Syene as face of battle: Heliodorus and late antique historiography.

Ross, Alan J.

Introduction

Heliodorus' Aethiopica stands apart from the other extant examples of the ancient Greek novel. It is the latest and longest example of the genre, and perhaps shows greatest awareness of the literary traditions that precede it. (1) Although the issue of the Aethiopica's date has not been resolved definitively, there is now at least a majority opinion that it is a product of the fourth century AD, and that in its ten books it responds to a wealth of canonical, classical literature, including epic, tragedy, and historiography. (2) Heliodorus' relationship to that lattermost genre, historiography, is the most important for establishing the text's claim to verisimilitude. More than thirty years ago, John Morgan convincingly argued that Heliodorus' adoption of a narrative pose commonly associated with the writers of history imbues Heliodorus' narrative with a realism that invites 'a certain kind of response from the reader, which involves equating the events of the novel with those of the real world--that is to say, an intensely emotional, sympathetic response'. (3) This is an especially literary form of verisimilitude; Heliodorus alludes to the practices of the genre that makes the greatest claim to represent events that take place in the 'real world'. Historiography mediates between the world of the novel and 'reality'. Heliodorus' verisimilitude is constructed not by writing in a way that is in itself directly mimetic of reality, rather it is constructed by imitation of the standard mode of writing about historical reality, historiography.

In this paper I argue that these issues of dating and intertextuality can be united in order better to situate Heliodorus within a late antique literary milieu. My approach focuses on one episode of the Aethiopica which best corresponds to a recognizable scene-type within historiography, namely the siege of Syene in Book 9. I will explore to what extent Heliodorus' narrative in this episode may represent a contemporary, fourth-century approach to siege narrative, allusion to which Heliodorus uses to strengthen the verisimilitude of his narrative and to guide his reader's response to the scene. Finally I will suggest where Heliodorus may have found his historiographic models.

Helidorus' historiographic pose revisited

The generic proximity of the novel and historiography has long been noted. (4) Both are prose genres which utilize narrative to offer a mixture mimesis and diegesis of human characters operating in an ostensibly real, or at least credibly real, world set in the past. (5) The novel's relationship with historiography, however, is closest in the form of its narrative, rather than its content. (6) Morgan's survey of Heliodorus' 'historiographic pose' identifies mannerisms, such as authorial uncertainty, particularly inclusion of alternative explanations for events, and the use of ecphrases and excursus, which give the Aethiopica an historiographic framing. (7) The erotic content, by contrast, rarely resembles that of historiography. (8) Book 9 of the Aethiopica, however, provides an exception. (9) The main protagonists, the young couple Theagenes and Charicleia, are marginalized, and the narrative in the first half of the book is devoted to a lengthy description of the siege of the Egyptian city of Syene, in which the Persian satrap Oroondates takes refuge with his army and which he subsequently defends against the forces of the Ethiopian King Hydaspes. (10) In this respect the opening narrative of Book 9 is historiographic in content: the siege scene was a standard element of historiographic prose. (11) Although the greatest siege of all is found at the beginning of classical literature within the genre of epic (Homer's Iliad), descriptions of siege soon became a definitive part of the fabric of historiography. (12) Indeed, siege descriptions proliferated to such an extent during the Hellenistic period that Polybius complained that writers would often write up sieges using stock elements where there was little historical accuracy underpinning their accounts. (13) The historians who formed the object of Polybius' criticism evidently viewed the siege scene as an essential marker of their genre, even if the inclusion of such scenes called for a degree of invention, which, as Polybius thought, stretched the generic conventions of historiography. Sieges would remain important elements of Greek historiography under the Roman empire, with notable examples of extended descriptions appearing in Josephus, Arrian, Cassius Dio and Herodian. (14)

Most scholars have argued that Heliodorus looked further back for his historiographic models than to the historians of the empire, especially to Herodotus. (15) The theme of Herodotus' Histories certainly chimes with Heliodorus' narrative of Persian aggression in Book 9, and the fictional period in which the Aethiopica is set, during the Persian occupation of Egypt between the 6th and late 4th century BC, (16) * reflects the period under discussion in the Histories. (17) The nature of these proposed Herodotean intertextualities ranges from large-scale imitations of narrative elements such as geographic and ethnographic digressions, (18) through similar episodes, (19) down to the use of lexical allusions and the reuse of Herodotean phrases. (20) However, the events at Syene have already been strongly linked to literature of the fourth century AD. The resemblance between the siege of Syene and the Persian siege of the city of Nisibis in Roman Mesopotamia in AD 350 is a central issue for dating the Aethiopica to the mid-fourth century. (21) There are strong similarities between Heliodorus' Syene and the narrative of the siege of Nisibis found in Julian's two panegyrics to his senior emperor, Constantius II, composed in the mid to late 350s. (22) At both Syene and Nisibis, the attackers construct dykes in order to form a concentric ring of earthworks around the city walls. They then divert a nearby river (the Nile at Syene and the Mygdonius at Nisibis) to flow into the enclosed space between wall and dyke, creating a flood upon which boats could sail. The wall collapses under the effects of the water, however, the attackers fail to press home their advantage, hampered by the mud left by retreating waters. In each case the wall is reconstructed.

Heliodorus responded to contemporary awareness of the historical events at Nisibis, and indeed to specific details of the siege as they were presented by Julian. As Morgan has pointed out, Helidorus' exploitation of the 'public awareness and interest' in Nisibis was an attempt to enhance the realism of his novel. (23) Of course, Julian's account of Nisibis was not set within a work of history, but in epideictic oratory. Nevertheless, the sections of the panegyrics relating to Nisibis have a distinctly historiographic flavour, using narrative as the medium to represent this (recent) past event. (24)

Heliodorus capitalises on his readers' awareness of a recent historical event to add verisimilitude to his narrative of Syene. Just as the siege scene as a whole is a major example of historiographic content in Heliodorus' novel, so the principal tactics and operations (the dykes and the flood) used at Syene add further historically credible details within the siege scene itself because they resemble those of the actual, contemporary siege at Nisibis.

Nevertheless, though Heliodorus and Julian coincide in the content of their narrative, they differ markedly in their form. Julian's narrative otherwise is selective in the scenes it narrates, confining its narrative focus only to fantastic events, and avoiding detailed explanation of tactics, or the motivations of either the commanders or troops. (25) His primary aim is to magnify the achievement of the subject of his two speeches, Constantius, rather than provide a clear exegesis of the siege. (26) Heliodorus, by contrast provides a far more complete narrative of his siege.

However, it is possible to show that Heliodorus also adheres to a distinct narrative form within the Syene scene, which, like the content of the episode, is recognisably historiographic, and therefore serves as a further point of intertextuality with the genre of historiography.

'Face of battle' narrative in Late Antiquity

Heliodorus' narrative of Syene exhibits many of the features that historians have identified as typical of late antique historiographic narrative of sieges. Particularly these features comprise a style of narrative that focuses on the experiences, especially the psychology, of the combatants and which has been defined as 'face of battle.' The term was developed not by a classicist but by the British military historian Sir John Keegan. (27) Keegan argued that a more effective method of assessing the progression and outcome of a conflict is to reconstruct the experience of front-line soldiers rather than to focus solely upon the point of view of the commander. His approach, therefore, examines psychology, motivation and human experience on the micro-level, rather than more traditional attempts to evaluate overall strategy and tactics at the macro-level of the commander. In The Face of Battle, Keegan applied this methodology to the battles of Agincourt, Waterloo and the Somme. Despite the limitations of the source material available to the historian of antiquity in comparison to that of more modern periods, this approach has found favour amongst some ancient historians. (28) In the absence of first-hand accounts of 'ordinary' soldiers on the front line--such as the letters and diaries of soldiers at the Somme that Keegan had used to construct his 'face of battle' histories of more recent conflicts--historians of antiquity have turned to narrative accounts of battle that had been written by former soldiers or could closely be linked to eyewitness sources.

