¹ It is being published by the Oxford University Press. The following abbreviations are used in this article. Hughes = Hughes, T. S., Travels in Sicily, Greece and Albania (London, 1820)Google Scholar; Kromayer = Kromayer, J., Antike Schlachtfelder in Griechenland II (Berlin, 1907)Google Scholar; De Sanctis = De Sanctis, G., Storia dei Romani IV (Torino, 1923)Google Scholar; Walbank, Philip = Walbank, F. W., Philip V of Macedon (Cambridge, 1940)Google Scholar; Walbank, Polybius = Walbank, F. W., A Historical Commentary on Polybius 1 (Oxford, 1959).Google Scholar References to Plutarch, Flamininus are to the Teubner edition (ed. Lindskog-Ziegler). Place-names are spelt as on the maps cited in n. 4 and n. 6; the heights are also from these maps.

² Travels in Northern Greece I, 32 and 385 f. (London, 1835).

³ II, 273. Both say the wall was near the foundations of the principal entrance to the Turkish fort (this was itself a ruin in 1931).

⁴ The best geological map is that of E. Nowack (1: 200,000), based on observations in 1922–1924.

⁵ Lent to me by the British School of Archaeology at Athens. Clarke travelled in Epirus in 1922 and 1923 and was drowned in an accident in the Gulf of Corinth.

⁶ The reference is to the War Office Map (1: 100,000), 1944, based on the Greek Staff Map.

⁷ Hughes II, 272 took some 14 hours by a devious route from Berat to Kelcyrë and 14 hours from Kelcyrë to Mertzani.

⁸ Hughes II, 254 reckoned it at 12 hours riding, but it is not clear that he is giving his own time. I walked from Byllis to Berat in 6½ hours; it is easier going; see n. 13 below.

⁹ It is described in detail in my book on Epirus.

¹⁰ C. Praschniker in Jahrb. d. öst. arch. Inst. Wien XXI-XXII(1922–24), Beiblatt, 1 ff. held that the plain was even more swampy in ancient times than today; because of this the Via Egnatia followed the foothills at the inland side of the plain.

¹¹ So Leake, Travels in Northern Greece I, 360 f.; 111, 325 f. and 485 f.; Hughes 11, 383; Patsch, C., Sandchak von Berat (Vienna, 1904)Google Scholar; Kromayer p. 10, n. 4.

¹² Studia Albanica I (1964), 184, Tirana: the masonry is illustrated there in Plates II, 2 and 3 and III, 2 on pp. 191 ff. A copy of this new Albanian periodical was kindly sent to me by the Albanian scholar, Professor Frano Prendi. Livy 31, 27 mentions the strength of the walls. The strong site of the fort is shown by the sketch in Hughes 11, 255.

¹³ The best crossing of this range is just south of the high point. I walked in 1931 from Byllis in the Aous valley via Han Ballsh and Metoh to Berat in 6½ hours (walking time) through wooded chalky foot hills; this area contained Christian villages, whereas those to west and north were all Mohammedan.

¹⁴ The only fortified site so far known in the area inland of Berat is at Kalaja Rrmait to the north of Mt. Tomor (Riv. d. Alb. 3, 157) in Gramsh just NE. of Mirakë (Buletin i Universitetit Shtetëror të Tiranës, seria Shkencat Shoqërore, 1963, 4, 3 ff., where the name is given as Kalaja e Irmajt). The strong walls of ashlar masonry (Buletin, figs. 3–7) are in the same style as the foundations at Berat. As the Illyrians did not build walled sites, these walls are due to Epirus or Macedonia. It is probably to be identified with Codrion. It lies north-east of Berat in the direction of Lake Ochrid.

¹⁵ Kromayer 10 considers the campaign by Apustius to have been a ‘Streifzug’. He does not appreciate the significance of his actions at Antipatrea and Codrion.

¹⁶ Kromayer II and Walbank, Philip 142, put these passes in the valley of the Vardar (Axius); but I doubt if Lyncus extended so far east.

