My thanks are due to Prof. Ch. Bouras, Dr. R. S. Cormack, Prof. P.-A. Février, Mr. M. F. Hendy, Mr. J. Humphrey, Dr. H. P. Laubscher Dr. Ph. Petsas and Dr. W. H. Plommer for assistance in various ways in the preparation of this article. Dr. Plommer and Prof. A. D. E. Cameron were kind enough to read it through in draft form and it has benefited as a result of their comments.

¹ Gibbon, E., The Decline and Fall of the Roman Empire, ed. Bury, J. B. (1900), 172–3.Google Scholar The sources are now discussed in detail by I. Hahn, Ή ἐξέργεση τοῦ 390 στὴν Θεσσαλονίκη καὶ τὸ ἱστορικό της πλαίσιο ByzNGJ XIX (1966), 350–372. See also R. Browning's attractive suggestion (CR n.s. XXI, 1971, 138) that Amphilochius Iconiensis, Iambi ad Seleucum 172 ff. might well refer to Theodosius' punitive measures in 390.

² ‘Observations sur l'histoire et sur les monuments de la ville de Thessalonique’, Histoire de l'Académie Royale des Inscriptions et Belles-Lettres XXXVIII (1777), 134.Google Scholar

³ Tableau de Commerce de la Grèce (Paris, 1800), 37, followed by Clarke, E. D., Travels in various countries of Europe, Asia and Africa, 4th ed., VII (London, 1818), 454.Google Scholar

⁴ Voyage dans la Macédoine (Paris, 1831), 34.

⁵ Moraïtopoulos, M., Τοπογραφία Θεσσαλονοίκης (Athens, 1883), 16–17Google Scholar (cited by Tafrali, O., Topographie de Thessalonique, Paris, 1913, 129Google Scholar).

⁶ ΄Καὶ ἕως μὲν καὶ τοῦ πολιτικοῦ καθ΄ ἡμᾶς ἱπποδρόμου τοιαύτη διάθεσις ἡμᾶς εὄθυνεν΄ De Thessalonica a Latinis capta (ed. Bonn, 463); cf. Theocharides, G., Hellenika XIII (Thessaloniki, 1954), 29Google Scholar, n. 1.

⁷ Ή Μακεδονία ἐν λίθοις φθεγγομἐνοις σωζομἐνοις (Athens, 1896), 412.

⁸ Byz. Zeit VI (1897), 538.

⁹ ‘Plan of Salonica’, 1: 10,000.

¹⁰ I have drawn out the various features mentioned in the text against a plan made before 1917 (kindly supplied by Professor Ch. Bouras) since it is much closer to the mediaeval, and hence the Roman, city plan.

¹¹ Dyggve, E., ‘Compte-rendu succinct des fouilles de Thessalonique 1939’, RACrist XVII (1940; below, = Dyggve, 1940), 150Google Scholar; id., ‘Kurzer vorläufiger Bericht über die Ausgrabungen im Palastviertel von Thessaloniki’, Laureae Aquincenses Vol. Kuzsinsky dicatae II, Diss. Pannon., 2nd ser., XI (Budapest, 1941; below, = Dyggve, 1941), 66.

¹² The interior of one of the vaults is illustrated in Dyggve, , ‘La région palatiale de Thessalonique’, Acta Congressus Madvigiani (1954) I (Copenhagen, 1958), 356. fig. 3.Google Scholar

¹³ Dyggve, 1941, 66 and pl. 5, fig. 16.

¹⁴ Dyggve, 1940, 150; Dyggve, 1941, 63; BCH LXIII (1939), 313, fig. 25; AA 1940, 255, fig. 66; Gnomon XVII (1941), 230.

¹⁵ cf. Petsas, Ph., Makedonika VII (1967), 297.Google Scholar

¹⁶ Papadopoulou, Ph., ADelt XIX (1964), B′3, fig. 4 opp. p. 330.Google Scholar

¹⁷ Drosoyanni, Ph., ADelt XVIII (1963)Google Scholar, B′2, 244–6, figs 4–6, pl. 276a. Petsas, , ADelt XXIV (1969)Google Scholar, B′2, 295–6, fig. 2, pl. 305c–d.