Although Keegan and the ancient historians who followed him have been primarily interested in pitched battle, in a recent article Noel Lenski has offered a face-of-battle analysis of late antique sieges, in an attempt to uncover the 'lived experience of ancient battle from the perspective of those engaged in fighting.' (29) Lenski's methodology takes two sieges by ancient authors who themselves exhibit many of the concerns of Keegan's face-of-battle style. These 'proto-face-of-battle' narrators are Ammianus Marcellinus, who narrates a siege of the Mesopotamian fortress, Amida, in 359 by the Persian king Sapor, and a group of authors who describe a siege of the same city, Amida, in 502-3 by a later Persian king, Kavad. The main source for this later siege is Procopius, with additional material in the chronicles of Pseudo-Joshua the Stylite and Pseudo-Zachariah of Mytilene. (30) Lenski selects this group of authors because he presumes, a priori, that accounts by eyewitnesses are necessarily more inclined to be sympathetic to the experience of individual combatants because of their authors' participation in the events narrated, and that they will encapsulate this sympathy in their narrative of the sieges. Ammianus had been amongst the defenders of Amida in 359. (31) Pseudo-Joshua the Stylite was a contemporary of the siege of 502/503, although based in nearby Edessa; Procopius writing later in the 6th Century may have made use of the contemporary history of Eustathius of Epiphaneia, as did Pseudo-Zachariah of Mytilene. (32)

Ammianus' credentials as an eyewitness and particularly as a proto-face-of-battle narrator had also been the subject of a study by Kimberley Kagan only a year before the publication of Lenski's article. (33) Like Lenski, she identifies Ammianus' description of Amida in 359 as an almost unique example of ancient 'face of battle' narrative:

   [Ammianus' military narratives] convey the experience of battle
   from the perspective of the participants, they explain the outcome
   of events on the basis of generic causal explanations, and they
   convey the impression of the reality of combat. Like Keegan,
   Ammianus accentuates some of the more mundane, although not
   necessarily less heroic, aspects of combat. He seems fascinated by
   psychological reactions to casualties. Ironically, his
   determination to convey the 'atmosphere of battle' ... undermines
   the military historian's ability to explain the outcome of these
   battles accurately. (34)

Like Lenski, Kagan identifies eyewitness participation as the cause of Ammianus' interest in the experiences of combatants and of his face-of-battle style, though she is more critical of Ammianus than Lenski, condemning Ammianus' failure to provide a clear tactical explanation of the fall of the city. (35)

The following section of this article argues that Heliodorus, like Ammianus, uses a 'face-of-battle' style at Syene, although he was writing a work of fiction and, despite the paucity of biographical detail, it is reasonably safe to assume he had never experienced siege-warfare first hand. (36) For the sake of a more coherent and direct comparison, I use only Ammianus' Amida as the primary example of this narrative style, especially since, according to Lenski and Kagan, he is fully illustrative of the entire range of face-of-battle elements. It also provides an opportunity to compare the works of two contemporary fourth-century authors.

Syene and Amida compared

In terms of tactical details, the sieges of Syene and Amida have little in common. At Amida in 359, Ammianus was amongst the defenders who held out against the Persian king Sapor II for seventy-three days. (37) According to Ammianus, Sapor took advantage of the Emperor Constantius II's absence from the front to invade Mesopotamia. Sapor was initially reluctant to devote time to a siege, preferring to push on into Roman territory. However, when the son of one of his client-kings was killed by a Roman ballista, he vowed to destroy the city. After a series of attacks by the Persians, a plague, and a sally by a troop of Gallic auxiliaries, the city's walls were finally breached. They were brought down by excessive pressure from an interior ramp, built by the defenders to counteract a similar mound which was constructed by the Persians on the outside. (38) Sapor slaughtered many of the defenders, taking some prisoner but executing the Roman commanders. Ammianus, however, managed to escape and thus survived to write up his account twenty-five years later. (39) Amida is not the only siege in Ammianus' work,40 but it is the longest, most detailed, and is clearly designed to be a purple patch. Although Ammianus constructs his authority in the scene via autopsy, especially heavily marked by sections of first-person narrative, (41) he also peppers the scene with Homeric exempla through which he has 'moulded' his account to the literary tradition of siege narrative begun by the Iliad (42) Ammianus also creates authority, therefore, via emulation of the literary tradition. (43)

During the Amida scene, Ammianus' face-of-battle technique largely takes the form of an awareness of the human perspective on and responses to conflict, concentration on certain types of operation which allow the narrator to describe the emotional effect upon the defenders and attackers, and how both groups responded in terms of physical limitations and psychological reaction. (44)

Operations

Although the tactical details of the sieges of Amida and Syene may at first seem quite different, both authors converge in number of respects. First is the relative size of the armies. Lenski points to the usual disparity between attacker and besieged. (45) At Amida the Persians outnumber the Roman troops ten to one. (46) Heliodorus presents Hydaspes' attacking army as vastly larger than the defenders: 'countless thousands of Hydaspes' men.... reduce[d] the open plains of Syene to a narrow crowded passage.' (47) For both authors the psychological impact of the disparity is an important cause for defeatism amongst the defenders. Ammianus and his colleagues 'beholding such innumerable peoples ... despaired of any hope for safety' (48) and to the Syenians the Ethiopians' 'numbers [were] so vast that the mere sight of them made resistance inconceivable.' (49) Ammianus may well be tapping into a 'barbarian horde' motif here. (50) Comparison with other sieges in the Res Gestae reveals in which context Ammianus chooses to draw his readers' attention to the numbers of combatants: during the Persian campaigns of 359/360 he willingly records the small numbers of Roman defenders with precise figures, (51) whereas the attacking Persians are left comparatively and vaguely huge, (52) but when the tables are turned and the Romans are the besiegers then no details are provided for either side. (53) If the Persians are the aggressors, then Ammianus deploys the horde motif, whereas if the Romans are in that position, then he elides references to the relative size of the armies altogether.

Comparing Ammianus' practice to Heliodorus, we may have expected Heliodorus to present Oroondates as openly hostile, and Hydaspes, as father to Chariclea (although this has not yet been revealed to either at this stage in the plot), a more sympathetic character. However, the latter's casting as the commander of the 'barbarian horde' (a definitely negative characteristic) leaves Hydaspes' intentions ambiguous to the reader. It is a point that receives corroboration elsewhere in the scene, (54) and helps build suspense during this long hiatus in the plot, during which the main protagonists are absent, caused by the description of the siege before Hydaspes can interact directly with Theagenes and Chariclea much later in Book 9. (55)

Lenski points to sorties as a particular type of operation that receive greater attention amongst eyewitness, 'face of battle' historians. (56) They provide a good opportunity to narrate individual acts of heroism and to concentrate on the disheartening effect upon the besiegers. As mentioned above, at Amida a group of Gallic troops make multiple attacks upon the Persian camp (19,5,2), often returning with 'diminished numbers.' Ammianus portrays the most significant of these sorties as a night raid in the guise of the Iliadic Doloneia. The situation in Heliodorus necessarily does not allow for such expeditions--the lagoon between the besiegers and the besieged kept the two groups firmly separated. Nevertheless, Heliodorus shows he is aware that sallies were a common feature of sieges by explaining why they did not happen at Syene: Hydaspes' besieging army was intimidatingly large. (57) Heliodorus thus suggests that he viewed sorties as an essential facet of siege narrative, and he is compelled to explain their absence here to his reader.

Human limitations

Ammianus shows a particular concern for the ways in which certain human failings act as significant causative factors during the course of events at Amida, notably the lack of visibility caused by volleys of missiles ('a thick cloud of arrows in compact mass darkened the air' (58)), or by darkness (a Persian sneaks into Amida under the cover of night; (59) the Gauls make their sortie 'taking advantage of a gloomy, moonless night' (60)). Additionally he focuses on the effects of wounds upon the defenders (who 'when wounded in their great ardour for defence fell with destructive results' (61)), the weariness sustained by the lengthiness of the siege (which effected both sides equally, 'a truce of three days was granted by common consent, we also gained time to take breath' (62)), and particularly by the physical labour involved in constructing and maintaining both offensive siege works and defences: 'constant toil and sleeplessness sapped the little strength that remained,' (63) and 'not a man anywhere through fear of death gave up his ardour for defence.' (64)

Heliodorus focuses his narrative on the same factors. He acknowledges the usefulness of darkness for carrying out covert operations. Hence the city wall, once breached, is rebuilt during the course of a night (9,8,1), and the Persian commander waits for the cover of darkness before making a secret escape. In addition to serving a tactical purpose (Heliodorus notes that when the Persians escape, the 'Syenians were sunk in a deep slumber' (65)), darkness has important psychological effects upon the defenders and attackers: the return of light and day 'dispelled the fog of doubt and fear.' (66)

Physical activity plays a similar role. Toward the end of the siege, 'physically [the Syenians] were exhausted by their terrible predicament.' (67) The cause was primarily the maintenance of the defences: when the Ethiopians first start constructing the dykes, within the city 'no one was idle: women, children, and old men alike all joined in the work, for mortal danger is no respecter of age or sex.' (68) Once again, Heliodorus provides his reader not just with the description of action, but its psychological motivation.

