Published online by Cambridge University Press: 24 September 2012
Since the hurried journey of Hogarth and Yorke in 1894, the Euphrates limes has remained virtually unexplored, and uncertainty has continued to surround its geography and organization.
The basic structure, established under Vespasian and consolidated by Hadrian, and many of the details of its course, can now be located on the ground. Three legionary fortresses—at Satala (Plate IV, 1), Melitene and Samosata, with legionary vexillations (replaced under Diocletian by a new legion) at a fourth at Trapezus—blocked the major strategic routes from the east, and are already well known. They were linked by an eight metre military road, whose remains I have managed to trace on foot, almost without a break, for a hundred miles north of Melitene, along the right bank of the Euphrates: barely a fifth of the total distance from the Black Sea to the southern foothills of the Kurdish Taurus. A series of auxiliary forts seems to have stood on, or at points east of, the road at intervals of a day's march. Most of the sites hitherto proposed—by armchair inspection of inaccurate maps—should be dismissed, but genuine sites have proved elusive. Only at Dascusa has excavation been possible, to reveal fourth-century work, rather than the original fort. But an important inscription in re-use attests military activity under Titus or Domitian: along with the Pompeius Collega milestone, it provides confirmation of scattered literary and numismatic evidence for Vespasian's hand in the building of the frontier.
1 Yorke, V. W., Geographical Journal 8 (1896), 317–335 and 453–472CrossRefGoogle Scholar. Some sections of the limes were briefly visited by Stark, Freya, Rome on the Euphrates (London 1966)Google Scholar. The article limes in the Dizionario epigrafico has been in suspension for over a decade.
2 Several seasons of fieldwork have been made possible by the generous support of the British Institute of Archaeology at Ankara, and of the Craven and Tweedie Exploration Committees.
3 Harper, R. P., Anatolian Studies 22 (1972), 27CrossRefGoogle Scholar f., and below, Inscription 8; ILS 8904, Melik Şerif, of early 76.
4 I hope to publish the inscriptions in Studia Pontica 3, fasc. 2; and to discuss the structure and history of the frontier more fully elsewhere.
5 Chrysanthos, , ‘Αρχ. Πóντ.’ 4–5 (1933), 380, n. 2:Google Scholar cf. Ballance, S., AS 10 (1960), 146–151,Google Scholar and plate XVIb.
6 D. M. Magie, Roman Rule in Asia Minor 1470, n. 6; Weber, Hadrianus 123, 264.
7 Periplus 1, 2.
8 Periplus 2, 1; F. Cumont, Studia Pontica 2, 366; Hamilton, W. J., Researches in Asia Minor, Pontus and Armenia 1 (London 1842), 162Google Scholar.
9 Respectively unpublished, and IGR 3, 106 and 1440.
10 Periplus 1, 3 f.
11 Arakelyan, B. N., VDI 118 (1971, no. 4)Google Scholar, 114–118; ILS 8801.
12 Compare Danoff, , R-E Suppl. 9, 1062Google Scholar f., s.v. Pontos Euxeinos, para. 17.
13 Tac, ., Ann. 13, 39;Google ScholarAntonine Itinerary (ed. Cuntz, 1939), 216, 4.
14 For example, Barkley, H. C., A Ride through Asia Minor and Armenia (London 1891), 347Google Scholar.
15 Tac, ., Hist. 3, 47;Google ScholarJosephus, , Bell. Jud. 2, 16,Google Scholar 4, (s. 367, Niese).
16 Periplus 16, 6; Chrysanthos, map, loc. cit.; Brant, J., JRGS 6 (1836), 189Google Scholar f.
17 Periplus 3, 1; Collingwood, R. G. and Richmond, I. A., The Archaeology of Roman Britain (London 1969), 85Google Scholar.
18 Periplus 3, 1–10, 4; cf. ILS 2660 (Abella). XII Fulminata was not present in full strength in Arrian's army against the Alani, , Extaxis 6, 15 and 24Google Scholar.
19 Procopius, , Bell. Goth. 8, 2, 16;Google ScholarPeriplus 9, 3 f. and 4, 1 f.; Pliny, , NH 6, 12Google Scholar f. Their sites have been identified by the presence of structures of a much later date, and only at Sebastopolis have first-and second-century remains been claimed. For Apsarus, at Gonio, five miles south of Batumi, see Levkinadze, V. A., Byzantinsk. Vremen. 20 (1961), 225–242;Google Scholar for Phasis, beneath Poti, de Montpéreux, F. Dubois, Voyage autour du Caucase 1 (Paris 1839), 65Google Scholar f., and Atlas, plate XVIII; for Sebastopolis, beneath Sukhumi, Levkinadze, , VDI 108 (1969, no. 2), 81Google Scholar ff.
