Notes

1 THE RISE OF THE SASSANIANS

1 The Cadusaei were originally a proto-Median tribe who dwelt on the south-west coast of the Caspian Sea. See also pp. 368, n. 54.

2 The genealogy of Ardashir presented by Agathias is different from that given by Shapur I for his father on his Great Inscription. According to the latter, Ardashir was the son of King Papak (S KZ, Gk line 1, see p. 35), a certain Denak was mother of King Papak and a Radak was mother of Ardashir, King of Kings (line 55). Sasan was simply honoured as a lord, his name coming before that of Papak (line 46) but specifying no relation. Frye (1983:116–17) suggests that Sasan might have been the natural father of Ardashir, but the latter was adopted by Papak either after the death of Sasan or that of his own son Shapur. The version of Agathias appears to have been derived from a tradition which is also found in the Ka mak i Artaxs e r i Papakan (Book of Deeds of Ardashir, son of Papak) and is transmitted to later writers like Tabari and Firdawsi through the Khwaday-namagh (Book of Lords)—a fairly official historical work (now lost) which covered Persian history from its beginnings to the end of Khusrau II’s reign (AD 628) and compiled under Yazdgird III (631–57). Cf. Cameron, 1969–70:112–17 and 136–7; Widengren, 1971:714–25; Frye, 1984a: 266–7; and Felix, 1985:25.

3 The most decisive of the engagements was fought at Hormizdaga n (near mod. Gulgayagan, between Isfahan and Nihawand, cf. Widengren 1971:743) where Artabanus was killed. According to the chronology established by Nöldeke, the battle took place some time in September, AD 224. A more precise date, and one which is often encountered, is given by the so-called Chronicle of Arbela (Chronicon Ecclesiae Arbelae) attributed to Msiha-Zkha. First published by Mingana in 1907, the relevant part of this Syriac work says (ed. Kawerau, CSCO 467, pp. 29-30=Mingana, pp. 28–9):

The Parthians showed themselves to be strong and powerful and proud that they sought only murder, but God who has said through his prophet:

Though you soar aloft like the eagle, though your nest is set among the stars, thence I will bring you down…’ (Obadiah

4), restrained them and brought about their overthrow. In former times, the Persians had desired to overthrow the Parthians and many times they had put their power to test in battle; but they were repulsed and proved unable to get the better of the Parthians. However, the Parthians were weakened by the great number of their wars and battles (i.e. against the Romans). The Persians and the Medes perceived this and they joined forces with Shahrat, King of Adiabene and with Domitianus, King of Karka de Bet Selok; and in spring they engaged the Parthians in a mighty struggle. The Parthians were defeated and their kingdom ceased to exist. Beginning from this, they hurled themselves against Mesopotamia, then Bet-Aramaye, then Bet-Zabdai (i.e. Zabdicene) and then Arzon (i.e. Arzanene). Within a year, they took all the regions. All the Parthian effort at resistance was useless—for their day had come and the hour had arrived. In the end, they all fled to the high mountains, leaving to the Persians all their land and all the treasures stored in the Cities (i.e. Seleucia and Ctesiphon). The young son of Artaban (i.e. Artabanus), called Arshak (i.e. Arsaces), was killed in cold blood by the Persians at Ctesiphon, and they took up residence there and made it their capital. The day which saw the end of the kingdom of the Parthians, children of the mighty Arshak (i.e. the Arsacids), was a Wednesday, the 27th of Nisan, the year of 535 of the Greek kingdom (recte 28 April, 224).

In recent years, however, considerable doubt has been cast by modern scholars on the authenticity of this Syriac work as major discrepancies have been found between the only surviving manuscript and Mingana’s printed version. The historical information it provides must, therefore, be treated with extreme caution. Cf. Brock, 1979–80:23–4; and Bivar, 1983:92.

4 Artabanus V was not the sole Parthian ruler. He only had control of the eastern part of the Parthian Empire. His brother Vologeses (Valagas) V had been ruler of Mesopotamia and Babylonia since 207/8. Cf. Widengren, 1971:741.

5 This excursus on the origins of Ardashir is not preserved in the Armenian original of Agathangelos. It is a later interpolation and shows clear influence of Karnamak material (see Note 2). The sympathetic and romantic depiction of Ardashir in this excursus is not typical of the manner in which he is portrayed in the main part of the history. Cf. Felix, 1985:29–30.

6 By Assyria the interpolator means the area coinciding roughly with the later Nestorian ecclesiastical province of Bet Aramaye (=Sassanian Assuristan) so as to distinguish the magnates from Parthian noblemen.

7 This might well have been based on a historical person as among the names of the Sassanian court listed on S KZ (Gk line 54) is a certain Zig (sic) with the title of (lit.: ‘the meal announcer’, i.e. ‘chief of ceremonies’. Cf. Back 1978:348; Felix 1985:30.

8 George was the author of several poems on the reign of Heraclius (AD 610–41). His Herac/ias, composed after 628, is a survey of the achievements of the emperor at home and abroad.

9 This should not be taken to mean that Ardashir moved against Media and Armenia after his initial failure to capture Hatra. Parthia and Media were already in his hands and he was attacking Hatra from Media Atropatene. Cf. Widengren, 1971: 757.

10 The correlation of the events in Armenia between classical and Armenian sources reveals major discrepancies in chronology and genealogy. The main Armenian tradition, as represented by Agathangelos and (Ps.) Moses Khorenats’i, gives the Armenian king Khosrov (Chosroes) (I) as the hero of the hour. Also known as Trdat (Tiridates) (II), he was a relative of the defeated Artabanus (referred to as ‘brother’ of the Parthian king in the Greek version of Agathangelos, 9a.32, ed. Lafontaine), and it was he who thwarted Ardashir’s design by his skills in battle and he even raided Assuristan, the heartland of the new Sassanian Empire. Some time after Shapur I had come to power, this Khosrov was murdered by agents of the Persian king. Armenia was then successfully invaded by the Sassanians but his son Trdat (i.e. the future Trdat III or Tiridates the Great) managed to flee to the Roman Empire. Cf. Chaumont, 1969:25–57; Chaumont, 1976: 158–71. However, this traditional interpretation of the sources has been called into doubt by Toumanoff who, in an influential article (1969:237ff.), draws attention to the claim by the Armenian historian Elishe (Eliseus) (tr. Thomson, 1982:123) that the father of Trdat (i.e. Tiridates the Great) was murdered by his uncles and that according to the chronological synchronism of the Armenian historian Sebeos (Eusebius) this event took place in 287. To this we may add that one of the Persian agents who instigated the murder had the unlikely name of ‘Anak’ which is nothing more than the Parthian word for ‘evil’, the implication being that Khosrov/Trdat could not have been defeated in battle. Cf. Russell, 1982:167. Toumanoff (1969:250) argues that this Khosrov (I) was a composite figure, ‘a hyperbolic memory preserved by the Armenian tradition’. At the time of the coming to power of Ardashir in Iran and Iraq, the ruling Armenian king was Trdat II who was a relative of Artabanus V. It was he who resisted Ardashir with the help of first some unsubdued Medes and later the Roman forces sent by Alexander Severus. He was succeeded by Khosrov II who was murdered by his kinsmen and the throne was then passed on to Trdat (III) who ruled Armenia as a Persian vassal. The son of Khosrov II, Trdat (IV) fled to Rome and was restored by Diocletian after 298/9. Khosrov I was a much earlier ruler who died in captivity in Rome between 216 and 217. See further pp. 374–5, n. 26.

11 The emendation from ‘fourteen’ to ‘ten’ is generally accepted even though it has no manuscriptal support. Herodian states elsewhere (e.g. VI, 9, 8) that Alexander Severus ruled for fourteen years—though most modern scholars would subscribe to thirteen (i.e. 222–35). It is difficult to imagine that both the Persian and German campaigns could have been undertaken in the space of one year, especially when the former clearly spanned at least one winter. The date of 231/2 for the Persian campaign is also suggested by the numismatic and epigraphic evidence. Cf. Cassola, 1968:284n.; Whittaker, 1970:88–9, n. 1; and Felix, 1985:32–3.

12 It is questionable whether Ardashir had intended so early in his reign to restore the Persian Empire to its Achaemenid frontiers. Herodian was likely to have tried to explain present Persian intentions in the light of traditional Graeco-Roman (especially Greek) historiography. The desire of Ardashir to establish Sassanian control over the frontier client kingdoms like Hatra and Armenia would have inevitably embroiled him in conflict with Rome, which saw them as an essential buffer zone for its defences in the east. Cf. Potter 1990:371–80.

13 Ensslin (1939:128–9) has suggested on the basis of this inscription that the southern and central columns (see 1.3.3.) probably advanced as far south as Palmyra in order for Alexander to disguise the main direction of his attack. Alexander might have personally led the southern column as far as Palmyra before leaving it to join the main force which was following the Euphrates route.

14 The northern column or ‘wing’ of the expedition under the command of Julius Palmatus probably took the road from Zela (modern Sille) to Sebastopolis (modern Sulusaray), as evidenced by the inscriptions translated on p. 352. The route links the Pontic port of Amisos and the metropolis of Amasia with the main thoroughfare to Armenia passing through Sebastopolis.

15 The route of this southern column appears to be designed to bypass the Sassanian capital-complex of Seleucia-Ctesiphon and to ravage Mesene and Elymias. It would later reunite with the main force under the emperor. The difficulty of ascertaining the precise route of this column is intrinsic to Herodian’s weak grasp of geography of the more outlying areas of his known world. Cf. Cassola, 1968:298.

16 The most natural route for the central column, the main thrust of Alexander Severus’ attack, would have been to cross the Euphrates at Zeugma and then march across Mesopotamia via Carrhae to the Tigris. Cf. Dillemann, 1962:209. However, the mention of a visit by Alexander to Palmyra (see p. 23) suggests a more southerly route, perhaps a feint march to convince Ardashir that the main attack would come in the south. The latter appeared to have swallowed the bait as it was against the southern column that he directed his main counterattack. The central column may have sailed down the Euphrates past Dura Europos where the holder of the recently created post of dux ripae may have been in charge of transit arrangements. The army then marched across the Syrian desert to Singara.

17 The presence of substantial numbers of women in Persian expeditionary forces is often noted by Roman writers. See, e.g., Zonaras XII, 23 (ii, p. 596, 1–4), Libanius, or. LIX, 100 and Julian, or. I, 27A. Cf. Lieu, 1986b:480.

18 The hostility of the Arsacid royal house in Armenia to the Sassanians would have undoubtedly assisted the passage of this column through the kingdom. The close connection between Rome and Armenia in this period is also attested by the appearance of ‘Arta[xata]’, an important Armenian city, in a list of satrapies inscribed on a Roman shield found under the Tower of the Archers’ at Dura Europos. Cf. Cumont, 1926:331 (Les Parchemins IX, line 12).

19 This blaming of the domination of Alexander by his mother Mamaea for the disaster is typical of Herodian’s understanding of the last year of the emperor’s reign. Cf. Whittaker, 1970:114, n. 1.

20 The tradition of the annihilation of the southern column appears to have come from a source hostile to Alexander Severus and is inconsistent with Herodian’s own statement that the Persians failed to exploit their victory and agreed to a truce. Cf. VI, 6, 5–6. It is possible that the battle was bitter but indecisive, causing serious losses to both sides. Cf. Welles, 1941:100–1.

21 This brief but heroic depiction of Alexander’s achievements by the author of the SHA differs considerably from the more detailed one by Herodian. For a comparative study of the main sources on the Persian campaign of Alexander Severus, see Roesger, 1978:172–4.

22 Maricq, op. cit., 1957, suggests that Singara was Alexander Severus’ advance headquarters. It would have given him easy access to the Tigris route.

23 Modern Salihye, former Macedonian colony and Parthian administrative centre for the region of Parapotamia, it became a Roman frontier outpost on the Euphrates under the Antonines. Cf. Watzinger 1940:149.7–169.31; PECS, pp. 286a–287a (Hopkins).

24 The Greek word used here, often found in inscriptions i Modern Salihye, former Macedonian colony and Parthian adminisAsia Minor, ‘denotes a child reared from infancy, and probably, though evidence here is lacking, was applied to the purchased as well as to the slave-born in the household or rescued from exposure’ (Cameron, 1939:53).

25 As commander of a frontier garrison watching over a certain district (the ripa), the office of dux ripae at Dura Europos seems to presage the dux limitis of the fourth century. However, unlike the latter, he was certainly subordinate to the legate of Syria Coele. In this period both Syrian legions were stationed in the north of the province, mainly for the defence of Antioch and the Cyrrhestica; the garrison at Dura played a major ‘tripwire’ role in frontier defence. Cf. Gilliam, 1941: 169–71. See also Lieu, 1985:65–6.

26 i.e. to be remembered by a deity. Cf. Rostovtzeff et al 1952:33.

27 Probus was probably his comic ‘side-kick’.

28 The last line presents considerable difficulty in interpretation. Cf. Rostovtzeff et al. 1952:35, where other epigraphical parallels from Dura are cited.

29 The Osrhoenean archers later plotted against Maximinus after the murder of Alexander Severus. The plot fizzled out after the execution of the praepositus (?) of the Osrhoeneans, a certain Macedo, by Maximinus. Cf. Herodian VII, 1, 9–11. The Roman camp at Zagurae (Ain Sinu, on the main route between Singara and the Tigris) might have been the recruiting and training ground for these mounted archers. Cf. Oates 1968b:91–2.

30 Jardé (1925:81–2) has drawn attention to an African milestone, part of which reads: ‘[I]mp. Caes. Aurelio [Ale]x[andro] invicto pio [f]elice Aug. Divi Magni Severi [nep.] Partico (sic) max. [Persico] max.’, which appears to confirm in part the salutations listed here. See also Whittaker, 1970:124–5. However, it must be remembered that after the death of Elagabalus in March 222, his memory was condemned and his name was erased from some of his monuments to be replaced by that of his successor, Alexander Severus (e.g. ILAl I, 3892), with the latter assuming his titulature. Elagabalus himself had assumed his titles from Caracalla, his alleged father. Alexander’s salutation of ‘Part(h)ico [max.]’ might have originally belonged to Caracalla whom Julia Mamaea claimed to have been the father of both Elagabalus and Alexander Severus. The restoration ‘[Persico] max.’ is also questionable as the salutation was not generally used before Philip the Arab. Cf. Gricourt, 1961:319; Kneissl, 1969:167–8; Kerler, 1970:131.

31 Loriot (1975a:763) suggests that the fall of Hatra was part of a general invasion of Roman Mesopotamia which resulted in the capture of Rhesaina, Singara and possibly also the garrison camp of Zagurae (Ain Sinu, cf. Oates 1968b:91–2) as the first two cities were later restored to Roman rule by Timesitheus. It is just as likely that these cities were lost under Maximinus Thrax during whose brief reign (235–8) there was a complete suspension of minting in Mesopotamia. Cf. Kettenhofen, 1983:155.

32 The source is a small papyrus codex containing sections of biographical accounts of Mani, the founder of the gnostic evangelical religion of Manichaeism. Cf. Henrichs and Koenen, 1970:97–216; Lieu, 1985: 30–1.

33 This form of Ardashir’s name is unique and, if genuine, must be related to the Sassanian claim that the dynasty was linked to the Achaemenids. Cf. Henrichs and Koenen, 1975:20, n. 40. See also Henrichs and Koenen 1970:121, n. 53. The absence of a Greek nominal ending suggests the form *rdks r in Aramaic script. The Greek is probably a shortened version of the mp. Daryaw-Ardaxsahr. Cf. Sundermann, 1986:293, 390–1. Such combined name forms are not unusual as Shapur’s son Hormizd appears as Hormizd-Ardashir in his father’s inscription (S KZ, line 40). Cf. Back, 1978:190.

34 The garrison was probably starved into surrendering to the besiegers as archaeology has not so far yielded any clear evidence of a violent end to the city. Cf. Milik, 1972:355; and Drijvers, 1977:827.

2 THE PERSIAN EXPEDITION OF GORDIAN III

1 This was the revolt led by a certain Sabinianus (perhaps governor of the province of Africa) against Gordian III. Cf. Pflaum, 1960:829–30.

2 On Timesitheus, see 2.2.1.

3 According to the CMC (18.1–10, see above 1.5.1), Shapur received the diadem as co-regent on the eighth day of Pharmuthi in the year in which he captured Hatra (i.e. 17/18th April, 240). The exact date of his accession to sole rule is not known, but a date in 242 is commonly held to be the most likely. The length of his reign varies in our sources between thirty-one years, six months and nineteen days (cf. Nöldeke, p. 42) and thirty years (Elias of Nisibis, opus chron. CSCO 62, p. 92). Cf. Felix, 1985:43–4, Cameron, 1969–70; 138 and Mosig-Walburg, 1980:122–4. However, according to a recently published inscription on a column capital found near Shiraz (cf. Tavoosi and Frye, 1990: 30–8), the Roman invasion of Parthia and Persis took place in the third year of the reign of Shapur I. This would confirm the year of accession of Shapur as 240/1 and not later as most scholars have thought. Many problems surround the interpretation of this important new text and readers are advised to consult the publication by Tavoosi and Frye (op. cit.)

4 On the provinces of the Sassanian Empire see Map 1, p. xiii. For description and identification, see especially Honigmann amd Maricq, 1953:39–110. See also Morony, 1982: passim; Morony, 1984: 125–68; Gignoux: 1971 passim; Sundermann, 1979:144–5; Sundermann, 1986:281–4.

5 Edessa (modern Urfa), the chief city of Roman Osrhoene, was recaptured from the Parthians by Avidius Cassius in 165. On the status of the city under the Severans, see especially Wagner, 1983: 103–15.

6 Edessa ceased to be the capital of the Abgar dynasty under Caracalla and became a Roman colony. It still possessed the title of ‘the renowned Antoniniana Edessa, Colonia Metropolis Aurelia Alexandria’ in the Syriac Deed of Sale from Dura Europos (PDura 28, cf. Welles et al, 1959:146) which is dated to May 243. The restoration of Edessa as an independent kingdom must have taken place shortly afterwards. Cf. Drijvers, 1978:878–82; Kettenhofen, 1982:28–9. The presence of Gordian at Edessa and the restoration of the Abgar dynasty are both well attested to in local coinage. Cf. BMC Mesopotamia, pp. 113–18. Furthermore, Carrhae, Nisibis, and Singara all resumed the minting of Roman coins under Gordian. Cf. ibid., pp. 89–90, 121, 135–6.

7 The battle was fought between 13 January 244 (date of the last extant imperial legislation of Gordian, cf. CJ VI, 10, 1) and 14 March 244 (date of the first edict of Philip, cf. CJ III, 42, 6). Cf. Loriot, 1975b: 789, contra Honigmann and Maricq, 1953:122, n. 1. On Meshike, see Note on Peros-Sabour below. The fact that this major defeat was not mentioned in any Late Roman source attests to the tight control over ‘media coverage’ of this event.

8 The word is presumably a rare passive form of or a mason’s mistake for ‘anerethe’. It may also be an impersonal form of the passive, leaving open deliberately the precise cause of Gordian’s death. Cf. Kettenhofen, 1983:165; Felix, 1985:49. Sassanian imperial reliefs depict a dead Roman emperor (identified by the wreath around the brows) under the charger of Shapur I. Cf. MacDermot, 1954: 76–80 and plate IV.

