles; see 3.23, 4.4, 6.3, 8.18.

(203) . On Julian’s imitatio Alexandri, see Lane Fox 1997, 247–52, with bibliography. On imitatio Alexandri more broadly, see Heuss 1954, esp. 99–101; Wirth 1975, 200–209; Meyer 1980, 414–17; Campbell 1984, 391–93; Potter 1990, 372.

(204) . It. Alex. 1, 5, 8–11. Julian also claimed to desire the total conquest of Persia; see Ephrem CJ 2.15; Lib. Ep. 1402; Or. 12.100, 15.2, 18.1, 282.

(205) . Lucullus 15; Pompey 16; Crassus 17; Ventidius Bassus and Marcus Antonius 18; Augustus 19; Nero and Trajan 20; Lucius Verus, Septimius Severus, and Caracalla 21; Severus Alexander, Gordian III 22; Valerian, Gallienus 23; Aurelian, Carus 24; Diocletian, Galerius 25; Constantine 26 (planned); Constantius 27; Julian 28; Jovian 29.

(206) . Lucullus 6.9–10; Pompey 6.12–14; Crassus 6.18; Ventidius Bassus 7.5; Mark Antony 7.6; Augustus 7.9; Trajan 8.3; Lucius Verus 8.9; Septimius Severus 8.18; Caracalla 8.20; Severus Alexander 8.23; Gordian III 9.2; Valerian 9.7; Aurelian 9.13; Carus 9.18; Diocletian, Galerius 9.25; Constantine 10.8 (planned); Constantius 10.10; Julian 10.16; Jovian 10.17. Ammianus (23.5.16–18) portrays Julian as offering a similar catalog of eastern commanders in his harangue to the troops before the Persian invasion of 363 (Lucullus, Pompey, Ventidius / Mark Antony, Trajan, Lucius Verus, Septimius Severus, Gordian III); cf. Jul. Or. 1.17d–18a.

(207) . Eutr. 8.2–5. See also Boer 1972, 149; cf. 199 and 205 on Festus and Trajan. Lightfoot 1990, 121–25, offers interesting insights on the glorification of Trajan’s Persian achievements during the fourth century; cf. Syme 1971, 91–112. Eutropius also devotes considerable attention to Aurelian (9.13–15), the restorer of Roman prestige in the east, whose accomplishments (Gothic victory, monetary reform, fortification) also mirror Valens’s own.

(208) . Eutr. 8.2; cf. Bird 1986, 13–14, and 1993, xxi. A comparison is perhaps also implied at Festus 26 between Valens and Constantine, who set out for the east: “recenti de Gothis victoria gloriosior.” Eutropius also notes instances where brothers campaigned simultaneously and triumphed in unison (4.4, 4.25, 6.10). The parallel he draws to Marcus Aurelius and Lucius Verus is picked up by Festus 21, who further develops the theme: “Antonini duo…pariter Augusti, imperium orbis aequata primum potestate tenuerunt. Sed ex his Antonius iunior ad expeditionem Parthicam profectus est;…ingenti gloria de Persis cum socero triumphavit.” See also Eutr. 8.10: “cum fratre eodem socero triumphavit.” The clear reference to the equality of the brother Augusti and to the necessity for the iunior colleague to campaign in the east were probably meant to reflect Valens’s own situation. For the theme of return to the days of Trajan, Marcus, and Antoninus under the Valentiniani, see Them. Or. 13.166b.

(209) . Tabari 814 (p. 3 Bosworth); cf. Christensen 1944, 59–62, and Jul. Epistulae, leges, poemata, fragmenta varia 205 (Shapur II as a descendant of Darius).

(210) . Dio Cass. 80.4.1; Herodian 6.2.1–2, 4.4–5. The origins of the Sassanian demand for the return of Achaemenid lands are much debated. Winter 1988, 26–44; Frendo 1992; and Lee 1993, 21–22, argue for an early origin for the Sassanian claims to Achaemenid ancestry and territory. Contra Kettenhofen 1984; Potter 1990, 370–80.

(211) . AM 17.5.3–8. Ammianus tells us that he does not reproduce the letter verbatim but follows its sense. There is every reason to believe he had at his disposal a copy of the actual missive; cf. Christensen 1944, 237–8; Jonge 1977, 133–34; Matthews 1989, 39–40 n. 2.