20 Vexilatio leg. XII Ful., CIL 3, 6745; medicus leg. XV Apo., ibid., 6747 (Trapezus).
21 Among τò συμμαχικóν, Arrian, Expeditio c. Alanos 7 and 14.
22 See below, Inscription 4.
23 ILS 9117 and 394 (Eçmiadzin).
24 The Zydreitae, whose territory evidently embraced the area in which Petra was later founded, were alone excluded from Hadrian's comprehensive system of clients among the coastal tribes between Trapezus and Sebastopolis. Inherited from Trajan, his policy was evidently continued under Antoninus, who installed a client among the Lazi, and appointed a praefectus orae gentium Ponti Polemoniani, presumably to administer and collect taxes from the coastal tribes. The Zydreitae were perhaps included in these later arrangements: they flanked the Lazi to the south, and had no doubt lost their Iberian support on the defeat of the Alani. Levkinadze, VDI, loc. cit., 87; Arrian, , Periplus 11, 2;Google Scholar Trajan, Magie, Roman Rule 1465, n. 32; Antoninus, SHA Pius 9, 6, and AE 1956, 124.
25 Zosimus 1, 33. See also below, Inscription 5.
26 Arrian, , Periplus 11, 1Google Scholar f.
27 Mitford, , Byzantion 36 (1966), 482Google Scholar ff.
28 ILS 639 (Trapezus). Cf. Bean and Mitford, Journeys in Rough Cilicia (Denkschr. Akad. Wien. 1970), no. 50, Colybrassus (?); Notitia Dignitatum Oriens 38, 9 and 15.
29 Studia Pontica 3, no. 34.
30 Unpublished, at Sebastopolis: Valeria Mais wife of Claudius Chareisios.
31 Unpublished, Satala. Birley, A., Life in Roman Britain (London 1964), 45Google Scholar f.
32 de Tournefort, P., Relation d'un voyage du Levant i (Paris 1717), 169,Google Scholar mentions the aqueduct. Sir Porter, R. Ker, Travels in Georgia, Persia, Armenia, Ancient Babylonia, etc., etc., 1817–20 (London 1822), 684,Google Scholar passed the village of Saddock.
33 ILS 8904, (Melik Şerif), of early 76; Cumont, F., Bull. Acad. royale de Belgique (Brussels 1905), p. 197Google Scholar f., from Sipdiǧin, three miles east of Refahiye, of 92–94.
34 Tac, ., Ann. 12, 45Google Scholar and 13, 39.
35 For the date, Reinach, T., ‘Le mari de Salomé et les monnaies de Nicopolis d'Arménie’, REA 16 (1914), 132CrossRefGoogle Scholar ff.; cf. B. Pick, ‘Une monnaie du κοινòν Άρμενίαϛ’, id., 283 ff. The coins do not exclude late 71.
36 Unpublished (Satala). CIL 3, 264 and 6766/7 (Ankyra).
37 Cohors III Ulpia ntiliaria Petraeorum equitata sagittariorum, AE 1931, 36 and 38 (Chellah); cf. Arrian, Ἔκταξιϛ I, and Notitia Dignitatum, Oriens 38, 27, stationed later at Metita, in the Kurdish Taurus. Cohors I Germanorum miliaria equitata, ILS 8868 (Prusias ad Hypium); cf. AE 1931, 36 and 38; Arrian, 2; and Notitia 30, stationed at Sisila in Armenia Minor. Ala I Augusta gemina Colonorum, AE 1926, 150 (Beirut); cf. Arrian, 1; Notitia 21, stationed at Chiaca, north of Melitene.
38 Dio 55, 24, 3.
39 Josephus, Bell. Jud. 7, 1, 3; Suetonius, Vesp. 8; Tacitus, Ann. 15, 6 f.
40 N. Marcius Plaetorius Celer, subsequently appointed to command the garrison at Apsarus, ILS 2660 (Abella). The legion was perhaps commanded during the war by L. Burbuleius Optatus Ligarianus, who returned to govern Cappadocia under Hadrian and Antoninus, ILS 1066 (Minturnae).
41 Fronto, p. 205; AE 1931, 36 and 38 (Chellah).
42 ILS 8795 (Harmozica), after 1st July, 75; AE 1951, 263 (Büyük Taş), above the Caspian, of 84–96. They blocked respectively the southern approaches to the Caucasian and Caspian Gates. With the latter should be associated the lost Latin inscription mentioning XII Fulminata, from the left bank of the lower Araxes, Jampolski, , VDI 31 (1950, no. 1), 182Google Scholar. The legion's activities may have a connection with the death in 91/2 of Rutilius Gallicus, Statius, Silvae 1, 4, 79Google Scholar.