9 The extant Roman and Byzantine sources on the death of Gordian can be divided into four categories according to their different emphases: (1) the tradition which mentions the opening of the gate of the temple of Janus by Gordian, his initial success against the Persians and his subsequent murder by Philip, cf. Aurelius Victor, Eutropius, Jordanes, and John of Antioch; (2) the success of the Roman arms under Timesitheus, the change of fortunes after his death and the death of Gordian in a mutiny incited by Philip and the burial at Zaitha, cf. Festus, Jerome, Ammianus, SHA, Epitome de Caesaribus, Orosius, Zosimus, Syncellus, and Zonaras; (3) Gordian died in battle after falling from a horse and crushing his thigh, cf. Malalas (ap. Sathas), Georgius Monachus, Cedrenus, and Zonaras (on Gordian II); (4) Gordian fell in ‘the ranks’ as a direct result of a colleague’s jealousy, cf. Oracula Sibyllina. None of the sources, with the possible exception of the Oracula, hints at the Roman defeat at Meshike. Cf. Loriot, 1975a:770–3; Potter, 1990:204–8.

10 Lines 19–20 of this oracle have long been a source of perplexity to the editors of the text as the first word of line 19: is clearly corrupt and the translation here follows Willamowitz’s suggestion of Still this is not entirely suitable to the context, nor will it scan. Among the various suggestions in Geffcken’s apparatus, makes the best historical sense. Potter (1990:166, 199–200) suggests an inversion of the two lines and gives as translation (ibid. 140): ‘he will fall in the ranks, smitten by burning iron because of jealousy and betrayed by a companion’. These two lines have been used by Honigmann and Maricq (1953:119) to prove that Gordian fell in battle against the Persians and was not victim to the conspiracy of Philip. However, as they stand, they imply that Philip deliberately brought about the death of Gordian in battle, which is not supported by other sources. Loriot (1975a:773) surmises that the fraus Philippi might have been a tactical error committed by the prefect, such as allowing the Roman army to fall into a position in which the redoubtable Sassanian cavalry could be deployed against it with the most devastating effect. The words here translated as ‘in the ranks’ can mean either ‘in battle’ or ‘in the camp’.

11 A votum ‘pro salute a[tque incolumitat]e et victor[ia]’ of Gordian (CIL III, 6763) dedicated by a legate of the XXII Legion at Mainz on 1 July 242, gives one of the earliest dates for the start of Gordian’s expedition. Cf. Townsend, 1934:128; Loriot, 1975a:759–60.

12 Zaitha is probably the name of a region and not merely of a place. According to the evidence of Ammianus and Zosimus (see p. 232), the monument would have been situated about twenty miles south of Circesium. The location of this monument in Persian-held territory may be significant as it might well have marked the actual place of his death and argues against the Persian claim that Gordian died in battle at Meshike. This was not the actual grave of Gordian, as his remains were taken to Rome (cf. Eutropius and Festus, trans. supra). Cf. Potter, 1985:174. For a more precise attempt to locate Zaitha, see Musil 1927:237–8.

13 Philip did not suffer damnatio memoriae in the reign of his successor, an alleged adversary (cf. Eutropius, IX, 2–3). This may imply that Gordian was the victim of a general uprising of the defeated Roman troops who were angry at the young and inexperienced emperor. Philip was not the most important man in imperial service at the time. His brother Julius Priscus would have preceded him in seniority in the praetorian prefecture. Philip’s elevation, as Potter (1990:204–12) has suggested, may have been due to the reluctance of those who held most power to select one of their members to rule when none of them was pre-eminent—witness the elevation of Diocles (later Diocletian) after the death of Numerianus and that of Jovian after Julian. The guilt of the murder was later shifted to Philip, the one who had apparently derived the greatest benefit from Gordian’s death. This ‘deformation’ of the facts in the sources probably took place during the reign of Decius, who was hostile to the memory of Philip, and was the work of the senatorial aristocracy who regarded the young Gordian as their creature. Cf. MacDonald 1981:507–8. See also Mazzarino, 1971a: 69–76; Kettenhofen, 1983:165–6, n. 67.

14 Reading ‘ac Sapore Persarum rege (post Artaxerxen) summoto [et post Artaxansen] …’. Cf. Soverini 1983, ii: 828. See also Zos. I, 18, 1.

15 The important role played by Timesitheus in the initial Roman victory at Rhesaina and the restoration of Carrhae (and Nisibis?) may also be commemorated in the Jewish Apocalypse of Elijah, which says in the second great Roman campaign against the Persians—the first being that of Alexander Severus—the prefects Philippus (reading ‘nph l pus for ‘phliphu) and Timesitheus (reading mistus bn priphqtus for dmitrus bn puriphus) will lead a force of 100,000 cavalry, 100,000 foot, and 30,000 men from ships. Cf. Krauss, 1902:363, 371; Krauss, 1903:627–33; Potter, 1990:194, n. 21. For a more cautionary note on Krauss’ emendations, see Honigmann and Maricq, 1953:119, n. 2. On the career of C.Furius Sabinius Aquila Timesitheus, see PIR2 III, 581; Howe, 1942:78–9; Pflaum, 1960:811–21; and Sartre, 1982:89.

16 A play on the words, reges and leges. Cf. Hohl, 1985, ii:319. The Persians might well have attempted to impose fire-worship on inhabitants of conquered Roman territories. See below, 3.3.3.

17 Carrhae and Nisibis were restored to Roman rule in 242 by Timesitheus before Gordian arrived in the main theatre of war.

18 The fact that the decisive battle was fought at Meshike/Pirisabora— later a key to the defence of Ctesiphon—bears out the claim of a deep penetration into Assuristan.

19 M.Julius Philippus (PIR2 III, 462) was probably vice-prefect with his brother C.Julius Priscus (see 2.3.2; cf. Howe, 1942:79, no. 46). The latter was already Praetorian Prefect while Timesitheus was still alive. Cf. CISem. II, 3932 (supra 1.3. 1), and Prentice 1908:312, no. 399. He may well have played a significant role in the elevation of this brother after the death of Gordian.

20 The retreating Persians probably carried out a scorched earth policy, as they would against Julian in 363. This would have accentuated the problems of supplies for the advancing Roman army which had begun its march much earlier than the normal campaigning season.

21 Reading for

22 This is almost certainly an error. The battle of Rhesaina was fought on the approaches to the Khabur, the main tributary of the Euphrates.

23 This fragment of Malalas (?) has been adduced as additional evidence by Honigmann and Maricq (1953:120, n. 1) for the death of Gordian in battle. However, one must bear in mind its similarity to accounts of the death of Gordian II in battle in Africa and of the death of Philip. MacDonald (1981:507) suggests that the topos of falling from the horse and fracturing the thigh may ultimately have its origin in the official report of the death of Gordian III submitted to the Senate by Philip. Cf. SHA Gordian 31, 2 and Zos. I, 19, 1. See also Mazzarino, 1971b:655–78 which traces the story back to the Anonymous Continuator of Dio Cassius. For a critical evaluation of the Byzantine chronographical sources on the reign of the three Gordians, see Patzig, 1896:41–2. Most of the relevant material is translated in Jeffreys, 1986:159–60.

24 He may have had to make his escape for fear of the mutinous troops. Cf. Oost, 1958:106–7; De Blois, 1978–9:13.

25 Philip probably paid a regular tribute to Shapur in addition to the 500,000 aurei mentioned by the S KZ. The burden of this expense would have been borne by the inhabitants of the eastern provinces. Cf. Potter, 1990:246. For the identification of the denarii of the S KZ with the aurei, see Guey, 1961:261–74.

26 Modern al-Anbar. It appears as Pirisabora in Roman sources. Cf. Ammianus Marcellinus XXIV, 2, 9; Zos. III, 16, 3 (Versabora). Its importance as a military depot is reflected in its Arabic name Anbar —‘the granary’. Cf. Morony, 1982:24; Morony, 1984:144–5. Known as Pumbedita in the Babylonian Talmud, it possessed an important Jewish community. Cf. Oppenheimer, 1983:351–68.

27 This must be distinguished from Lesser Armenia which had been part of Cappadocia since the first century AD and had as its provincial capital Sebasteia after the reforms of Diocletian. The division of the Kingdom of Armenia into Greater (or Pers-)Armenia and Lesser Armenia was consequent to the treaty between Theodosius I and Shapur III (c. 384). Besides imprecision of geographical terminology, Evagrius’ statement contradicts the assertion by Zosimus (III, 32, 4, trans. supra) that no Roman territory was lost by Philip. Cf. Chaumont, 1969:46; Kettenhofen, 1982:34–5, n. 72.

28 This appears to imply that Philip did not concede any territory to the Persians. Zosimus stressed this to underscore the loss of Nisibis by the Christian Jovian in 363.

29 The part of the treaty between Philip and Shapur regarding the position of Mesopotamia is also uncertain. The Roman colony of Rhesaina, liberated by Timesitheus, minted Roman coins throughout the reign of Decius (cf. BMC Mesopotamia, ‘Resaina’, pp. 127 (no. 10)-133 (no. 41). The appointment of Julius Priscus, his brother and Praetorian Prefect, as Prefect of Mesopotamia would indicate that the province was not entirely ceded to the Persians. Furthermore, the fact that Shapur had to precede his ‘second’ campaign in 253 with the capture by force of Nisibis and the first objectives of the campaign being Roman garrisons like Anatha on the Euphrates argues against the possibility of total abandonment. Cf. Kettenhofen, 1982:36, n. 72. On the other hand, withdrawal of Roman support for Armenia and Hatra might have been seen by some as significant contraction of the sphere of Roman influence. Cf. Wirth, 1980–1:334–5, n. 75. Diocletian’s acceptance of the Singara-Nisibis line as the boundary between the two empires even after Galerius’ great victory in 297/8 may indicate a partial abandonment by Philip of some outlying areas which were not regained by Timesitheus. Cf. Oates, 1968:89.

30 These two lines (28–9) are rendered here as traditionally interpreted by most scholars. Potter (1990:225–7), however, asserts that these lines only make sense in terms of events during the 240s if in line 28 is understood as a dative of advantage and the passage rendered, ‘but when the wolf should make oaths to the white-toothed dogs with respect to the flock’ (or, less prosaically, ‘with an eye to the flock’).

31 Lines 21–34 refer to events before the mid-summer of 247. Cf. Potter, 1990:229.

32 Pacatianus was an officer of the Danubian army; hence the ‘German Ares’. He rebelled against Philip in 247 but was murdered two years later. His rebellion was both disastrous to Philip and opportune to the Persians. Cf. Felix 1985:53.

33 The date of his appointment to the prefecture (or governorship) of Mesopotamia is hard to pin-point despite the evidence of Zosimus. An acephalous inscription found at Rome (CIL, VI 1638=Dessau 1331) gives the cursus of an equestrian official which completes a distinguished career in imperial service in both the prefecture of Mesopotamia and the praetorian prefecture. This unknown official had also performed the duties of praepositus of a legionary vexillation at the behest of one of the Gordians. The identification of him with C.Julius Priscus is a tempting one. Cf. Domaszewski, 1899:159–60; Howe, 1942:79, no. 46. Pflaum (1960:833–6; cf. PIR2 III, 488), however, has earlier argued against this and given a post-249 date to the praetorian prefecture of this unknown official on the grounds that some of the many posts held by him, especially the iuridicus Alexandriae vice praefecti Aegypti, could not be attributed with confidence to Priscus. More recently, Potter (1990:214) has drawn our attention to the fact that the province of Mesopotamia was ordinarily described as provincia Mesopotamiae et Osrhoenae after the abolition of the dynasty of Abgar at Edessa in 214 by Caracalla; this official would have been praefectus Mesopotamiae during the period (240–2) when the kingdom was temporarily restored by Gordian. See above, 2.1.4. See also Duncan-Jones, 1969:229–33; Duncan- Jones, 1970: 107–9. This may imply that if the official of the acephalous cursus was indeed Priscus, he would already have held the prefecture of Mesopotamia during the early part of the Persian campaign of Gordian, either prior to or concurrent with the praetorian prefecture.

34 The lines refer to the harsh administration of Julius Priscus who remained in charge of the east after the departure of Philip. According to Zosimus (I, 20, 2), the eastern provinces declared (M.F.R. (u.)) Iotapianus as emperor in 248 or 249. The revolt was dated to the reign of Decius (249–51) by Aurelius Victor (29, 1), and placed by Polemius Sylvius (Laterculus I, p. 521, 37–8, MGH) in Cappadocia but under Philip.

35 The precise powers of this office are unknown but the title certainly resembles that of ‘restitutor (or corrector) totius orientis’ granted later to Vaballathus (and Odaenathus?) by Gallienus. (See below, 4.5.5.) Cf. Pflaum, 1960:835; Paschoud, 1971:145, n. 46.

36 An official rating 200,000 sesterces.

37 Miliariae, indicated by a numerical symbol on the inscription.

38 The unit was originally recruited in the Near East and had spells of garrison duty in Germany and Pannonia. Cf. CIL XIII, 7323= Dessau, 9148. It was part of the growing field army when it returned to the east under Gordian III and retreated to Bostra after his death. Cf. Speidel, 1977:705.

39 The Legio VIII ‘Augusta’, or at least a detachment of it, was stationed at Syria at various times since Augustus. At first it was stationed mainly near Beirut and, under Philip, moved to Baalbek. Cf. Prentice, 1908:129.

3 THE SECOND AND THIRD CAMPAIGNS OF SHAPUR I AGAINST THE ROMAN EMPIRE

1 The date adopted here is that given by al-Tabari (eleventh year of the reign of Shapur: i.e. 251/2), but Nisibis is not among any of the lists of captured cities in the S KZ, nor is its capture stated unequivocally in Roman sources. Its recapture by Odaenathus for the Romans, however, is well known. Cf. SHA Gallieni Duo 10, 3 and 12, 1 and triginta tyranni 15, 3. The occupation of Nisibis would have been a logical preliminary step to the invasion of Armenia in 252, but a later date cannot be ruled out as the same consideration would have applied to Shapur’s ‘third’ campaign in 260, which began in Mesopotamia and was probably launched from Nisibis. Cf. Kettenhofen, 1982:44–6; Potter, 1985:329–30.

2 This passage is not found in the more recent edition of Eutychius by Breydey. 1985. See below, p. 394, n. 3

3 I.e. the reign of Trebonianus Gallus (251–3). This late date for the murder of Khosrov and the flight of Tiridates (c. 252) is accepted by Hewsen (1978–9b:99) on the basis of the traditional length of the reign of Khosrov (thirty-five years, i.e. 216/7–252). See also Kettenhofen, 1982:38–43.

4 The original manuscript gives ‘children’. The emendation to follows the suggestion of Markwart on the evidence of a letter from a group of Armenian priests and nobles to Theodosius the Great, which mentions the education on Roman soil of the young Tiridates after his escape from the hands of ‘cruel uncles and parricides’, preserved in the Armenian history of Elishe (trans. Thomson 1982:23). Cf. Chaumont, 1969:54; Felix, 1985:54.

5 The asylum given by the Romans to Tiridates probably constituted a breach of some form of ‘non-intervention’ clause in the treaty between Philip and Shapur and was therefore used as a pretext for launching an invasion which would pre-empt any Roman attempt to interfere in Armenia. Cf. Chaumont, 1969:58–60. For other suggestions, see Olmstead, 1942:257; Sprengling, 1953:84–5 and Ensslin, 1949:98.

6 Barbalissos (modern Qal‘at Balis) is listed among the Roman garrisons on the Euphrates in the Not. Dig (Or. XXXIII, 25). This defeat, like that of Meshike under Gordian III, finds no echo in Roman sources. The Roman army was probably the permanent garrison of Syria commanded by the legate of the province. Cf. Ensslin, 1949:100; Gracey, 1981:87.

7 The list gives the cities in groups, which suggests that after Barbalissos the main Persian army was divided into detachments for the purpose of raiding. It describes therefore a number of simultaneous rather than continuous actions.

8 Anatha, modern ‘A na, fortified island on the Euphrates. Cf. Musil, 1927:6–7; Kettenhofen, 1982:50–1; Kennedy, 1986b; Kennedy and Northedge, 1989. (Birth)- Arupan, modern Qreiye, a fortress in Syria. Cf. Poidebard, 1934:87; Honigmann and Maricq, 1953:162; Back 1978. (Birth-) Asporakan, modern Halebiye, a fortress in Syria, later the site of Zenobia on the Euphrates. Cf. Honigmann and Maricq, 1953:162–3.Gignoux, 1972:47; Back, 1978:191; Kettenhofen, 1982: 52. Sura or Soura, modern Suriya, near al-Hammam, city on the west bank of the Euphrates, important for its situation at the head of the Strata Diocletiana. Cf. RE IV A/1, cols 953–60 (Honigmann); Poidebard, 1934:71–88. These cities were clearly captured by the main Persian force on its march up the Euphrates before encountering the main Roman army at Barbalissos.

9 After its victory at Barbalissos, the Persian army probably advanced as far as Hierapolis (modern Menbij), an important route centre in Syria between Beroea and the Euphrates. Cf. RE Suppl. IV, cols 733–42 (Honigmann); and Cumont, 1917: 20. Using Hierapolis as a base, a Persian army group turned south and ravaged cities in the vicinity of Chalcis: Beroea, or Ber(rh)oia, modern Halab or Aleppo, cf. PECS, p. 150b (Rey-Coquais); Chalcis (ad Belum), ruins at Qennisrin, near modern ‘Is, important city in the Syrian limes, cf. Mouterde and Poidebard, 1945: 17, 21–3; Apamea, modern Qal’at el-Mudig, later metropolis of Syria II, cf. PECS, pp. 66b–67b (Rey-Coquais); and Rephanea, modern Rafniye, in Syria, former legionary headquarters of III Gallica, cf. Mouterde and Poidebard, 1945:29–31. All these cities He along the same main road from Hierapolis and the main objective of this group might have been the neutralization of the threat posed by the military base at Rephanea.

10 From Hierapolis another army group moved north towards Zeugma (Seleucia on the Euphrates), modern Balqiz, an important river crossing and legionary headquarters of IV Scythica (cf. Wagner, 1976: passim, especially pp. 165–288; PECS p. 1000 (Rey-Coquais); Mouterde and Poidebard, 1945:212–13), where it joined the main road from Edessa to Antioch, capturing also Ourima (?), modern Horum Huyuk, between Belkis and Rum Kale on the Euphrates (cf. Hellenkemper, 1977:467; Frézouls, 1977:167). This northern army group, consisting mainly of cavalry under Hormizd, then headed towards Antioch, capturing en route’. Gindaros, modern Gindaris, a large village in Syria administered from Antioch, (L) armenaza, modern Armenaz, and Seleucia. One would expect the last mentioned place-name to be that of Seleucia Pieria, modern el-Kabusaye, port of Antioch on the Orontes, cf. PECS p. 822 (Rey-Coquais), indicating an attempt by the Persians to cut off Antioch from the sea. However, it is difficult to envisage how the Persians could have accomplished their drive to the sea without first capturing Antioch. The first Seleucia could therefore be Seleucia (ad Belum?=Seleukobelos?, modern Seluqiye on the Orontes, cf. Ensslin, 1949:101; Kettenhofen, 1982: 67) which was captured by the Persian force advancing on Antioch from Apamea. Cf. Potter, 1985: 298–9, n. 2.

