Unless the length of a source needs to be specified, I cite only the first of its relevant pages. In addition to editorially authorized abbreviations, the title stated in n. 27 is shortened in subsequent notes to ‘Landmauer’ and ‘Courtauld’ refers to my own negatives, now the property of the Courtauld Institute of London University; they include all the photographs illustrated except that of Pergamon, for which I thank the donor. I also thank the owners of copyrights for permission to reproduce figures.

¹ W. Karnapp and A. M. Schneider, Die Stadtmauer von Iznik (1938).

² Milet ii 3: von Gerkan, A., Die Stadtmauern (1935) pl. 14.Google Scholar

³ Thompson, , JRS 49 (1959) 61 fig. 1Google Scholar; Frantz, , Hesperia 48 (1979) 202 fig. 3.CrossRefGoogle Scholar The wall used to be called Valerian's because of a misleading statement by Zosimus; most has now been demolished.

⁴ Traquair, , BSA 12 (1905–6) 417 pl. viii.Google Scholar

⁵ Alt. Ägina i 2: Wurster, W. W., Die spätrömische Akropolismauer (1975) 9Google Scholar Beilage 1–2 pls. 1–2.

⁶ ADelt 16 2 (1960) Chron. 129 plan 3 pl. 105a; Neue deutsche Ausgrabungen (1959) 276.

⁷ de Villard, Monneret, Archeologia 95 (1953) 96 pl. 34.Google Scholar

⁸ von Petrokovits, , JRS 61 (1971) 178–218Google Scholar analyses late Roman fortifications in Europe, relying largely on tower shapes. The fan shape (184 n. 15 fig. 29.7) is not found in Asia or Africa and seems almost confined to the 4th cent.; Procopius records as anomalous its use in a fort built for Justinian in Thrace, (Aed. iv 8)Google Scholar, and probably this was due to imitation of a Roman example, such as could be seen at Abritus.

⁹ Brünnow, R. E. and Domaszewski, A., Die Provincia Arabia ii (1905) 49 figs. 619–36.Google Scholar

¹⁰ The grooves were cut before the blocks were laid, but so carelessly that the width varies between 13·5 and 15 cm.

¹¹ Gerber, W., Forschungen in Salona i fig. 244.Google Scholar

¹² Harper, , AS 21 (1971) 10, 22 (1972) 27.Google Scholar

¹³ The word is a grammatically undesirable variant of the Hellenistic tetrapyrgia, which may have had a less restricted meaning.

¹⁴ According to Strabo, the river was navigable up to Seleucia (now nine miles inland) but this may have ceased to be the case by the 4th cent, owing to alluvium, which has formed an extensive plain around the mouth, accompanied by shoals on its west side. The steep east shore may already have been preferable as a landing-place when the new town was built at it.

¹⁵ Mansel, A. M., Die Ruinen von Side (1963) plan in pocketCrossRefGoogle Scholar; Beaufort, Francis, Karamania (1818) 147.Google Scholar

¹⁶ Lanckoronski, K., Niemann, , and Petersen, , Städte Pamphyliens und Pisidiens i (1890) 33 fig. 49.Google Scholar

¹⁷ Ibid. 9 fig. 4; the Turkish plan is discussed below with reference to the early 10th cent.

¹⁸ MAMA ii: Herzfeld, E. and Guyer, S., Meriamlik und Korykos (1930) fig. 177 etc.Google Scholar; MAMA iii: Keil, J. and Wilhelm, A., Denkmäler aus dem Rauhen Kilikien (1931) 102 fig. 133 pl. 42Google Scholar; Müller-Wiener, W., Castles of the Crusaders (1966) 79 pls. 111–14Google Scholar; Courtauld.

¹⁹ Theoretically, the fortress might have been built at any time between 391 (or earlier if the emperor had granted special leave for demolitions) and the rise to power of an Isaurian self-named Zeno, who succeeded his imperial father-in-law in 474 and reigned to 491. Anastasius I then launched the first of the campaigns that crushed the Isaurians.

²⁰ The floor of the ditch is all above sea-level but too shelving to have held water unless dammed.

²¹ Beaufort, op. cit. 242.

²² Migne, Patrologia Graeca 100 col. 940.

²³ Alexiad xi 10C.

²⁴ Ibid. 10D. Eustathius was ordered also to restore defences at Seleucia/Silifke that are likely to have originated before the wars with the Isaurians but been strengthened during them.