(212) . Inheritance: 10 Asia; 11 Bithynia; 13 Libya. Conquest: 10 Syria, Lydia, Caria, Hellespont, Phrygia; 11 Pamphylia, Lycia, Pisidia, Pontus; 12 Syria, Cilicia, Isauria; 13 Egypt; 14 Syria, Phoenice, Arabes et Iudaei. Request: 10 Rhodes; 11 Galatia; 13 Cyprus.

(213) . AM 17.5.6; cf. 5.11, 14.1; Lib. Ep. 331; Or. 59.71.

(214) . Eutr. 6.9 also notes Lucullus’s capture of Nisibis.

(215) . Festus 14; cf. 25.

(216) . Festus 16; cf. 14. Eutr. 6.13–14; cf. 8.3. At Festus 19, Pompey’s settlement of the east after the Mithridatic war is again cited as the basis of Rome’s eastern claim in the context of Gaius Caesar’s diplomatic exchange with the Parthians in A.D. 1 (“ex instituto Pompei”). On the importance of Pompey in Festus’s narrative, see Boer 1972, 206–7. Similar historical arguments were already at play in the late first century A.D. (Tac. Ann. 13.34, 15.13–14).

(217) . Rome seems to have reached a nonintervention pact respecting Armenia in 244, but it was not until 252–53 that Shapur I actually invaded and imposed his son as king of Armenia. See most recently Winter 1988, 80–123, with full bibliography, and Potter 1990, 221–25. Most of the literary and epigraphical sources are collected at Dodgeon and Lieu 1991, 45–47.

(218) . Rather than focus on the empire’s losses under the treaty terms of 244, Festus 23 and Eutr. 9.10–11 shift the emphasis to Shapur’s successful invasions of the Roman frontier under Gallienus, which Festus asserts represented Persia’s first effort to lay claim to Mesopotamia. Even here, both historians are quick to point out that Shapur I’s efforts in the 250s and 260s were forestalled by the counterstrikes of the Palmyrenes.

(219) . Eutr. 2.9, 12–13, 27, 3.21–22, 4.2, 4, 7, 17, 5.7, 6.13, 7.9, 10.7, 17. Eutropius does not mention the terms of the 299 treaty with Persia.

(220) . Eutr. 2.12–13, 3.22; cf. 5.7.

(221) . Eutr. 2.9: “pax tamen a senatu et populo soluta est quae cum ipsis propter necessitatem facta fuerat.” Cf. Eutr. 10.17: “pacem…necessariam quidem.”

(222) . Eutr. 4.17, 26.

(223) . Eutr. 10.17: “nam et Samnitibus et Numantinis et Numidis confestim bella inlata sunt neque pax rata fuit.” AM 25.9.11 cites exactly the same examples in making the same argument. Matthews 1989, 187, points out that the constitutional situations were totally different between these republican treaties—which were negotiated by a field commander and could later be ratified or rejected by the Senate—and Jovian’s treaty—for which the emperor was solely responsible. This misses the point: Eutropius’s and Ammianus’s case for dissolution is all the more pointed because of the weakness of the comparison. For the rhetoric of the pax ignobilis, see AM 25.7.13: “quo ignobili decreto firmato”; 27.12.1: “pudendae pacis icta foedera”; cf. Zos. 3.32.4: . Valens and the Eastern Frontier on 244.

(224) . For the name, see Shahid 1984a, 172 n. 127. Ibid. 172–74 offers a religious rationale for the Maratocupreni uprising, a suggestion with no basis in the sources.

(225) . AM 28.2.11–14. For the rationalis [rei privatae], see A. H. M. Jones 1964, 412–13.

(226) . Them. Or. 8.117a: . Valens and the Eastern Frontier. Heather and Matthews 1991, 32 n. 59, suggest a reference to the Mavia revolt, but this was much later; cf. pp. 208–9. Instead, the Maratocupreni seem to be at issue (Lenski 1999a, 311 n. 19). For the raids in Britain, see P. Salway 1981, 374–93.

(227) . Lib. Or. 48.36; cf. Ep. 1385.