11 Cyrrhus, modern Qal’at Nebi Hum, chief city of the region known as the Cyrrhestica, NE of Antioch, was an important Roman military base. Cf. Frézouls 1977: passim. It was probably bypassed by the Persian column advancing on Antioch from the Euphrates and was not captured till later.

12 Either Seleucia Pieria, the port of Antioch, or Seleukobelos. See above. Note 10.

13 After the fall of Antioch, an army group moved into northern Syria via Alexandria (ad Issum), modern Iskenderun, a city on the Syrian coast founded by Alexander near the site of his famous battle, cf. PECS p. 38 (MacKay) and raided Nikopolis, modern Islahiya, in Cilicia Campestris, while another army group marched from Antioch up the Orontes, capturing Sinzara (?) (identified by Henning 1939:827 as Larissa on the Orontes in Syria, modern Saizar), Chamath (modern Hama, an ancient city in Syria on the Orontes, refounded as Epiphaneia by the Greeks, cf. Honigmann 1923:184 (no. 170); Sauvaget, 1941, i:36–53; Kettenhofen, 1982:70, Ariste (modern ar-Rastan, also known as Arethousa, cf. Honigmann, 1923:163 (no. 65); Kettenhofen, 1982:70). This particular expedition probably withdrew after its defeat by a local militia at Emesa. See below, Note 30.

14 Dichor, city or settlement of uncertain location probably south of Doliche in Euphratensis (modern Zevkir?) (cf. Honigmann and Maricq, 1953:154–5), and Doliche, modern Duluk, north of Gaziantep (cf. Kettenhofen, 1982:76–7), were probably raided by another Persian contingent pushing out from Antioch.

15 The listing of two place-names on the Euphrates in a south to north order: Dura Europos, see Chapter I, Note 23, and Circesium, modern el-Busaira, situated at the junction of the Euphrates and the Khabur, suggests that they were captured by a special invasion force from Pirisabora. Cf. Kettenhofen, 1982:83.

16 The appearance of Germanicia, could not save image in zone modern Mar’as , city and important communication centre in Cilicia Campestris, linking routes from Armenia to Syria and Cappadocia (cf. RE Suppl. IX, col. 70 (Treidler)), after Circesium returns the reader’s attention to the north. As the northernmost city listed on the S KZ for this campaign, it probably marks the point from which the Persians made their withdrawal. This seems to be confirmed by the immediately following mention of two place-names on the Euphrates route south.

17 There are at least two places with the name of Batna(e) which could have been captured by the Persians on their southward retreat. One is in Osrhoene, modern Sarug, an ancient city renamed Anthemusis by the Greeks which was capital of an eponymous region (i.e. Anthemusia), cf. Segal, 1970:34, 137. The other is in Syria, modern Tell Batnan, which was visited by Julian on his march from Antioch to the Euphrates in 363. Cf. Cumont. 1917:20. The first seems more likely to be the Batna mentioned here because we know that its walls were repaired by the prefect Aurelius Dasius some time after 256. The Batna in Syria was a only a large village and would not have merited mention as a city ‘with its surrounding territory’ in the S KZ. Cf. Petersen, 1977:277–8.

18 Chanar has been identified with caution by some scholars as Ichnai in Osrhoene. Cf. Honigmann and Maricq, 1953:155–6; Kettenhofen, 1982:77. This, together with the identification of Batnae with Sarug, points to the Persians withdrawing along the eastern bank of the Euphrates.

19 Most scholars are agreed that the following Cappadocian cities were raided by a Persian force which operated out of Armenia, commanded perhaps by the crown prince Hormizd (cf. Ensslin, 1949:104; Chaumont, 1969:63): Satala, modern Sadak, an important military base on the Cappadocian limes (cf. PECS p. 810 (Harper); Crow, 1986:84); Doman(a), city, probably in Armenia Minor and situated in the locality of Kose, (cf. Honigmann and Maricq, 1953:155–6); Artangil, city (?) of uncertain location (in Armenia Minor?) which must not be confused with Artaxanses in Armenia (cf. Honigmann and Maricq, 1953:156); Souisa, out-post(?) located near modern Kelkit/ Ciftlik, probably to be identified with Suissa, a statio in Cappadocia on the road between Satala and Nicopolis (cf. RE IV A 1, s. v. ‘Suissa’, col. 722 (Rugé); Kettenhofen, 1982:87); Souid(a), city(?) of unknown location (in Cappadocia?) (cf. Kettenhofen, 1982:87); Phreata, city(?) of unknown location, perhaps the Phreata placed by Ptolemy (Geog. V, 6, 13, ed Müller, p. 878) in Garsauritis (cf. Honigmann and Maricq, 1953:156).

20 Shapur’s invasion of Syria is firmly placed by Zosimus in the reign of Trebonianus Gallus, i.e. 253, as the passage earlier mentions events which took place towards the end of the reign of Decius. Cf. Felix, 1985:56.

21 See Note 2 above.

22 The name Mariades is probably a slightly Graecized form of the Syriac: Maryad’a, which means ‘my Lord knows’. Cyriades is a partial Greek translation of the name. Cf. Felix, 1985:59.

23 The metropolis of Antioch was probably the main Persian objective in the ‘second’ campaign of Shapur as it would have provided the most plunder and numbers of captives. The way to it was opened by the Persian victory at Barbalissos, and the fact that its capture is celebrated in the S KZ means that the exact date of the event is central to any attempt to date the whole campaign. Mentioned in the same campaign is Doura (Europos), the final capture of which could not have taken place earlier than 256 because of the presence on the corpses of Roman soldiers of Antoniniani of the second emission of Antioch for Valerian and of coins of the next-to-last issue from the same mint (both dated to 256) found in the grave of a hastily buried victim of the siege. Ammianus dates the capture of Antioch through the treachery of Mariades to the time of Gallienus. Though he was Augustus from 253 to 268, his sole rule did not begin till after the capture of Valerian in 260. The date of 253 for the ‘first’ capture of Antioch is, however, unequivocally stated by Zosimus and suggested by a break in the emission of its coins between the second issue of Trebonianus Gallus and Volusianus and the first issue of Valerian (spring 254). The second campaign (agoge) of Shapur as outlined on the S KZ could have covered more than one year of campaigning. Cf. Alföldi 1937:56; Kettenhofen 1982:65.

24 Rzach reads for i.e. someone who was summoned, an ally.

25 Most probably a corruption of Oromastes, the Graecized form of the Persian name Hormizd. The commander mentioned here might well have been the crown prince Hormizd-Ardashir. Cf. Felix, 1985:60.

26 The phrase ‘through the limes of Chalcis’ does not necessarily imply the existence of an inner defensive zone in the region of Chalcis for the protection of Antioch as interpreted by Mouterde and Poidebard, 1945, i:1–7. It may simply mean ‘by the Chalcis-route’. Cf. Liebeschuetz 1977:487–9.

27 The date of 314 (Gk ) Antiochene Era (=AD 265/6) given in the Bonn text of Malalas is an emendation from the impossible . The altered date is nevertheless still too late for the Mariades episode. Müller (FGH IV, p. 192) has proposed emending the figure to 304 (Gk =AD 255/6). Cf. Stauffenberg, 1931: 366. Fora detailed discussion of this passage, see Felix, 1985:56–8.

28 Cf. Downey, 1961:589–90, Nock, 1962:306–9; Lieu, 1986a: 44.

29 I.e. the siege of Bezabde in the reign of Constantius. See Chapter 8.

30 Emesa (modern Homs) was the capital of a local sheikdom which enjoyed a considerable measure of independence until the Flavians. However, the dethroned dynasts, the Sampsigerami, continued to be mentioned in inscriptions as members of a local aristocracy. Cf. Sullivan, 1977:219. The Sampsigeramus mentioned here is almost certainly the Emesene usurper, L.Iunius Aurelius Sulpicius Uranius Antoninus. His activities are attested mainly on the coins which were issued after his victory over the Persians. Though he adopted the titles of ‘imperator’ and ‘Augustus’, he at first struck coinage with local reverses and as his ambition grew he regarded himself as the junior colleague of Valerian and Gallienus. After Valerian’s victory over Aemilianus in 253, a compromise appears to have been reached between him and Uranius who renounced the imperial title, but, in exchange, ‘adopted a long legend, expressing his illustrious descent, and he retained the imperial bust with the laurel wreath.’ (Delbrueck, 1948–9:28). His subsequent fate is unknown. Whether the name Uranius is a Hellenized form of Sampsigeramus (=Syr.: Sismosgram (?) which means ‘the sun has decided’) is disputable. Domninus, Malalas’ source, might well have tried to conceal the identity of the usurper behind the more acceptable figure of the heroic priest of Aphrodite. Cf. Felix, 1985:61–3. On Uranius Antoninus, see especially Delbrueck, 1948; Baldus, 1971:236–55.

31 Domninus’ claim that Shapur met his death at Emesa is not supported by any other evidence. It may be that the commander of the southern Persian army group was killed in the manner described at Emesa. Emesa is conspicuous by its absence from the S KZ. Cf. Baldus, 1971: 240.

32 Malalas/Domninus is referring probably to the main Sassanian invasion of Armenia in 252/3.

33 Olmstead (1942:408) has interpreted these hardly legible and enigmatic graffiti as commemoration for the victory of a local hero over a major enemy. The stated date of 252/3 is certainly suggestive of the events described by Malalas. Cf. Baldus, 1971:250–1.

34 For this interpretation of the phrase ‘tute[lae] gratia’ see Speidel, 1977:

35 Aelius Aurelius Theon was imperial legate in Arabia from 253 to 259 as attested by inscriptions from Bostra (IGLS 9078–80). Cf. Sartre, 1982:92. The transfer of troops to Arabia might have been a direct reponse to the impending Persian threat.

36 Carrhae, modern Harran, ancient city and Macedonian colony in Osrhoene. Cf. PECS pp. 200b–201a (Segal); RAC XIII, 634–50 (Cramer).

37 This was almost certainly Successianus, who had earlier (c. 254) repelled an attack of the Borani against Pityus on the Black Sea and on account of this was made Praetorian Prefect and assisted Valerian in the restoration of Antioch (c. 256–7) after the earthquake. His complete disappearance from Roman sources after the capture of Valerian strongly suggests that he was taken prisoner along with his emperor. Cf. Howe, 1942:80–1.

38 Samosata, modern Samsat, former chief city of Commagene, then of Euphratensis, was an important route centre. Cf. PECS pp. 803b– 804a (Serdarogu); RE IA/2, cols. 2220–4 (Weissbach). The decision of Shapur to make straight for Samosata is completely understandable as it was there that Valerian had stationed his reserves. See 3.3.4. The main Persian army might then have moved on to Antioch-on-the- Orontes. Its capture is not listed in the S KZ but it is difficult to imagine how the Persians could have penetrated as deeply into Asia Minor as they did on this campaign with Antioch in Roman hands; unless the capture of Valerian and the destruction of the field army had such a demoralizing effect on the metropolis that its garrison refrained from interfering with the Persian raids.

39 Alexandria=Alexandria ad Issum, see above, Note 13. Like nearby Rhossos, modern Arsuz, the capture of which is known to us from Philostratus (ap. Malalas) but is not recorded on the S KZ, Alexandria probably fell to a Persian raiding party from Antioch. Cf. Kettenhofen, 1982:102–3.

40 After the capture of Samosata, the main Persian army under the command of Shapur headed for the Cilician coast at Katabolos (Kartarayya), city(?) of unknown location (in Cilicia, near modern Burnaz?) and made for the important city and route centre of Tarsus, modern Tarsus, cf. PECS pp. 883b–884a (Gough) and capturing en route in Cilicia Campestris: Aig(e)ai, modern Ayas, a port and naval base, Mopsuestia, modern Misis (cf. PECS pp. 593b–594a (Gough)), Mallos, a Roman colony in Cilicia Campestris of uncertain locality (probably Bebeli near Karatas), (cf. PECS p. 547b (Gough); Honigmann and Maricq, 1953: 159), and the inland cities of Adana, later Antioch ad Sarum, modern Adana (cf. PECS, p. 8a (Gough)), and A(u)gusta or Augustina, colony (?) in Cilicia(?). The latter is mentioned only in the Middle Persian version (line 17) of the S KZ: ‘gns (t)yn’ y, and is difficult to locate. Honigmann (1939:37) has suggested that it might have been a translation of the Persian for Sebastia. See also Maricq, 1958:341; Back, 1978:180.

41 After capturing Zephyrion, modern Mersin, city in Cilicia Campestris (cf. PECS pp. 999b–1000a (Gough)), this particular Persian army group probably continued westwards and added to their tally of cities: Sebaste, modern Ayas, a Roman colony and important naval station in Cilicia Campestris (cf. PECS, s.v. ‘Elaeusa’, pp. 294a–295a (Gough); Kettenhofen, 1982:111), and Korykos, modern Korgos, ruins near Kizaklesi, a city in Cilicia Aspera (cf. PECS, pp. 464b–465a (MacKay)). It fell victim to a seaborne attack led by Callistus (cf. Zonaras XII, 23, trans. infra 3.3.1.) at Pompeiopolis (Soloi), a coastal settlement immediately to the west of Zephyrion. Sebaste and Korykos were then possibly abandoned by the Persians as a consequence of their defeat by the Romans at Pompeiopolis.

42 After the unexpected turn of events at Pompeiopolis, the main army group, commanded probably by Shapur himself, withdrew through eastern Cilicia capturing the following six cities in the eastern part of the province: Agrippiada, modern Anavarza, Roman colony, also known as Anazarbos (cf. PECS, pp. 53b– 54a, s.v. ‘Anazarbos’ (Gough)) and Kastabala, modern Bodrum (cf. PECS p. 392, s.v. ‘Hierapolis Castabala’ (Gough)); Neronias (later Irenopolis?), a city in eastern Cilicia in the region of modern Yarpuz, east of Anazarbus (cf. Maricq, 1958:356; Kettenhofen, 1982:112); Flavias, modern Kadirli, a Roman colony (cf. PECS, p. 330, s.v. ‘Flaviopolis’ (Gough)); Nikopolis, modern Islahiya, (see above, Note 13) and Epiphaneia/ Oiniandos, modern Gozene near Erzin (cf. PECS p. 315 (Gough)).

43 The next series of six cities in the S KZ, all situated on the coast of Cilicia Aspera (later Isauria), poses considerable problems to the historical reconstruction of the events. Its position on the list after the cities captured by the main Persian force in Cilicia suggests that the cities were captured by an army group, probably a vanguard which had advanced beyond Sebaste before the Roman counterattack or a rearguard which had remained at Sebaste after Shapur had withdrawn his main force. From Sebaste the group proceeded to capture Kelenderis, modern Gilindire (cf. PECS, p. 445 (Mitford)) and Anemurium, modern Anamur (cf. PECS, p. 58 (Russell)), and reached the westernmost point of penetration by any Persian force in Shapur’s campaigns at Selinos, modern Selinti (cf. PECS, p. 823 (Bean)). It then divided itself into two groups to attack on their return the cities which had been bypassed along the coast. One group took Myonpolis, city of uncertain location, perhaps near modern Iskele (cf. Kettenhofen, 1982:116), while another went for Antioch (ad Cragum), modern Endiseguney (cf. PECS, p. 63 (Bean)), before returning east via Seleucia (ad Calycadnum), modern Silifke (cf. PECS, pp. 821–2 (MacKay)). There is no support for the suggestion of Downey (1961 : 589, n. 7) that the Antioch and Seleucia mentioned here signify the second capture of the two cities with the same names on the Orontes in Syria.

44 Dometioupolis, modern Dinbebol(?), city in Cilicia Aspera. The Parthian version (line 14) reads: miustynprws. Cf. Honigmann and Maricq, 1953:160; Gignoux 1972:58, s.v. ‘mzdnprws’. This city could have been captured by a splinter-force which advanced upon it from Selinus, as suggested by Sprengling (1953:105)—a route which would have involved travelling over difficult terrain, or by a force using the more natural route from Seleucia (ad Calycadnum).

45 After sending out a column to invest Pompeiopolis, the main Persian force withdrawing from Selinus then headed north for the important Roman colony of Tyana, modern Kemerhisar and later metropolis of Cappadocia II (cf. PECS, p. 942 (Harper)). There it split into two groups, one advancing deeper into Cappadocia towards Armenia Minor and the other westwards into Lykaonia. The first group captured the metropolis of Caesarea, formerly Mazaka, modern Kayseri, (cf. PECS, p. 182 (Harper)), probably after heavy fighting. From Caesarea, a minor force raided Comana, modern Sar (cf. PECS, pp. 233–4 (Harper)), and advanced as far north as Sebastia, modern Sivas, on the River Halys and later capital of Armenia Minor, (cf. PECS, p. 816 (Firatti)). The second group headed southwest from Tyana for Kybistra, modern Eregli (we have no option here but to abandon the S KZ order of Comana—Kybistra—Sebasteia), thence Birtha (=Barata? in modern Madensehir) (cf. Honigmann and Maricq 1953:152) and Rhakoundia, a city or settlement of uncertain identification but located probably near Barata (cf. Honigmann and Maricq, 1953:157–8). A subsidiary force might have advanced from Kybistra to Laranda, modern Karaman, while the main battle group advanced on their main objective of Iconium, modern Konya. Cf. Kettenhofen, 1982:117–22.

46 The Roman sources on the capture of Valerian give three contradictory versions: (1) he was captured after defeat in open battle, thus agreeing with the Persian account (cf. Eutropius and the Epitome de Caesaribus); (2) he was treacherously seized while he was trying to negotiate with Shapur (cf. Zosimus and Petrus Patricius); and (3) he sought protection from the Persian king from his mutinous troops at Edessa (cf. Zonaras). A major Roman defeat is the most likely as Valerian, like Crassus before him, had unwisely chosen to fight on the open plain between Edessa and Carrhae, which would have given the Sassanian cavalry considerable advantage. Cf. Potter, 1985:336.

47 It is unlikely that Lactantius here has in mind the scene of the captured Valerian kneeling in supplication before a mounted Shapur as depicted on the Sassanian rock reliefs at Naqs-i-Rustam. The main inspiration behind this alleged comment of Shapur must have been Psalm CX, 1, chosen here as the fulfilment of a prophecy against an emperor who was a persecutor of Christians. The paintings referred to by Lactantius may be official wall-paintings depicting scenes from imperial campaigns. Cf. SHA Max. Thrax 12, 10, Tac. 16, 2 and 25, 4 and Aur. 10, 2. For a full discussion of this passage, see Schwartz, 1978:99–101; Felix, 1985:69. One cannot, however, rule out completely the fact that returned Roman captives who might have helped with the sculpturing of the Sassanian rock reliefs could have contributed their knowledge to the Roman official paintings.

48 ‘Gallienus could very effectively plead that the acute problems of central defence during his reign precluded any possibility of considering vengeance on Persia; but he may have wished to dissociate himself from the demoralizing effects of his father’s capture’ (Creed, 1984:86.)

49 Lactantius’ account of the treatment of Valerian at the hands of Shapur agrees in general with that of Tabari (pp. 826–7, Nöldeke, pp. 32–3, see Appendix 1, pp. 282–3.)

