by David F. Graf Between Bostra and Philadelphia Between Kerak and Aqaba

The Madaba Mosaic Map is a jewel without a crown. For all of Byzantium's reputation for other matters, its contribution to the science of cartography is dismal at best (cf. Dilke 1987: 258). The brilliant exception is the Madaba cartographic representation of the Holy Land. Unfortunately, everything north and south of the region of Charachmoba (modern Kerak) is missing. The purpose of this study is a hypothetical reconstruction of those missing areas of Transjordan. The creation of the Madaba mosaic map is dependent on earlier sources of information, particularly the Onomasticon of Eusebius. It is generally assumed that a Roman road map provided the initial structure for the Onomasticon (Donner 1992: 2126). Afterwards, a pictorial representation of his work provided vignettes of the prominent towns and churches. In the sixth century, additions were made, perhaps reflecting itineraries used by pilgrims before the execution of the Madaba Mosaic Map sometime late in the reign of Justinian, between probably A.D. 542 and 565 (AviYonah 1954: 26-32; Donner 1992: 26; cf. Piccirillo 1993: 2634, 8195). Since most of the region of Transjordan is lost, any reconstruction of the sites in this region is dependent on the available sources for the Roman road system, namely the Antonine Itinerary, the Peutinger Table, and possibly the Ravenna Cosmography. But they are of limited value for understanding the Madaba Map's depiction of Transjordan. The Itinerarium Provinciarum Antonini Augusti preserves listings of 225 land and sea routes ostensibly associated with the travels of emperors of the Antonine dynasty (Miller 1916; Cuntz 1929; Berchem 1937), with later additions extending to the Tetrarchy (Reed 1978). Neither Transjordan nor Egypt are represented, although the latter is the terminus of several routes. These gaps and the nature of the Antonine Itinerary as an official military document excludes it from consideration as a primary source for the Madaba Map. The Tabula Peutingeriana depicts only one northsouth artery for Transjordan, the Via Nova Traiana between Bostra and Aila on the Gulf of 'Aqaba. Between Bostra and Philadelphia (Amman), the Peutinger Table lists Thantia, Hatita, and Gadda, identified with Thughrat elJubb, Khirbet elSamra, and ElHadid (Parker 1986: 32-34). But between Philadelphia and Petra are only Rababatora, Thorma, and Negla, identified with Rabba, Thawana, and the environs of Shobak (Bowersock 1983: 174175). Philadelphia alone is mentioned of the Decapolis cities in northern Transjordan and nothing is listed between Philadelphia and Rabba including Madaba. Clearly, other sources than the Peutinger Table must be the basis for this region on the Madaba Map. The Ravenna Cosmography produced by an anonymous Latin author in the seventh or eighth century lists over 5300 cities and 300 rivers and seas with their respective islands (Funaioli 1920: 306). Some of the sources cited by the author are legendary and fictitious, disguising the fact that the author drew his information from a Roman road map. It appears that the author derived some of his information from the Peutinger Table or a common source (SchillingerHäfele 1963: 238), but the listings are more extensive and in different sequences than the Peutinger Table, suggesting that more sources are at issue than just a road map. The author shows familiarity with Ptolemy the Geographer (who he confused with one of the Macedonian dynasts) and mentions three Gothic authors (Athanarid, Heldebald, and Marcomir), perhaps reflecting the scientific interests at the court of Theodoric the Great (Staab 1976: 58). The conflation of his itinerary with these diverse sources suggests that a road map is only a secondary source on which other sources have been imposed (cf. Dillemann 1976: 170). For Transjordan south of Amman, only the Arnon River (Wadi Mujib), Rabba, Petra, and Aila (Aqaba) are listed (Schnetz 1990: 20), although a few of the Decapolis cities are mentioned in Northern Transjordan (Schnetz 1990: 25). The archaeological evidence derived from milestones and roads is of limited value for reconstructing the Byzantine roadsystem in Transjordan. The majority of the epigraphic evidence from milestones dates to the earlier Roman imperial era. The latest milestones for Syria and Phoenicia are from the reign of Constantine, for Palestine the end of the fourth century, and for Arabia the reign of Julian (361363). It has been suggested that the Byzantine emperors left the roads in a state of unrepair. Oxcarts purportedly went out of style as the camel gradually replaced the wheel and paved roads for transporting goods by the reign of Diocletian and afterwards. As a result it is assumed there was a general decline of the Roman road system in the East during the fourth to the sixth century (Bulliet 1975: 2226). But there are also clues that suggest the Roman road system in the Levant continued to function into late antiquity up to the Islamic conquests and perhaps later. Among them, milestones played an important part for pilgrims in the location of holy sites during the early Byzantine period. For example, Eusebius locates Mount Nebo by the sixth milestone of the EsbousLivias road (On. 136.6). Egeria's pilgrim travelogue to the Holy Land at the end of the fourth century uses the same locus (10.8) and locates the memorial pillar of Lot's wife by the sixth milestone from Zoar (12.7). Milestones were still of vital use to pilgrims traveling across the Holy Land in the sixth century. The primary evidence for reconstructing the Byzantine roadsystem must be the physical evidence, supplemented by any insights provided by documents reflecting pilgrim traffic (Donner 1992: 26). For example, the itineraries preserved in Theodosius' Topography of the Holy Land produced probably in the reign of Anastasius (A.D. 491-518) mention Livias east of the Jordan, as well as Madaba, Philadelphia, Gerasa, Bostra, Gadara, Abila and Capitolias in Arabia (145147 = Wilkinson 1977: 6970). It is possible that Theodosius utilized an archetype pilgrim map as the basis for his topography and the same source may have been at the disposal of the creators of the later Madaba Map (Tsafrir 1986: 136139; cf. Donner 1992: 14). This pilgrimage map must have mentioned road stations, accounting for such otherwise irrelevant information being included on the Madaba Map (Donner 1992: 25). In order for the map to be useful, it must have included the major Christian communities in Palestine and the sites that linked them, such as the ones depicted in the mosaic in the Church of St. Stephen at Umm al Rasas (Piccirillo 1987: 196204; 1993: 3637, 218 239). Christianity is well attested in each of them. But we know of additional Christian communities in the same region on roads that connected these same sites, such as at Dhiban, Dhat Ras and Nitl, which are not mentioned on the Umm alRasas mosaic. Such settlements are ideal candidates for inclusion on the Madaba mosaic map. From this hypothesis, we can then begin to reconstruct the other missing portions of Transjordan.