(228) . Isaac 1992, 64–65; cf. 1984. On brigandage in the ancient world more generally, see Shaw 1984 and 1993.

(229) . Recent secondary work on Isaurian banditry can be found at Minor 1979; Shaw 1990; Lewin 1991; Lenski 1999b.

(230) . Houwink ten Cate 1965, 190–201; Shaw 1990. See also Er 1991 on the continuity of culture reflected in Isaurian art.

(231) . See SHA Tyr. Trig. 26.6. On topography, see Mutafian 1988, 14–20; Hild and Hellenkemper 1990, 22–29.

(232) . Lenski 1999b, 431–39.

(233) . On Shapur I’s invasion of Isauria, see RGDS ll. 27–31, with Kettenhofen 1982, 111–20. On Isauria as a breakaway territory from the third century onward, see Rougé 1966; Lewin 1991, 172–80. Shaw 1990 contends that the Isaurian hinterland never broke away, because it was never truly controlled by the Roman state in the first place, but see Lenski 1999b; cf. Mitchell 1999.

(234) . Lenski 1999b, 439–46.

(235) . For Isaurian revolts in Ammianus, see esp. Santos Yanguas 1977; Matthews 1989, 355–67; Hopwood 1999.

(236) . AM 14.2.1–20.

(237) . On the cities surrounding the hinterland, see Hellenkemper 1980; A. H. M. Jones 1971, 207–14; Lenski 1999a, 320–22.

(238) . Even imperial troops seem to have been reluctant to venture into the Taurus (AM 14.2.5–7, 19.3.1–2, 27.9.7; cf. Joh. Ant. fr. 214b5).

(239) . AM 19.13.1–2. The commander sent by Constantius, Bassidius Lauricius, and his subordinate? Aur. Ious[tus] built secured structures along the route through the hinterland in the wake of this uprising; cf. ILS 740 = CIL 3.6733; Bean and Mitford 1970, no. 231.

(240) . AM 27.9.6–7.

(241) . Eunap. Hist. fr. 43 (Blockley). On diogmitae, see the bibliography at Robert 1994, 91.

(242) . CIG 4430 = OGI 580 = MAMA 3: 102 n. 1, with pp. 102–7.

(243) . Tarsus: CIL 3.13619–21; cf. CIG 4437. Iconium: CIG 3992. Pisidian Antioch: Levick 1965, 59–62 = AE 1965, 15.

(244) . Lenski 1999, 319–20.

(245) . On limitanean legions, see Not. dign. or. 29.7–8; cf. 7.56. See also AM 14.2.14; VM Thecla Mir. 28. Other passages indicate the presence of other imperial units as well (AM 14.2.5, 8, 12–14, 19.13.1). On civic police forces, see Hopwood 1989; Lenski 1999a.

(246) . AM 27.9.7: “per indutias pacem sibi tribui poposcerunt”; cf. VM Thecla Mir. 19: . Valens and the Eastern Frontier.

(247) . Zos. 4.20.1–2; cf. Eunap. Hist. fr. 43.4 (Blockley), with Blockley 1981–83, 2: 141 n. 97.

(248) . Ep. 200, 215, 217 canons 55–57, 84, with Lenski 1999a.

(249) . Lycia: Zos. 4.20.1. Central Anatolia: Basil Ep. 215. Later Isaurian raids also closed communication routes in central Anatolia, Joh. Chrys. Ep. 69, 70, 135, 136 (PG 52.616, 647, 693–94); VM Thecla Mir. 16.

(250) . Basil Ep. 200 confirms that troops were sent through Iconium, north of the Taurus. VM Thecla Mir. 13 with Dagron 1978, 117, shows that Saturninus led comitatensian units from the south as well. Two funerary inscriptions from Seleucia (CIG 9207, 9230) may record the deaths of soldiers involved in this conflict; cf. Lenski 1999a, 316 n. 33.

(251) . Milestones make it clear that Valens rebuilt the road between between Colonia Archelais and Iconium after November 375 (RRMAM nos. 637, 639, 648–51, 655, 660).

(252) . Isauria’s exemption from the military clothing tax in 377 (CTh 7.6.3) confirms that the wounds opened up in 375 were still festering two years later.