¹⁸ Petsas, , ADelt XXIII (1968)Google Scholar, B′2, 332–4, fig. 6, pl. 279; id. Makedonika VIII (1969), 151, pls. 50–3.

¹¹ Clearly visible, ibid., pl. 51a.

²⁰ Diocletian's decision to retire to Split can only have been taken at the end of the Persian wars (297) at the soonest and the palace, or more properly villa (cf. Duval, N., Urbs IV (Split, 1961–1962), 70)Google Scholar, must have been incomplete when the emperor first occupied it in May, 305 (cf. Kähler, H., Mullus, Festschrift für Th. Klauser = JbAC Ergh. I (1964), 176–7).Google Scholar The Tetrarchic mint at Thessaloniki seems to have opened in c. 298/9 (RIC VI, 501), which implies that work on the palace had begun then. Diocletian is thought to have attended the dedication of Galerius' triumphal arch on his way west from Nicomedia to Rome in c. 303 (Seston, W., Dioclétien et la Tétrarchie 1 (1946), 392Google Scholar), and I hope to show elsewhere that work on the palace at Thessaloniki was still in progress in 311.

²¹ Hébrard, E. and Zeiller, J., Spalato, le palais de Dioclétien (Paris, 1912)Google Scholar, figs on pp. 66, 76, 78, 100. The Porta Aurea socles are discussed by Kähler, , Das Fünfsäulendenkmal für die Tetrarchen auf dem Forum Romanum (1964), 6–7Google Scholar, fig. I, and the substructures of the palace by Marasović, T., ‘Gli appartamenti dell'Imperatore Diocleziano nel suo palazzo a Split’, Acta of the Norwegian Institute in Rome IV (1969), 35–37, pls 10, 12, 13a, 15b.Google Scholar

²² Luxor, , Monneret de Villard, U., Les couvents près de Sohag (1926), fig. 168.Google Scholar Philae, id., La Nubia romana (1941), 8–9, figs. 4–7.

²³ ‘The starting gate for chariots at Olympia’, GR, n.s. XV (1968), 113–26.

²⁴ Gerasa, Müller, E. B., ‘The Hippodrome’, in Kraeling, C. H. (ed.) Gerasa, City of the Decapolis (1938), 85–99Google Scholar, plan VI. Circus of Maxentius, Nibby, A., Del circo detto di Caracalla (Rome, 1825)Google Scholar, plan (the most up-to-date one !—reproduced in Crema, L., ‘L'architettura romana’, Enciclopedia classica, Sez. III, XII, I (Turin, 1959), 597, fig. 788)Google Scholar; aerial view in Boethius, A. and Ward-Perkins, J. B., Etruscan and Roman Architecture (1970), pl. 258Google Scholar; useful photographs, but inaccurate plan, in Popoff-Béboutoff, G., ‘Der Circus des Maxentius; eine vernachlässigte Ruine Roms’, Antike Welt I, 1 (1970), 28–31.Google ScholarMérida, Mélida, J. R., ‘El circo romano de Mérida, memoria de las excavaciones practicadas de 1920 a 1925’, Junta superior de excavaciones y antigüedades, Num. gral., 72, Num. 2 de 1924–5 (Madrid, 1925)Google Scholar; id., ‘Excavaciones de Mérida, el circo, etc., memoria de los trabajos practicados en 1926 y 1927’, ibid., Num. gral., 98, Num. 6 de 1927 (Madrid, 1929); useful aerial photograph in Wiseman, F. J., Roman Spain (London, 1956), pl. II, opp. p. 136Google Scholar, Toledo, Comisión de Monumentos historicos y artisticos de Toledo, ‘Excavaciones en Toledo: Memoria de los trabajos efectuados en el circo romano’, Junta superior de excavaciones, Num. gral., 96, Num. 4 de 1927 (Madrid, 1928).Google Scholar Antioch, W. Campbell, ‘The circus’, Antioch-on-the-Orontes 1, 34–41, plan 35, fig. 4.