Ammianus and Heliodorus share least in common when describing wounds. Whereas Ammianus frequently turns to the description and impact of injury, both physical and psychological, the overall tactical situation at Syene allows Heliodorus few opportunities for the Persians and Ethiopians to come to blows because they were separated by the artificial lake caused by the flooding. (69) Only when the Persian envoys approach the walls by boat in order to negotiate do the two sides become close enough to engage. The Syenians mistake the intentions of the Persians and start shooting, 'for when one is in an extremity of danger, nothing is without fear or suspicion.' (70) Once again Heliodorus provides a psychological perspective on causation. He subsequently includes details of the wounds suffered by the Syenians as the Persians retaliated and 'shot dead a couple or more, some of whom were catapulted headfirst from the walls ... by the force of an unexpected impact.' (71)

Psychology

Throughout my discussion of operations and human limitations the central aspect of Ammianus' and Heliodorus' 'face-of-battle' style has not just been the description of the actions of individuals but the role given to their psychological reactions and motivations as a prime explanatory tool for the narrator.

Both authors focus on the effects of sights and sounds. The image of the troops arranged outside the city strikes fear into the defenders. For Ammianus 'beholding such innumerable peoples ... we despaired of any hope of safety.' (72) Heliodorus refers to 'numbers so vast that the mere sight of them made resistance inconceivable.' (73) The focalization here is that of the defenders. The sounds of the sieges, the rushing of water and the crash of the collapsing dykes at Syene, (74) and the whirring of ballistae and the trumpeting of elephants 'whose clamor and immense bodies the human mind can conceive nothing more terrible' (75) at Amida, provide a rich aural sense of conflict in each narrative and strike fear into the combatants.

At Syene, Heliodorus frequently notes the defenders' (negative or defeatist) psychological reaction to these sights and sounds. Thus the 'thunderous and ear-splitting din' of the diverted river as it surrounded the city 'was enough to bring home to the people just what a desperate predicament they were in.' (76) Yet such a psychological reaction could have positive results: all the defenders quickly rushed to strengthen the defences. (77)

Ammianus and Heliodorus are in remarkably close alignment not only in recording psychological motivation, but also the types of event which provoke those responses. For example, there is a striking similarity in the initial episode of each siege scene. At Amida, Sapor approaches the wall on a magnificent and conspicuous horse, believing that 'all the besieged would be paralyzed with fear at the mere sight of him' (78) and would surrender immediately. Instead the defenders target him with missiles, and Sapor, described as 'raging' and 'outraged,' vows to destroy the city. The same sequence of over-confident commander, audacious riposte by defenders and the commander's psychological reaction of rage, which then prompts the siege and planned destruction of the town is played out in Heliodorus. Hydaspes 'attacked Syene, which he expected to capitulate with its walls intact before a blow was struck.' Instead the defenders assail him with 'outrageous and exasperating verbal abuse,' and 'furious' at this rebuff and refusal to surrender, he plans 'to destroy the city utterly and quickly.' (79)

Comparing these two accounts shows Hydaspes again characterized in a role which more naturally fits the commander of the side which is less sympathetic to the primary narratee. Above we noted how Hydaspes was the leader of the 'barbarian horde,' here we find him both arrogant and driven into the siege by negative emotions. It is typical of how Ammianus presents Sapor at Amida and elsewhere, (80) and finds repetition in Procopius. (81) Once again Hydaspes, despite being the father of Chariclea, is depicted in an ambiguous fashion.

The use of siege narrative in Heliodorus and Ammianus

The accounts of Amida and Syene share much in common, but a comparison of these two sieges also reveals some noticeable divergences, which have important ramifications for the understanding of each text. As already indicated, in Heliodorus Hydaspes is not presented as a sympathetic figure. Although he fights the Persians, who have already proved to be threatening to Chariclea and Theagenes, Hydaspes is cast in the typical role of the hostile aggressor: he is commander of a 'barbarian horde,' and is driven by rage not just to take the stronghold but to destroy it utterly. (82) These traits help sustain an air of ambiguity which surrounds Hydaspes up until this point in the narrative. He had only appeared fleetingly within embedded narrative prior to Book 9. (83) In Book 4, fear of her husband's anger and accusations of adultery had prompted Persinna, Hydaspes' wife, to expose her unusually white-skinned daughter Chariclea. (84) Later, Hydaspes engaged in the war with the Persians for territorial gain, and at the opening of Book 9 he plans to use Theagenes and Chariclea as victory sacrifices if the Ethiopians defeat the Persians. (85) Only later in Book 9 does his kindly disposition towards his longlost daughter become clear. (86) This technique of withholding information from the his primary narratee is typical of Heliodorus' primary narrator elsewhere in the Aethiopica, and is a good demonstration of how, despite being quite a self-contained scene, Syene is integrated with the narrative concerns of the rest of the novel. (87)

Conversely the comparison of the siege of Amida to that of Syene shows that, for all Ammianus' purported interest in individual experience, he fails to draw a distinction between two different groups within Amida: the indigenous civilian population and the Roman troops, who had temporarily taken refuge in the city. Heliodorus had made the tension between the Syenian population and the Persians a key element of his tactical and emotive explanations. The Syenians are painfully aware that they are trapped between two warring armies, thus after the wall partially collapses, they plead with Oroondates to seek a truce. (88) And once the Persians sneak away under the cover of darkness, they worry about Ethiopian reprisals for their perceived 'treachery by conniving at the Persian escape.' (89) Although Ammianus notes that 'within the limits of a city that was none too large there were shut seven legions, a promiscuous throng of strangers and citizens of both sexes, and a few other soldiers, to the number of 20,000 in all,' (90) he makes no attempt to distinguish the civilians and troops during the course of the siege. (91)

It is a curious omission from an author who, it has been claimed, is attuned to the sufferings and experiences of the besieged precisely because he was there amongst them and witnessed them. Ammianus' enumeration of the besieged clearly demonstrates that he was aware of the differing groups within the city, but chooses not to exploit this in his narrative. Perhaps this omission served a patriotic purpose--he wished to present a united front of 'we' Romans who 'burned, not with the desire of saving our lives, but...of dying bravely.' (92)

Face of battle and late antique literature

So far we have seen how Heliodorus and Ammianus share a narrative approach when they incorporate siege scenes within their respective works. Even if the tactical details of each scene may differ, both authors choose to narrate in a way that explains causation in terms of human experience and the psychology of conflict. There are even points at which the content as well as the form of the narrative coincide, particularly the riposte of the attacking commander at the opening of each scene.