(253) . Recent work on the Arabian frontier has been abundant. See esp. Bowersock 1976; 1983; Isaac 1992; Millar 1993, 387–436; MacDonald 1993.

(254) . Matthews 1984; Millar 1993, 319–36.

(255) . Peters 1978; D. Graf 1989, 366–80; Villeneuve 1989, with modifications by MacDonald 1993, 311–52.

(256) . The nature of the nomadic menace in Arabia has been hotly debated. Cf. the views of Parker 1986; 1987; 1997; D. Graf 1989; 1997; MacDonald 1993; Millar 1993, 428–36; Isaac 1998, 444–52.

(257) . Bowersock 1987, 72–75.

(258) . D. Graf and O’Connor 1977; cf. D. Graf 1978, 10–15. Shahid 1984b, 123–41, has argued against this etymology, but see the defense offered at Bowersock 1986, 113. Further arguments at Sahas 1998.

(259) . Peters 1978, 324–26; Bowersock 1983, 131–37; Matthews 1989, 348–53. On Palmyra after the 270s see Pollard 2001, 298–300.

(260) . On sheikhs, see MacDonald 1993, 368–77. On tribes, see D. Graf 1989, 359–66.

(261) . The inscription was originally published by Dussaud 1902, 409–21; cf. id. 1903, 314–22. I follow here the reading at Bellamy 1985. Debate about this and other readings is ongoing. For interpretation, see esp. Sartre 1982, 136–39; Bowersock 1983, 138–42; Shahid 1984a, 31–53.

(262) . Tabari 822, 834 (pp. 22, 44 Bosworth).

(263) . AM 24.2.4; cf. AM 25.1.3; Zos. 3.27.1. See esp. Bowersock 1980, 485; Shahid 1984a, 119–23; contra Sartre 1982, 139–40.

(264) . See Shahid 1984a, esp. 353–417, and the more digestible summaries at D. Graf 1978, 15–19; Sartre 1982, 134–53; Isaac 1992, 235–49; Millar 1993, 431–35.

(265) . Pan. Lat. 3 [11].5.4: “oppressumque captivitatis vinculis Sarracenum”; cf. 7.1. For the date and Diocletian’s participation, see Barnes 1982, 51.

(266) . Sartre 1982, 64–75; Bowersock 1983, 144–47.

(267) . Parker 1986, 135–45; 1989; Millar 1993, 180–89; Konrad 1999; cf. Speidel 1977, 717–27.

(268) . E.g., AE 1974, 661 (a. 333) with Zuckerman 1994; AE 1933, 171 (a. 349) and AE 1933, 170 (a. 351); cf. Parker 1986, 145.

(269) . AE 1948, 136 (a. 334); AM 14.4.1–3 (a. 354); cf. AM 14.8.13; Exp. tot. mun. 20 (a. 359); Jer. Vita Malchi 4 (PL 23.57); Nilus Narrationes (PG 79.589–693), passim, with Mayerson 1963.

(270) . Jul. Or. 1.19a; cf. 20b and Shahid 1984a, 74–100.

(271) . Jul. Or. 1.21b. The Notitia dignitatum indicates that by the late fourth century, several Saracen groups had also been organized into regular limitanean units; cf. Not. dign. or. 28.17, 32.27–28; cf. 34.22.

(272) . AM 14.4.1–7.

(273) . AM 23.3.8, 5.1; cf. AM 24.1.10; Jul. Epistulae, leges, poemata, fragmenta varia 98 (401d); Tabari 840–2 (pp. 59–61 Bosworth). Bowersock 1983, 142, suggests that the Lakhmids also played a role in Diocletian’s and Galerius’s Persian invasion of 298.

(274) . AM 25.6.9–10. For sources and debate on the role of the Saracen federates in Julian’s death, see Bowersock 1978, 116–18; Shahid 1984a, 124–32. Persia’s Saracen allies also attacked Julian’s army in 363; cf. AM 24.2.4, 25.1.3, 8.1.

(275) . See Ammianus’s assessment, 14.4.1: “Saraceni tamen nec amici nobis umquam nec hostes optandi”; cf. Josh. Styl. 79.

(276) . Lib. Ep. 1236: . Valens and the Eastern Frontier; cf. Ep. 1127. On Ulpianus, see Sartre 1982, 104, no. 71.

