¹ Cagnat, R., Mém. Acad. Inscr. XXXIX (1912), 77Google Scholar, incorporated verbatim in the same author's L'Armée romaine de l'Afrique² (1912), 11, 524–568Google Scholar. Subsequent references are to this edition. See also A. Merlin, CRAI 1921, 236-248.

² For the ancient topography of the coast of the Greater Syrtis see Cerrata, L., Sirtis (Avellino, 1933)Google Scholar. Cerrata's identifications of the Roman road-stations are to be treated with reserve.

³ Nouv. Archives des Missions scientifiques x (1902), 246–277Google Scholar; XII (1904), 1–80; and XIII (1906), 73–102.

⁴ For this road see Romanelli, P., Epigraphica 1 (1939), 110Google Scholar, where a military character is postulated; and Goodchild, R., Roman Roads and Milestones in Tripolitania (Tripoli, 1948) 11–13Google Scholar (cited hereafter as Roman Roads), where the economic significance of the route is stressed.

⁵ Inscriptions of Roman Tripolitania (publication now in preparation by the British school at Rome), 854. Hereafter cited as IRT.

⁶ Cagnat-Merlin, Inscr. lat. d'Afrique 30.

⁷ CIL VIII 11048 (Gasr Ghelan) and Cagnat-Merlin, op. cit. 26 (Sidi Mohammed ben Aissa).

⁸ SHA, Severus 18, 3; Aur. Victor, Caes xx, 19. For the general frontier policy of Severus in Africa cf. Picard, G.-Ch., Castellum Dimmidi (Alger-Paris, 1946) 55–6Google Scholar.

⁹ Guey, J., Mél. d'Arch, et d'Hist. LVI (1939), 243.Google Scholar

¹⁰ Itineraria Romana (ed. Cuntz, ), 1 (1929), 11Google Scholar. For the identification of Vezereos, cf. Merlin, loc. cit.

¹¹ Cagnat, op. cit. 531.

¹² Notitia Dignitatum (ed. Seeck), 186–7 (Occ. XXXI). The four sectors of the Tripolitanian Limes, which occur both in the Notitia and the Itinerary are as follows: Talatensis (= Talalati); Tentheitani (= Thenteos); Bizeretani (= Vezereos); and Tillibarensis (= Tillibari).

¹³ CIL VIII, 22763 (Tibubuci); IRT 565 (Lepcis).

¹⁴ Diehl, L'Afrique byzantine (1896), 228-230, argued effectively from the evidence of Corippus and from other literary indications against a Byzantine occupation of the interior of Tripolitania. No structures of official character and post-Justinian date have yet been identified in the interior.

¹⁵ Cagnat, op. cit. 553-7, where the plan of the fortress of Bu-Ngem is inaccurate in detail; and Bartoccini, R., Afr. Ital. 11 (1928-1929), 50–8Google Scholar.

¹⁶ Cagnat, op. cit. 534, 546. Extensive stretches of ‘fossata’ have also recently been discovered in Algeria, cf. Leschi, L., Rev. africaine, nos. 412–3 (1947), 201–212Google Scholar; but their purpose and relationship to the limes are still far from clear. See, however, Baradez, , Vue-aerienne de l'organisation romaine dans le Sud-Algerien. Fossatum Africae (Paris, 1949)Google Scholar, passim.

¹⁷ On the accompanying map (fig. 15) are indicated the differing types of ‘desert’ which lay outside the Limes Tripolitanus. It will be seen that the Roman outposts avoided the completely sterile areas of the Grand Erg and the Hammada el-Hamra.

¹⁸ Roman Roads 11–13 (inscriptions 12–13).

¹⁹ ibid. 14-20 (Central Road), and 21-3 (Upper Sofeggin road).

²⁰ This church and others recently discovered and surveyed in Tripolitania will be published in a separate paper.

²¹ The altar was first seen in 1925 by Tenente M. Orano. Prof. R. Bartoccini, formerly Superintendent of Antiquities in Tripolitania, courteously communicated the report to the British School at Rome, and the site was investigated by the writers in 1948. The survey (fig. 16) was carried out by members of the BSR Sabratha expedition, under the direction of Mr. A. Wells.

²² On Severan inscriptions from Bu-Ngem (IRT 913-15) Geta is similarly described as Augustus while still in fact only Caesar.

²³ We owe this suggestion to Mr. Eric Birley, who refers us to Gsell, Inscr. latines de l'Algérie 3675. There are difficulties in accepting an alternative reading H(e)m(esenorum).

²⁴ J. Carcopino, Syria, 1925, 30-57 and 118-149; 1933, 20-55.

²⁵ Cylisterium, not otherwise recorded in epigraphy, is presumably derived from the Greek κυλιστήριον, a place in which one rolls. Its exact significance in the context of the Ain Wif bath-house, is obscure.

²⁶ The road-station at Medina Doga lies precisely at the 42nd Roman mile from Lepcis : it is slightly larger than Ain Wif, but of the same general character. Surveyed in 1949.