50 Coins of Valerian continued to be struck in Egypt until Aug./Sept. 260 but the earliest official documents of the usurpers Macrianus and Quietus appeared in late Sept. (PFlor. II 195:26 Sept. and POxy. XII 1476:29 Sept.). The last dated document of Valerian (with Salonina and Gallienus) on papyrus was issued on 28 Aug. 260 (POxy 2186). Cf. Felix, 1985:66. The capture could have taken place as early as late spring. Cf. Christol, 1975:819–20.

51 The main part of the vita of Valerian in the SHA has not survived but the extant portion gives the impression that it was positive and even laudatory, very different from the Christian accounts of his reign.

52 Reading: ‘…Sapori[s] rex regum Velsolus:’ (ed Hohl) for ‘…Sapori regi regum vel soli:’ (ed. Magie). Velsolus is otherwise unknown. The title of King of Kings was normally the preserve of the King of Iran and Non-Iran but it had earlier been allowed (with the right to exhibit it on coinage) to the kings of Pontus and of Armenia. Cf. Alföldi, 1939:175–6.

53 This letter which appears to be the reply from an otherwise unknown vassal king to Shapur’s message of victory is almost certainly apocryphal. The purpose of this falsification may have been a pagan riposte to the Christian accounts of Valerian’s capture, as exemplified by that of Lactantius, and underscores the fact that Romans were no less formidable and respected even in defeat. Cf. Alföldi, 1963:1–3.

54 The Cadusii or Cadusaei were mentioned in Xenophon (inst. Cyr. V, 2, 25 etc.) as a warlike people living on the south-west coast of the Caspian Sea. We are grateful to Sir Ronald Syme for drawing our attention to the fact that, in this period, this ethnic term features mainly in works of fiction such as the Life of Apollonius of Tyana by Philostratus (I, 19).

55 ‘Artavasdes rex Armeniorum’ should not be seen as a ‘classicization’ of Hormizd- Ardashir, the son of Shapur and the Sassanian king of Armenia.

56 The Great Inscription of Kirder the Mobed (KKZ lines 12–13, see 3.3.3.) speaks of the Sassanian invasion of Iberia, Albania, and other Caucasian and Caspian lands, which seems to imply that these territories had not yet been fully brought into the Sassanian politicial sphere. Cf. Toumanoff, 1969:255.

57 Shapur I had a reputation for cruelty and harshness to his enemies which even someone as devoted to his cause as the prophet Mani could not readily brush aside. Cf. Sundermann, 1981:107 (11.2, lines 1660–7).

58 The heavy fighting round the city gave currency to the rumour that Shapur slew 12, 000 Jews in Caesarea-Mazaka. Cf. Neusner, 1966: 45–8.

59 On Mani’s service in the entourage of Shapur and his travels in the frontier regions, see Lieu 1985:58–9, 63; Sundermann, 1981:95; Sundermann, 1987:64–5.

60 Alexander was a Neo-Platonist philosopher writing at the beginning of the fourth century.

4 THE RISE AND FALL OF PALMYRA

1 Septimius Odaenathus, the architect of the meteoric rise of Palmyra in the years after Valerian’s capture, was born c. 220 and was elevated to senatorial rank probably under Philip the Arab. His ancestry and early history are far from clear. Cf. PLRE I, pp. 638–9, Gawlikowski, 1985: 261 and Potter 1990:381–94. On the history of Palmyra under the High Empire, see esp. Drijvers, 1977:837–46 and Matthews, 1984: 158–73.

2 This inscription has been given an early date in this collection because of the absence from it of the more grandiose titles which Odaenathus enjoyed after his victory over the Persians and the Roman usurpers.

3 The safeguarding of the trans-continental trade, so vital to the prosperity of Palmyra, might have lain at the heart of this attempt at a treaty by Odaenathus with Shapur. Earlier, Palmyra’s trade had greatly benefited from her close connection with the Romans. The city became a civitas libera under Hadrian and a colonia under Septimius Severus. Archaeological evidence has shown that Palmyrene troops were stationed on the island of Ana on the Euphrates under the Parthians and continued to do so after the island had passed into Roman hands c. 208 and their presence is recorded as late as 252. Cf. Kennedy, 1986b:103–4 and Invernizzi, 1986:359. A date after the defeat of Valerian is commonly suggested for Odaenathus’ attempt at a treaty with Shapur, but he could have come to realize the precarious position of Rome after the fall of Ana in 253 and of Dura in 256.

4 The vagueness of this title probably reflects the confused state of the Roman East after the death of Gordian III. Cf. Gawkilowski, 1985: 261.

5 Prior to the discovery of this inscription, it was generally assumed that Odaenathus the Great was the son or nephew of another Odaenathus —the Odaenathus Senior mentioned by the Anonymous Continuator of Dio Cassius (trans. infra 4.5.1.). Upon the murder of this Odaenathus ‘Senior’, power was briefly held by his son Septimius Haeranes (c. 251) who was the grandson of Vaballathus Nasor. He was soon succeeded by Odaenathus ‘Junior’, i. e. Odaenathus the Great and husband of Zenobia. The late date of this new inscription dedicated to an Odaenathus who was the son of Vaballathus Nasor effectively rules out the existence of an Odaenathus ‘Senior’ who held power in Palmyra till 251. Odaenathus the Great was the same Odaenathus who was the son of Haeranes and grandson of Vaballathus Nasor. Septimius Haeranes was his son and not his brother or uncle. Cf. Gawkilowski, 1985:257–61.

6 Sherira Ben Hanina (c. 906–1006) was gaon of Pumbedita from 968 to 1006. His famous epistle, Iggereth Rav Sherira Ga’on, which chronicles the origins of the Mishnah and of the Talmud, was composed in 987.

7 Neusner (1966:48–9) believes that the date should be altered to 262 or 263. On Odaenathus and the Jews of Babylonia, see De Blois, 1975: 12–16 and the documents translated in Neusner, 1966:43–52.

8 On the identification see Neusner, 1966:50 and De Blois, 1975:12.

9 Nehardea (modern Tall Nihar, cf. Oppenheimer, 1983:287) was an important Jewish settlement in the Middle Euphrates and the destruction of its famous academy is well documented in Rabbinic sources. The incident provides clear indication of the southward penetration of Palmyrene forces under Odaenathus. For a more cautious view, see Neusner, 1966:49–52.

10 The location of Sekansiv is unknown.

11 Mahoza was a suburb of Veh-Ardashir on the east bank of the Tigris and an important centre of Jewish life in the Sassanian period. See map in Oppenheimer, 1983:233. The name Mahoza is also applied to the area covered by a large conurbation comprising: (1) Mahoza (= Maizomalcha i.e. the ‘Royal Fort’ or ‘Royal Port’ mentioned by Ammianus), (2) Aksak (=ancient Opis), (3) Veh- Ardashir (=Coche) and (4) (the ruins of) Seleucia. Cf. Oppenheimer, 1983:179–93.

12 Pumbedita is the Jewish name for Pirisabora (al-Anbar).

13 T.Fulvius Junius Quietus (Aug. 260–261) was the younger son of Macrianus and brother of T.Fulvius Iunius Macrianus. He was a tribune under Valerian. Cf. PLRE I, pp. 757–8.

14 The allegation that Ballista murdered Quietus and seized the throne for himself is not supported by other evidence. Cf. PLRE I, p. 146.

15 The imperial title granted to (or assumed by) Odaenathus is reflected in Manichaean missionary texts. Cf. Sundermann, 1981:42 (3.3, lines 451–2).

16 The person is otherwise unknown; the reading of Quintus (=Quietus) by Müller and the suggestion of Mai for Carinus (= Macrianus) are both rejected by Boissevain.

17 He was almost certainly the same person as Herodes, the son of Odaenathus by an earlier marriage. His victory over the Persians on the Orontes might have caused them to withdraw from Antioch.

18 He is likely to have been the same person as Aurelius Vorodes who had been honoured as early as 258/9. See above 4.2.4. The large number of dedications to him shows that he probably held the reins of power in Palmyra during Odaenathus’ periodic absences from the city.

19 A certain Vorod ‘the agoranomos’ is listed on the S KZ (Gk, line 67) among those who had submitted to Shapur’s authority. If this was the same person as Septimius Vorodes, then it would suggest that he was the leader of a pro-Persian party in Palmyra. Or he might have been the leader of the Palmyrene delegation which had unsuccessfully negotiated with Shapur. Cf. Schlumberger, 1972:339–41. Such an act of homage demonstrates the need for a trading city like Palmyra to keep her channels of communication open with both super-powers on which her trade depended.

20 We know from a Palmyrene inscription (CISem II, 3971, trans. infra 4.5.5.) that Zenobia is called Septimia Bath Zabbai and Bath Antiochus. The latter might have been Antiochus IV Epiphanes (175–64 BC) whose wife Cleopatra Thea was the daughter of Ptolemy VI Philometer (181–145 BC) or Antiochus VII Sidetes, the third husband of Cleopatra Thea. Cf. Ingholt, 1976:137.

21 Herodianus had already shared some of his father’s more important titles since 251 and was clearly groomed as his successor. This might well have angered Zenobia, who had designs for her son Vaballathus Athenodorus.

22 By ‘the younger Odaenathus’, the anonymous author must have meant Vaballathus Athenodorus. Cf. Mommsen, 1894:436, n. 2.

23 The allegation rests solely on her possible favour to Judaism and the Judaizing character of Paul of Samosata’s teaching. Cf. Millar, 1971: 13.

24 Iuppiter Hammon was the tutelary god of Bostra. Legio III Cyrenaica was then stationed at Bostra and it is highly probable that it was the standard bearer and the hornblowers (i.e. those responsible for the religious life of the legion) of this legion which took an active part in the plundering of the Temple of Bel at Palmyra (see below 4.9.4.) in revenge for this earlier act of sacrilege.

25 Commercial interests, rather than dynastic claims, lay behind the Palmyrene invasion of Egypt. The Blemmyae in S.Egypt had revolted and come from below the Second Cataract of the Nile (i.e. between Semna and Buhen) to occupy much of the Thebaid, thus threatening the caravan trade which used the Red Sea ports. Furthermore, L. Mussius Aemilianus, the prefect of Egypt from 258, had supported the usurpers Macrianus and Quietus and minted their coins at Alexandria. He was defeated and captured by Aurelius Theodotus, who then commanded the troops who had remained loyal to Gallienus and succeeded as prefect c. 262. Aemilianus might have been opposed to Odaenathus because of their destruction by the Palmyrene prince at Emesa. Cf. Schwartz, 1976:145–6.

26 Aurelius Timagenes probably belonged to a party in Alexandria which supported Odaenathus out of commercial interests and because he was acting on behalf of Gallienus when he suppressed the rebels. Cf. Schwartz, 1976:148–9.

27 Babylon (modern Kasr-Ash-Shama) was a major Roman fortress on the Nile Delta and the base of the Legio XIII Gemina (Not. Dig., Or. XXVIII, 15). It was the scene of a famous siege during the Arab Conquest. Cf. Butler and Fraser, 1978:238– 48.

28 Some Roman units, especially those of the cavalry, might have been pressed into service alongside the Palmyrenes in Egypt. Cf. Speidel, 1977:724.

29 This title has been translated by some scholars (e.g. Gawlikowski, 1985:256 as ‘corrector orientis’. However, a more exact equivalent of ‘corrector’ in Palmyrene is its transliterated form as held by Vaballathus. See below 4.5.5. Cf. Millar, 1971: 10.

30 The main text of the document, which is an application by the citizens of Oxyrhynchus to the keepers of the archives for the formal entry into their registers of a right of inviolability, is here omitted.

31 The author of the HA appears to be using the term ‘Saraceni’ here in the same way as Ammianus to mean the Scenitae Arabes. He was probably embroidering his third century source, the lost (so-called) ‘Kaisergeschichte’, with anachronistic material. Cf. Bowersock, 1987: 78–9.

32 This arrogant letter, allegedly translated by Nicomachus, but in fact the work of Longinus, demonstrates the eagerness of the author of the HA to exonerate Zenobia. Cf. Bowersock, 1987:78.

33 The titles ‘Sarmaticus’, ‘Armeniacus’ and ‘Adiabenicus’ have not yet been discovered on inscriptions of Aurelian. Cf. Kneissl, 1969:177. ‘Adiabenicus’, however, was part of the titulature of Vaballathus. See above, 4.7.3.

34 The ‘Third Legion’ was the Legio III Cyrenaica stationed at Bostra. See above, note 23.

35 Cf. Rey-Coquais, 1978:60.

36 Knowledge of the properties and villas in the suburbs of Rome is a characteristic of the author of the HA. Cf. Bowersock, 1987:78–9.

37 Anatolius was a native of Cilicia and was senator of Constantinople c. 390–393. Cf. Seeck, 1906:69, ‘Anatolius VI’ and PLRE I, pp. 61–2, ‘Anatolius 9’.

38 Cf. Seeck, 1906:146, ‘Eusebius XXXIV’ and PLRE I, p. 306, ‘Eusebius 27’.

5 FROM PROBUS TO DIOCLETIAN

1 According to Elias of Nisibis (CSCO 62, p. 46, 21–2), Bahram II came to the throne in 588 Sel. which begins on 1 Oct. 276, and was succeeded by Bahram III in 605 Sel. which begins on 1 Oct. 293 (CSCO 62 p. 96, 18–19) after a reign of seventeen years as stated by Agathias. Cf. Cameron, 1969–1970:142 and Felix, 1985:96–7.

2 On the revolt of the Blemmyae in Upper Egypt under Gallienus, see above Ch. 4, n. 25.

3 Bahram II is the only Persian king whose reign paralleled that of Probus (Jul. 276– Oct. 282). Narses was the fourth son of Shapur I and was the king of the Sakas under Shapur I. Bahram I, who succeeded his brother Hormizd after his short reign of one year, may have been the son of a minor queen or even of a concubine, and Narses objected to his being succeeded by his son Bahram II. However, there is no sign that he actually rebelled during the relatively long reign of Bahram II. He became king of Persian-held Armenia some time before 293 and therefore enjoyed the title of ‘King of Kings’. Cf. Frye, 1983:127–8 and Skjaervø 1983(ii): 10–11. Narses’ subsequent accession to Shahanshah of Persia was the outcome of a power struggle between two political factions, one supporting him and the other supporting Bahram III. Still, it is difficult to see how Bahram II could have entrusted the negotiations with the Romans to him, unless of course the negotiations had to do with Armenia. (On this, see below, note 26). The author of the SHA might have confused an embassy from Narses to Diocletian (infra 5.4.2.) with one from Bahram II to Probus. Cf. Felix, 1985: 97–8.

4 There is no numismatic evidence to support such a proposed campaign, but Probus’ peace treaty with the Persians might have been a precursor to military action with the restoration of Tiridates to the throne of Armenia as the possible aim. See below SHA Car. 7, 1, in 5.1.5. Cf. Chaumont, 1969:98–9 and Felix, 1985:98.

5 Hormizd, the brother of Bahram II, was the governor of Khorasan. With the aid of Kushans and the Gelani, he established an independent state in the East. This insurrection was suppressed and Bahram II crowned his son the future Bahram III as ‘King of the Sakas’. Cf. Agathias IV, 24, 7–8, trans. infra 5.3.2. The need to undertake such a major military campaign in Central Asia might have weakened the defences of Iran’s western frontiers, thus enabling Carus to capture Ctesiphon in 283. Cf. Christensen, 1944:227–30.

6 ‘Rufii’ is probably a misspelling or copyist’s error for ‘Cussi’—the people of the Central Asian kingdom of Kushan. Cf. Markwart, 1901: 36, n. 1.

7 The Gelani were inhabitants of a region on the S.W.coast of the Caspian Sea.

8 Coche is the name of a hillock around which Ardashir founded the city of Veh- Ardashir to replace the nearby city of Seleucia which had fallen into ruins after two important changes in the course of the Tigris towards the end of the first century had deprived the former Seleucid capital of its important river frontage. Veh- Ardashir is generally referred to as Seleucia in western sources. It was also the seat of the Catholicos of the Christians in Persia. Cf. Fiey, 1970b:44 and Oppenheimer, 1983:223–5. See above, Ch. 4, note 11.

9 Jerome’s dating is incorrect as nearly all our sources place the death of Carus in 283.

10 Carus bears the title of ‘Persicus maximus’ in his inscriptions and coins. See below Appendix 4, p. 337.

11 Gaius Sollius Apollinaris Sidonius (c. 430–479) was a Gallo-Roman of noble birth. After holding a number of minor offices at Rome and in Gaul, he was consecrated bishop of Auvergne. He was both famous for his skills as a poet and panegyrist and for his heroic leadership of his people in resisting the Goths.

12 Niphates is the name of a mountain in Armenia which is situated to the south of the headwaters of the Euphrates and to the north of Lake Van. Cf. Chaumont, 1976: 184. It is far too out of the way of the normal itinerary of an invasion directed against Ctesiphon, unless, like Trajan, Carus first marched into Armenia. See also the reference to Armenia in a story concerning Carinus (trans. infra 5.2.2.).

13 M.Aurelius Carinus was appointed emperor by his father Carus in 282 and placed in command of the western provinces. He was killed in battle against Diocletian at Margum in Moesia in 285. Cf. Felix, 1985:103.

14 Carrhae (ancient Harran) did not receive its name from Carus nor was it turned into a major fortress. It was abandoned by the Romans in 359 because of the weakness of its fortifications. Cf. Amm. XVIII, 7, 3.

15 Malalas’ source for the claim that Carus died fighting the Huns might have been the Anonymous Barbari (Chron. Min. I, p. 388, 15, MGH). Cf. Stauffenberg, 1931: 394–5.

16 Numerianus, the son of Carus, was allegedly a highly reputable poet and was said to have competed for literary honours against Olympius Nemesianus, the author of this poetic treatise. Nemesianus’s poem on the art of hunting is the only partially surviving one of three poems which were known to have been well received in the provincial cities. Cf. SHA Car. 11, 2. It was composed after the death of Carus but before that of his sons. Cf. Clinton, 1845:323. His praises for the deeds of the sons of Carus in the East are more literary than historical, but they may also reflect contemporary uncertainty over the precise nature of Carus’ death and the prosecution of the war. Cf. Felix, 1985: 103–4.

17 In the majority of the traditions concerning the martyrdom of Babylas, Philip the Arab, rather than Numerianus, was the emperor who ordered his execution. Cf. Lieu, 1986a:52–3.

18 This is followed in the account of Malalas by the story of the Saints Cosmas and Damian (pp. 304, 13–306, 6) which was set in Cyrrhestica in the winter period of truce after Carinus’ victory in a bitterly contested battle and prior to Carinus’ departure for Persia.

19 Numerianus was made Augustus two months before the death of his father. Cf. PLRE I, p. 634.

20 Synesius (c. 370–c. 414) was a native of Cyrene and was chosen bishop of Ptolemais c. 410. His speech de regno, which depicts the ideal Roman emperor, was delivered before the emperor Arcadius in 399.

21 Carinus could not possibly have taken part in the Persian expedition of 283. Chaumont (1976:101) proposes taking ‘Carinus’ as a confusion for ‘Carus’ and that this episode, fictional though it may first appear, suggests Carus’ intention to restore Trdat to Armenia—a suggestion which is supported by the Armenian historian Moses Khorenats’i (II, 79, Thomson, p. 227, see Appendix II, p. 318) who says that Tiridates served in the army of Carinus and, when the latter was defeated and killed by the Persian general Kornak, he fled across the Euphrates to Licinius.