(277) . Them. Or. 13.166c, not noted in previous scholarship.

(278) . For the inscriptions, see app. A. On the forts, see Parker 1986, 145–46; De Vries 1986; 1993; 1998, 131–47, 229–31; Kennedy and Riley 1990, 179, 183–84.

(279) . Not. dign. or. 34.35: “Ala secunda felix Valentiana, apud Praesidium”; 34.42: “Cohors secunda Gratiana, Iehibo”; 37.29: “Ala prima Valentiana, Thainatha”; 37.30: “Cohors secunda felix Valentiana, apud Adittha”. Thainatha is probably not to be identified with Umm el Jimal (De Vries 1998, 36–37; contra Shahid 1984a, 415–16).

(280) . As assumed at Bowersock 1976, 223–26; Speidel 1977, 726.

(281) . Soz. 6.38.1: . Valens and the Eastern Frontier.

(282) . Soc. 4.36.1: . Valens and the Eastern Frontier; cf. Soz. 6.38.1: . Valens and the Eastern Frontier . Valens and the Eastern Frontier.

(283) . Soc. 4.36.1: . Valens and the Eastern Frontier; Ruf. HE 11.6: “Saracenorum gentis regina”; cf. Theoph. a.m. 5869; Theod. Lect. Epit. 185.

(284) . The most reliable sources on Mavia’s revolt are Ruf. HE 11.6, Soc. 4.36, and Soz. 6.38. See also Theod. HE 4.23; Theoph. a.m. 5869; Nicephorus Callistos HE 11.47 (PG 146.32–33); Theod. Lect. epit. 185. For secondary work, see Bowersock 1980; Mayerson 1980b; Thelamon 1981, 123–47; Sartre 1982, 140–44; Shahid 1984a, 138–202; Rubin 1990, 182–89.

(285) . The agglomeration of fanciful details in later versions has led Mayerson 1980b to extreme skepticism, but Bowersock 1980, 478–83, has clarified most of the problems by establishing the source tradition.

(286) . Ruf. HE 11.6: “Mavia…Palaestini et Arabici limitis oppida atque urbes quatere vicinasque simul vastare provincias coepit.”

(287) . Soz. 6.38.1. This finds confirmation in the notice at Theod. HE 4.23.1 and Ruf. HE 11.6 that the monk Moses lived on the border between Palestine and Egypt. Since Soz. 6.38.5 tells us that Moses lived in the desert near Mavia, one must assume that her activity skirted Egyptian territory.

(288) . The text of the Relatio Ammonii survives in both Greek (ed. Combefis, 1660) and Syriac (ed. and trans. Lewis, 1912). The Greek purports to be a translation from the Coptic original of Ammonius (Greek, 132; cf. Syriac, 14). The authenticity of this narrative had been questioned by Devreesse 1940, 216–20, but Mayerson 1980a and Shahid 1984a, 308–15, have both defended it. The Relatio also reports a simultaneous raid by a group of bandit Blemmyes on the monastic community at Raïthou, discussion of which is omitted here (cf. Mayerson 1980a, 141–48).

(289) . Relatio Ammonii, Greek, 88; Syriac, 1, 14.

(290) . Relatio Ammonii, Greek, 91: . Valens and the Eastern Frontier Syriac, 2: “[T]he king of the Saracens had died, he who was the guardian of the desert.”

(291) . The latter officer is clearly spelled out at Soz. 6.38.2, although the first (. Valens and the Eastern Frontier . Valens and the Eastern Frontier) is nowhere else attested. For speculation about his title and identity, see Shahid 1984a, 150, and Woods 1998, 328–34.

(292) . Demandt 1970, 704; Shahid 1984a, 151, based on CIL 3.88 = ILS 773 and AM 31.16.8; cf. Zos. 4.26.1–9.

(293) . Soz. 6.38.4.

(294) . Ruf. HE 11.6; Soc. 4.34.4; Soz. 6.38.5; Theod. HE 4.23.1.

(295) . Soc. 4.34.12 alone reports this. Details at Shahid 1984a, 158–69. The temptation to see this daughter in the Mavia of AE 1947, 193, should be resisted. Shahid 1984a, 222–27, actually identifies the dedicant as Mavia the rebel queen herself, also unlikely given the date (a. 425).