²⁵ Circus Maximus, P-W III, 2, 2573–4, s.v. ‘Circus’; Nash, E., Pictorial Dictionary of Ancient Rome I (1961), 236–40.Google Scholar

²⁶ Byzantium, Mamboury, E. and Weigand, E., Die Kaiserpaläste von Konstantinopel (1936), pl. 102.Google Scholar

²⁷ Cyrene, however, is slightly exceptional, for there the carceres, though askew, are not curved but straight: Stucchi, S., ‘Cirene, 1957–1966’, Quaderni dell' Istituto Italiano di Cultura di Tripoli III (Tripoli, 1967), 44Google Scholar, plans pp. 15 and 36–7; idem, ‘First outline for a history of Cyrenaican architecture during the Roman period’, Libya in History, Historical Conference 1968 n.d. (c. 1971), 228.

²⁸ Mamboury, , ‘Les fouilles byzantines à Istanbul’, Byzantion XI (1936), 271–2.Google Scholar Cf. Mamboury, , Byzantion XXI (1951), 455–9Google Scholar; Naumann, R. and Belting, H., ‘Die Euphemia-Kirche am Hippodrom zu Istanbul und ihre Fresken’, Istanbuler Forschungen XXV (Berlin, 1966), plan p. 17, fig. 1Google Scholar; and Guilland, R., ‘Les hippodromes de Byzance; l'hippodrome de Sévère et l'hippodrome de Constantin’, Byzantinoslavica XXXI (1970), 182–8.Google Scholar

²⁹ Another Spanish circus that looks as if it might have been the same size is the one formerly visible at Calahorra (Calagurris Julia Nasica) which, according to Cean-Bermúdez, J. A.. Sumario de las antigüedades romanas que hay en España (Madrid, 1832), 138Google Scholar, 1, was ‘de largo 489 pasos comunes, 116 de ancho y las paredas 22 pies de grueso, y se señalan las gradas en que se sentaban los espectadores’. The ‘muchos acueductos de plomo’ found ‘en su recinto’ were probably from the euripus (on which see below, p. 30 f.). The circus at Miróbriga dos Celticos in Portugal was similar in many ways: the spina was c. 230 m long, and the track 76 m wide, but only 356 m long (Cruz e Silva, J., Arquivo de Beja 111, 1946, 341, fig. 6Google Scholar; de Almeida, F., ‘Nota sobre os restos do circo romano de Miróbriga dos Celticos (Santiago do Cacém)’, Revista de Guimarães LXXIII, 1963, 147–54Google Scholar).

Two African circuses have similar dimensions: Cherchel, (400 m X 90 m; Gsell, S., Les monuments antiques d'Algérie 1, 1901, 204)Google Scholar, and Sétif (400 m X 80 m; Romanelli, P., ‘Topografia e archeologia dell'Africa romana’, Enciclopedia classica, Sez. III, vol. X, 7 (1970), 168).Google Scholar

³⁰ e.g. in the Circus of Maxentius and at Toledo. R. Guilland has assembled the literary evidence for such a gateway at Byzantium, , ‘Études sur l'hippodrome de Byzance’, Byzantinoslavica XVIII (1957), 39–76Google Scholar. Relevant mosaics occur at e.g. Gerona, , Taracena, B., Ars Hispaniae (Madrid, 1947), 71, fig. 49Google Scholar; Balil, A., ‘Mosaicos circenses de Barcelona y Gerona’, Boletin de la real Academia de la Historia CLI (1962), 257–351, pls 24–37Google Scholar; Reinach, S., Répertoire de peintures grecques et romaines (1922; cited below as RPGR), 291, 1Google Scholar. Lyons, , Grenier, A., Manuel d'archéologie gallo-romaine III (1958), 981, fig. 322Google Scholar; RPGR 291, 2. Carthage, , Constans, L. A., ‘Mosaïque de Carthage’, RA, 5e sér. III (1916), 247–59, 248, fig. 1Google Scholar; RPGR 293, 2.