22 The panegyrist is Mamertinus (cf. PLRE I, ‘Mamertinus 1’, pp. 539–40) and the panegyric was delivered on 21st April, 289, in honour of the emperor Maximianus.

23 This does not imply, as suggested by Christensen (1944:227), that the Romans withdrew in 283 after a peace had been concluded with the Persians. The reasons for Bahram II’s desire to negotiate must be found in the internal history of Persia. Cf. Ensslin, 1942:9 and Barnes, 1982:50–1.

24 These were not likely to have included territorial concessions. Shapur II would later name only Narses as the Shahanshah who had conceded territory to the Romans (Amm. XVII, 5, 6.). Cf. Ensslin, 1942:13–14 and Felix, 1985:105.

25 A literary topos which would be repeated almost verbatim by Pacatus in his panegyric to the Emperor Theodosius (Pan. Lat. XII/2, 22, 5).

26 The date of 287 (‘the third year of the reign of Diocletian’) is given by Moses Khorenats’i for the restoration of Trdat. This was carried out, according to Agathangelos (46, trans. Thomson, p. 61) with the aid of a large Roman army. Such a view, Toumanoff (1969:153–265) argues, may be over-simplistic. The negotiations between Probus and ‘Narses’ (see above Note 3) might have been over the recognition of the de facto division of the Kingdom of Armenia into a Roman (western and smaller) and a Persian (eastern and larger) zone. This would explain why, on his inscription at Paikuli (section 89, ed. Skjaervø, 1983:71), a monument raised between 293 and 296, Narses was congratulated by a ‘Trdat the king’. As Narses himself was the Sassanian king of Armenia until 293, Trdat could not have ruled over the whole kingdom from 287 onwards. The discrepancy has led Toumanoff (1969:253–9) to suggest that the Trdat of the inscription might have been one of the murderers of Khosrov II, the Armenian king who succeeded Trdat II and who was restored to suzerainty over western Armenia under the treaty negotiated between Narses and Probus in 279/80. Upon the accession of Narses as Shahanshah, Trdat (III) became king over the whole of Armenia, but ruling as a vassal of Persia, and it was in this capacity that he was listed among the congratulants on the Paikuli inscription. This state of affairs lasted until the defeat of Narses in 298 when the Romans restored Trdat (IV), the exiled son of Khosrov, as king of the whole of Armenia. Trdat II vanished from the records to be confused sometimes with Khosrov. The accounts of the latter’s death had subsequently been whitewashed by Armenian historians to make Khosrov the victim of a Persian-instigated plot. The restoration of Trdat (IV) the Great was virtually unattested in classical sources and ‘Armeni(a)cus Maximus’ did not become part of the imperial titulature till after 298.

27 The panegyrist is again Mamertinus and the panegyric, also in honour of Maximianus, was delivered in 291 (probably on 21 July).

28 This particular campaign can be dated with reasonable accuracy. Diocletian was in Emesa on 10 May, 290 (CJ IX, 41, 9) and the first extant legislation of his after his return to Sirmium (discounting CJ VI, 30, 6) issued from ‘Sirmi’ (sic) is dated 21 Sept. 290. Cf. Ensslin 1942:15. His motives for such a campaign are harder to discern. The destruction of Palmyra by Aurelian had led to a realignment of Arab tribes and Diocletian’s campaign may have been directed against those who were potential allies of the Persians and posed a threat to his plans for the reorganization of the frontier by raiding new foundations such as Diocletianopolis in S.Palestine. Cf. Felix, 1985:108. Diocletian took the title of ‘Persicus Maximus’ rather than ‘Arabicus Maximus’ in the same year (CIL III, 5810= Dessau 618), probably for reasons of prestige and propaganda. On the use of the word ‘Saraceni’ in the Latin Panegyrics, see Bowersock, 1987:73.

29 This attempt by Agathias to explain the origins of Bahram’s title is erroneous. The Saganshah was the Sassanian governor (with royal rank) of Sistan—a title which Narses also held. See the important discussion in Sundermann, 1987:60 and n. 141.

30 Agathias seems to have confused the length of Narses’ reign with that of Hormizd II. Narses reigned for nine years (293–302). Cf. Felix, 1985:110.

31 The Quinquegentianae were tribal inhabitants of the Atlas Mountains with modern Setif as their main centre.

32 Jerome’s dating is at least two years too early for the renewal of the war by Narses.

33 The fragment, probably from an epic composed in the reign of Diocletian, was discovered by Reitzenstein in the papyri collection of the University of Strasburg.

34 Enyo-Bellona was the Roman god of war.

35 The region of Nisaya in Media was famous for its horses and was therefore the traditional recruiting ground for Persian cavalry.

36 The more legible fragments of what must have been an epic poem on the achievements of the Tetrarchs appear to refer to the defeat of Diocletian and Galerius in Mesopotamia in 296. Cf. Barnes, 1976a: 182–3.

37 Constantius I (Chlorus) was the ruler who was occupied in Britain. The other ruler who was detained in Spain would have almost certainly been Maximianus. This Spanish campaign is otherwise unattested. The poet follows the Latin panegyrists in comparing the military exploits of Diocletian and his associates with those of their celestial patrons. Cf. Cumont, 1902:39.

38 A clear reference to Diocletian who was accompanied on his campaign by Galerius, his Caesar in the Tetrarchy.

39 The traditional order of events, viz. Diocletian left Alexandria for the eastern frontier in March 297 after the suppression of the revolt of Achilleus, has been revised by Barnes (1982:54) on new papyrological evidence. Diocletian went to Egypt after and not before the initial defeat of Galerius.

40 Though control of Arzanene and Gordyene would have given the Romans considerable territory east of the Tigris, the latter, especially its middle course, remained the de facto boundary between the two empires till 363. Cf. Amm. XVIII, 5, 3 and XXV, 6, 10 and 7, 8–9.

41 Galerius’ running before Diocletian’s carriage might have been a form of Tetrarchic court ceremonial rather than a sign of humiliation. Cf. Eadie, 1967:147– 8. Barnes (1981:17) suggests that Diocletian and Galerius took the field together and the junior colleague was made to shoulder the blame after the defeat.

42 The route of Narses’ advance is hard to discern. The venue of the battle gives the impression that the Persians were taking the Tigris-route with the intention of invading Osrhoene or that Narses was invading Mesopotamia from Armenia. Galerius was probably attacked (or tricked into attacking) before the arrival of the main force under Diocletian.

43 The Goths were pacified by Galerius before 297. Cf. Pan. Lat. IV, 10, 4.

44 Shapur I was the father, not the grandfather, of Narses.

45 Malalas has confused Maximianus ‘Herculius’, the then Augustus of the West, with Galerius, whose full name was C.Galerius Valerius Maximianus.

46 Archapet, i.e. Hargbed (mp. hlgwpt, pth. hrkpty) a very high ranking Sassanian official. In Sassanian inscriptions, his position comes before even that of the royal princes. Cf. Skjaervø, 1983 (ii):39.

47 The original manuscript reads ‘had the command of Sumia’ . The translation follows Peeter’s (1931:27) emendation from to . Cf. Ensslin 1942:51.

48 The five ‘provinces’ or satrapies ceded by Narses gave the Romans strategic control of the Upper Tigris, but this does not altogether seal off Armenia from Sassanian interference as it might have appeared at first sight. Intelene: region centred on Karkathiokerta (mod. Egil), 40 km. north of Amida, and later (Roman) Armenia IV after the partition of Armenia at the end of the fourth century, cf. Dillemann, 1962: 120–1; Sophene: region on the river Arsanias and centred on Asmosaton, cf. Dillemann, 1962:121–6; Arzanene: region east of the river Nymphios (mod. Batman Suyu) with capital at Cefa since the time of Constantius, cf. Whitby, 1983: 205–7; Cordyene: region south of Arzanene and north of the Tur Abdin. Cf. Dillemann, 1962: 110–12; Zabdicene: is the region surrounding the city of Bezabde. See below pp. 389–90, n. 19. For a detailed study of the territorial implication of this treaty to both sides, see esp. Blockley, 1984:29–36 and Winter, 1988:171–84.

49 The frontier agreed upon between Narses and Sicorius ran roughly along the Tigris in the north-east down to Bezabde (Amm. XX, 7, 1) and a number of regions beyond the Tigris (the Transtigritanian regions) came under Roman influence. From the Tigris the frontier ran south-east to include Singara and westwards to the Euphrates, following the river until Circesium where it met the Khabur. Cf. Gracey, 1981:96.

50 Some scholars have suggested that Zintha is a miscopying of Ziatha (cf. Amm. XIX, 6, 1), a fortress in Intelene, 156 km. north of Amida, probably to be identified with Kharput which according to Arab sources was earlier called Hisn Ziyad. Cf. Bivar, 1986:11. See also Ioannes Scylitzes, synop. hist. p. 316, 18–22, CFHB. Such an identification, however, would place the boundary too far to the west. Zintha as stated by Petrus must have been situated in Media (i. e. Media Atropatene). Cf. Chaumont, 1969:106–7. On the other hand, Winter (1988:181–2) argues for Zintha=Ziatha on the grounds that the place marked the boundary between Armenia and Rome and not between Armenia and Persia.

51 The designation of Nisibis as the sole venue for transaction between the two empires was clearly an attempt to limit the scope of espionage by cross-frontier traders. Cf. Felix, 1985:127 and Lieu, 1986b:491–2.

52 Constantine served on the staff of Galerius until his return to his father’s court under controversial circumstances in 305. If this statement was not an empty boast, then one may infer from it that Galerius undertook an invasion of Sassanian Mesopotamia on one of his two campaigns against Narses. Cf. Barnes, 1981:18 and Barnes 1982:63.

53 This has traditionally been assigned to the Armenian campaign of Maximinus Daia (infra pp. 145–6). Castritius (1968–1969:102) has conjectured that Verinus (Lucer)’s service should go back to the Persian War of Galerius and that his friendship with Constantine, therefore, goes back to 298. Cf. Barnes, 1982:118–19.

54 The oration was composed in the spring of 298.

55 The edict was issued in 302 and not in 297 as had previously been suggested by scholars. Cf. Barnes, 1982:169.

56 The Collatio legum Mosaicarum et Romanarum is an anonymous compilation composed between AD 390 and 428 with the explicit aim of comparing selected Roman legal norms with the Laws of Moses. It preserved the texts of a number of important imperial edicts from the Tetrarchic period.

57 War-psychology and political propaganda as well as religious conser-vatism were undoubtedly the main motives behind this edict. However, there is little to support the theory that Narses had hoped to use the Manichaeans as a fifth-column. The Manichaeans were a persecuted sect under Bahram II and most of the Manichaean missionaries in the Roman Empire would have been refugees from Sassanian Mesopotamia. Cf. Lieu, 1985:91–5.

58 I.e. the Elect members of the sect who were roving ascetics and missionaries.

59 The following is no more than a representative selection of the large number of inscriptions found in the frontier regions which are dated to the Tetrarchy. For other examples see IGLS 59, 499, 9058, 9060–9061 and Littmann, Magie, and Stuart, III.A, nos. 114, 205–207 and III.A. 2, Appendix V,A and B, VI,B, VIII,A and IX,A.

60 Palmyra became the base of the Legio I Illyricorum (cf. Not. Dig., Or. XXXII, 30) after its occupation by the Romans.

61 The Strata Diocletiana was a military road, constructed, or largely rebuilt, by Diocletian to act as the main line of defence for Syria Phoenice. The principal line ran south-east from Sura on the Euphrates to Palmyra and then along the Jebel Rawaq to Khan abu Khamat, near Dmeir (Thelsee). It continued from there to the Hauran and joined a similar road along the Arabian frontier. The Strata made use wherever possible of low hills both partly to enable the defenders to see better and partly because the higher ground helped to attract precipitation. Forts were built at regular intervals (about 30 Roman miles from each other) with intermediate signalling stations and watch-towers. Heavier units were stationed at key road-junctions to the rear of the Strata. The idea of basing a system of defence on a main road is not new, and in its Arabian section, the Strata Diocletiana ran for some distance parallel to the Via Nova Trajana. Cf. Poidebard, 1932:73–84 and Gracey, 1981:100.

62 The road which follows the depression of the Wadi Sirhan linked Damascus and Bostra with the Persian Gulf and its protection from raiding Saracens and Persians would have been the essential task of the Roman army in Arabia. Cf. Bowersock, 1983:95–109 and Speidel, 1987:213. On the threat of attacks from Saracens, see below 6.2.4.

63 The word ‘praetensio’ is a hapax and Speidel (1987:219–20) believes it is derived from the verb praetendere, ‘to be stationed out in front, beyond the main line of defense’. The praetensio at Basiensis was probably one of a chain of outposts responsible for patrolling and intelligence gathering.

64 This may indicate that an earlier structure, a praetorium, was replaced in 306 by a fort. Cf. Parker, 1986:54.

65 The dispute was between the Arab tribes allied to Rome and those under the leadership of al-Mundhir (Alamoundaras) allied to Persia.

66 The third factory was at Antioch. Cf. Not. Dig., Or. XI, 22–3.

67 We may presume that the walls of Edessa were restrengthened or rebuilt by Diocletian.

68 Qaryatein was Roman Nazala which was the base of a legionary cavalry unit (Equites promoti indigenae) according to the Not. Dig. (Or. XXXII, 23).

6 ROME AND SHAPUR II: SHAPUR AND CONSTANTINE

1 Adarnases (Adurnarseh) was the son of Hormisdas (Hormizd) II and not of Narses. His reign lasted for only 40 days. Cf. Felix, 1985:128.

2 This third son might have been the Shapur who was called the king of the Sakas in two inscriptions. Cf. Frye, 1983:133.

3 Cf. Christensen, 1944:234–5 and Frye, 1983:131–2.

4 Galerius Valerius Maximinus Daia was Augustus from c. 309 to 313. Cf. PLRE I, p. 579.

5 The origins of this particular campaign in Armenia are completely obscure. It might have been undertaken to assist Trdat in suppressing rival claimant(s) to the throne who might have had the support of the Christians. It was about this time (312–13) that the titles of’Persicus Maximus’, of ‘Medicus’ and ‘Armenicus Maximus’ were recorded for Constantine and Licinius. See Appendix 4, p. 339.

6 The importance of Christianity in Armenia prior to the conversion of Constantine has recently found additional attestation in the Manichaean account in Sogdian (TM 389d R 1–V 40) of the visit of Gabryab, one of the more important early disciples of Mani, to (E)revan where he debated with Christian leaders at the court of the local king. Cf. Sundermann, 1981:45–9 and Lieu, 1985:77–8. Trdat IV then was not a Christian and later tradition attributed to him the authorship of a letter to Diocletian in support of the latter’s religious policy. Text in Garitte, 1946:37, trans. Thomson, 1976:441, n. 82.

7 M.Aur. Val. Maxentius was the son of Maximianus ‘Herculius’ and was Augustus from 306 to 312. He never had a major command in the East. Cf. PLRE I, p. 571. Malalas appears to have confused him with Maximinus Daia.

8 The creation of a sizeable field army by Constantine was an important change in military policy. The central striking force was placed under the command of two newly-created officers, the magister peditum and magister equitum. Units of this field army (the comitatenses) were given privileges superior to those of the ripenses, i.e. the soldiers of the frontier legions, in a law of 325 (CT VII, 20, 4). The formation of the comitatenses probably dates back to Constantine’s campaigns against Maxentius in 312. Cf. Jones, 1964 (i):97 and Hoffmann, 1969(i): 186–93 and 226–36.

9 An official by this unusual name is recorded as quaestor and praetor at Rome before 287/8. Cf. PLRE I, p. 299 and Barnes, 1982:187. The details of these early ‘Persian’ (or probably Armenian) campaigns of Constantine are unknown.

10 On the career of Hormisdas (Hormizd), see esp. PLRE I, p. 443 (Hormisdas 2). His potential as a rival (and pro-Roman) candidate to the Sassanian throne was not exploited by the Romans until Julian’s invasion in 363.

11 Licinius was Augustus in the east until his defeat by Constantine in 324.

12 Basia is probably the same place as ‘Basiensa’ found on an inscription recorded by Stein at Qasr al-Azraq (cf. Kennedy, 1982:179–85) or ‘Basiensis’ of the praetensio inscription (supra 5.5.5.)

13 There had been sporadic persecutions of Christians in Persia since the reign of Bahram II, but widespread persecution did not begin until after 337 when Simeon bar Sabba’e, the ninth Catholicos, refused to collect a double tax from the Christian community in aid of Shapur’s war effort (Narratio de Simeone 4–5, PS 2.789–98). Cf. Decret, 1979: 125–35, Brock, 1978:167–8 and Brock, 1982:3–8.

14 The letter should probably be dated very shortly after October, 324. Cf. Barnes, 1985:131–2. The reason for their visit may have been related to the asylum given to Hormisdas by Licinius or Constantine.

15 I.e. lamps used in pagan rites. Cf. Lampe, p. 791.

16 I.e. the emperor Valerian. See above 3.3.1.

17 Michael has clearly telescoped the main events in Romano-Persian relations in the last decade of Constantine’s reign. The first siege of Nisibis in which Jacob played a heroic part began in 337, shortly after the death of Constantine. See below 7.1.3.

18 For earlier publications of this inscription, see Bowersock, 1971:241 and AE 1974, 661.

19 Fl. Severinus is otherwise unknown. He may have been either the garrison commander or the provincial governor. Cf. Kennedy, 1982: 91.

20 What sort of building had collapsed and therefore needed repair work is not clear from the inscription. Bowersock (1971:241) suggests reading: aedem (?) inc]uria vetustate/parietu]m ruina conlapsam/ refici (?) iu]ssit et […].

21 I.e. the future Constantine II and Constantius II.

22 The reconstruction of the events in Armenia in the last years of Constantine is beset with problems and uncertainties. Until recently, most scholars have accepted the hypothesis of Baynes (1955:187–9, first published in 1910) that the account of the abduction of Tiran (Tigranes)—an event which is placed in the reign of Valens by Faustus Buzandats’i (III, 21–2)—actually took place in c. 335. Baynes’ redating rests on his belief that Faustus had wrongly placed the consecration of Nerses, the first Catholicos of Armenia, in the reign of Valens—an event which in fact took place under Constantius, thereby post-dating the events of the latter’s reign. In Bayne’s schema, Trdat the Great died c. 318 and his successor Khosrov II reigned probably until 327. He was succeeded by Tiran who was deposed after being betrayed to the Persians by one of his satraps and was blinded (c. 335). This was followed by a Persian invasion of the kingdom which was repelled by a group of Armenian nobles with Roman assistance. It was at this stage that Constantine made Hannibalianus king of Armenia and of the Pontic regions. Shapur II, realizing the reluctance of the Armenians to accept a non-Arsacid as king, released Tiran’s son Ars ak (Arsaces II) who returned as king of Armenia and a Persian ally. See below p. 309. The comprehensive redating of the history of the third century Arsacids by Toumanoff (1969) has inevitable effects on their fourth century successors. Accepting Toumanoff’s dating of the reign of Khosrov II to c. 278–287 (i.e. before rather than after the reign of Trdat IV), Hewsen (1978–79b: 99–108) dates the death of Trdat IV (the Great) to c. 330 instead of c. 317, and argues that events attributed to the reign of Tiran in fact took place under Trdat III. Faustus’ (III, 20) naming of the Persian king who ordered the blinding of Tiran as Nerseh (i.e. Narses) is indicative of a chronological error. The murder of Trdat in 330 was followed by a period of unrest until the accession of Ars ak (Arsaces II) who ruled from 337/8 to 367. ‘Tiran’ was not a personal name but a kind of title, referring to the Armenian sun-god Tir (Helios) which an Armenian king might have prefaced to his name. ‘Tiran’ and ‘Trdat’ (Tiridates), both being theophoric names, are virtually synonymous and could have been confused in the Armenian sources. Cf. Hewsen, ibid. pp. 104–5.