What exactly is the relationship, then, between Ammianus and Heliodorus, and how may it explain their shared approaches to siege narrative? One issue should be addressed immediately, that Ammianus' 'face of battle' style is not a necessary result of his participation at Amida. Although Ammianus' presence may have prompted him to be sympathetic towards the experiences of the combatants at Amida, the existence of narrative patterning which shows similar concerns within a work of fiction demonstrates that this was as much a literary choice as a necessary result of autopsy. Equally one need only to turn to Polybius to see how another siege-participant could narrate sieges in an overly tactical method, which allows little space for the actions, let alone the emotions of individuals. (93)

For Heliodorus, too, this was an important choice of narrative presentation. Although there are several allusions to Herodotus throughout this most historiographic section of Heliodorus' narrative, Heliodorus did not draw upon Herodotus as a model for siege narrative. (94) Sieges are regular occurrences in Herodotus' Histories, but they are rarely treated as large-scale setpieces. (95) When Herodotus does lavish more narrative space on a siege, it is never on the scale of Heliodorus or Ammianus, and he is more concerned with tactical objectives and the method by which the stronghold was defended or captured than with the experiences of the combatants, such as we had seen in Heliodorus. (96) The intertextual relationship between the historiographic sections of Heliodorus and Herodotus have been well established by Morgan and Elmer, but it would appear that Heliodorus looked elsewhere for a model of narrative style of sieges. (97)

Could this be a localized phenomenon, confined only to Heliodorus and Ammianus, in which the later of these two authors had read the earlier, and subsequently applied the narrative patterning to his work? Ammianus is securely dated to the early 390s, whereas the all the evidence in favour of Heliodorus' fourth-century date points to the 350s, so Heliodorus seems not to post-date Ammianus. (98) The reverse is more plausible (though still unlikely) that Ammianus, as a native speaker of Greek, could have read Heliodorus, especially if the latter were a supporter of Julian, the hero of the Res Gestae, though it is unclear why Ammianus should wish to draw upon the narrative style of a Greek novel in writing Latin historiography, especially in an episode which has closer generic associations to epic and historiography. (99)

I offer one, admittedly speculative, theory to account for the relationship between these two texts. Both Heliodorus and Ammianus may be independent witnesses to a wider historiographic phenomenon, which developed in the later Roman Empire specifically within Greek historiography. Both authors share an important, if peripheral relationship with the genre of Greek historiography. As Morgan argued, Heliodorus' narrator relies upon some of the conventions of historiographic narrative to lend verisimilitude to his novel. Heliodorus not only incorporated tactical details of a recent historical siege (Nisibis 350) into his narrative, (100) but did so within a narrative format which would have been recognizably historiographic to his intended readership. By representing this aspect of his novel in an historiographic way, Heliodorus gains additional verisimilitude.

Ammianus, although a native Greek, (101) chose to write a large-scale history of the empire in Latin, and in so doing created the only extant Latin historiographic work since the early second century. (102) His was a consciously innovative resurrection of a dormant genre. By modelling his narration of Amida to a standard formula, Ammianus, nevertheless establishes his work's presence within a tradition of historiography, mapping his experiences onto a standard literary pattern.

This theory of course presumes that this style of siege-narrative had far more examples than only these two authors and was recognizable to Heliodorus, Ammianus and their intended readership as a narrative topos. (103)

If we assume, as I think we must, that Heliodorus predates Ammianus, it is now difficult to judge who Heliodorus' historiographic models were. Although Greek historiography in the late empire had not suffered the same fate as the genre in Latin, the works of the late third- and fourth-century Greek historians survive in either fragmentary form or as testimonia. (104)

However, if we look back to the last fully extant work of Greek historiography of the third century, Herodian, we find some of the same 'face of battle' elements of siege narrative (though in a less developed fashion) as Heliodorus or Ammianus. Herodian composed his Roman history in the middle of the third century, a culminating event of which is the siege of the northern Italian city of Aquileia by the forces of the emperor Maximinus in 238. (105) The city sided with the senate in a revolt against Maximinus, and the emperor was killed by his soldiers during the course of the siege, the first assassination in what was to become the year of the six emperors.

Ammianus has already been identified as a close reader of Herodian, and importantly Sotinel has argued that Ammianus used Herodian's description of the siege of Aquileia in 238 as an intertextual source for his description of a siege of the same city in 361 (Amm. 21.1 Iff.), though mostly in terms of the presentation of the protagonist than narrative style. (106) It is an important indication that Ammianus was conscious of the Greek historiographic tradition when he composed his work of Latin historiography. (107)

Although Herodian lacks the full range of face-of-battle material, he understands the role of morale, specifically the importance of success in generating hope and failure in causing despondency (dysthumia); (108) the potentially demoralizing effect of the huge numbers of attackers; (109) and he also focuses on wounds. (110) Importantly, Maximinus, whose role as victim of a rebellion but the enemy of the senate is ambiguous in Herodian, is driven to attack the city because of anger at being rebuffed by the defenders, a sequence familiar from Heliodorus and Ammianus. (111)

Additionally, the fragmentary evidence for the period after Herodian provides some insight to the continued scope and importance of siege narrative. Classicizing histories with political and military subject matter continued to be produced throughout this period, such as Rufus' Roman History in the second or third century. (112) In the early third century Asinius Quadratus wrote about the Persian wars of the second century in his Parthica. (113) Siege description makes up a large proportion of the fragments of the early-third-century historian Dexippus. (114) The siege of Marcianopolis shows some concerns for the experiences of the civilian inhabitants, dwells on the large numbers of 'barbarian' attackers and uses their emotions as explanatory devices. (115) The Scythian attackers begin the siege assuming 'they would easily take the city by assault' (.2), but running out of ammunition 'despaired of ever capturing the city painlessly and became dejected' (.4). By contrast the Romans' psychology develops in the opposite direction: 'taking heart from the ineffectual attack of the barbarians on the day before, [the defenders] let out a shout and fired such arrows and stones as they had.' (.5).

A recently discovered fragment of Dexippus also contains poliorcetic material. The fragment narrates an attempt to take a city by a series of plots, including a feigned retreat by the attackers. The attackers are aware of the usefulness of darkness for covert action, like Heliodorus' Persians when escaping Syene, or the Gauls when making their Doloneia at Amida. The attackers in Dexippus 'refrained from kindling a fire at night; it was necessary for them not to be visible.' (116)

Sieges were a common feature of wars against the Persian empire. (117) The inscription of Shapur I at the Kaaba of Zoroaster, commonly known as the Res Gestae Divi Saporis, boasts a catalogue of besieged strongholds in the Romano-Persian wars of the mid-third century, a culminating event of which was the capture of Antioch. (118) There was certainly enough material for these third- and fourth-century historians to include several siege scenes within their works. In the early fourth century, two remaining fragments of a historian named Eusebius also show a predilection for sieges. (119)

Conclusions

Although hypothetical, it seems likely that between Herodian and Heliodorus, the continued threat from the Persians coupled with the frequency of poliorcetic warfare that could penetrate uncomfortably far into areas of the Greek-speaking eastern empire (Antioch, for example) prompted historians of the period to develop a topos of narrative style, which favoured the description of the experience and psychological reaction of combatants and inhabitants over a tactical approach that was typical of Herodotus, Thucydides and Polybius.

By the second half of the fourth century, the topos was significantly well established that two authors deployed it within their respective works, although neither were works of Greek historiography.

On one level, intertextual engagement with this topos enabled Heliodorus to invest his Syene scene with an additional layer of verisimilitude. Not only were the events modelled on those of a recent historical siege, but the narrative form too was recognizably that of recent examples of the genre which was most closely associated with the representation of historical events. Heliodorus could then use his readers' familiarity with those narrative conventions to guide their interpretation of certain characters, especially Hydaspes.

Although this argument on its own is not an independent justification for Heliodorus' fourth-century date, and at the risk of circularity, it nevertheless helps corroborate a date subsequent to the siege of Nisibis in 350--Heliodorus strove to combine recognizably historiographic narrative content with recognizably historiographic narrative form. Heliodorus was attentive as much to developments in recent literary culture, as he was to the established place held by Herodotus within the tradition of historiography.

Participation in an event alone does not provide the literary ability to narrate it effectively. (120) Ammianus, a Greek resurrecting the genre of classicizing historiography in Latin, drew upon the same traditions of third- and fourth-century historians as Heliodorus when he narrated an event in which he participated twenty-five years earlier. No doubt he was drawn to face-of-battle style not only because it provided a legitimizing continuation and even appropriation of the conventions of recent Greek historiography, but also because it allowed him to incorporate his recollections of Amida in greater abundance and detail.

Even if these suggestions about the development of the siege topos are hypothetical, it remains that Heliodorus should be read within a late antique literary milieu, and not just as a late reader and appropriator of classical Greek literature. Equally, although Ammianus' appropriation of literary Latin is impressive, (121) his access to Greek literature (rare for a Westerner by this period) (122) was as productive an influence upon his creation of the Res Gestae. (123)

Bibliography

Ash, R. 2007. 'Tacitus and the battle of Mons Graupius', in: J. Marincola (ed.), A companion to Greek and Roman historiography, London: Blackwell, 434-440.