²⁷ Gasr Anessa (or. Hanesh), equated by De Mathuiseiulx with Thenadassa, lies near Uestata and appears to be of early Islamic date : the distances do not accord with those of the Itinerary.

²⁸ IRT 887, reported in 1925, but the full text is still to be recorded. A dedication to the Emperor at this remote site is almost certain to be military in character. Attempts to retrace this inscription in 1949 were unsuccessful.

²⁹ Nouv. Archives XIII (1906), 73–102. For the Roman milestones (Caracalla and Maximinus), first observed by Coró, cf. Roman Roads 21–3.

³⁰ We are indebted to A/Supt. J. Colman, of the Tripolitania Police, for drawing our attention to the inscriptions at Gasr Duib. The site is mentioned briefly by Coró in Atti del II Congresso di Studi Coloniali II (1935), 69–75.

³¹ The name ‘gasr’ (plural ‘gsur’) is applied locally in Tripolitania to any castle-like building, whether it be a Roman mausoleum or a Turkish fort. As the great majority of the ‘gsur’ in the southern area are fortified dwellings of the Roman period, uniform in plan and constructional technique, we use the word in this paper to denote these buildings.

³² The obliteration of the name following centenario appears to be intentional, and may indicate that the outpost was named after a member of the imperial house.

³³ De Mathuisieulx, Nouv. Archives XII (1904), 13–14. Although the inscription appears to confirm the identification of Zintan with Thenteos, it must be noted that the tribal name Zintani occurs as far south as Mizda.

³⁴ Carcopino, Syria 1925, 31.

³⁵ PIR ² C 1265. Cf. CIL VII, 2611.

³⁶ L. Leschi, CRAI 1941, 163-170, and Rev. africaine, nos. 394–5 (1943); Goodchild, R., ‘The centenaria of the Limes Tripolitanus,’ Reports and Monographs of the Antiquities Department in Tripolitania, no. 2 (1949Google Scholar, forthcoming).

³⁷ Small groups of this type occur at Gasr es-Suq el-Oti in the Wadi Bosra, Faschiet el-Habs in the Wadi Merdum, and Gasr Chanafes in the Wadi Scetaf. The extensive site at Ghirza in the Wadi Zemzem, best known for its elaborate mausolea, is a group of some thirty ‘gsur’ set close together without communal planning or defences : it is, in fact, a reductio ad absurdum of the whole ‘gasr’ system.

³⁸ None is precisely dated, but the palaeography affords a clear terminus post quem. The only inscription in the frontier zone which might be earlier than the third century is from a small temple at Tininai, between Mizda and Beni Ulid (IRT 888). The Ammonium at Tarhuna (above p. 8) shows, however, that the construction of Romanized sanctuaries for the deities of the Libyan tribesmen was undertaken long before extensive limitanei settlements were planned.

³⁹ Plate XIII, 2, illustrates a well-built and exceptionally well-preserved ‘gasr’ at the junction of the Wadis Dreder and Sofeggin, near Bir Scedeua, on the road from Beni Ulid to Mizda. Other ‘gsur’ of similar construction exist in the same area and would appear to belong to a relatively early phase in the settlement of limitanei.

⁴⁰ Angle-moulding of the ‘gasr’ at Gheria esc-Scergia illustrated in P. Romanelli, Il Limes romano in Africa (Quaderni dell'Impero : Il Limes romano x (1939)).

⁴¹ SHA, Sev. Alex, LVIII, 2Google Scholar. Cod. Theod. VII, 15, 1, contains a rescript of Honorius and Theodosius II addressed to the Vicarius of Africa ordering the ejection of tenants of limitaneus land who do not observe their military obligations.

⁴² IRT 889. The word centenare also occurs in an inscription of the same type from the Breviglieri area of Tarhuna (IRT 899).

⁴³ The Ghirza reliefs are illustrated, in part, by De Mathuisieulx, Nouv. Archives XII (1904), pi. vi-xiv, and by Bauer, , Afr. Ital. VI (1935)Google Scholar, figs. 2–14. A complete and definitive illustration of these important documents of limitaneus life, which have suffered, injury and dispersal during the last century, is much to be desired.

⁴⁴ The title tribunus occurs frequently in Latino-Libyan inscriptions from Africa (cf. IRT 886; Chabot, Receuil des Inscriptions libyques (Paris, 1940) 85, 145, 252). Further examples have been discovered during 1949 in a Romano-Libyan cemetery at Bir ed-Dreder.

⁴⁵ Ammianus, Hist, XXVIII, 6.

⁴⁶ Corippus, , Johannis 11, 80Google Scholar. mentions the former Limes station Tillibari; and the names Macurasen (IV, 955) and Cullen (IV, 791, 961), both of which seem to appear on the Ghirza mausolea (IRT 898).

⁴⁷ The problem of climatic changes in the interior of Tripolitania is too complex to be discussed here. The writers would merely place on record their opinion that there has been, in historic times, a small but significant reduction of rainfall.