³¹ And also at Cherchel (Gsell, loc. cit). Prof. P.-A. Février kindly informs me that at Sétif he found a monumental gate at the curved end of the circus, through which a series of steps led up to the roadway outside.

³² Nash, op. cit. (n. 25), 1, 240, fig. 278. This arch appears on a frieze of the Tomb of the Haterii (Helbig,⁴ 1, 1076) now in the Vatican Museum.

³³ The less emphatic bend made by this street immediately to the west of the Arch is well attested from excavation (Hébrard, E., ‘Les travaux du service archéologique de L'Armée d'Orient à l'Arc de Triomphe “de Galère” et à l'église Saint-Georges de Salonique’, BCH XLIV, 1920, 9, fig. 3).Google Scholar

³⁴ ADelt XVIII (1963), B′2, 243–6, figs 3 and 6, pl. 275. More stretches had been found further to the south in 1950 (BCH LXXV, 1951, 116 and Makedonika 11, 1941–52, 597), 1951 (BCH LXXVI, 1952, 227), and 1952 (BCH LXXVII, 1953, 224), but were incorrectly attributed to Galerius.

³⁵ Vickers, M., ‘The date of the walls of Thessalonica’, Istanbul Arkeoloji Müzeleri Yilligi XV–XVI (1969), 313–8Google Scholar; id., ‘The date of the mosaics of the Rotunda at Thessaloniki’, PBSR XXXV (1970), 183–7.

³⁶ Miracula S. Demetrii I, ch. 10.

³⁷ See A. Cameron, Circus Factions (forthcoming).

³⁸ See Vickers, M., ‘The Stadium at Thessaloniki’, Byzantion XLI (1971), 339–48.Google Scholar It is probably relevant that a new palace was built fairly near the stadium in the mid-fifth century, the Tetrarchic palace having in all likelihood gone out of use: see my ‘A note on the Byzantine palace at Thessaloniki’, BSA LXVI (1971), 369–71.

³⁹ Floriani Squarciapino, M., Leptis Magna (1966), 132.Google Scholar

⁴⁰ loc. cit. (n. 28).

⁴¹ C. Edson, Inscriptiones Graecae X, II, 1, no. 41. Euripus clearly does not mean ‘canal”, as suggested there. Cf. C. H. V. Sutherland, RIC VI, 89, n. 1.

⁴² Dion. Hal. III, 68; Suet., Caes. 39. Cf. P-W VI, 1, 1284–5, s.v. ‘Euripos’.

⁴³ Adv. Hermogen. XXXI.

⁴⁴ Mango, C. A., ‘L'Euripe de l'Hippodrome de Constantinople’, RÉB VII (1949), 180–93Google Scholar; Guilland, R., ‘Études de Topographie de Constantinople byzantine’, Berliner byzantinische Arbeiten, XXXVII (Berlin, 1969), 447Google Scholar (= Jahrbuch der Österreichischen byzantinischen Gesellschaft VI (1957), 29–32).

⁴⁵ e.g. Malalas (ed. Bonn, 174); Chron. Pasch. (ed. Bonn, 208); Cedrenus (ed. Bonn I, 258).

⁴⁶ Thouvenot, R., ‘Mosaïque de Volubilis représentant une course de chars’, CRAI, 1954, 344–8CrossRefGoogle Scholar; id., Maisons de Volubilis (1958), 66–9, pl. 16.

⁴⁷ B. Taracena, op. cit., 71, fig. 48; Balil, op. cit., pls. 38–48; Taradell, M., Arte romano en España (Barcelona, 1969), 12–13Google Scholar, fig. 2; 16, fig. 5; 197–8, fig. 169; RPGR 291, 3.

⁴⁸ Carthage, see n. 30 above. Piazza Armerina, Gentili, G. V., ‘Le gare del circo nel mosaico di Piazza Armerina’, Bolletino d'Arte XLII (1957), 9, fig. 2.Google Scholar

⁴⁹ Diocletian attended the dedication of the hippodrome at Nicomedia in 304, which suggests that the palace and the hippodrome there had been put up together (Lact. De mort. pers. XVII, 2–9).