²⁵ Exceptional dimensions in the citadel at Old Cairo (Toy, S., History of Fortification (1955) 54Google Scholar) could have provided for the occasional depletion of the garrison to subdue trouble in the provinces; it was the headquarters of the army in Egypt.

²⁶ Cf. n. 4.

²⁷ Die Landmauer von Konstantinopel i (1938) by F. Krischen—generalities and drawings restoring original condition of Theodosian walls; ii (1943) by B. Mayer-Plath and A. M. Schneider—piecemeal survey.

²⁸ There is no evidence for recrudescence in Parthian times; the outer line at Hatra seems a mere proteichisma, 1·30 m thick, with only one visible salient.

²⁹ AJ 55 (1975) 36; 57 (1977) 255; Excavations at Carthage. The British Mission i 1 (forthcoming)

³⁰ Wilkes, J. J., Dalmatia (1969) 360, 418.Google Scholar

³¹ Vickers, , Makedhonika 12 (1972) 228CrossRefGoogle Scholar; BSA 68 (1973) 292.

³² Tsigaridas, , ADelt 28 (1973)Google Scholar Chron. 479 fig. 1 pls. 434–42.

³³ Diehl, Ch., Tourneau, , and Saladin, , Les Monuments chrétiens de Salonique 30–114Google Scholar summarize Tafrali's survey, Topographie de Thessalonique; his scale has been falsified in reproduction of the plans.

³⁴ ADelt 26 (1971) Chron. 374 fig. 4; BCH 98 (1974) 505 fig. 2.

³⁵ Ovčarov, , Arheologia 15 4 (1973) 11Google Scholar, (in French) 23.

³⁶ Ivanov, , Arheologia 15 4 24Google Scholar, (in French) 34, figs. 5·6.

³⁷ Ibid. 15 fig. 4.

³⁸ Ibid. 178 figs. 6–8.

³⁹ Ibid. 14 fig. 3; Milčev, and Damjanov, , BIABulg 23 (1972) 263, (in French) 276, figs. 1 8.Google Scholar

⁴⁰ Bace, , Monumentet 9 (1975) 5, (in French) 29Google Scholar; Rey, , Albania I (1925) 33.Google Scholar

⁴¹ Harrison, , Archaeologia Aeliana ⁴ 47 (1969) 33.Google Scholar

⁴² Aed. iv 9.

⁴³ Aed. ii 1 4; Bell, i 10 13–14, ii 13 17–18.

⁴⁴ An unwalled town, Sufetula/Sbeitla, contained five residential towers dispersed on the approaches to the fort. They vary in size up to some 20 m square and were two or three storeys high; each floor was divided into a large number of rooms, entered through open-fronted porches from a colonnaded court. Pringle, D., The Defence of Byzantine Africa (1981) 142 figs. 48a, bGoogle Scholar; Diehl, Ch., L'Afrique byzantine (1896) fig. 66.Google Scholar

⁴⁵ Data, plans, and illustrations were assembled in Diehl, op. cit. and Gsell, S., Monuments antiques de l' Algérie (1901)Google Scholar; Leschi, L., L'Algérie antique (1952)Google Scholar has the best photographs. Creswell, K. A. C., A Short Account of Early Muslim Architecture (1958) 178Google Scholar disputes Diehl's accounts of Thignica/Ain Tounga and Ksar-Bellezma. Data on Libya were collected by Good-child, , Corsi di Cultura sull' Arte Ravennate bizantina 13 (1966) 232–43Google Scholar; his fig. 2 restores a gateway at Leptis Magna (with lintel) between rectangular towers. Pringle op. cit., is an authoritative historical and archaeological summing-up, which I did not see till my text was unalterable, but it would have entailed no changes, only additions.

⁴⁶ It is drawn on the extreme right of Saladin's generally trustworthy restoration (Julien, Ch. A., Histoire de l' Afrique du Nord (1931) fig. 155Google Scholar; Diehl, Manuel d' art byz. fig. 90). This north end of Haidra, facing the road that ascends to the Algerian frontier, was remodelled by the Tunisians in the 16th cent, as a strong fortification in the style then prevalent. The Byzantine enceinte, which is otherwise well preserved, must have been intended only for the same purpose, of repelling invaders, because the surrounding country is practically uninhabitable, growing nothing but scrub. That no intensive siege was anticipated is obvious from the inadequate wall-thickness; on three sectors it is reduced also between a series of buttresses, which are linked to support the walk, some by arches, some by lintels. Diehl, , L'Afrique byz. 196 figs. 34 5 pls. v–viGoogle Scholar; von Petrokovits, , JRS 61 (1971) 202 n. 45Google Scholar; Pringle, op. cit. 180 fig. 18 pls. i–iv.