(296) . Soc. 5.1.4–5; Soz. 7.1.1; AM 31.16.5–6; Eunap. Hist. fr. 42 (Blockley); Zos. 4.22.1–3; cf. Shahid 1984a, 175–83. Woods 1996 offers an idiosyncratic reading of the incident, on which see p. 335 n. 94.

(297) . Them. Or. 34.20 (a. 384/85). This reference has not previously been noted.

(298) . Itin. Eg. 1–10 (CCSL 175.37–51) with Rubin 1990, 177–82; cf. Hunt 1982, 58–60.

(299) . Pan. Lat. 2 [12].22.3: “Dicam a rebellibus Sarracenis poenas polluti foederis expetitas?” Cf. Shahid 1984a, 210–14.

(300) . Shahid 1984a, 152–8, 185–90.

(301) . Shahid 1984a, 143, 155, 330–45, collects the evidence for Arab bishops from episcopal lists, but he cannot prove that these were Saracen nomads (as opposed to sedentarists), let alone the Tanukhid Saracens whom he argues Mavia led. He also falls well short of proving his Arab bishops were staunchly Nicene—two were certainly Homoian. See further remarks at Bowersock 1986, 114–16.

(302) . The evidence is collected at Trombley 1993, 143–73.

(303) . Soz. 6.38.9; Theod. HE 4.23.5; Theod. Lect. Epit. 185; Theoph. a.m. 5869.

(304) . Ruf. HE 11.6; Soc. 4.34.6–7; Soz. 6.38.6–8.

(305) . Soc. 4.34.3 tells us that Moses was . Valens and the Eastern Frontier and Soz. 6.38.5 that he lived in the desert near Mavia’s people. He would thus have established trust with Mavia and her Saracens; cf. Theod. HE 4.23.1. Similar conversions of Saracen groups by local holy men are reported at Jer. Vita Hilarionis 25 (PL 23.42) and Soz. 6.38.14–16. Rubin 1990, 182–91, makes a strong case that the Moses who converted Mavia’s Saracens was the same Moses who converted the sedentarized Pharanites near Raïthou in the southern Sinai (Relatio Ammonii, Greek, 99–101; Syriac, 4–5). Though Shahid 1984a, 186, entertains the possibility that the two Moses were identical, he refuses to see the fatal consequences of this for his argument at 142–50 that Mavia’s home was in the north, in the Syrian Hauran. In fact, if Theod. HE 4.23.1 is correct that Moses lived near the border between Palestine and Egypt, his neighbor Mavia and her Saracens cannot have been based in Syria; cf. Sartre 1982, 142.

(306) . Soc. 4.36.1; Soz. 6.37.15–38.1; cf. Bowersock 1980, 485–87. Sartre 1982, 143–44 (followed by Greatrex 2000, 38 n. 20) dates the revolt to 376; Shahid 1984a, 183–84, who offers further bibliography on the question, dates it to ca. 375–78.

(307) . See p. 205. See also Theod. HE 4.23.2: . Valens and the Eastern Frontier; Ruf. HE 11.6: “cumque frequentibus bellis Romanum attrivisset exercitum.”

(308) . The sources directly mention arrest; cf. Ruf. HE 11.6: “captus Moyses”; Soc. 4.36.5: . Valens and the Eastern Frontier; Soz. 6.38.6: . Valens and the Eastern Frontier.

(309) . Theoph. a.m. 5869.

(310) . There is some confusion over the date caused by the author’s attempts to make the massacre in Sinai and Raïthou fall on the same day, January 14. The Sinai massacre seems, however, to have occurred “in Tybi on the second,” i.e., December 28 (Relatio Ammonii, Greek, 95; Syriac, 14; cf. Greek, 129). Mayerson 1976; 1980a, 141–42, is led to mistaken assumptions because he notes only the last reference.

(311) . A fortlet erected at private expense at Moujedil in the Hauran in 377 may provide further confirmation of trouble in this year (Clermont-Ganneau 1888, 8–10).

(312) . Valens’s false start may be reflected in the fact that Socrates mentions his departure from Antioch in three separate contexts (4.35.2–3, 36.1, 37.1).

(313) . Lenski 1999b, 437–38.