Bonamente, G. 2003. 'Minor historians of the fourth century AD', in: G. Marasco (ed.), Greek and Roman historiography in Late Antiquity, Leiden: Brill, 85-126.

Bowersock, G.W. 1978. Julian the apostate, London: Duckworth.

Bowersock, G.W. 1994. Fiction as history: Nero to Julian, Berkeley: University of California Press.

Bowie, E. 2008. 'Literary milieux', in: T. Whitmarsh (ed.), The Cambridge companion to the Greek and Roman novel. Cambridge: Cambridge University Press, 17-38.

Cameron, A. 2011. The last pagans of Rome. Oxford: Oxford University Press.

Coello, T. 1996. Unit sizes in the late Roman army. Oxford: BAR IS 645.

Dodgeon, M.H. & S.N.C. Lieu 1991. The Roman eastern frontier and the Persian wars (AD 226-363): A Documentary History. London: Routledge.

Drake, H. 2012. '"But I digress ...": rhetoric and propaganda in Julian's second oration to Constantius', in: N. Baker-Brian and S. Tougher (edd.), Emperor and author: The writings of Julian the apostate, Swansea: Classical Press of Wales, 35-46.

Drijvers, J.W. 2011. 'A Roman image of the 'barbarian' Sasanians', in R. Mathisen & D. Shanzer (edd.), Romans, barbarians, and the transformation of the Roman world, Farnham: Ashgate, 67-76.

Due, B. 1996. 'Xenophon of Athens: the Cyropaedia', in: G. Schmeling (ed.), The novel in the ancient world, Leiden: Brill, 581-599.

Elmer, D. 2008. 'Heliodoros's "sources": intertextuality, paternity, and the Nile River in the Aithiopika', TAPA 138, 411-450.

Elton, H. 1996. Warfare in Roman Europe ad 350-425, Oxford: Oxford University Press.

Gera, D. 1993. Xenophon's Cyropaedia: Style, genre and literary technique, Oxford: Oxford University Press.

Goldsworthy, A. K. 1996. The Roman army at war, 100BC-AD200, Oxford: Clarendon.

Greatrex, G. 1998. Rome and Persia at war, 502-532, Leeds: Francis Cairns.

Gruskova, J. 2010. Untersuchungen zu den griechischen Palimpsesten der osterreichischen Nationalbibliothek : codices historici, codices philosophici et philologici, codices iuridici, Vienna: Verlag der Osterreichischen Akademie der Wissenschaften.

Hanson, V.D. 1989. The western way of war: infantry battle in ancient Greece, London: Hodder and Stoughton.

Hilton, J. L. 2012a. 'The sphragis of Heliodorus, genealogy in the Aithiopika, and Julian's Hymn to King Helios', Agora 14, 195-219.

Hilton, J. L. 2012b. 'The cult of Neoptolemos at Delphi in Heliodoros' Aithiopika', Acta Classica 55, 57-68.

Hinds, S. 1998. Allusion and intertext, Cambridge: Cambridge University Press.

Hunter, R. 1994. 'History and historicity in the romance of Chariton', ANRW II.34.2, 1055-1086.

Hunter, R. (ed.) 1998. Studies in Heliodorus, PCPhS Supplement 21. Cambridge: Cambridge Philological Society.

Janiszewski, P. 2006. The missing link: Greek pagan historiography in the second half of the third century and in the fourth century AD. (trans. D. Dzierzbicka), Warsaw: Raphael Taubenschlag Foundation.

Jonge, P. de 1982. Philological and historical commentary on Ammianus Marcellinus XIX, Groningen: Bouma's Boekhuis.

Kagan, K. 2006. The eye of command, Ann Arbor: University of Michigan Press.

Keegan, J. 1976. The face of battle: A study of Agincourt, Waterloo and the Somme, London: J. Cape.

Kelly, G. 2008. Ammianus Marcellinus, the allusive historian. Cambridge: Cambridge University Press.

Keydell, R. 1966. 'Zur Datierung der Aethiopica Heliodors', in: P. Wirth (ed.), Polychronion: FestschriftF. Dolger, Heidelberg: C. Winter. 345-350.

Kim, L. 2008. 'Time', in: T. Whitmarsh (ed.), The Cambridge companion to the Greek novel, Cambridge: Cambridge University Press.

Kraus, C.S. 1994a. Livy ab urbe condita Book 6, Cambridge: Cambridge University Press.

Kraus, C.S. 1994b. '"No second Troy": topoi and refoundation in Livy, Book V', TAPA 124, 267-289.

Kuch, H. 1996. 'A study on the margin of the ancient novel: "barbarians" and others', in: G. Schmeling (ed.), The novel in the ancient world, Leiden: Brill, 209-221.

Kulikowski, M. 2012. 'Coded polemic in Ammianus book 31 and the date and place of its composition', JRS 102, 79-102.

Lacombrade, C. 1970. 'Sur l'auteur et la date des Ethiopiques', Revue des Etudes Greques 83, 70-89.

Lendon, E. 1999. 'The rhetoric of combat: Greek military theory and Roman culture in Julius Caesar's battle descriptions', ClAnt 18, 273- 329.

Lenski, N. 2007. 'Two sieges of Amida (AD 359 and 502-503)', in: A. Lewin & P. Pellegrini (edd.), The late Roman army in the Near East from Diocletian to the Arab conquest, Oxford: Archaeopress, 219-236.

Levithan, J. 2008. 'Emperors, sieges, and intentional exposure', in: E. Bragg, L. I. Hau & E. Macaulay-Lewis (edd.), Beyond the battlefields: new perspectives on warfare and society in the Graeco-Roman world, Newcastle: Cambridge Scholars, 25-45.

Lightfoot, C.S. 1988. 'Facts and fiction: the third siege of Nisibis (AD 350)', Historia 37, 105-125.

MacQueen, B. D. 2008. 'The stepchildren of Herodotus: the transformation of history into fiction in Late Antiquity', in: J. Pigon (ed.) The children of Herodotus: Greek and Roman historiography and related genres, Newcastle: Cambridge Scholars Publishing, 329-348.

Marincola, J. 1997. Authority and tradition in ancient historiography. Cambridge: Cambridge University Press.

Martin, G. and J. Gruskova 2014. '"Scythica Vindobonensia" by Dexippus (?): New Fragments on Decius' Gothic Wars', GRBS 54, 729-754.

Matthews, J. F. 1989. The Roman Empire of Ammianus, London: Duckworth.

Millar, F. 1969. 'P. Herennius Dexippus: the Greek world and the third-century invasions', JRS 59, 12-29.

Morgan, J.R. 1982. 'History, romance, and realism in the Aethiopika of Heliodoros', ClAnt 1, 221-265.

Morgan, J.R. 1989. 'An Ethiopian Story' in: B.P. Reardon (ed.) Collected ancient Greek novels, Berkeley: University of California Press, 349-588.

Morgan, J.R. 1996. 'Heliodoros', in: G. Schmeling (ed.), The novel in the ancient world, Leiden: Brill, 417-456.

Morgan, J.R. 2004. 'Heliodoros', in: I. de Jong, R. Nunlist & A. Bowie (edd.), Narrators, narratees, and narrative in ancient Greek literature, Leiden: Brill, 523-543.

Ogilvie, R. M. 1965. A commentary on Livy books 1-5, Oxford: Oxford University Press.

Paschoud, F. 1989. '"Se non e vero, e ben trovato": tradition litteraire et vertite historique chez Ammien Marcellin', Chiron, 19, 37-54.

Paschoud, F. 1999. 'Utrum in Amidae obsidione narranda Ammianus veritati magis an argutis commentis studuerit', in: K. Sallman & J. Blansdorf (edd.), Loquela vivida: donum natalicum Nicolao Sallman, Wurzburg: Konigshausen & Neumann, 81-86.

Paul, G. 1982. 'Urbs capta: sketch of an ancient literary motif, Phoenix 36, 144-155.

Perry, B.E. 1967. The ancient romances: a literary-historical account of their origins, Berkeley: University of California Press.

Rees, R. D. 2010. 'Form and function of narrative in Roman panegyrical oratory', in: D. Berry and A. Erskine (edd.), Form and function in Roman oratory, Cambridge: Cambridge University Press, 105-21.