⁴⁷ The deserted town of Bagai regained inhabitants when a somewhat irregular area of over 300 m each side was surrounded by a Justinianic wall, which remained, when surveyed, distinct in outline but so ruined that no gateways were recognizable. One rectangular and three round towers stood at the corners; the twenty-one intermediate towers, all rectangular, varied in size for no perceptible reason. The site occupies a slight eminence amid a plain. On the summit, an apparent citadel of 74 × 63 m might possibly have been an initial defence for the builders. It lay behind both the wall and a shallow parallel outwork, suggestive of a proteichisma, which might have been entered at a little salient in front or else laterally through a tower-like excrescence from the wall. Probably there was a gateway in the wall midway in the interval (15 m) between a pair of towers (about 7 m square) that projected into the outwork, off the corners of a court (26 m square) that was contained within the citadel and may have formed an inner entrance. Diehl, op. cil. 152, 292, figs. 5, 31–2; Pringle, op. cit. 183–5 fir. 21.

⁴⁸ Diehl, op. cit. pl. iv, before restoration; Leschi, op. cit. 60; Leschi, op. cit. ii fig. 154 pl. xcvi.

⁴⁹ Grand houses near the Syrian Desert were already safeguarded by single machicoulis, either rectangular or half-cylindrical (Mattern, J., Villes mortes de haute Syrie 34 fig. 9 pl. xxxi 2Google Scholar; AASOR 25–8 (1925) 8 fig. 6).

⁵⁰ Welkov, , Germania 19 (1935) 149 figs. 2–3, 13–15.Google Scholar

⁵¹ Ivanov, , Arheologia 15 4 (1973) 24, (in French) 34.Google Scholar

⁵² Decline and Fall ch. xl.

⁵³ Welkov, op. cit. 149.

⁵⁴ Goodchild, , JRS 43 (1953) 75.Google Scholar

⁵⁵ Goodchild, , JRS 41 (1951) II fig. 3 pl. 1Google Scholar, also Corsi di Cultura (n. 45 above). It is questionable whether the fortified headland that projects from the town-site was really a citadel; it might antedate both Justinian's town-wall and the synagogue which he converted into a church.

⁵⁶ Dyggve, E., Recherches à Salona i (1928) 18 plan BGoogle Scholar; Gerber, W., Forschungen in Salona i fig. 1 plan of 1907.Google Scholar

⁵⁷ Ivanov, , BSocIABulg 7 (1919–1920) 88 fig. 66Google Scholar; Bobčev, , BIABulg 24 (1961) 115Google Scholar (an article profusely illustrated with small plans of fortifications in the Balkans).

⁵⁸ The tower at the broader entrance was 5·60 m wide and projected 4·50 m; the other, 5 m wide, projected 3 m. Both could as easily have been placed on the opposite side of the gateways instead of on the enemy's left, but the actual design may have been preferred because it would protect the unshielded right of sortie parties.

⁵⁹ Procopius, , Aed. v 96 11.Google Scholar

⁶⁰ Béquignon, , RA 4 (1934) 18Google Scholar; Mackay, , AJA 67 (1963) 241, 252.CrossRefGoogle Scholar

⁶¹ Zosimus i 29; Aed. iv 1 27.

⁶² Broneer, , Antiquity 32 (1958) 80 fig. 1.CrossRefGoogle Scholar

⁶³ Illustrated reports on American excavations in the fortress become more informative after 1967. Jenkins, and Megaw, , BSA 32 (1931–1932) 68 pl. 26.Google Scholar

⁶⁴ RE xvii.i (1936) s.v. cols. 513 14—plan; cf. sketch-plans by Leake, W. J., Travels in Northern Greece i (1835) 187Google Scholar and Chris, . Wordsworth, , Greece (1839) 230.Google Scholar

⁶⁵ The Roman wall on the north seems to have been rebuilt to match the new west side. The irregular Roman wall on the east may have been left almost unaltered; outworks would have been advisable, it being mainly without towers, but none has been noticed.

⁶⁶ EA 1961 44; Courtauld.

⁶⁷ Washington, , AJA ¹ 6 (1890) pl. 33Google Scholar; Courtauld A63/3590.

⁶⁸ Procopius, Aed. iv 9.