⁵⁰ CIL VI, 1145. I am grateful to Mr. Michael Hendy for this reference and for the conclusions to be drawn from it. He discusses the question of the date of Catafronius in ‘Aspects of coin production and fiscal administration in the late Roman and early Byzantine periods’, NC, 7th ser, XII (1972), (forthcoming).

⁵¹ AÉ 1927, 165. Mr. J. Humphrey has kindly shown me some photographs of the elaborate water-works, of uncertain date, found in the spina at Antioch by the Princeton expedition of 1935.

⁵² Nash, op. cit. (n. 25), 236, fig. 271.

⁵³ Piganiol, A., Second Int. Congr. Byz. Studies (Belgrade, 1929), 56 ff.Google Scholar; Vogt, A., ‘L'hippodrome de Constantinople’, Byzantion X (1935), 482–8Google Scholar; Guilland, , ‘Études sur l'hippodrome de Byzance; le palais du Kathisma’, Byzantinoslavica XVIII (1957), 39–76.Google Scholar

⁵⁴ See n. 24 above and Lugli, G., Bull. Com. LII (1925), 126–34, Pls 1–2.Google Scholar

⁵⁵ Downey, G., A History of Antioch in Syria (1961), 318–23, 648 and fig. 11.CrossRefGoogle Scholar

⁵⁶ The circus, De Capitani d'Arzago, A., Il circo romano (Milan, 1939)Google Scholar; Calderini, A. in Storia di Milano I (Milan, 1953), 530–5.Google Scholar The palace, Cagiano de Azevedo, M., ‘Admiranda palatia’, Boll, del Centro Studi di Architettura XIV (1959), 5.Google Scholar

⁵⁷ ‘Der Circus des römischen Trier’, Trierer Zeitschrift XVIII (1949), 149–69. von Massow takes the dimensions of the Circus of Maxentius as the standard for Roman circuses without apparently realizing that it was exceptionally large. At Trier, Egbertstraβe (p. 162, fig. 6) runs along the east (right) side of the hippodrome and has a striking westward bend which recalls the oblique sector in the right hand side of the hippodrome at Byzantium (Duyuran, R., Istanbul Arkeoloji Müzeleri Yilligi V (1952), 23–38Google Scholar, plan, fig. 1, and ibid., VI (1954). Plan, fig. 2), and lies at more or less the same distance from the curved end. This suggests that the circus at Trier might have had the same overall length.

⁵⁸ e.g. Brusin, G., Aquileia e Grado, guida breve, 7th edn. (Padua, 1967), plan.Google Scholar

⁵⁹ Mirabella Roberti, M., ‘L'edificio romano nel “patriarcato”’, Aquileia nostra XXXVI (1965), 45–78Google Scholar, esp. 62. Dr. Robin Cormack has drawn my attention to Frazer, A., ‘The iconography of the emperor Maxentius' buildings in Via Appia’, Art Bulletin XLVIII (1966), 385–92CrossRefGoogle Scholar, who discusses the principles underlying the planning of Tetrarchic palaces in some detail (see esp. figs. 4–9; nos 8 and 9 should be transposed) without, however, observing that such palaces always lie to the left of the carceres of their neighbouring hippodrome.

Postscript. Since the above was written, Dr. Hans Peter Laubscher has kindly sent me a copy of von Schoenbeck's notebooks which deal with the survey of 1935. S. was able to follow the seating foundations (presumably A-B on Fig. 1) for a distance of 300 m. The vaulted substructure was 4·50 m wide and consisted of two concentric brick vaults resting on a deep stone-built foundation. There were windows in the side of the vault towards the arena placed at regular intervals, while on the other side there were doorways more frequently spaced (12·50 m apart). Although he found no direct evidence for the semicircular end, S. argued from the irregular streets towards Odos Egnatia that it must have lain in that area. The brickwork of the extant vault he compared with that of the Arch of Galerius.