43 Dio 68, 19,2; and see Inscription no. 4. A move to Samosata under Hadrian is less conveniently explained, but cannot be ruled out.
44 Dio 55, 24, 3; cf. Ptolemy 5, 14, 8; ILS 1142 (Rome).
45 Satala is usefully described only by Taylor, , JRGS 38 (1868)Google Scholar, (Hogarth, and) Yorke, , GJ 8 (1896)Google Scholar, and Cumont, Studia Pontica 2. Biliotti's account of his excavations, which failed to recover the body of the bronze head of Aphrodite, is preserved in a consular report in the British Museum, and will be published shortly in Anatolian Studies.
46 CIL 3, 4491, perhaps the latest record of XV Apollinaris at Carnuntum.
47 Arrian, Ἔκταξιϛ 5, 15 and 24; cf. CIL 11, 383.
48 Tac, ., Ann. 15, 25Google Scholar f.
49 From Moesia Superior, VII Claudia took part in expeditione [P]artica et Ar(meniaca), AE 1905, 163; cf. Arrian, Parthica frag. 80, and ILS 2083. From Moesia Inferior, a vexillation of I Italica, which had fought in Dacia, was present at Artaxata, B. N. Arakelyan, VDI, loc. cit.; while from Germania Superior XXII Primigenia served in bellum Armenia-cum et Parthic(um), AE 1962, 311. Some perhaps spent the winter of 113/4 in Ankyra, , IGR 3, 173Google Scholar.
50 Ala I Flavia Aug. Britannica miliaria c.R., attested continuously in diplomata of Pannonia Inferior from 110 to 167, was described on 1st September, 114 as missa in expeditionem, CIL 16, 61; it must have returned to the Danube by 123, for discharges in 148 included Reidomarus, an Eraviscus, , CIL 16, 179/80Google Scholar. During this absence it left epigraphic trace of its deployment at Amaseia and perhaps at Nicopolis, important stages on the road to Satala, Studia Pontica 104 and AE 1908, 23. Ala I Claudia nova appears to have been similarly transferred from Dacia or Moesia Superior, outside which it is known only at Amaseia, Studia Pontica 105.
51 praef. vexillation. eq. Moesiae Infer, et Daciae … in expeditione Parthic., ILS 2723 (Rome).
52 Dio 68, 19; cf. Fronto, p. 205. IV Scythica, attested at Artaxata in 116, may be considered to have garrisoned the eastern part. Trajan left troops in strategic places, Dio 68, 21, and Arakelyan, loc. cit. Satala lies at 5,900 feet.
53 Dio 55, 23, 5; Antonine Itinerary 183, 5; Notitia Dignitatum, Orient 38, 13.
54 cf. Yorke, V. W., JHS 18 (1898), 321, no. 36Google Scholar.
55 CIL 3, 242, 268, 6761; there may also be a tribune, id., 6752.
56 Studia Pontica 34.
57 It is probably of Trajan, of the mint of Caesareia. The same countermark appears on a coin from Nicopolis, of 113/4, Reinach, T., Recueil général des monnaies grecques d'Asie Mineure, I (ed. 2, 1925), 136,Google Scholar n. 2, and REA 16 (1915), 136Google Scholar f., fig. 5 f.
58 See above Inscription 2. A road was apparently constructed from Satala to Kainepolis: CIL 3, 13627A, a milestone fragment perhaps of Commodus, was found among the north western foothills of Mt. Ararat, some twenty miles south west of Kainepolis.
59 CIL 14, 5357, Imperatore Gallieno Augusta V et Nummio Fausiano cons 2. Faustianus recurs in SHA, Gallienus 5, 2;Google Scholar and in Fasti Theonis Alex. 3, 378. The variations Faustinianus and Faustinus are also known from later Fasti, PIR 2 125. For the Nummii, brought to prominence by Gallienus' patronage with the consulship, cf. Diz. Epigr. 2, 1015.
60 Magie, 1568, n. 29. CIL 3, 1418414, where the reason for his repairs is unstated. The bridge, 30 m long, rested on foundations of solid rock. Its larger Severan counterpart still carries jeeps and trucks across the Chabina; see below, Inscription 9.
61 Zosimus 1, 33. The extra soldiers defeated at Trapezus must have been drawn from supporting garrisons and from XV Apollinaris itself, μνρίων may indicate no more than an indefinitely large number.
