¹ Diodorus xx, 41. There are in fact no known caves in the area in question.

² Diodorus, loc.cit. for Ophellas; Plut. Cato Minor, 56.

³ Sallust, Jug. 79, where the featureless character of the terrain (neque flumen neque mons erat) is exaggerated.

⁴ ProfessorCary, , in his Geographical background of Greek and Roman history (Oxford, 1949), p. 219Google Scholar, dismisses the Syrtica with the statement that ’a 500-mile strip of desert separates Cyrenaica from the coastal oasis of Tripolitania’.

⁵ Cella, P. Delia, Viaggioda TripolidiBarberia (Genoa, 1819), p. 90Google Scholar.

⁶ Beechey, H. W. and Beechey, F. W., Proceedings of the expedition to explore the northern coast of Africa (London, 1828), p. 210Google Scholar.

⁷ Holmboe, Knud, Desert Encounter (London, 1936), pp. 95–176Google Scholar.

⁸ RE xix, 2098, s.v. ‘Philaenorum Arae’.

⁹ The evidence of the Stadiasmus Maris Magni shows clearly that Aspis lay near the modern Buerat el-Hsun, west of the town of Sine, the ancient Euphranta or Macomades.

¹⁰ H. W. and F. W. Beechey, loc. cit.

¹¹ Strabo ii, 123; xvii, 836.

¹² Muller, C., Geographi Graeci Minores (Paris, 1882), i, 456–7Google Scholar; and Tab. XX.

¹³ The official description of the Italian coast-road and its arch (La strata litoranea delta Libia, Mondadori, 1937Google Scholar) states (p. 134): ‘A pochi passi dell'Arco eretto sulla Litoranea, sono infatti gli avanzi di antichissimi ruderi, che hanno la forma di tomba, entro la quale una tradizione mai interrotta, dal periodo preromano ad oggi, vuole siane stati sepolti i fratelli Fileni,’ There is no other evidence of this remarkable ‘tradition’, nor of any ancient tomb near the site of ‘Marble Arch’. The only archaeological feature visible to-day is a rough field-boundary wall, of a type to be encountered throughout the Syrtic region.

¹⁴ I am indebted to Group Captain J. C. Larking, D.F.C., of Air Headquarters, Malta, for air photographs of this and other Syrtic sites. The identification of Tugulus with Gasr Haddadfa was first made by Cerrata in his Sirtis (Avellino, 1933), p. 220, in which the Roman and early Islamic remains are not differentiated.

¹⁵ Windberg, , RE xix, 2098Google Scholar, where Banededari is accepted as a Libyan form of the place-name, and not as a corruption from a Greek or Punic name.

¹⁶ The wells between Nofilia and Bescer are all more or less brackish, but they are none the less used by the local bedouin, whose palates are less sensitive than those of Europeans.

¹⁷ H. W. and F. W. Beechey, op. cit., frontispiece.

¹⁸ The ancient sites of the Agheila and Bir Umm el-Garanigh areas were visited during the 1951 field campaign of the Map of Roman Libya Committee. Transport and other amenities were generously provided by the British military authorities at Headquarters Cyrenaica District. I am especially indebted to Lieut. A. Weston-Lewis, 16/5 Lancers, and to 2/Lieut. G. Carpenter, R.E., who accompanied our party; and to my archaeological assistants, Messrs. R. M. Bradfield, V. Hancock, D. Strong, and D. Smith. The last-named has kindly allowed me to reproduce his photographs (pls. XVIII, 2; XIX, 1).

¹⁹ Cerrata, op. cit., pp. 227-9. The importance of this discovery appears to have escaped the attention of the builders of ‘Marble Arch’ who, as we have seen, selected a site near Ras Lanuf as the ‘traditional’ burial-place of the Philaeni.

²⁰ Cerrata's account is vague, especially as regards the character of building A, and the period of buildings C and D. Although the latter are patently post-Roman, he thought that they represented the two structures shown on the Pentinger Map.

²¹ Very few rural shrines have hitherto come to light in Libya. The small sanctuary of Jupiter Ammon, at Ras el-Haddagia in Tripolitania (P.B.S.R. xix (1951), pp. 51–56Google Scholar) may be cited as a parallel.

²² Schlumberger, D., ‘Bornes frontières de la Palmyrene‘, Syria, xx (1939), p. 43CrossRefGoogle Scholar.

²³ Lacau, P., ‘Inscriptions latines du temple de Louxor’, Annales du Service des Antiquités de l'Egypte, xxxiv (1934), pp. 17–46Google Scholar. Cf. de Villard, U. Monneret in Archaeologia xcv (1953)Google Scholar.

²⁴ Forschungen in Ephesos (Vienna, 1906), I, pp. 132–40Google Scholar. The Ephesus columns appear to have been erected in the Christian period, but it may be conjectured that in other instances columns of the pagan Tetrarchs may have been re-dedicated to the Evangelists after the triumph of Christianity.