Rolfe, J.C. 1935. Ammianus Marcellinus with an English translation. Vol 1, Cambridge, MA: Harvard University Press.

Rohrbacher, D. 2002. The historians of Late Antiquity, London: Routledge.

Rosen, K. 1968. Studien zur Dargellungskunst und Glaubwurdigkeit des Ammianus Marcellinus, Diss.: Heidelberg.

Ross, A.J. 2014 forthcoming. 'Constantius and the sieges of Amida and Nisibis: Ammianus' relationship with Julian's Panegyrics', Acta Classica 57.

Roth, J.P. 2006. 'Siege narrative in Livy: representation and reality', in: S. Dillon & K.E. Welch (edd.), Representations of war in ancient Rome, Cambridge: Cambridge University Press, 49-67.

Sabbah, G. 1978. La methode d'Ammien Marcellin: recherches sur la construction du discours historique dans les Res Gestae, Paris: Belles Lettres.

Sabin, P.A.G. 2000 'The face of Roman battle', JRS 90, 1-17.

Seyfarth, J. 1968. Amminaus Marcellinus: romische Geschichte, t.2, Berlin: AkademieVerlag.

Sidebottom, H. 1998. 'Herodian's historical methods and understanding of history', ANRW II.34.4, 2775-2836.

Sivan, H. 1992. 'The historian Eusebius (of Nantes)', JHS 112, 158-163.

Sotinel, C. 2005. Identite civique et Christianisme: Aquilee du IIIe au IVe siecle. Rome: Ecole Franjaise de Rome.

Swain, S. 1996. Hellenism and empire. Language, classicism and power in the Greek world, AD50-250, Oxford: Oxford University Press.

Szepessy, T. 1975. 'Die Neudatierung des Heliodorus und die Belagerung von Nisibis' in: Actes de laXIIe Conference International d'Etudes Classiques: Eirene, Bucarest: Editura Academiei Republicii Socialiste Romania, 279-287.

Szepessy, T. 1976. 'Le siege de Nisibe et la Chronologie d'Heliodore', Acta Antiqua Academiae Scientiarum Hungaricae 24, 247-276.

Tantillo, I. 1997. La prima orazione di Guiliano a Constanzo, Rome: "L'Erma" di Bretschneider.

Tomlin, R. S. O. 2000. 'The legions in the Late Empire', in: R.J. Brewer (ed.), Roman fortresses and their legions, London: Society of Antiquaries of London, 159-181.

Tougher, S. 2012. 'Reading between the lines: Julian's First Panegyric on Constantius II', in: N. Baker-Brian and S. Tougher (edd.), Emperor and author: The writings of Julian the apostate, Swansea: Classical Press of Wales, 19-34.

Tritle, L. 2006. 'Warfare in Herodotus', in: C. Dewald & J. Marincola (edd.), The Cambridge companion to Herodotus, Cambridge: Cambridge University Press, 209-223.

Trombley, F. & J.F. Watt 2000. The chronicle of Pseudo-Joshua the Stylite, Liverpool: Liverpool University Press.

Van der Valk, M. 1941. 'Remarques sur le date des Ethiopiques d'Heliodore', Mnemosyne 9, 97-100.

Whately, C.C. 2009. Descriptions of battle in the Wars of Procopius, Diss: Warwick.

Whitmarsh, T. 1999. 'The writes of passage: cultural initiation in Heliodorus' Aethiopica', in: R. Miles (ed.), Constructing identities in Late Antiquity, London: Routledge, 16-40.

Wifstrand, A. 1944-5. Eimra. Emendationen und Interpretationen zu grieschischen Prosaikern der Kaiserzeit. V: zu den Romanschriftstellern, Lund: Gleerup.

Alan J. Ross

University of KwaZulu-Natal

(1) Hunter 1998, iv, and borne out by the contributions by Bowie, Hardie and Morgan in the volume which he introduces.

(2) Elmer 2008, 418.

(3) Morgan 1982, 262.

(4) Even if the 'biological' model that the novel evolved directly from historiography is now discredited. Perry 1967, 32ff.

(5) MacQueen 2008, 340.

(6) Hunter 1994.

(7) Morgan 1982, 227-234.

(8) Cf. Kim 2008, 146-147. An exception could be Xenophon's Cyropaedia, which focuses on a historical character and contains some distinctly erotic content, notably the story of Panthea (Gera 1993, 221-244). However, even by Cicero's time it was considered to be prose fiction rather than a work of history or biography. Cic. Q.fr. 1,1,23, Due 1996, 588.

(9) Morgan 1996, 439.

(10) Heliod. 9,1-13.

(11) Paul 1982, 145.

(12) Examples of sieges in Herodotus: Sardis 1,80-86; Samos 3,54-56; Babylon 3,151-155; Barce 4,200; Paros 6,133-135; the Acropolis at Athens 8,52; Thebes 9,86-88. And in Thucydides: Corcyra 1,21-29; Plataea 2,15-18; Melos 5,144-156.

(13) Polybius 29,12,4. Cf. Paul 1982 for a survey of the development of the representation and use of the urbs capta motif in classical literature, which focuses on the after-effects of sieges.

(14) E.g. Josephus, Jotapata BJ 3,141-339 and Masada BJ 1,215-406; Arrian, Alexander's siege of Tyre 2,18-24; Cassius Dio, Byzantium besieged by Severus 15,11; Herodian, Aquileia 8,2-5.

(15) Morgan 1982, passim, esp. 221. Elmer 2008, 419-429. Morgan 1996, 439 also suggests allusions to Xenophon and Plutarch.

(16) Morgan 1996, 434-35.

(17) Elmer argues that in many places Heliodorus' allusions to Herodotus are too specific to create merely a historiographic pose, but instead invoke a specifically Herodotean pose, 'a deliberate and unmistakable appropriation of a specifically Herodotean attitude and manner of expression.' This, Elmer argues, undermines Morgan's thesis of realism. 2008, 420421.

(18) E.g. the Nile digression, Heliod. 9,9.

(19) E.g. Arsake's 'sexual predation' towards Theagenes in Book 7 has a strong thematic similarity to the Gyges story in the first book of Herodotus. Elmer 2008, 421.

(20) E.g. Morgan 1982, 233-234 notes that the first sentence of Book 10 seems to be an imitation of Herodotean phraseology which closes a digression. 'It is as though Heliodorus has marked the preceding narrative, the account of the siege of Syene, as a Herodotean logos.' Elmer 2008, 422.

(21) It is not the only argument for Heliodorus' fourth-century date. Hilton has recently identified intertextual references in Heliodorus to the works of Julian relating to solar theology (Hilton 2012a) and the representation of hero-cult at Delphi (Hilton 2012b). Whitmarsh (1999, 33 n.2) also points to Chariclea's 'martyrdom' as a further indicator of a fourth-century date, as has Wifstrand (1944-5, 36-41) on linguistic grounds.

(22) Julian Or. 1,27A-29A and Or. 3,62A-67A. The panegyrics were composed in 355 (Or. 1) and 356 or 357 (Or. 2); Tougher 2012, 21, Bowersock 1978, 43 n.10, cf. Drake 2012, 39. Van der Valk first pointed to the similarities between the two episodes in 1941 and suggested Heliodorus drew on Julian. This argument was endorsed and reinforced by Keydell 1966 and Lacombrade 1970. However, Szepessy 1975 and 1976 argued that significant discrepancies between Julian's account and that found in the Syriac hymns of Ephrem of Nisibis (an eyewitness to the siege) suggested that Julian took inspiration from Heliodorus, thus turning the date of Nisibis (350) from a terminus post quem to ante quem for the Aethiopica. Bowersock 1994, however, has convincingly overturned this view, demonstrating that Szepessy had relied on incorrect Latin translations of Ephrem, which distorted the supposed disparities between Julian and Ephrem. He also points to other evidence which corroborates the fourth-century date. There now exists something of a consensus over Bowersock's conclusion (his view has been accepted by: Morgan 1996, 418-9; Tantillo 1997, 305-11; Whitmarsh 1999, 33; Elmer 2008, 426; Hilton 2012b, 59), though there are still some dissenting voices (Swain 1996, 423; Bowie 2008, 32-35). See Ross 2014 for the difficulties in treating Julian's narrative as if it belonged to the genre of historiography rather than panegyric.