¹⁷ I walked from Elbasan on the Genusus to Shtërmen (see Map II) in the sink north of the bend in the Apsus river, and on the way I saw the piers of the ancient bridge which has been described by Praschnicker and Schober, ‘Archäolog. Forschungen in Albanien u. Montenegro’, Schrift. d. Balkankommission VIII (1919), 59 f. Its position shows that the route through this sink was a main route, probably in the time of Pyrrhus rather than in that of the Roman occupation.

¹⁸ The Greek position in the winter of 1940–1941 defended this route against the Italians. The Greek line ran from Pogradec on the southern shore of Lake Ochrid to Grabovë-e-Krishterevë; the sector from this Grabovë to Sqimar on the river Tomorricë cuts the tracks eastwards from Berat to Strelcë and to Voskopoj. From Sqimar the line ran south between Mt. Tomor and Mt. Zaloshnyë to Fratar, just east of Kelcyrë. See Papagos, A., The Battle of Greece, 1940–1941, 290 (Athens, 1949).Google Scholar The proposed Greek offensive (pp. 339 ff.) shows the importance of Berat.

¹⁹ If the Romans advanced up the Genusus valley to the northern end of Lake Lychnitis as Kromayer and Walbank suppose, one cannot see how Philip failed to make contact; for this entry must have been an expected one, and it is also a narrow one. Nor is Dassaretis likely to have extended north of the Lake. Polybius described the route of the Via Egnatia with some detail. The description survives in Strabo (C 323, cf. C 327), but no mention is made of Dassaretis or of the river Bevus or of the town in Macedonia called Beve (Steph. Byz., Βεύη πόλις Μακεδονίας καὶ πρὸς αὐτῇ Βεῦος ποταμός τὸ ἐθνικὸν Βευαῖος). The river and the town are likely to have lain south of Lake Lychnitis. Lyncus is of course the name of the canton occupied by the Lyncestae; Sulpicius camped near the frontier of Lyncus and foraged in Dassaretis. Kromayer, 16 f., seeing the difficulty of Philip failing to find Sulpicius, argued that Philip was still at Pella when Sulpicius was already in Lyncus in the plain of Monastir; but, if so, it is even more odd that his cavalry could not find Sulpicius in that plain.

²⁰ Thus he did not enter Epirus, as he would have done if he had gone via Leskoviq. Leake followed this route in September 1805; see n. 24 below.

²¹ The famous march of Alexander from Pelium to Pelinna (near Tricca) in Thessaly (Arrian, Anab. I, 7, 5) followed such a route.

²² Kromayer 35, ‘die Verluste der Römer in Obermakedonien waren so bedeutend gewesen, die Überzeugung, hier nicht zum Ziele kommen zu können, muss bei den Römern so allgemein gewesen sein, etc.’

²³ Kromayer 37 ‘Wenn alle Berechnungen trogen und die Römer sich doch direkt gegen Makedonien in Marsch setzen, … sich ihnen vorzuschieben und … dem Gegner den Marsch auf Pella zu verlegen.’

²⁴ Leake travelled in the reverse direction, from Koritsa to Berat, in 28½ hours in wet weather in September (1, 343 f.); Philip therefore would not have been able to overhaul a Roman army which had at least 15 hours start.

²⁶ De Sanctis 59.

²⁶ Philip 148 f.; Holleaux, CAH VIII, 168, has much the same view.

²⁷ One Italian column in 1940 took this route and got into serious difficulties. Kromayer described the upper Aous valley as ‘das sich über 100 Kilometer südöstlich ins Gebirge hineinziehende, verhältnismässig weite und bequeme Tal der Wiossa’! He certainly had no idea of its nature above Konitsa.

²⁸ This is where De Sanctis 60, n. 117, put Philip's position: he was not aware that it does not fit Plutarch's description at all.

²⁹ Walbank, Polybius I, 156, realizes this: ‘the στενά must be the short gorge on that river (the Drin) immediately south of its confluence with the Viossa, despite the more common application of the term to the gorge of the Klisoura on the Viossa itself’. But the gorges are five miles or so apart, on different rivers and giving access to different areas; they were hardly likely to have shared a common name.