62 Honigmann, E., Maricq, A., Recherches sur les Res Gestae Divi Saporis (Brussels 1952)Google Scholar, para 18. See also ‘Res Gestae Divi Saporis’, Syria 35 (1958), 295:Google Scholar Satala in 1. 18.
63 Notitia Dignitatum, Oriens 38, 13; A. H. M. Jones, LRE 3, 381, dates the whole document ‘basically to 408’: but the armies of the eastern frontier are in it preserved largely unaltered since the time of Diocletian.
64 For instance at Ankyra, , IGR 3, 206;Google Scholar and perhaps at Dorylaeum, Magie, 1567, n. 28.
65 Taylor, J. G., JRGS 38 (1868), 287Google Scholar f.; cf. CIL 3, 141843
66 See above, Inscription 2.
67 Compare Studia Pontica 81, 157B, and, for the formula, 221.
68 Compare ILS 8880, βφ… καὶ κορνικονλάριοϛ καὶ ἑκατóνταρχοϛ γενομἑνοϛ τῆϛ ἡγεμο (νίαϛ). And CIG 3, 4453.
69 Dobson-v. Domaszewski, Rangordnung 29 and 38.
70 IGR 3, 115; Studia Pontica 96.
71 Compare the consul of 142, L. Cuspius Pactumeius Rufinus; and L. Catilius Severus Julianus Claudius Reginus, governor of Armenia Major under Trajan, and associated with a Fabius (?) at Zimara, in Inscription 7.
72 The surviving letters suggest, for example, [λεγ. ιε´Ά] [π]ολ.
73 Compare at Trapezus Inscription 1, above; and at Kainepolis the tombstone erected by Valens, P. Aelius, tribune of λεγ. ιε´Ἀπολ, SEG 15, 839Google Scholar.
74 Arrian, , Periplus 1, 2Google Scholar. His army included three Greek officers in command of cohorts: Daphnis of Corinth, Demetrius, and Lamprocles: Arrian, Ἔκταξιϛ 1 and 3.
75 Titianum, suggested by the nomen Fabium as well as by the erasure, cannot be supported from the letter traces, ILS 741.
76 Inscriptions of Nero, CIL 3, 6741, 6742, 6742A, Kesrik and Kowank, in the plain of Elaziğ, below Harput; of Marcus and Commodus, ILS 9117 and 394, at Kainepolis; of Severus, AE 1908, 22, at Carsaga (Melik Şerif); of Decius, , CIL 3, 14184Google Scholar14, at the Karabudak bridge, fourteen miles east of Zimara. In each the average line length is of 16–20 letters, distributed over 7 or 8 lines. the Aurelian inscription at Satala, mentioned under Inscription 5, is too fragmentary for comparison; and the Severan on the Chabina bridge are cut on stelae, CIL 3, 6709–11. the fragmentary inscription from Dascusa (below, Inscription 8) appears to be the only exception. In eastern Cappadocia and Pontus, such Latin texts are entirely lacking.
77 Compare at Kainepolis ILS 9117 and 394.
78 ILS 1041 (Antium); cf. Groag, PIR 2 C 558.
79 Compare at Risingham C. Julius Marcus, governor of Britain in 213, RIB 1235, cf. 1202.
80 ILS 1141/2 (Rome).
81 Inscription 8.
82 Dio 68, 19; Magie, Roman Rule 1464 f., n. 31.
83 Perhaps including IV Scythica at Kainepolis, and XV Apollinaris at Satala, Dio 68, 21 and Exc. 75, 9, 6; and see Inscription 4, above.
84 SHA, Hadr. 9, 2, and 5, 10;Google Scholar cf. Fronto, p. 206.
85 Pliny, , NH 5, 83;Google ScholarYorke, , GJ 8 (1896), 454;Google ScholarWünsch, J., AEMO 8 (1884), 239Google Scholar ff. From Pingan, goat skin rafts now carry corn through the Antitaurus gorge to Kemaliye.
86 Cf. CIL 3, 12218 and ILS 263. For the imperial titles, cf. ILS 268 (Ankyra). The cohort, which formed a part of the early and permanent auxiliary garrison of Cappadocia, is supplied exempli gratia.