(23) Morgan 1996, 419.

(24) See Ross 2014 for an analysis of Julian's use of narrative for an encomiastic purpose, and Rees 2010 for the use of narrative in imperial Latin panegyric.

(25) As noted by Lightfoot 1988, 118.

(26) For which Julian has, perhaps, been unduly criticized, e.g. Lightfoot 1988, 123.

(27) Keegan 1976.

(28) Hanson pioneered the approach with Greek historical texts (Hanson 1989). The approach has also been applied to Roman history notably by Goldsworthy (1996) and Sabin (2000). Lenski (2007, 219) criticizes their tendencies 'to reduce investigations of battlefield experience to overviews of battlefield operations and to substitute analogies with modern comparanda for eyewitness accounts by the soldiers themselves.'

(29) Lenski 2007, 219.

(30) Ps-Joshua Chronicon 49-50; Procopius Pers. 1,7,5-35; Ps-Zachariah Historia Ecclesiastica 7,3-4.

(31) Signaled clearly in his narrative by the use of first person narrative. He enters Amida just before the Persians arrive (18,8,13) and makes a timely escape as the Persians take the city (evado 'I got away'19,8,5).

(32) Greatrex 1998, 62-67; Lenski 2007, 220 n.6. For Pseudo-Joshua, see Trombley & Watt 2000, xxi-xxii & xxvi. Here Lenski stretches the terms of his methodology--only Ammianus is a true eyewitness to the events he narrates, 'though the 502-503 siege is not reported in eyewitness accounts nor indeed the accounts of soldiers, it is described in three sources close to the event.' (Lenski 2007, 220). Lenski acknowledges that the absence of eyewitness sources for 502-503 'runs the risk of compromising the integrity and specificity of the evidence for the Amida sieges and thereby rendering a pastiche of battles, rather than a soldier's-eye account of a specific battle.' (2007, 219). But it is a risk he is evidently willing to take. See 'Face of Battle and Late Antique Literature' below for an alternative explanation.

(33) Kagan 2006. Neither author appears to have been aware of the work of the other, though they share a similar attitude to Ammianus' eyewitness narrative.

(34) Kagan 2006, 28.

(35) She contrasts Ammianus' style with that of Caesar, favouring the latter's 'intelligent, command-centered approaches' to combat narrative. Kagan 2006, 181.

(36) See Morgan 1996, 417-420 for a good overview of the evidence for Heliodorus' biography.

(37) Amm. 18,9-19,9. For an analysis of the scene, see Matthews 1989, 57-66, who largely believes that the narrative accurately reflects Ammianus' experiences as a participant.

(38) Gallic sally: 19,6; plague: 19,4; collapse of wall: 19,8.

(39) Sabbah (1978, 579-582) surveys the structure of the scene, noting the careful oscillation between Persian attack and Roman riposte. Kulikowski 2012, 81-82 offers a good overview of the arguments in favour of 390 as Ammianus' date of publication.

(40) There are three sieges of other Mesopotamian fortresses in the subsequent book, Singara (20,6) and Bezabde twice (20,7 & 20,11).

(41) For the importance of autopsy for the construction of authority in Roman historians, see Marincola 1997, 76-78.

(42) The verb is Kelly's, who catalogues Amida's Homeric exempla, 2008, 59-61. Amongst others he identifies a comparison of the death of the son of Sapor's client king Grumbates to that of Patroclus (19,1,9), and the Gallic sally (19,6) to the Doloneia of Iliad 10. The practice is familiar from Livy, who presents the siege of Veii in such a way as would recall Troy, Ogilvie 1965, 269, Kraus 1994b, 272. Livy, however, did not use a 'face of battle' style in his siege narratives, Roth 2006, 58. Paschoud has cast doubt on the veracity of Ammianus' account of Amida, arguing that the Homeric episodes are largely the historian's 'affabulation.' Paschoud 1999, 81, cf. Paschoud 1989. A similar view had been advanced by Rosen 1968, 68.

(43) Ammianus models aspects of his presentation of Sapor and Constantius at Amida on Julian's depiction of the same two figures in his narrative of the siege of Nisibis in 350, which is contained in his two panegyrics to Constantius. Although not directly involved, Constantius is brought into the narrative in similar ways and given a similar role by both authors. Ross 2014.

(44) I take the headings from Lenski 2007, 224-234, who surveys 'operations,' 'human limitations,' and 'psychology.'

(45) 2007, 225.

(46) Amm. 18,9,3-4, 19,2,14 & 19,6,11.

(47) Heliod. 9,1,2. All translations of Heliodorus are from Morgan 1989, occasionally slightly adapted.

(48) Amm. 19,2,3. All translations of Ammianus are from Rolfe 1935.

(49) Heliod. 9,1,2.

(50) For the 'barbarian horde' motif in historiography, see Kraus 1994a, 131 & Ash 2007, 439. It was common also in the later empire, e.g. Lact. mort. pers. 9,7.

(51) Two legions (c.2,500-3,500 men) at Bezabde and three at Singara (c.3,500-5,000 men). Amm. 20,6,8 & 20,7.1. The figures are Lenski's based upon Coello's calculations of unit sizes (1996). For a more conservative suggestion of about half those numbers, see Tomlin 2000, 169-173.

(52) Sapor begins the siege of Singara armis multiplicatis et viribus, 'having increased his arms and power' (20,6,1)

(53) E.g. The second siege of Singara in 360, narrated in Amm. 20.11.

(54) See under 'Psychology' below.

(55) Heliod. 9,24.

(56) Lenski 2007, 226.

(57) 'No one had the courage to make a sortie from the city to attack an army so immeasurably strong.' 9,3,2.

(58) sagittarum creberrima nube auras spissa multitudine obumbrante. Amm. 19,2,8, Lenski 2007, 227. De Jonge (1982, 40) suggests this is also an allusion to Virgil, Aen. 2,621: dixerat et spissis noctis se condidit umbris. It is also likely to be a reference to Dieneces' laconic quip at Thermopylae, that the Spartans would willingly fight in the shade caused by the numbers of Persian arrows (Hdt. 7,226).

(59) Amm. 19,5,5.

(60) Amm. 19,6,7. Lenski 2007, 227-228.

(61) Amm. 19,2,9. Lenski 2007, 229.

(62) Amm. 19,6,13. Cf 19,2,14, 19,8,1 & 19,8,6. Lenski 2007, 229.

(63) Amm. 19,2,14.

(64) Amm. 19,8,2. Cf. 19,6,6 & 19,7,1. Lenski 2007, 229.

(65) Heliod. 9,10,2.

(66) Heliod. 9,8,4.

(67) Heliod. 9,10,2.

(68) Heliod. 9,3,8.

(69) Heliodorus draws his reader's attention to the distance between wall and rampart, by describing successive bowmen's failure to shoot an arrow from the walls to Ethiopians on the other side of the water (9,5,2-3).

(70) Heliod. 9,5,7.

(71) Heliod. 9,5,8.

(72) Amm. 19,2,4.

(73) Heliod. 9,1,2.

(74) Heliod. 9,3,5 & 9,8,3.

(75) Amm. 19,7,6.

(76) Heliod. 9,3.5-6. Other examples of psychological reaction to sights and sounds include: the collapse of the wall 'was greeted with a wail of despair' 9,5,1; the appearance of the Ethiopian boats made the Syeneans 'distraught with terror at their perilous plight', 9,5,6; and the breaching of the retaining dyke during the night 'was sufficient to fill men's hearts with dread.' 9.8,3.

(77) Heliod. 9,3,8.

(78) Amm. 19,1,4.

(79) Heliod. 9,2,3.

(80) A similar sequence of events enrages Sapor at Bezabde (Amm. 20,7,2).

(81) At the siege of Antioch in 540, the Persian king Chosroes responds with rage when the Antiochenes 'heaped insults' upon him and shot one of his emissaries before the beginning of the siege (Pers. 2,8,7). In his survey of 'intentional exposure' in sieges and battles, Josh Levithan (2008, 35) concludes that Roman commanders used risky appearances in sieges and battles as a motivational ploy to inspire their soldiers, regardless of the risks to themselves. In applying this formulaic action to hostile, Persian commanders, Ammianus and Procopius neatly reverse its outcomes.