³⁰ Philip, facing p. 148, no. 1.

³¹ See Livy 32, 12, 1–3. Similarly Leonidas fought in advance of his strongest point at Thermopylae; see my History of Greece 234. Defence in depth has obvious advantages.

³² See his Map 3 based on the Austrian Staff Map, which is inaccurate. I have put his positions for the Macedonian forces on my Map IV.

³³ Five miles in the other direction would put the Roman camp south of Lekel (Antigonea) and would have left Philip cutting Villius' lines of supply from Apollonia. Leake 1, 388 put the Roman camp near Lekel.

³⁴ Livy 32, 6, 3 presents such a withdrawal as an alternative to attack.

³⁵ Lyncus is mentioned in order to bring out the precedent afforded by Sulpicius.

³⁶ The reinforcements came either via Phoenice and the Drin valley or via the Skarficë pass through the Kurvelesh, the latter perhaps being more likely. Philip evidently had no cavalry with him, his army of phalangites, mercenaries and light-armed troops was very mobile, and he must have left large forces in Macedonia to hold back his numerous enemies.

³⁷ Plutarch uses imperfect tenses, which implies several attacks. Walbank takes Livy's forty days ‘sine ullo conatu’ (32, 10, 1) to be forty days of attacks, but this is not justified by the Latin text. The first attack is described at 32, 10, 9. He also reverses the order of events in putting the parley after his supposed forty days of attacks; this is an arbitrary rearrangement of Livy.

³⁸ The question whether Livy and Plutarch drew only on Polybius or on other authors cannot be properly treated within the limits of this paper. But it is worth noting that there are considerable differences between the account in Livy and the account in Plutarch. I list some of them. Livy has one shepherd; Plutarch has several shepherds. In Livy the shepherd was sent by Charops; in Plutarch the shepherds came of their own initiative and referred Flamininus to Charops (Livy 32, 11, 1; Plut., Flam 4, 4). When the turning party started off, Livy reports that for two days attacks were delivered from every side; Plutarch says that the Roman army rested for two days (Livy 32, 11, 6; Plut., Flam 4, 8). On the morning of the third day Livy says that the Romans saw the smoke-signal before they attacked; Plutarch says that they saw it after the attack (Livy 32, 12, 1; Plut., Flam 4, 12). In Livy 32, 12, 3–4 when the impetuous Romans ran into danger the shouts of the turning party threw the Macedonians into a panic; in Plutarch, Flam. 4, 10 all three Roman columns were held up before the smoke-signal was seen. Thus Livy's account is not only much more laudatory of Rome than Plutarch's account, but it also has substantial differences of detail and of timing. If Livy and Plutarch used only one source, namely Polybius, as Kromayer 11, 41, with n. 2, following Nissen, and others believe, then we must recognize that either one (and, if one, Livy) or both treated the account of Polybius with considerable freedom. The possibility that a Roman source who was acquainted with Ennius' Annales was also used, at least by Livy, whose mention of Valerius Antias at 32, 6, 5–8 only to refute him is peculiar, should not be excluded from consideration.

³⁹ So too Leake 1, 388.

⁴⁰ BSA 32, 145 and in my forthcoming book, Epirus.

⁴¹ E.g. by Kromayer 11, 36, Walbank Polybius 1, 163 and Holleaux and others in CAH VIII, map 7, p. 117.

⁴² Römische Geschichte 1, 709.

⁴³ Kromayer 11, 36.

⁴⁴ Eph. Arch. 1914, 239.

⁴⁵ Hughes, op. cit. 11, 253 gives a noble description of the mountain.

⁴⁶ This area was a most important one in the Greek position in 1941 when the Italians made their spring offensive against the sector between Mt Trebeshin and Bubesi. See Papagos, A., The Battle of Greece 1940–41 (Athens, 1949), 300.Google Scholar The division in the Greek line between the Army of Epirus and the West Macedonian Army was made on the line of the Osum just south of the Tomor massif (p. 304).