⁶⁹ Ibid, iv 2 5.

⁷⁰ Richmond, I. A., The City Wall of Imperial Rome (1930) 264Google Scholar; Procopius, , Bell, v 14–15.Google Scholar

⁷¹ Lawrence, , Opuscula Romana 4 44Google Scholar; Boethius, ibid. 36 figs. 9, 10; Courtauld.

⁷² Forma Italiae i/1.i G. Lugli Auxur-Terracina.

⁷³ Procopius, , Bell, ii 610–13, 8 819.Google Scholar

⁷⁴ Probably Justinian did not appreciably alter this entrance, to judge from the engravings of views from both back and front by Cassas, L. F., Voyage pittoresque de la Syrie… (1797–1799) xi 1.Google Scholar The inner portion resembled Theodosian entrances to Constantinople except for a greater thickness of the flanking towers and an inward prolongation of the passage by means of spurs. A low outwork seems to have been a proteichisma, doubled to contain two successive archways.

⁷⁵ Over-romanticized in the sketch by W. H. Bartlett engraved for Fisher's Views (Carne, J., Syria, The Holy Land, Asia Minor, etc. i (1836) 63Google Scholar).

⁷⁶ Cassas, op. cit. i 7, xiii 2.

⁷⁷ Carne, op. cit. iii (1838) 11.

⁷⁸ Ibid, i (1836) 24.

⁷⁹ I know of Chesney's account only from quotations by some Victorian author.

⁸⁰ Procopius, , Bell, ii 1317–18Google Scholar; Aed. ii 1 14–25; Crow, , Yayla 4 (1981) 12 figs. 2–12.Google Scholar

⁸¹ Aed. iii 2.

⁸² Ibid, ii 93–9; Karnapp, W., Die Stadtmauer von Resafa in Syrien (1976).Google Scholar

⁸³ Aed. ii 88–25; Karnapp, op. cit. 27 figs. 100–9; Lauffray, , Ann Arch Syrie i (1951) 41 pl. ivGoogle Scholar—plan questioned by Karnapp's n. 99; Poidebard, A., La Trace de Rome dans le désert de Syrie (1934) pls. 82–4.Google Scholar

⁸⁴ But T. Ivanov ascribes to Justinian the total blocking of three out of the four main entrances and of all the posterns yet found at Abritus, (Abritus i 239, 241, 248).Google Scholar

⁸⁵ Bullough, , BSR 24 (1956) 14 pl. v.Google Scholar The only distinct feature is a tower, less than 4 m square but 16 m high, set back from a tall concrete base above which it is faced with stone; the courses alternate between half and full height as was customary in north Africa.

⁸⁶ Hill, Geo. F., History of Cyprus i (1940) 284, 285, 326–9Google Scholar; Antiquities Dept. Salamis (1966) 5.

⁸⁷ Mansel, A. M., Die Ruinen von Side 40, figs. 12, 24–5Google Scholar, plan in pocket; Courtauld; Knoblauch, P., Die Hafenanlagen und die anschlieβenden Seemauern von Side (Ankara 1977)—not seen.Google Scholar

⁸⁸ The junction with the harbour wall is thickened into forming a miniature tower, as though it had flanked a lost gateway. The corresponding junction towards the open sea has fallen.

⁸⁹ Foss, , Zeitschrift für Papyrologie und Epigraphik 26 (1977) 172.Google Scholar

⁹⁰ Brook, , JHS 18 (1898) 187.Google Scholar

⁹¹ Borchhardt, J. F. C., Myra (1975) 45 fig. 2 Beil. 1 pls. 4, 5. 79.Google Scholar

⁹² Ibid. 87 fig. 21.

⁹³ Tsakos, , AE 1979 Parar. 11Google Scholar; AR 1981–2 50 fig. 111.

⁹⁴ Hood, , ILN Jan. 30 1954 159 figs. 1–2.Google Scholar

⁹⁵ Müller-Wiener, , IstMitt II (1961) 36 fig. 8.Google Scholar

⁹⁶ Id., IstMitt 17 (1967) 282 fig. 1.

⁹⁷ Foss, Cl., Ephesus after Antiquity (1979) 192.Google Scholar

⁹⁸ Müller-Wiener, , IstMitt II (1961) 86 fig. 21Google Scholar; Foss, op. cit. fig. 35.

⁹⁹ So called because of mistaking the subject of an earlier relief built into the arch.