23 The historicity of his peregrinations cannot be doubted as, according to Rufinus (Historia Ecclesiastica I, 9, PL 21.478–80), they inspired those of a certain Meropius. Cf. PLRE I, p. 601, Lightfoot, 1981: 17–18 and Lieu, 1986b:492–3.

24 The Persians had always regarded the re-exportation of silk and luxuries from India and China as an important source of wealth to their state. Cf. Pigulewskaja, 1969: 152.

25 Ensslin (1936:106, n. 1) believes that Narasara may be identified with the station Nararra (Nehar Harre) on the road from Amida to Tigranocerta, about 13 Roman miles east of Amida. Peeters (1931: 44–5), on the other hand, believes that the ‘pugna Narasarensi’, where Narses was killed, was part of the battle of Singara.

26 Jacob the Recluse was a monk who began his ascetic calling in Egypt in the reign of Julian. After a period of wandering, he came with some companions to Hesn-Kef via Amida. At Hesn-Kef they were later captured by Shamir, the Persian general, and one of Jacob’s companions suffered martyrdom with ten of his disciples. According to the vita (unpublished Syriac manuscript, Add. 12, 174, fol. 417a– 419a in the British Library), the Persian army was immediately afterwards dispersed and destroyed by a hailstorm and earthquake. Jacob built a small convent at Shilloh or Salah called the Convent of the Recluse. He died in 421. Cf. Wright, 1872 (3): pp. 1135–6, DCCCCCLX, and Bell and Mango, 1982:10–13 and 147.

27 He was the son of Dalmatius and a nephew of Constantine. Cf. PLRE I, p. 407, ‘Hannibalianus 2’.

28 The anonymous Excerpta Valesiana are preserved on one manuscript in Berlin. They were first published by the great classical scholar Henricus Valesius in 1636. The first part, composed about 390, an incomplete biography of Constantine the Great from 305 to 337 which is based on reliable sources.

29 Barnes (1985:132) theorizes that Constantine was planning to install Hannibalianus as Christian Shahanshah upon his successful invasion of Persia. This may explain why the exiled Hormisdas, who presumably was a Zoroastrian, was not put forward as a rival claimant to Shapur II. However, our sources are too scanty for us to discern clearly Constantine’s motives behind his appointment. The fact that he was positioned in Caesarea, often the first port-of-call of Armenian exiles, may indicate that he was appointed to represent the interests of the exiled Armenian royal family and to gather valuable information. His title ‘King of Kings’ was appropriate for one who had to defend Roman interests in Armenia, Iberia and the Pontic region in general as the epithet had historically been part of the titulature of the kings of Pontus and of Armenia. See above Ch. 3, n. 52.

30 The next verse and the whole of the next chapter (i.e. 57), with the exception of the chapter heading, are missing from most manuscripts because of a lacuna in the text. The translation given by Richardson of the linking material (here omitted) is based on an interpolation found in a seventeenth-century edition of the Vita.

31 Libanius had clearly minimized the events in Armenia to exaggerate the Persian build-up in 337. Nevertheless, it is odd that he made no mention of Constantius’ achievements in Armenia in a panegyric which was partly dedicated to this particular emperor.

32 The shortage of high-grade iron ore on the Iranian Plateau is a known geological feature. There is also a lack of wood to treat the mineral when extracted Cf. Lightfoot, 1981:37, n. 116. The Roman ban on the export of weapon-grade iron to Persia is well attested. Cf. Expositio totius mundi et gentium 22, trans, infra p. 221, and Procopius, de bello Persico I, 19, 25–6.

33 For a detailed but albeit fictionalized description of the Persian cataphracts, see below 7.6.4.

34 This last-minute effort by the Persians to avoid war is also recorded by Eusebius in the now lost portion of his vita Constantini (IV, 57) but the extant heading reads: ‘How he (i.e. Constantine) received an embassy from the Persians and how he kept the vigil (lit. spent the night) with the others at the Feast of Pascha.’ If Shapur was sole aggressor in 337, then his efforts to make peace would hardly have been sincere. Constantine might well have broken the peace by his desire to invade Persia, using the plight of the Christians there as his pretext. Cf. Seeck, 1911:24–5 and Barnes, 1985:32.

35 Constantine died on 22 May, 337. Cf. Barnes, 1982:80.

36 The name Aphrahat is Persian and he was probably the son of a Roman Christian captive. To him is attributed the authorship of twenty-three treatises (or demonstrations) in Syriac. The first ten demonstrations can be dated to 336–337. Cf. Murray, 1975:29, Barnes, 1985:126–8 and Brock, 1982:7–8.

37 Pusai was martyred along with the Catholicos, Simeon Bar Sabba’e in 339. On his career in general, see esp. Brock, 1982:4.

7 ROME AND SHAPUR: THE EARLY WARS OF CONSTANTIUS II

1 Constantine the Great had a large number of male relatives, mainly half-brothers who were sons of his father’s second marriage to Theodora. These were the main victims of the massacre, which took place not long after the funeral of Constantine (?Aug., 337) when most of his relatives were still in residence at Constantinople. The massacre left the three sons of Constantine, i.e. Constantine II, Constantius II and Constans, as the only claimants to the throne. Cf. Seeck, 1911: 6–9. The future emperor Julian and his half brother Gallus, sons of Julius Constantius who perished in the massacre, were spared because of their young age, and, in the case of Gallus, also his apparent frailty.

2 Julian is our sole witness in classical sources of an anti-Roman revolt in Armenia c. 337. For this reason, his statement has been variously interpreted, especially in the light of Baynes’ redating of the abduction of Tiran by Shapur II to the reign of Constantine. See above Ch. 6, n. 22. Peeters (1931:14–20) has proposed that with the murder of Hannibalianus disappeared the only realistic hope of some form of a Pontic confederation headed by a Roman prince. Roman victory in Armenia while Constantius was Caesar (see above 6.3.4) might have given rise to a pro-Roman faction in Armenia, led by the nobles who had appealed to Rome for help after the abduction of Tiran. In the meantime Shapur, realizing the difficulty of controlling Armenia without an Arsacid on the throne, had raised Ars ak (Arsaces) as vassal-king and returned him to Armenia. His return, coming so soon after the murder of Hannibalianus, might have occasioned the desertion of Rome’s allies in Armenia, who naturally preferred an Arsacid as their monarch, as described here by Julian. Cf. Dodgeon, 1967:51–3. Warmington (1977:512), however, believes that Julian might have exaggerated in a flattering manner the difficulties in Armenia. Under the new chronological scheme of Hewsen, however, the events of these years are virtually left blank. Lightfoot (1981: 327–8) suggests that the unrest referred to by Julian might have been the revolts of Sanatruk and Bacour (Faustus Buzandats’i III, 9, see below pp. 302–3).

3 The three royal brothers met in Pannonia during the late summer or early autumn of 337 and partitioned the empire. Cf. Barnes, 1980: 160–1.

4 The sources give the impression that the siege began in 337, soon after the death of Constantine and before the end of August, as the Persian army would have been in a high state of alert anticipating Constantine’s expedition. Cf. Barnes, 1985:133. However, a date of 338 cannot be ruled out as August was well into the campaigning season.

5 Jacob was the first recorded bishop of Nisibis and was one of the few bishops from sees east of the Euphrates who attended the first Ecumenical Council of Nicaea in 325. Most of the surviving stories concerning his early life found in Theodoret and Faustus Buzandats’i are of a legendary nature. Cf. Peeters, 1920:286–95 and Drijvers, 1981:28–30.

6 For the river to be used as a battering ram, the dam constructed by the Persians would have to be of enormous height in order for the water to acquire sufficient pressure prior to its release onto the plains and thereby to lose momentum. The historicity of the account has therefore been much questioned, partly on grounds of scientific improbability and partly because it reads like an embroidered account of the third siege in which the Mygdonius was also utilized for siege purposes by the Persians. However, though there is little comparable evidence in Roman sources, the diversion of rivers for the undermining of defences was used by the Chin Shih Huang-ti (259–210 BC), the ‘First’ Emperor of China in his campaigns to unify Northern China. Cf. Tzu-chih t’ung-chien 7.228 (CHSC edn.). The effect of water against foundations of battlements in a dry climate could be devastating.

7 Nisibis (Nusaybin) had a reputation among Arab travellers in the Middle Ages for the fierceness of its gnats. Cf. Le Strange, 1905:95.

8 For a comparison of the two accounts of the siege by Theodoret, see esp. Peeters, 1920:297–300. Both are probably derived from Syriac sources. In the Syriac version of the vita of Jacob by Theodoret, the description of the siege is closer to that of the Historia religiosa rather than the Historia ecclesiastica. Cf. Schiwietz, 1938: 80–3.

9 Constantius arrived at Antioch at the end of 337, too late to play any significant part in the siege. Cf. Barnes, 1980:162.

10 This is incorrect as Julian never visited Nisibis. However, he did deny the Nisibenes the special protection which they had demanded from him. See above pp. 231 and 268.

11 Shapur had obliged Ars ak to give military assistance against the Romans and Armenian forces took part in the siege of Nisibis as allies of the Persians. Cf. Faustus Buzandats’i, Hist. Arm. IV, 20 and the more abbreviated version in Procopius, Pers. I, 5, 10–15, where the common enemy were some unspecified barbarians rather than ‘Greeks’. After Shapur’s apparent defeat before the ramparts of Nisibis, a pro-Roman and anti-Persian faction emerged around Vardan the Mamikonian. When Constantius arrived at Caesarea in early 338, he had the choice of supporting either this pro-Roman faction or the more popular Ars ak. His sound political sense led him to give an amnesty to Ars ak for the support he gave to the Persians in return for Armenian friendship and neutrality. This was gained at the expense of the pro-Roman faction, viz. those who, in Julian’s words (21 A), ‘were better pleased with their present condition than with their previous power.’ Cf. Peeters. 1931:39–41 and Dodgeon, 1967:52–4.

12 Constantius was at Emesa in 338 as he issued an edict from there on 28 October. Cf. CTXII, 1, 25. It is quite likely that both the Armenian Settlement and the treaty with the Arabs were negotiated during his sojourn at Emesa.

13 Imru ‘al-qais, Rome’s most significant ally among the Arab chieftains, had died in 328. This, together with the death of Constantine in 337, appears to have led to the dissolution of the pro-Roman alliance (foedus) among the Arab tribes. Shahid (1984:76–9) believes that their Christianization and their subscription to the anti- Arian creed of the Council of Nicaea led to disenchantment with the Roman alliance in the last years of Constantine when the pro-Arian party was in the ascendant. They might have even transferred their alliance to Persia with the renewal of the conflict in 337. The Arabs referred to here were located to the east of the province of Syria, beyond the Roman frontiers. The statement that Constantius received a series of embasies suggests that he was dealing with an organized federation and not with individual tribes. See also Barcelo, 1981:84–5 and Lightfoot, 1981:263–71.

14 On the development of the Roman heavy cavalry under Constantius and Julian, see the important discussion in Hoffmann, 1969(i): 265–77.

15 The Itinerarium Alexandri is an anonymous brief history of Alexander (the Great)’s expedition against Persia based mainly on the Anabasis Alexandri of Arrian and the Res Gestae Alexandri Macedonis—a Latin version of the Greek Ps.- Callisthenes (Romance of Alexander). The fact that the Codex Ambrosianus P 49 (9th–11th C.) contains both a Latin version of Ps.-Callisthenes by Julius Valerius and the text of the Itinerarium has given rise to the hypothesis that the author of the latter might have been Iulius Valerius Alexander Polemius (consul of 338, cf. PLRE 1, 709–10, ‘Polemius 3 and 4’) whose name is in fact given (almost) in full (Iuli Valeri Alexandri Polemi v(iri) c(larissimi)’) in one manuscript version of the Res Gestae Alexandri Macedonis. It would have been highly appropriate for a politician of the prominence of Polemius to be the author of a contemporary dedication to Constantius who was then about to launch his campaign. However, although most modern scholars accept the consul Polemius as the translator of Ps.- Callisthenes, they are on the whole more sceptical in attributing the authorship of the Itinerarium to Polemius. Cf. Herzog, 1989: 212–15, esp. 215. On the importance of the Itinerarium for Romano-Persian relations, see esp. Barnes (1985: 135–6) who argues that the comparison of Constantius to Alexander and Trajan reflects the mood of the times which expected him to undertake an offensive war against the Persians—i.e. the continuation of a war which he had inherited from his father Constantine.

16 Constantius was at Edessa in August 340, as indicated by an edict issued from there on 12 August (CT XII, 1, 30, reading ‘Edessae’ for ‘Bessae’, cf. Seeck, 1919:188).

17 On these raids in the early years of the reign of Constantius, see also Libanius, or. XVIII, 206, trans. infra, p. 226. Hierapolis, about one hundred miles to the northeast of Antioch, was probably Constantius’ base for this campaign. Cf. Bowen, 1982:75–6.

18 There were no major Persian cities within easy reach of the Romans across the Tigris with the exceptions of Arbela and Karka de Bet Selok. More likely was the ancient city of Nineveh which was still inhabited under the Seleucids and even enjoyed the status of a Greek polis. Cf. Oates, 1968:61–2. It was mentioned by Ammianus (XVIII, 7, 1) as ‘an important city of Adiabene’. This would also explain the title of ‘Adiabenicus Maximus’ which Constantius had included in his titulature by 354 (CIL III, 3705=Dessau 732).

19 A reference to the enslavement of the Eretrians by Darius I who settled them near Susa. Cf. Herodotus VI, 101 and 119.

20 The date of the battle has long been a source of dispute because of a conflict in the ancient sources. While Jerome and other chroniclers who follow his lead place it under 348, Julian says that it took place ‘about six years’ prior to the revolt of Magnentius (Jan., 350). For a full survey of the arguments on the date, see esp. Stein and Palanque, 1959:138; Barnes, 1980:163, n. 13. See also below note 25.

21 The Persian objective was most likely the capture of Singara rather than a full scale invasion of Mesopotamia in the manner of the campaigns of Shapur I. Singara (mod. Beled Sinjar) lies at the southern foot of Jebel Sinjar and was well positioned to observe Persian military activities on the Middle Tigris, and therefore changed hands regularly between the two sides (Amm. XX, 6, 9). On Singara, see esp. Oates, 1968:97–106.

22 For a detailed comparison of the sources on the battle, see esp. Seeck, 1920:2337. 24–2338.52, Peeters, 1931:43–5 and Dodgeon, 1967: 72–6. Libanius’ account of the battle follows closely the rules of rhetoric for describing the military achievements of the Roman emperors as recommended by Menander the Rhetor (II, 373.15–25, ed. and trans. Russell and Wilson, pp. 84–6). Libanius’ panegyric on the achievements of Constantius and Constans was composed c. 348/9. Cf. Gladis, 1907:4–5.

23 The death of the Persian crown-prince at the battle may explain why Shapur II was eventually succeeded by his ‘brother’ (more likely his nephew, cf. Frye, 1983:140– 1) after his death in 379.

24 Libanius (or. LIX, 107) gives 150 stades, i.e. about 30 kilometres.

25 Barnes (1980:163, n. 13) proposes two battles at Singara, as a solution to the problem of dating one in summer/autumn 343, and a nocturnal one at Eleia in 348. However, Libanius’ description of the battle makes it clear that the fighting at Singara began at mid-day and drifted into dusk and it was in the night engagement that the Romans threw away through indiscipline what could have been a major victory. Nevertheless, there must have been a number of engagements at or near Singara because of its strategic importance to both sides, and it is entirely possible that the chroniclers had combined two battles into one; the nocturnal battle mentioned here by Festus may be the same engagement in which the Romans, under the leadership of the comes Aelianus, made a successful nocturnal raid on the Persian camp. See below 7.5.3.

26 Festus is the only source to name the Persian prince killed at the battle of Narasarensi. Theophanes (supra 6.3.4) mentions the death of a Narses, brother of Shapur, in Armenia in the reign of Constantine. The Byzantine chronicler might have transferred the name in error to the earlier event.

27 Peeters (1931:44) has shown that the ‘Eliensi prope Singaram’ in Festus is Eleia which was situated in a gorge at the foot of the Jebel Sinjar, on a water-course which still bears the name of Nahr Ghiran, which may be the ‘Narasarensi’ or ‘Nararensi’ in Festus.

28 The Fasti Hydatiani is a composite consular chronicle (284 BC-AD 468) mistakenly attributed to the fifth century Spanish bishop Hydatius. It falls into three parts and the second part, covering the period from 330 to 389, was compiled in Constantinople. Cf. Bagnall et al., 1987:54.

29 Constantius issued an edict from Nisibis on 12 May, 345 (CT XI, 7, 5).

30 This does not necessarily mean that Constantius personally directed the siege or led the relief-force. All it means is that Babu was bishop of Nisibis at the time of Constantius’ visit shortly after the end of the second siege. Cf. Barnes, 1980:163, n. 14.

31 On the role of the bishops in the redemption of captives in the frontier zone, see Lieu, 1986b:487–90.

32 Silvinianus was dux Arabiae from 348 to 351. Cf. PLRE I, p. 842 and Sartre, 1982: 103.

33 For a detailed analysis of the sources of this, the most important and best documented of the three sieges of Nisibis, see Sturm, 1936: 742.60–746.59, Dimaio, 1977:285–90 and Lightfoot, 1981:94–103.

34 Syr.: nsybt, a play on the name Nisibis (Syr.: Nsybyn).

35 Bidez-Cumont (1932:36–7, n. 2) believes that this brief mention of a Roman victory in Mesopotamia refers to the battle of Singara while Sturm (1936:743.43) suggests the third siege of Nisibis. As the panegyric was composed in 350, the latter seems more likely.

36 The Mygdonius (Dschaghschagh) flowed past (not through) the city. Its course is still marked by a bridge of Roman date. The historicity of Julian’s description of the siege has been called into question as a sea-battle on land taxes one’s credulity, and the account itself shares a number of stylistic and topological similarities with the fictional account of the siege of Syene in the Aethiopica of Heliodorus, a work which was published in the second half of the fourth century. For the towers of Nisibis to be ‘just visible above the waters’, the earthen mound needed to hold the moat would have to be of enormous strength and height. The Persian use of the Mygdonius against the defences cannot be denied because of the innumerable references to flood-water and its effects in the poems of Ephrem who was a witness to the events of the third siege.

37 Flavius Julius Constans, the Augustus of the western half of the Empire, was murdered by the usurper Magnentius in 350.