(82) Both Ammianus and Julian attribute anger and desire for destruction to Sapor's motivations in the conduct of his wars against Constantius II in the 350s (Jul. Or. 1,28C). Drijvers (2011, 71) notes that Sapor possesses characteristics which are the inverse of those of the ideal Roman emperor. The antithesis between Hellene and 'barbarian' is also a theme running throughout all the Greek novels (Kuch, 1996). The ambiguity surrounding Hydaspes is further stressed by a comparison to Julian's panegyrics: Hydaspes as the attacker takes on the role played in the panegyrics by the hostile Sapor.

(83) Points made by Benjamin McCloskey in his paper 'Allusions to Alexander in Heliodorus: Guiding the Reader's Expectations in Aithiopika 9,3-22' at the 107th Annual meeting of CAMWS in 2011.

(84) The primary motivating event in the plot, which sees Chariclea's exile and return. Heliod. 4,8.

(85) Heliod. 9,1,4.

(86) Heliod. 9,25,3.

(87) For other examples of the primary narrator's controlled revelation of information to his primary narratee, see Morgan 2004, 526-533.

(88) Heliod. 9,5,1.

(89) Heliod. 9,11,3.

(90) Amm. 19,2,14. Rolfe's translation (which follows Clark's text here) has been adapted according to Seyfarth's and Sabbah's removal of Clark's insertion to ad usque numerum milium <centum> viginti cunctis inclusis. See Seyfarth 1968 ad loc. and de Jonge's endorsement (1982, 54).

(91) Although the inhabitants of Amida are also Romans like the troops who defended them (whereas the Egyptians are forced to share Syene with Persians), Ammianus elsewhere shows his awareness of the friction that can occur when troops are placed in proximity to urban populations, even Roman ones (e.g. 21.5.8 & 22,4,6). He downplays this distinction here.

(92) Amm. 19,2,13. His own, rather ignominious conduct at the end of the scene (in which he hid 'in a secluded part of the city' before making a get-away, leaving other officers to be captured by the Persians) rather contradicts the more noble sentiments he attributes to himself and his colleagues earlier in the siege. 19,8,5. Cf. Kelly 2008, 61 who suggests Ammianus' escape narrative is an allusion to the capture of Troy in Aeneid 2. There are also similarities to Josephus' account of his capture after the siege of Jotapata, BJ 3,340-392, especially since both men hide within the city after it has fallen.

(93) Perhaps this is not surprising from an author who also wrote a (lost) tactical treatise (Polyb. 9,20,4). Polybius had been present at the siege of Carthage in 146 bc. This section of his work exists only in fragments, but it appears that he included himself as an historical actor in his narrative. For Polybius' overly tactical approach to military narrative elsewhere, see Lendon 1999, 282-285 and Whately 2009, 89-91.

(94) Herodotus had been identified as major source for Heliodorus' historiographic pose by Morgan 1982, 231 & 235.

(95) More often than not a siege is mentioned in passing, without any details other than the outcome. E.g. Azotus, 2,157; Memphis 3,13; Naxos, 5,34; Miletus 6,18.

(96) Examples of longer sieges: Darius at Babylon, 3,151-157; Miltiades at Paros, 6,133-135; Themistocles at Andros, 8,111-112; Thebes besieged by Greek forces for supporting Persians, 9,86-88. Herodotus may deploy some 'face of battle' elements in his battle narrative (Tritle 2006, 210-213), but this does not extend to siege scenes. Instead, he seems particularly interested in how sieges are brought to a conclusion, e.g. Miltiades' abandonment of the siege of Paros after suffering an injury (6,135) or the treachery of Zopyrus at Babylon (3,155).

(97) Cf. n. 17.

(98) For the dating of Ammianus and Heliodorus, see n.39 and n.21 respectively. The later a date posited for Heliodorus, the more diminished the contemporary relevance of the intertextuality with Nisibis becomes.

(99) Although Ammianus knew another writer of prose fiction, Apuleius, Ammianus' allusions to Apuleius are mostly lexical, comprising single phrases, e.g. fortunae saeuientis procellae Amm. 14,1,6 ~ procellae saeuientis fortunae Apul. Met. 10,4, rather than extended narrative tropes. For a complete catalogue of possible 'borrowings' from Apuleius, see Fletcher 1937, 26-27, not all of which are convincing as allusions. Cf. Kelly 2008, 168.

(100) Cf. n. 22 above.

(101) Amm. 31.16.9.

(102) A series of breviary histories was published in the 360s and 370s by Aurelius Victor, Eutropius, and Festus (see Rohrbacher 2002, 42-63), with whom Ammianus placed his expansive, classicising history in rivalry (Kelly 2008, 240). Cameron (2011, 627-690) has thoroughly quashed the more extravagant theories surrounding the nature and importance of the Annales of Nichomachus Flavianus, which were most likely as breviary in nature as the other fourth-century epitomes, and probably published after Ammianus.

(103) A topos is a literary commonplace which does not necessarily require the author of one example of the topos to be aware of each or every other example of the same topos in the works of other authors. As Stephen Hinds puts it, 'rather than demanding interpretation in relation to a specific model or models ... the topos invokes its intertextual tradition as a collectivity, to which the individual contexts and connotations of individual prior instances are firmly subordinate.' Hinds 1998, 34.

(104) Millar 1969, 14-16 gives a classic overview of the 'renaissance' in Greek literature, particularly historiography, in the second and third centuries. Janiszewski 2006 provides an indispensable survey of the fragmentary Greek historians of the third and fourth centuries, which takes into account more recent developments in textual, literary and historical analysis of these authors. Brill's New Jacoby (BNJ), provides an updated version of F. Jacoby's Die Fragmente der griechischen Historiker I-III including revised and additional texts, translation, critical commentary and a brief encyclopaedia-like entry for each historian.

(105) For Herodian's novelistic tendencies, see Sidebottom 1998, 2827-2830.

(106) Hdn. 8,4,1. Sotinel 2005, 57-59.

(107) Kulikowski 2012 also argues that Book 31 was originally composed as a war monograph on the battle of Hadrianople in Greek, and was later revised and incorporated into the Latin Res Gestae.

(108) Hdn. 8,5,1-2 & 8,5,8.

(109) Hdn. 8,3,5.

(110) Hdn. 8,4,11.

(111) Hdn. 8,3,2-4,1.

(112) BNJ 826. Janiszewski 2006, 77-84.

(113) BNJ 97. Janiszewski 2006, 85-91.

(114) BNJ 100 F 25, F27 & F29, and a new fragment published in Gruskova 2010.

(115) BNJ 100 F 25. Cf. Millar 1969, 12-29 for Dexippus' trait of focusing his narrative on local populations rather than the army.

(116) [TEXT NOT REPRODUCIBLE IN ASCII] Vind. Hist. gr. 73, fol. 195r vv.7-8. Published in Gruskova 2010. See Martin and Gruskova 2014, 746748 for a tentative identification of this town as Philippopolis.

(117) Although sieges were relatively uncommon against other foreign enemies in the fourth century particularly, Elton 1996, 357.

(118) In either 253 or 260. For a catalogue and description of the sources, Roman and Persian, see Dodgeon & Lieu 1991, 53-54.

(119) One possible identification is Eusebius of Thessalonica. Janiszewski 2006, 54-77. BNJ 101 F1 & F2, contra Sivan's (1992) suggestion of Eusebius of Nantes.

(120) Lendon 1999, 274.

(121) As Kelly's (2008) analysis of Ammianus' allusive practices makes clear.

(122) Cameron 2011, 528-535.

(123) Earlier versions of this paper were presented at the 30th Biennial Conference of the Classical Association of South Africa in Bloemfontein, and at the Institute of Classical Studies, London. It was subsequently read in written format by Bernhard Kytzler and Benjamin McCloskey, whom, together with the journal's referees, it is a pleasure for me to thank for their numerous helpful suggestions.

Alan J. Ross currently holds a Marie Curie Intra-European Fellowship at University College Dublin. Previously he completed a doctorate at the University of Oxford and held a postdoctoral fellowship in South Africa. His research focuses mainly on late antique historiography and he has a forthcoming monograph on Ammianus Marcellinus.