¹⁰⁰ Müller-Wiener, , IstMitt II (1961) 97, 102 fig. 23 Beil. ii–iii.Google Scholar

¹⁰¹ Altertümer von Pergamon i 2 305 pl. iii Beib. 62.1–2.

¹⁰² Karnapp, W. and Schneider, A. M., Die Stadtmauer von Iznik (1938) 12, 16 fig. 5.Google Scholar

¹⁰³ Ramsay, W. M. and Bell, G. L., The Thousand and One Churches (1909) 152, 542, 545 fig. 117.Google Scholar

¹⁰⁴ Jerphanion, , Mélanges de l'Université S. Joseph 13 (1928) 144Google Scholar; Mamboury, E., Ankara Guide Touristique (1923) 73Google Scholar with general plan; Courtauld.

¹⁰⁵ Gregoire, , Byzantion 4 445.Google Scholar

¹⁰⁶ Whenever these passages opened high above ground they must have been reached by wooden steps or a fixed ladder. One of the few still intact and accessible is 85 cm wide, and 5·35 m long to the inner face of the thick vault.

¹⁰⁷ For barbicans of the Assyrian period see my Greek Aims in Fortification (1979) 23, 25 fig. 10. Every entrance to Hatra was approached through a comparable barbican, probably before A.D. 200 (Andrae, W., Die Ruinen von Hatra ii (1912) figs. 25–7, 30–2Google Scholar). Descriptions of the Round City of Baghdad, which was completed about 766, indicate that some such work existed outside one or more of its entrances (Creswell, K. A. C., Early Muslim Architecture (1958) 162Google Scholar). Many undated instances in Soviet Turkestan are relevant because of the initial unity of the Islamic conquests, and Mogul barbicans of the same type followed the tradition of that region.

¹⁰⁸ Jerphanion's restoration (his pl. 92) misrepresents the height relation of barbican and main wall.

¹⁰⁹ No valid argument can be based on the fact that the stone facing of the entire wall came from buildings destroyed in 630, because these must have been 500 m distant from the new town and there was no reason to clear away their ruins except when need arose for the material.

¹¹⁰ Landmauer ii 118 pl. 40.

¹¹¹ Geoffroy de Villehardouin states that the Byzantines spent most of Lent adding these wooden storeys to the towers, before the Crusaders attacked from shipboard.

¹¹² Ramsay and Bell, op. cit. 491 fig. 366.

¹¹³ Landmauer ii fig. 32 pls. 40–2.

¹¹⁴ See n. 17 above.

¹¹⁵ I am indebted to Dr. G. L. Lewis and Professor Seton Lloyd for kindly reading the third line of the Turkish key as ‘İç Kale’, meaning an internal fortress (though its defences on the north and east, if not also on the west, were formed by the city wall). My photographs of the gatehouse from northwest and south-west (Courtauld A51/408–9) are foreshortened.

¹¹⁶ Hesperia 37 (1968) 204.

¹¹⁷ Ducoux, and Lemerle, , BCH 42 (1938) 4 pls. ivA, vi–vii.Google Scholar

¹¹⁸ The masonry was tied by many wooden beams laid flat in chases packed with cement, through from the inward to the outward face. That method of reinforcement had been common in Hellenistic walls, even of large blocks; prevalence in late Byzantine walls may be deduced from use in the medieval Ottoman castle at Kalccik.

¹¹⁹ A later doorway formed the sole entrance to a chapel that was added, utilizing the shorter of the partitions for a side-wall.

¹²⁰ Food for several months could have been stored in the tower. The rainfall on both tower and barracks must have been conserved in the cistern between them.

¹²¹ Antiquities Dept. Kyrenia Castle (1961) 81.Google Scholar

¹²² The defences of Ani were repeatedly altered between the terminal dates of 783 and 1312, and investigation has been restricted owing to the proximity of the Soviet frontier.

¹²³ Tchalenko, G., Villages antiques de la Syrie du Mord (1953) i 242, ii pls. 79, 82, 209Google Scholar; Courtauld.

¹²⁴ Andrews, Kevin, Castles of the Morea (1953) 206 figs. 218–19 pl. 36—Venetian plan.Google Scholar

¹²⁵ The roughly contemporary Venetian plan calls the hollow structure a mill, although no one would have transported grain so far across uneven rocky ground when there was a wind-swept cliff-top immediately above the town. The same identification as windmills has been accepted by Welsh countrymen to explain a chain of much smaller ruins that were actually the bases of beacons to give warning of pirate raids in years around 1600 (Lloyd, , Archaeologia Cambrensis 113 (1964) 150Google Scholar).