38 Flavius Claudius Constantius Gallus was proclaimed Caesar on 15 March, 351. Cf. PLRE I, p. 225.

39 This means he was a centurion of the primi ordines, probably in the wars of Constantius against Shapur II. Cf. Speidel, 1977:722.

40 Speidel (1977:722) gives AD 351 for the conversion of the date.

41 On this equestrian dux Arabiae, see esp. Sartre, 1982:103.

42 Heliodorus was a Phoenician from Emesa and wrote his novel, the Aethiopica, towards the end of the fourth century. Cf. PLRE I, p. 411, ‘Heliodorus 3’.

8 ROME AND SHAPUR II: THE LATER WARS OF CONSTANTIUS II

1 Philostorgius was an Arian church-historian and was favourable to Gallus because, like Constantius, Gallus was an ardent Arian. Nevertheless, based at Hierapolis, Gallus probably made ‘what was at least a highly successful demonstration against him (i.e. the Persian commander)’. Cf. Thompson, 1947:57.

2 The Artemii Passio draws its historical material heavily from the now very imperfectly preserved ecclesiastical history of Philostorgius and therefore shows the same pro-Arian bias towards Gallus. However, it was not blind to the faults of Gallus as an administrator. Cf. ch. 6, pp. 204–5, ed. Kotter. On the relationship between the Passio and the lost work of Philostorgius, see esp. Bidez’s introduction to his GCS edition of Philostorgius (pp. XLIV–LXI).

3 On Ammianus as a source for the Saracens, see esp. Shahid: 1984: 83–5 and Lightfoot, 1981:269–75.

4 The continuing importance of the September Fair at Batna (mod. Sarug) shows that the stipulation of Nisibis as the sole venue of exchange between Rome and Persia in the treaty between Diocletian and Narses (supra 5.4.3.) was not strictly enforced.

5 On the career of Ursicinus (Mag. Equ. 349–59 and Mag. Ped. 359–60), a key figure in the defence of the Eastern Frontier in the later years of Constantius, see esp. PLRE I, pp. 985–6 (‘Vrsicinus 2’). See also Thompson 1947:42–55.

6 The Chionitae were a Hunnic tribe who had occupied Transoxania in Central Asia. The most detailed study on them, drawing from both classical and oriental sources, remains Enoki Kazuo, 1955:757–80.

7 Ammianus has almost certainly preserved the original text of the letter. The titulature of the Shapur corresponds in part to that of his father in his inscription at Hajjiabad. Cf. Christensen, 1944, 237–8. Ammianus probably obtained a copy of it through his friendship with Libanius. The latter was a close friend of Spectatus (see below) who took part in the negotiations. Cf. Sabbah, 1970:173, n. 49.

8 Cf. PLRE I, pp. 74–5, ‘Antoninus 4’ and Sabbah, 1970: p. 196, n. 165. The latter’s identification of this Antoninus with a decurion who ran into financial problems at Antioch (cf. Libanius, ep. 210, 1) is by no means certain. Antoninus’ defection was a major boon to Shapur because as a protector he would have taken part in recruitment and his former role as accountant (or quarter-master) to the dux of Mesopotamia meant that he would have known the exact troop-strength of the Roman garrisons on the frontier as well as their disposition. The lengthy negotiations were needed to establish his credentials and the value of his information. Cf. Lieu, 1986b:494–5 and Matthews, 1986: 556–7 and 1989:39–47.

9 Cf. PLRE I, pp. 517–18, ‘Lucillianus 3’ and pp. 742–3, ‘Procopius 4’. The former defended Nisibis successfully against Shapur in 350.

10 The Expositio is a geographical handbook for commercial travellers composed by someone with first-hand knowledge of the Roman East. The fact that it mentions Nisibis as a Roman city which had heroically withstood sieges by the Persians suggests that it was composed in the 350’s, before the city was ceded to the Persians in 363.

11 On the distinguished career of this important official, see PLRE I, pp. 605–8, ‘Modestus 2’.

12 A new and distinctive feature of this campaign is the successful use by the Persians of mobile cavalry units advancing ahead of the main army for the purpose of disrupting the Roman preparations for defence and confusing the defenders as to the exact direction of the main attack. Cf. Lightfoot, 1981:142 and Matthews, 1986: 551–2.

13 Ostensibly a description of the ambitions of Shapur, the message was coded to give the Roman commanders on the frontier some idea of Shapur’s intended invasion route. By the rivers ‘Granicus’ (site of Alexander’s first victory over Darius in 334 BC) and ‘Rhyndaces’ (site of Lucullus’ victory over Mithridates in 74/73 BC) was meant the rivers Tigris and Anzaba (the Greater Zab) which Shapur was about to bridge. The ‘successor of Hadrian’ was, of course, a reference to the deserter Antoninus. Cf. Matthews, 1986:558 and esp. 1989:48–56.

14 Though some elements of this episode may seem fantastic, there is little doubt that the reconnaissance visit by the future historian did take place. Cf. Lightfoot, 1981: 142–3 and Matthews, 1986:559–61.

15 Sabinianus (PLRE I, p. 789, ‘Sabinianus 3’) was consistently portrayed in an unfavourable manner by Ammianus, who considered him as an unworthy successor of his patron Ursicinus. However, Sabinianus’ decision to remain at Edessa was a sound one as the main field force under his command would have blocked off Shapur’s most direct route to the Euphrates. Cf. Lightfoot, 1981:143 and Matthews, 1986: 555–6.

16 Warmington (1977:515) has adduced from Antoninus’ admonition and the subsequent events that the Persian plan of attack in 359 was to cross the Tigris north of Singara then head straight for the Euphrates, bypassing Nisibis. This was foiled by the flooding of the Euphrates. However, Ammianus himself probably deduced the Persian change of plan from the actual events of the campaign. It seems odd that the Persians who were experts in hydrodynamic engineering could not bridge a flooded river. The presence of the main Roman field army at Edessa could have been an equally strong factor in Shapur’s decision to head north towards the Taurus Mountains.

17 The classic study of Amida and its Byzantine and medieval defences remains Gabriel 1940(i):85–205. See esp. 90–2 (topography) and 175–82 (estimated size of the city at the time of Constantius). Ammianus’ description of Amida needs to be reordered for it to agree with basic topographical factors. The Tigris washes the eastern (and not the southern) side of the city and the plains of Mesopotamia lay to the south (not to the east). However, Ammianus is partly correct in his alignment of the city in that there was a tower on the south-eastern side of the city which overlooked the Tigris. Cf. Lightfoot, 1981:82 and Matthews, 1989:57–66.

18 On the Chionitae at the siege of Amida, see esp. Bivar, 1955:200–6. However, Bivar’s identification of the royal figure described in Amm. XIX, 1, 3, with the Kushano-Sassanian ruler Bahram II on the similarity between the crown of Bahram as shown on his coins and the helmet won by the Persian commander as described by Ammianus (p. 202), is almost certainly groundless. Cf. Lightfoot, 1981:161–3.

19 Most modern scholars have assumed that Bezabde stood on the same site as Jazirat Ibn-Omar (mod. Cizre) originally founded on the eastern bank of the Tigris, but not on the western bank because the river took a more direct route through an artificial channel to the east of the city. A recent archaeological survey in the region directed by Algaze (1989:249–52 and 265, fig. 9 and 267, fig. 11) has located a major late Roman site at Eski Hendek, 13 km north of Cizre, and south-west of Fenik on the opposite bank of the Tigris, which fits in well with our knowledge of Bezabde from literary sources. For an earlier attempt to identify the site south of Cizre, see Lightfoot, 1981: 85–92; 1983:189–204.

20 Virtha (Birtha)-Makedonopolis has been identified by Cumont (1917: 144–50) as Biredjik on the Euphrates on the basis of local inscriptions and lists of bishops at the Council of Chalcedon (451) in which Daniel of Makedonopolis appears in the Syriac list as Daniel of Birtha. However, Shapur’s route across Mesopotamia after the fall of Bezabde is hard to retrace. His failure at Virtha may in part be due to his having over-extended his supply lines. Cf. Dillemann, 1962:299.

21 The fate of the Christian prisoners taken from Bezabde was in stark contrast to that of the nuns who were well treated by Shapur after their capture at Reman and Busan in the campaign of 359 (Amm. XVIII, 10, 4). The ability of the Christian clergy to organize and raise the morale of the prisoners was clearly the main cause for their execution rather than mere religious intolerance. Cf. Lieu, 1986b:496–7.

22 On the use by the Roman authorities of philosophers as envoys, see Lieu, 1986b: 492–3. On the journey of Spectatus, see esp. Sievers, 1868:239–40 and the other sources, mainly the letters of Libanius, cited in PLRE I, p. 850.

23 In this campaign, Nisibis was picketed by the Persians rather than besieged. Cf. Amm. XVIII, 6, 8–16.

9 ROME AND SHAPUR II: THE PERSIAN EXPEDITION OF JULIAN

1 The extant text of Julian’s letter to Arsaces (ELF 202=ep. 57, Wight), composed in a haughty tone and intimating major loss of Roman territory, is generally acknowledged to be a forgery.

2 Mod. Elterib, a site of great antiquity in the region of Chalcis on the road from Antioch to Beroea. Cf. Cumont, 1917:2.

3 The dates in bold type are those given by ancient authorities, those in plain are suggested by Brok, 1959:257–8.

4 Mod. Menbij, about 30 km. west of the Euphrates. The infantry was assembled here so as not to alert the Persians of Julian’s intention to use the Euphrates as his main invasion route.

5 Cf. PLRE I, ‘Procopius 6’, pp. 742–3 and ‘Sebastianus 2’, pp. 812–13.

6 Deception was the key to Julian’s plan, as his false march towards Nisibis was to mislead the Persians into thinking that his main force was to use the Tigris route, thereby drawing the main Persian army under Shapur up the Royal Road and away from Assuristan. The use of a large transport fleet was to allow the Roman infantry to make a swift and unencumbered descent down the Euphrates towards Seleucia (Veh-Ardashir) and Ctesiphon. Cf. Ridley, 1973:318–19, Bowersock, 1978:119–20 and Lieu 1986a:92–3.

7 Cf. PLRE, I, p. 517, ‘Lucillianus 2’, not to be confused with ‘Lucillianus 3’.

8 On Zaitha, see above, Ch. 2, note 12.

9 See above Ch. 3, note 8. The island was linked by a stretch of sand with the riverbank at low-water. Cf. Musil, 1927:345 and Fontaine, 1977, ii:137–8, n. 300.

10 Almost certainly to be identified with modern Telbes, a rocky islet 14 km. south of Ana. Cf. Fontaine, 1977, ii:141.

11 Probably situated somewhere in the vicinity of modern Haditha, about 75 km. down river from Ana (or 45 km. as a crow flies). Cf. Musil, 1927:239.

12 According to Musil (1927:239), ‘Barax’ may be a corruption of the Arabic root farad ‘ford’ and Baraxmalcha, ‘The Royal Crossing’.

13 Identified as mod. Hit by Musil (1927:239) on the grounds that Diacira means ‘bitumen-giving’, as a place called Is on the Euphrates is mentioned by Herodotus (I, 179) as the source of bitumen for the walls of ancient Babylon. Cf. Fontaine, 1977, ii: pp. 142–3, n. 317 and Brok, 1959:111. Is, however, appears more likely to be the modern name of Sitha, a place in this region mentioned by Zosimus (III, 15, 2).

14 Ozogardana (‘Zaragardia’ in Zos, III, 15, 3) is identified by Musil (1927:239) with mod. Sari-al-Hadd.

15 ‘Suren’ is the hereditary title of one of the seven Persian families, charged with the second-in-command of the Persian armed forces. Cf. Fontaine, 1977, ii:143, n. 319.

16 Podosaces was emir of the Assanite Saracens who were allied to the Persians. Though Julian requested the envoys of the Arab federates to join him after initially refusing to pay them their usual bribe, Arab irregulars, essential for scouting and raiding, feature far more often in the accounts of the campaign fighting on the Persian rather than the Roman side. Cf. Fontaine, 1977, ii:143–4, n. 320 and Shahid, 1984: 132–5. On Julian’s lack of accurate military intelligence, see further, Lieu, 1986a:93–4.

17 The mention by Ammianus (XXIV, 2, 6) of the remains of a rampart at Macepracta has helped Musil (1927:240) to locate it near the beginning of an ancient rampart at Ummu-r-Rus, which stretches from the north bank of the Euphrates northward as far as the Tigris.

18 By Naarmalcha (the Royal Canal) Ammianus means a water-way beginning at Pirisabora on the Euphrates which was divided into two branches; the upper branch passed through the Hellenistic metropolis of Seleucia which no longer joined the Tigris at the time of Julian, while the Parthian port-city of Vologesias was situated on the northern bank of the lower branch which was called Maarsares by Ptolemy (Geog. V, 20) and was probably dug by the Parthian King Vologeses. The latter was probably the one Julian found to be blocked by the Persians and the same waterway was used previously by Trajan and Septimius Severus for conveying their fleets to the Tigris. On the Royal Canal, the Arabic geographer Ibn Serapion (c. AD 900) wrote: ‘From the Euphrates also is taken a canal called the Nahr-al-Malik. Its point of origin is five leagues below the head of the Nahr Sarsar. It is a canal that has along it numerous domains and fertile lands. There is also a Bridge-of-Boats over it: and further many villages and fields (along its banks). From it branch numerous other canals, and its lands form a District of the Sawad. Finally it flows out into the Tigris on its western bank, three leagues below al-Madain (trans. Le Strange, 1895: 70).’ Cf. Fiey, 1967:13–14 see also Musil, 1927:272–4 and Dillemann, 1961:153–8.

19 On Pirisabora (mod. Al-Anbar), see above, p. 358, note 26.

20 Identified by Musil (1927:240–1) with modern ‘Akar an-Na Jeli, lying about two kilometres from the Euphrates on the left bank of the canal.

21 Ridley’s translation of the first phrase of III, 19, 3 (p. 60) as ‘Advancing further and crossing the river, he came to…’ may be misleading as an equally plausible translation is: ‘While advancing along the course of the river, he arrived at…’. Cf. Paschoud, 1979:36 and 139, n. 50.

22 Probably Nehardea, the great centre of Jewish learning in Babylonia. See above 4.3. 1. Cf. Oppenheimer, 1983:276–93, esp. p. 290.

23 The location of Bithra or Birtha (=fort, a common name in this region) is not entirely certain. Musil (1927:241) has tentatively identified it with Ibrahim al- Halil, about 25 km. south-east of Phissenia. The mention by Zosimus of a palace there suggests that it was not the same place as the abandoned Jewish settlement mentioned in Amm. XXIV, 4, 1. Cf, Musil, 1927:241.

24 The name in Aramaic means ‘royal fort (or port)’, so called because of its position on the Naarmalcha. Musil (1927:241) noticed a ruin mound (present Han az-Zad) about ninety stades (i.e. 18 km.) west of Ctesiphon. The “ma’oz” part of the name is probably derived from Mahoza, the name given by Jews and Christians to a suburb of Veh-Ardashir. Cf. Oppenheimer, 1983:179–93.

25 The ruins of the Hellenistic metropolis of Seleucia were now used as a place of execution of criminals (including Christian martyrs) by the Sassanian kings. Cf. Fiey, 1967:8.

26 Meinas Sabath was identified by Maricq (1959:264–76) as the site of the Parthian river port of Vologesias (Walasapat). However, Oppenheimer (1983:390–3) believes it should be identified with a suburb of Veh-Ardashir.

27 On Coche (Zokhase in Zos. III, 23, 3), see above, Ch. 5, note 8.

28 Modern site unknown. Seen from the context of the narrative, it must have been east of Ctesiphon. Cf. Sarre and Herzfeld, 1911:86, n. 2 and Paschoud, 1979:180, n. 72.

29 Sarre and Herzfeld (1911:86) suggests identifying it with Djsir Nahrawan near the river Tamarra.

30 On identifying the Diyala, an important tributary of the Tigris, with the Douros of Zosimus, see Dillemann, 1961:146 and Paschoud, 1979:188.

31 Identified by Herzfeld (1948:65) with ‘Ukbara, about 50 km. north of Baghdad. Founded by Shapur II, its Syriac name was Buzurg Sabur. See also Brok, 1959: 172. It is vitally important to remember that the Tigris in this period followed a more direct north-south course below Samarra than it does now. Cf. Le Strange 1905:37 and map facing p. 25, Adams, 1965:90–1 and Oppenheimer, 1983:452–6.

32 Regarded by Dillemann (1961:145–6) as a confusion by Zosimus for Mischanabe— the latter itself a corruption of Meschana.

33 Identified by Herzfeld (1948:65) with Tell-Hir, about 54 km. from the modern city and former Abbasid capital of Samarra.

34 Site unknown, probably situated not far south of Samarra. Cf. Paschoud, 1979:200, n. 83.

35 Cf. PLRE I, p. 61 ‘Anatolius 5’.

36 Identified by Herzfeld (1948:67) with Karkh Fairuz, about 11 km. north of Samarra.

37 Probably Dur Arabaya, about 5 km. further north of Charcha. Cf. Herzfeld, 1948: 67.

38 Almost certainly the same place as Thebeta of the Tab. Pent, in the desert regions of Mesopotamia between Singara and Nisibis. Cf. Dillemann, 1962:312 and Fontaine, 1977:269, n. 678. 39. The words in brackets are supplied by the editor.

39 The CSCO edition has ‘RMNY’, clearly an error for ‘RMY’. (S.P. Brock).

40 The ms. reading is HRMN. Brooks takes it as ‘RMN (i.e. Armenia, but this would normally be ‘RMNY’) with note suggesting Rehimene.

41 Cf. Malalas, XIII, pp. 336, 21–2.

42 For fuller commentary on these verses, see esp. Lieu, 1986a:127–33 and Griffith, 1987:247–58.

43 See below Soz. V, 3, 5: p. 268.

44 One of the earliest references to Julian being tricked by the Persian ‘traitors’ into setting fire to his transport fleet.

45 Cf. Amm. XXV, 9, 1.

46 One of the Cappadocian Fathers, Gregory met Julian at Athens where they both studied. His two orations against Julian, published within a few years of the death of the apostate emperor were the earliest surviving Christian polemical works of the reign.

47 Greek: (courage) and (overconfidence).

48 The motif of the barbarian guard being responsible for the death of an emperor is also found in SHA Sev. Alex. 62, 4–5. Cf. Straub, 1980: 242.

49 A student of Libanius, John Chrysostom (‘the golden-mouthed’) was consecrated deacon in his native city of Antioch in 381. He became Patriarch of Constantinople in 398 against his wishes and at the time of his death in 407, he was one of the most outspoken critics of the Empress Eudoxia. On his works against Julian the Apostate, see Lieu, 1986a:59–63.

50 Julian may have encountered a geological phenomenon, common in that part of Syria, where linear outcrops of lava had been weathered down by the natural processes of erosion into something resembling a badly paved road. Cf. Cumont, 1917:8.