87 Vias provinciarum Galatiae, Cappadociae, Ponti, Pisidiae, Paphlagoniae, Lycaoniae, Armeniae Minoris stravit, ILS 268 (Ankyra); cf. Groag, , PIR 2 C 170Google Scholar. He is attested on five milestones, cited in chronological order, on roads radiating from Ankyra. Under Titus to the west to Dorylaeum, ILS 263; and south to Derbe, on the borders of Lycaonia and Pisidia, CIL 3, 12218. And under Domitian on the road to Tavium, AS 4 (1954), III ff., no. 8; at Ankyra, ILS 268; and on the road leading south east to Parnassus and Caesareia, CIL 3, 14184.48
88 PIR 2 J 241; CIL 3, 250 (Ankyra).
89 Carsaga, ten miles east of Refahiye, where a milestone of Cn. Pompeius Collega records the construction of the frontier road in early 76, ILS 8904.
90 The opus cochliae, built in 72/3 by III Gallica, ILS 8903 (Aini). Presumably as a part of the overall frontier system, M. Ulpius Trajanus constructed the road leading north east from Palmyra to the Euphrates at Sura in 75, Bowersock, JRS 63 (1973), 133Google Scholar ff. Support cannot readily be offered for his view, suggested largely ex silentio by Syme, Tacitus 31, n. 1, that Trajanus may also have organized the frontier against Armenia. It was an immense task. The limes stretched across country as difficult as any in the Empire, for five hundred miles north from Samosata. Two kingdoms required organization—Armenia Minor, annexed in 71/2 (see Inscription 3 above, with n. 35), and Commagene, added after July 72. And Trajanus is attested in Syria as governor by 73/4. Even for the Emperor's father, two years at most were not enough.
91 Above, Inscription 7.
92 Either by L. Antistius Rusticus, who died in office in Cappadocia in 93, Ramsay, , JRS 14 (1924), 182;Google Scholar or by (L. Caesennius) Sospes, the praetorius who apparently relieved him at short notice, ILS 1017(Pisidian Antioch); cf. Syme, , Hermes 85 (1957). 493,Google Scholar n. 2.
93 See especially CIL 3, 1418423, in the Lycus valley west of Neocaesareia, and an unpublished stone in the Sivas museum, perhaps from the road to Zara; and Magie, Roman Rule 1453, n. 10.
94 Pliny, , NH 5, 83Google Scholar and 6, 26; Orosius 1, 2, 23.
95 AS 22 (1972), 27Google Scholar f., and fig. 5.
96 AS 21(1971), 49Google Scholar.
97 In Pindar, and SEG 3, 400.
98 As dux exercitus Illyrici expeditione Asiana item Parthica, ILS 1140 (Tarraconnensis); cf. Stein, , PIR 2 C 823Google Scholar.
99 Sometimes with Parthicus. For Severus as Imperator V–VII, all while trib. pot. III in 195, see ILS 417 and 438; and RIC 4, 1, 60 and 98 f., nos. 58 f.
100 Magie, Roman Rule 1540 ff., nn. 22, 25, 26.
101 A solo restituerunt et transitum reddiderunt, CIL 3, 6709–11. Hogarth and Yorke reported an erased inscription of Vespasian on the north east column of the bridge, JHS 18 (1898), 315Google Scholar. Of their imagined letters, not one can be traced with any confidence on any drum of the column. In other inscriptions their accuracy is demonstrably poor, and they appear to have been misled by artificial marks on the lowest drum.
102 Fifteen milestones of Aelianus, C. Julius Flaccus, PIR 2 J 311;Google Scholar cf. Magie, op. cit. 1349 f., n. 3. The road can be traced eastwards from Elbistan, closely following the line of the modern road to Malatya, as far as the final summit four miles west of the junction for Darende and Malatya. There is a group of two badly worn milestones at the summit, above Topalı Yazı Köy, and half a mile east of the village is a third, of Diocletian.
103 The road is known only from the Antonine Itinerary 207, 10 and 215,7, and must be distinguished from any that may have followed the precipitous right bank of the Euphrates gorge through the Kurdish Taurus. Of the road per ripam no trace survives. The stations along the river can only have been linked by tracks possible at best for pack animals, involving a continuous switch-back of exhausting climbs and descents. The section describe by Freya Stark is altogether easier than the northern half of the gorge (Rome on the Euphrates 170 f.). The latter is wholly unsuitable for the passage of an army. The direct route was followed by Constantius in 360, on a journey from Caesareia to Edessa, Ammianus Marcellinus 20, 11,4; and presumably by Corbulo, with army and baggage train, nearly three centuries before, Tac, ., Ann. 15, 12Google Scholar. Lacotena is known only from the Antonine Itinerary, loc. cit., which dates perhaps from the time of Caracalla; and from the visit of Constantius. the earliest coin known from the site is an aes of Commodus, with on the reverse the legend ὑπἑρ νικ. Ῥωμαίων.