¹²⁶ Velkov, , GNMP Annuaire du Musée National Plovdiv ii (1950) 176, (in French) 183, fig. 13.Google Scholar

¹²⁷ Boase, T. S. R., Castles and Churches of the Crusading Kingdom (1967) 49, 51Google Scholar; Smail, R. G., Crusading Warfare (1956) 236 fig. 6Google Scholar; Müller-Wiener, W., Castles of the Crusaders (1966) 10, 44, 96–7 pls. 12–13, 17–19.Google Scholar

¹²⁸ Deroko, A., Srednjevekovni Gradovi u Srbiji, Crnoj Gori i Makedoniji (1950) 194 figs. 36, 40Google Scholar; Courtauld.

¹²⁹ The maximum dimensions of the towers are approximately: 6 m projection, 5 m width, height (slightly reduced) 13 m, internal length 5 m and width 3 m.

¹³⁰ See n. 95 above.

¹³¹ Zdravković, and Jovanović, , Actes du XII Congrès internat. d'études byz. Ohrid 1961 ii (1964) 423 fig. 2Google Scholar; Jovanović, , Starinar 13–14 (1962–1963) 137, (in French) 150, fig. 6.Google Scholar

¹³² Antiquities Dept. St. Hilarion Castle (1950).

¹³³ Another apsidal tower with an emphatic taper is apparently the only Byzantine relic in the otherwise Seljuk fortress at Anamur, within sight on the opposite coast of Asia Minor; it stands next to the gateway (Courtauld A51/354). It must have been built as a defence against Moslem aggression, perhaps rather earlier than 1092, but looks too like the towers of St. Hilarion for the resemblance to be dismissed as coincidental.

¹³⁴ Landmauer ii figs. 28–9.

¹³⁵ AR 1978–9 67; Altertümer von Pergamon i (1885) pl. iii; i. 19; i.2 307 Beib. 63–4; AM 29 (1904) pls. viii, x–xi.

¹³⁶ Müller-Wiener, , 1stMitt 17 (1967) 285 fig. 3.Google Scholar

¹³⁷ The trebuchet was developed by a gradual process of improvement, probably in France. The Byzantines merely transliterated the word into Greek. One of the Arabic names means ‘Frankish mangonel’ while the other calls it the maghrabi, i.e. ‘Western’, mangonel. The dissemination of the weapon should have been rapid because of the simplicity of both its construction and its operation.

¹³⁸ Karnapp, W. and Schneider, A. M., Die Stadtmauer von Iznik figs. 7–10 pls. 3–7, 13, plan at end.Google Scholar

¹³⁹ Rice, Talbot, JHS 52 (1932) 47CrossRefGoogle Scholar; Janssens, E., Trébizondeen Colchide (1967) 238 figs. 14, 31, 56.Google Scholar

¹⁴⁰ Hoepfner, W., Herakleia Pontike—Eregli (1966) 42 5 figs. 16–17 pl. 4bGoogle Scholar; the supposedly Byzantine citadel is not datable.

¹⁴¹ Orlandos, A., Archeion tōn Byzantinōn Mnimeiōn tis Ellados 2 (1936) 151.Google Scholar

¹⁴² Dakaris, , Dodoni 6 (1977) 201.Google Scholar

¹⁴³ My description was written after a visit in 1979.

¹⁴⁴ Orlandos, op. cit. 9 (1961) 54 figs. 2, 4; Courtauld.

¹⁴⁵ Landmauer ii 17.

¹⁴⁶ Woodward, , BSA 23 (1919–1920) 98.Google Scholar

¹⁴⁷ Dakaris, , ADelt 19 (1964) B3 314 pl. 353.Google Scholar

¹⁴⁸ Guide de Thasos (1967) 16.

¹⁴⁹ Italian as well as Byzantine influence on the Despots of Serbia (a Turkish vassal-state) resulted in two imposing fortresses: the wall around Manasija monastery at Resava, in 1407–10 (Deroko, op. cit. figs. 130–1 pl. 33; Courtauld), and the city wall and castle at Smederevo, in 1428–30 (Deroko figs. 139 41 pls. 12, 14, 30; Starinar 2 (1951) 59, and 7–8 (1956–7) 181; Courtauld).

¹⁵⁰ It is arguable that the style of facing adopted by the Palaeologi, which consisted of small regular blocks (Landmauer ii pl. 21), may effectively have localized damage caused by impact.