51 The passages here omitted include Sozomen’s refutation of a section of Libanius’ panegyric on Julian (or. XVIII, 268) and the intimation of the emperor’s death received by Didymus the Blind. The latter is a common theme in hagiographical literature. See similar accounts in [Pachomius] Epistula Ammonis 32–4 (trans. A.Veilleux, Pachomian Koinonia II, Kalamazoo, 1981, pp. 102–4), Palladius, historia Lausiaca 4, 4 (trans. R.T.Meyer, Washington, 1964) p. 36, Theodoret, historia religiosa 14 (trans. Russell, Kalamazoo, 1985, pp. 29–30) and Theodoret, hist. eccl. III, 24, GCS (trans. Jackson, p. 105). A later tradition names Mercurius as the ‘Christian’ executioner of Julian. Cf. Frend, 1986:69–70. See also Malalas, XIII, pp. 333, 18–334, 12.

52 This Magnus is very probably the Magnus of Carrhae who wrote an account of Julian’s campaign. (See above, Malalas). The incident referred to here is the capture of Pirisabora.

53 For a detailed historical commentary on this section of Zonaras, see esp. Dimaio, 1977:420–47.

APPENDIX 1

1 Abu Dja ‘far Muhammad b. Djarir, al-Tabari, was born c. 839 C.E. at Amul in Tabaristan where he devoted himself to serious study at a precociously young age. His later travels took him first to Baghdad, then to Syria and Egypt. He returned to Baghdad in 872 C.E where he remained as a celebrated scholar until his death in 923 C.E. His most important work is the history of the world (Ta’rikh al-Rusul wa’l-Muluk) which begins with the history of the Patriarchs and ends in July, 915 C.E. It was afterwards continued by other historians. The translation of this extract on the section on the Sassanians is the joint work of Ahmad al-Issa and Doris Dance, with generous acknowledgement to the outstanding German translation of Th. Nöldeke.

2 The sections on the history of the desert kingdom of Hira are here omitted.

3 Sa’id b.al-Bitrik (Eutychius) was born at Fustat in 876 C.E. and was Patriarch of Alexandria from 933 to 939 C.E. He was the author of several medical and historical works. The best known is his Arabic chronicle, Nazm al-Djawhar (String of Pearls), which records main events from the time of Christ to 938 C.E. The work was first published in the West by Pocoke in 1658/9 and a critical edition was prepared for CSCO by Cheikho in 1906. However, Breydey published in 1985 a new CSCO edition based on a shorter version preserved in the Sinaiticus Arab. 582 which the editor believes to be the author’s autographed manuscript.

4 Compiled shortly after 1036 C.E., the Nestorian Chronicle of Se’ert, which draws on early Syriac material, is a source of primary importance for the study of Christianity in Persia.

5 On the importance of these exiles to the diffusion of Christianity in Persia, see Peeters, 1924:294–8 and 308–14, Decret, 1979:110–11 and Lieu, 1986b:481–2.

6 This is a problematic section and Sher’s French translation (p. 224): ‘Avant ce second exil, et après le premier exil de Demetrius, Paul de Samosate était devenu patriarche d’Antioche’, has been used by scholars to support the view that Antioch was twice captured by Shapur I (256 and 260). Peeters (1924:296) gives: ‘Hic (sc. Demetrius) enim iam ante secundam hanc captivitatem (Antiochia) excesserat. Et Antiochiae patriarcha, postquam captivus abductus est Demetrius patriarcha, factus Paulus Samosatenus,…’; and Felix (1985:57) suggests: ‘…er war schon vor diese 2. Gefangennahme fortgeschafft worden, und Patriarch von Antiochia nach der Gefangennahme seines Patriarchen wurde Paulos von Samosata.’ A more radical departure has been suggested to us by Prof. Ze’ev Rubin (unpublished): ‘He (sc. Demetrius) had already been taken out of [the city] before this second captive (sc. Azdaq, i.e. the second to be taken out according to the chronological order of events). And Paul of Samosata became Patriarch over Antioch after Demetrius, its Patriarch, had been taken prisoner.’

7 Firdawsi (Abu ‘l-K sim), the famous Persian poet, was born in 932 C.E. at Tabaran in Khorasan. He lived on a small estate left to him by his father. A dihka n (i.e. landed proprietor) whom he knew gave him a Book of Kings (see above, Ch. 1, note 2) to put into verse and it was this that inspired him to compose the Sh hn mah (Book of Kings), an epic of about 60,000 lines—a task which took him thirty-five years to complete, by which time he was already in his 80’s.

8 Shapur I, son of Ardashir, is here confused with Shapur II, son of Hormizd, and the wars of the two Shahanshahs are also confused.

APPENDIX 2

1 The History of Faustus of Buzanda traces the fortunes of the Kingdom of Armenia from 327 to 387 (the year in which Armenia was partitioned between the Romans and Persians). Earlier suggestions that he was a contemporary of the Catholicos Nerses the Great (353–373) rest on a misidentification. The work, as it stands, was compiled in the second half of the fifth century. The author, whose identity is unknown (Buzanda is probably Podandus in the Taurus), uses as his sources the Life of Mashtots’ by Koriun, the Life of St. Basil and the Anaphora of the same saint as well as popular epics recounting the deeds of Armenian kings.

2 See above Ch. 6, n. 22.

3 The ‘Greek emperor’ is commonly identified as Caesar Constantius. However, Libanius’ silence on any military achievement by the Caesar in Armenia prior to the death of Constantine in his long panegyric on Constantius and Constans (or. 59) is significant. Perhaps the Roman intervention in Armenia was brought about by the revolt of Bacour (cf. Faustus, III, 8 and Moses Khor. (III), 4). The Roman commander may have been Flavius Eusebius who was Magister Equitum et Peditum under Constantius and was Consul in 347. His household, together with that of Arsaces, king of Armenia, was granted special exemption from taxation by Constantius in 360. See above Ch. 8, p. 219. This was probably a belated token of respect for Eusebius’ unpublicized services in Armenia. Cf. Lightfoot, 1981:30–1, n. 80.

4 Faustus’ account of the Persians’ defeat in Armenia by the ‘Greek King’ is clearly influenced by Roman accounts of Galerius’ victory against Narses in 298. The topos is of the Roman commander spying on the disposition of the Persian encampment and the capture of the Shahanshah’s harem. Cf. Ensslin, 1936:104–5.

5 The History of Agathangelos has as its main theme the reign of Trdat (Tiridates the Great) and the life and mission of Gregory the Illuminator. It exists in Armenian, Greek and (in part) Arabic versions and many recensions, not all of which are derived from the extant text of the Armenian. The author purports to be an eyewitness of the events of the reign of Trdat, but the work is almost certainly a tendentious fifth-century compilation. Cf. Thomson, 1976:xi–xvii and Winkler, 1980: passim.

6 Moses of Khorene, the name given to the author of an Armenian chronicle which covers the history of Armenia from the Biblical Age of the Patriarchs to the death of the Armenian Patriarch Sahak (early 5th C), is regarded in the Armenian literary tradition as the ‘Father of History’. However, his birthplace of Khorene is unknown and his claim to have been a contemporary of Mesrop (the inventor of the Armenian alphabet) and Sahak is unsupportable because the chronicle utilizes material later than the fifth century. Thomson (1978:60–1) believes that the work fits most appropriately into the first decades of Abbasid control over Armenia. For the Early Sassanian period it cites Agathangelos and Faustus among its authorities but there are considerable differences between ‘Moses’ version of the events and those of his alleged sources. See esp. Thomson, 1978:40–9.

7 There is little doubt that Arsak (Arsaces II) was King of Armenia during the short reign of Julian. The root of the error in synchronism lies in the confusion by ‘Moses’ (or some intermediate source) of the date of the year when Constantius became Caesar (AD 323) with that of his accession to the imperial title (337), thereby advancing events of the fourth century by fourteen years and in doing so making Julian a contemporary of Tiran (-Trdat IV). Cf. Hewsen, 1978–79b:113–15.

APPENDIX 3

1 This figure is most probably that of the total for the infantry and excludes the cavalry and dromedarii.

2 Fink, 1971:187: ‘Iulius Rufianus, tribune, sent the password (chosen) from the Seven Planets: Mercury s(anctus ?)’.

3 His title was perhaps circitor (an inspector of the watch) abbreviated as ci. Cf. Fink, 1971:185, n. 7.

4 This lower total figure compared with that of the earlier document may indicate the participation of the unit in the skirmishes against the Persians in which its tribune Julius Terentius was killed. (See above 1.4.4.)Cf. Welles, 1959:30.

5 In the original document the year is given by the names of the consuls in office.

APPENDIX 4

1 Cf. Kneissl, 1969:232–48, Homo, 1904:352–61, Barnes, 1976a: 182–6 and 191–3, idem, 1976b: 149–55, idem, 1980:162–6 and idem, 1982:254–8.

2 See above Ch. 1, note 30.

3 ‘ARAM. M.’ is almost certainly an error for ‘ARM(ENICUS) M.’ . Cf. Barnes, 1982:18.

4 ‘ARAB. M.’ may also be an error for ‘ARM(ENICUS) M.’ . Cf. Barnes, 1982:18.

5 See above, note 3.

6 See above, note 3.

7 For an attempt to date the inscription to the last months of 336 and the first part of 337 and therefore linking the title of Persicus to the early campaigns of Constantius, see Arce, 1982:247–9.

APPENDIX 5

1 Site not securely identified. Dussaud (1927:268–9) suggests Sadad, on the route between Damascus and Salamiye.

2 Site not securely identified. Dussaud (1927:269) tentatively suggests Deir ‘Atiye, south of Sadad.

3 Dussaud (1927:269) suggests ‘Atni, north east of Damascus which appears to have been the site of a Roman camp.

4 Placed in the Tab. Peut, on the route between Palmyra and Damascus, probably mod. Nebek. Cf. Van Berchem, 1952:16.

5 Site not securely identified but most likely to be located in the region of Jebel Qalamoun, north of Damascus. Cf. Van Berchem, 1952:16.

6 Site not securely located. Honigmann (1923:153) suggests Hadata in the neighbourhood of Hauwarin.

7 This implies that the Legio I Illyricorum was stationed at Palmyra.

8 The Tab. Peut, gives Danoua as one of the stops between Damascus and Palmyra, perhaps at Méhin. Cf. Poidebard, 1932:41. The presence of the Legio III Gallicorum at ‘Danavae Damasco’ is also confirmed by an inscription from Sistov (Nicopolis, Moesia): CIL III, 755.

Dussaud (1927:271) conjectured Jebel Seis and Musil, 1928:43ff. suggests Abu Sindâh, between Homs and Palmyra.

9 Perhaps ‘Ayn Wou‘ôul or Harbaqa, both beyond the Jebel Rawaq. Cf. Poidebard, 1932:49.

10 Dillemann places Rene (1962:239) in Mesopotamia, south of Dara, and argues that the unit has been listed in error under the ducate of Phoenice.

11 Many scholars believe that the ancient name of Dibsi Faraj was Athis. Harper (1977:457) believes that, though founded as Athis, its name was changed to Neocaesarea under Diocletian in honour of Caesar Galerius.

12 Seeck suggests: Ala prima Iota constituta. Böcking: (perhaps) Ala prima quingenaria Iota constituta.

13 Thomsen (1906:125) suggests identifying Tarba with Jebel umm Tarfa between el- Mesadd and el-Kwera.

14 Location uncertain. Thomsen (1906:125) proposes identifying it with Joppa, Flavia Joppe, mod. Jaffa, if situated in the north, and Wadi Guweibe, west of el-Busera, if in the south.

15 Avi-Yonah (1976:78) believes that the cohort was stationed at Maledomnei.

16 May have been the same place as Canaba in the Itin. Anton. 189, 3, p. 25, ed. Cuntz, situated between Mediana (see below) and Zeugma.

17 Identified by Dillemann (1962:229) with Banasymeon in Procop. de aed. II, 4; mod. Qartemin, S.S.E. of Midyat. See, however, Honigmann, 1961:13, n. 4.

18 Probably the Monithilla of George of Cyprus (ed. Honigmann, 1939: 25, no. 900) and the Thiolla of Procop., de aed., II, vi, 14.; perhaps mod. Tell Taouil. Cf. Dillemann, 1962:108.

19 Perhaps the same as the ‘In Medio’—the halfway point on the main road from Edessa to Zeugma as given in the Itin. Anton. (189, 4, p. 25, ed. Cuntz). Cf. Lightfoot, 1981:302, n. 15.

20 Identified by Dillemann (1962:232) as Redje (or Radjil), about 5 km. S.S.E. of Hisn Kef beside the Tigris.

21 Most scholars accept Seeck’s suggestion that the missing unit is Legio III Parthica.

22 Located by Poidebard (1932:134) to Tel Fedein (=Apphadna in Ptolemy). See also Fink, 1971:15 and Röllig and Kuhne, 1977:120–1 and 133..

23 Probably the Thilaticomum of the Itin. Anton. (192, 1, p. 26) on the main road between Batna (Serug) and Hierapolis (Membij). Located by Miller (1916: col. 776) at mod. Tell el Ghara.

24 Bintha is most likely a copyist’s error for Birtha/Macedonopolis (mod. Birecik). The unit was probably stationed on the western bank of the Euphrates. Cf. Diz. Epig. IV, p. 1343.

25 Probably the Phichas in Procop., de aed. II, iv, 14, mod. ‘Aïn el Beida. Cf. Dillemann, 1962:187.

26 Rhesaina appears to be the base of two units, one from the Mesopotamian and the other from the Osrhoenean command. Cf. Gracey, 1981:106.

27 The unit would have been stationed on the Lower Khabur where numerous sites have yielded evidence of Late Roman occupation. Cf. Röllig and Kuhne, 1977:121– 5 and 128.

28 Identified by Dillemann (1962:108 and 225) with the fortress of Zamarthas and the monastery of Samarthê in Procop., de aed., II, vi, 14 and V, ix, 32, both restored by Justinian.

29 Probably the same as Apatna above (Or. XXXV, 25).

30 Dillemann (1962:239) prefers identifying it with the Monocarton in Theophylactus Simocatta I, 13, 3, and locates it at Tell Armen near Constantina. See, however, Honigmann, 1939, no. 903.

31 The Equites Pafenses may have derived their name from Tell Fafan (Til) at the confluence of the Tigris and the Bohtan-su between Cepha and Bezabde. Cf. Sarre and Herzfeld. 1911 (i):148 and Lightfoot, 1981:298.

32 Assara may perhaps be amended to Massara, placed by Dillemann (1962:230) at the village of Maserte (perhaps the Matzaron of Theoph Sim. II, 18, 17) about 10 km. W. of Fafa in the Tur Abdin to the north of Dara. For other identifications see Diz. Epig. IV, p. 1332.

33 The unit appears to have been raised in the region of Thebeta (Amm. XXV, 9, 3) on the road between Singara and Nisibis (Miller, 1916: col. 771) and located by Dillemann (1962:212) at El Qoinet in the valley of the Radd.

34 Identified by Dussaud (1927:491, n. 5) as Tell Bisme, W. of Mardin and 50 km. N.E. of Viransehir at the confluence of the Khabour and the Djurdjub, perhaps the same place as the Bimisdeon in Procop., de aed. II, iv, 14.

35 Established by Septimius Severus, the Legio I Parthica was originally stationed at Singara. Its epithet ‘Nisibena’ in the Notitia suggests a long association with the city of Nisibis where it was probably based to counter the repeated attacks of Shapur II. The legion, or some of its elements, also took part in the defence of Singara in 360. Cf. Amm. XX, 6, 8.

36 The legion, or elements from it, also took part in the defence of Bezabde in 360. Cf. Amm. XX, 7, 1.

37 Perhaps a copyist’s error for Carcha (Syriac word for ‘fort’) and therefore alternative name for the. Birtha (which also means ‘fort’) on the Tigris, between Bezabde and Amida. Cf. Dillemann, 1962:235 and 238, map.

38 The title of the unit implies that it was recruited in the early part of the fourth century in Cordyene, one of the five regiones, ceded to Rome by Narses.

39 Site unknown; Dillemann (1962:239, n. 3) believes it was listed in error under the ducate of Mesopotamia.

40 Dillemann (1962:226) identifies Bethallaha with Thallaba on the Tab. Peut. (Miller, 1909: col. 781) and locates it at Tell Brak, site of Byzantine remains on the Dschaghschagh between Nisibis and Thannouris.

41 The title of the unit implies that it was raised during the Tetrarchy in Zabdicene, one of the five new Transtigritanian regiones. Those who served in this and similar units (see e.g. line 34) were treated as traitors by the Persians. After the fall of Amida in 359, survivors of these units were rounded up and executed without distinction. Cf. Amm. XIX, 9, 2.

42 For a detailed study of the history of the Roman army in Arabia, see Speidel, 1977: passim.

43 Deir el-Khaf (Khaf=‘cavern’) is commonly sugested as the site of Speluncis. Cf. GRP, p. 98. However, Kennedy (1982:299) expresses caution on the grounds that the Deir el-Khaf is more suitable as a base for an infantry rather than a cavalry unit.

43a GRP, p. 80 suggests Na’afa but Kennedy and Riley (Rome’s Desert Frontier from the Air, in press, London, commentary to Fig. 135) think the large rectangular fort at Umm er-Resas, Al-Asimah, Jordan, 31° 30 N and 35° 54 E is a more likely location as one of the four later churches built against its walls has yielded a mosaic of AD 785 (first published by Piccirillo) identifying the site as ‘Kastron Mephaa’. We are extremely grateful to Dr Kennedy for allowing us to draw on this yet unpublished information.

44 The name of the legion may reflect Galerius’ special relation with Mars, his protective deity. This suggests a date shortly after 293 for the creation of this legion —a suggestion which is now supported by numismatic evidence from the site. El- Lejju n is also close to Areopolis, the ‘City of Mars’. Cf. Parker, 1986:62–3.

45 Brünnow (1909:71) believes that Dia-Fenis should be sought in the S. Hauran, probably at Qasr el-Azraq but pottery from el-Azraq seems to stop in mid-fourth century and epigraphical evidence from the site suggests that its Roman name was Basienis. Cf. Kennedy and MacAdam, 1985:100–2 and Speidel, 1987:215–19. GRP, p. 100 conjectures Umm el Jimal but Kennedy (1982:152) suggests locating it at or near the XXIII mile-station on the Via Nova Trajana.

46 Brünnow (1909:73) locates the unit to Qal’at ez-Zerqa—where there is an Islamic fort which may have Roman foundations as it was the site of the discovery of a Latin inscription c. 260. See above 3.2.5.

47 Aditta is commonly identified with mod. el-Hadid. Cf. GRP, p. 63. Kennedy (1982: 149), however, points out that such an identification would involve inverting Gadda and Hatita (sic) on the itinerary as given in the Tab. Peut. (X, 1, col. 818) and that Gadda is an equally suitable candidate for el-Hadid. Cf. Miller, 1916: col. 818.

48 On the remains of Roman military buildings in this region, see esp. Parker, 1986: 48ff.

49 Located by Wagner (1977:678, map) to the west bank of the Euphrates between Zimara and Dascusa.

50 Perhaps to be identified with Sule at the source of Harmut-su near Kalecik. Cf. Adontz, 1970:81.

51 Better known as Pithyus, probably the Thia of the Itin. Anton. (217, p. 31)—a Black Sea port between Trachea and Sebastopolis. Cf. Wagner, 1977:673, map.

52 Pre-Roman Dioscurias in Colchis, mod. Sukhumi. Cf. PECS, p. 277, Bernhard.