²⁵ Ptolemy iv, 3, 4; Muller, C., Geographi Graeci minores (Paris, 1882), i, 85Google Scholar (Scylacis periplus, 109). The text of Scylax is corrupt, and is uncertain whether ἀλοῦς should be read as ἂλσος. It is clear, at least, that there was a sanctuary of ‘Syrtic Ammon’ in the neighbourhood of Arae Philaenorum. Were it not for the remains at Gráret Gser et-Trab, one might prefer to site it at Maaten Bescer where there were the ᾽μμωνίου Πηγἀς recorded in the Stadiasmus.

²⁶ Strabo iii, 171; xvii, 836. Pliny, H.N. V, 4 (exharena sunt hae).

²⁷ It is interesting to note that Dr.Smith, William, in his Dictionary of Greek and Roman Geography (London, 1854)Google Scholar, anticipated the conclusions to which the new archaeological evidence leads. He stated (i, p. 186, s.v. ‘Arae Philaenorum’) that the legend of the Philaeni ‘has all the character of a story invented to account for some striking object, such as tumuli’, and that Gebel ‘Allah’ (= Ala) seen by the Beecheys ‘has very likely as good claims … to be considered one of the so-called Altars, as any other hill or mound seen or imagined by the ancients.’

²⁸ Strabo xvii, 836. The date at which this frontier was effective is uncertain. (Strenger, F., Strabos Erdkunde von Libyen (Berlin, 1913), pp. 121–2.Google Scholar)

²⁹ Supplementum Epigraphicum Graecum, ix (Leyden, 1938), 1Google Scholar.

³⁰ In the Roman period the mausolea of the farmers who cultivated the fertile area around Macomades were of the obelisk form that prevails in inner Tripolitania but is completely absent in Cyrenaica.

³¹ The problem of post-classical desiccation in Libya is too complex to be discussed here; but there is reason to believe that the marshes of the Syrtic region were wetter in antiquity.

³² RE ii, 2604, s.v. ‘Automala’.

³³ Geographi Graeci Minores, Tab. XX.

³⁴ Schmidt in RE i, 2016, s.v. ‘Anabucis’.

³⁵ H. W. and F. W. Beechey, op. cit.

³⁶ Goodchild, R. G., ‘Boreum of Cyrenaica’, J.R.S. xli (1951), pp. 11–16Google Scholar.

³⁷ Arae Philaenorum to Automalax: 185 stades = 32·8 km. Actual sea-distance from Ras el-Aáli to Bu Sceéfa = 34 km. Fontes Ammonis to Automalax: 180 stades = 31·9 km. Actual sea-distance from Bescer to Bu Sceéfa =31.5 km.

³⁸ Goodchild, loc. cit.

³⁹ The Greek mariners whose information is embodied in such documents as the Stadiasmus seem to have been somewhat arbitrary in giving their own names to landmarks in preference to local names. The naval hydrographers of the nineteenth century were similarly inclined, and named Geziret el-Maracheb, in the Gulf of Bomba, ‘Seal Island’.

⁴⁰ The Peutinger Map gives the place-name as Zagazaena, but Zacasama, as listed in the Ravennas Cosmography (iii, 2) seems closer to the Sacamaza of Ptolemy (iv, 3).

⁴¹ Air photographs of the area of El-Agheila have so far failed to reveal any indications of a Roman fort; but it must be admitted that blown sand could have obscured such traces.

⁴² Windberg in RE, loc. cit. I am indebted to Mr. K. R. Butlin for information relating to the sulphur deposits of the Syrtic Gulf. The sulphur is formed by microbiological action, which is now being studied in English laboratories.

⁴³ Cato's army used donkeys in its epic march through the Syrtica (Plut., Cato Minor, 56), but by the third century A.D. the camel was widely used in Libya.

⁴⁴ G. A. Freund, travelling from Benghazi to Tripoli in May 1881, observed the wheel-ruts of an ancient road at a place called ‘Egelte Sania’ situated between the wells of Bescer and the Sebcha Mugtaa (Pioneri Italiani in Libia, Milan, 1912, p. 168). Although this report has not been subsequently confirmed and the exact site is uncertain, Freund was an acute observer, whose archaeological notes are generally reliable. In the Cyrenaican gebel ancient wheel-ruts are commonly encountered.

⁴⁵ The Antonine Itinerary gives the distance from Macomades (Sirte) to Boreum (Bu Grada) as 221 Roman miles (=32 7 km.). The actual distance, as measured on the latest maps, is 343 km. In such terrain, and on so ill-defined a route, the margin of error is astonishingly small.

⁴⁶ Of this campaign we have only the record of Zonaras (xi, 19), and a brief allusion in the geographical poem of Dionysius Periegetes (210). That Suellius Flaccus marched through the Syrtic region is confirmed by a boundary-inscription found near Sirte (IRT, 854).