Between Bostra and Philadelphia Missing from the Umm alRasas mosaic are the Decapolis cities with the exception of Philadelphia (Amman), but the others must have appeared on the Madaba mosaic map. The archaeological and epigraphic evidence for Christianity is abundant for JerashGerasa, Pella, CapitoliasBeit Ras, and Abila. In addition, Christianity and churches are known at Bostra, Salkhad, Der'a, Umm alJimal, Sabha, Umm alQuttayn, Qasr alHallabat, and Khirbat alSamra. These communities were all linked together by the Roman roadsystem, which is now better known from recent archaeological work. The Via Nova Traiana is the only road listed by the Peutinger Table running northsouth through Transjordan, but other routes emerged in late antiquity throughout Transjordan. West of the Via Nova Traiana, the various Decapolis cities of Transjordan were connected by a complex of roads. Gerasa (Jerash) was the nexus of a number of routes that connected it to Philadelphia (Amman) in the south, Pella to the northwest (Mittmann 1970: 152159; McNicoll 1992: 122-124), Adraha ( Der'a) in the north (Mittmann 1970: 133138; Agusta-Boularot, Mujjali and Seigne 1997), and evidently Mafraq to the northeast (Kennedy and al-Husan 1996: nos. 2-5). Gadara (Umm Qeis) was also linked to a route that led through Capitolias to Adraha (Mittmann 1966). Milestones and stations recently discovered for the Gerasa-Amman route (Rasson-Seigne and Seigne 1995) reveal that more is yet to be discovered for the road-system in the region. East of the Via Nova Traiana was another complex of roads between Bostra and Azraq. One of these runs northeast of Umm al-Quttayn to the Azraq Oasis, paved and accompanied by milestones and military posts like the Via Nova Traiana (Kennedy and al-Husan 1996: 258; 1997). The nexus of the others is Umm al-Jimal, where churches attest a flourishing Christian community. One runs south through al-Qihati to Qasr al-Hallabat (Kennedy and al-Husan 1996: nos. 6-18) and another to Qabbar Dawwas (Kennedy 1982: 162; cf 1997). Another road runs east-west between Deir el-Kahf through Umm al-Jimal to Umm al-Quttayn and Sabha in the southern Hauran (Kennedy 1997). A third route is indicated by a milestone suggesting Umm al-Quttayn was connected with Salkhad east of Bostra. Finally, further to the east, another route connected the Roman forts at Deir el-Kahf and Azraq (Kennedy 1982: 169-186). As the Via Nova Traiana sweeps west from Amman to Madaba, there is another complex of roads south of Amman. At Wadi al-Hinu, a branch road headed south through Khirbat al-Suq and Yadudah towards Jiza (Graf 1997), after which it may have run west to Nitl and then straight south pass er-Rumeil to Umm al-Rasas (cf. Zayadine 1992: 229). Milestones further south near the Roman legionary camp at Lejjun, Qasr al-Hasa, Jurf al-Darawish, and Udhruh have been connected into a virtual Via Militaris, but it is doubtful that a Roman road linked these sites (Graf 1997). These outlying sites rather seem to be connected internally to the settled region in the west. For example, the Peutinger Table probably links the legionary fortress at Lejjun (Bettoro) to the administrative center of Rabba (Rababatoro representing a conflation of Rabba and Bettoro) by a road leading westwards through Thamaro to Jerusalem (Finkelstein 1979: 31). It is doubtful that this road was still functioning in the sixth century. Rabba (Areopolis) continued to be occupied after the Islamic conquests, but Lejjun was abandoned long before (Schick 1995b: 434435; 388389)

Between Kerak and Aqaba For Transjordan south of the Wadi alHasa, the Peutinger Table lists Thornia (Tuwaneh), Negla (Shobak), Petris Zadagatta, Hauarra, Praesidio, and Haila (VIII, 45). In the sixth century, the Beersheba Edict lists the settlements at Adroa (Udhruh), Auara (Humayma), Zadakatha (Sadaqa), and Ammatha (alHammam near Ma'an) for the same region (Alt 1921: 8). Thornia (Tuwaneh) and Negla (Shobak) are both well known ancient sites (Bowersock 1983: 174175), but neither has revealed any evidence of Christian presence before Islam. In contrast, Udhruh (as Augustopolis) and Ammatha appear in the papyri archive found recently in the sixth century Byzantine church at Petra (Koenen 1996: 178, 181), and were connected on a route leading east from Petra. Both settlements may have appeared on the Madaba Mosaic Map. The other settlements mentioned on the Peutinger table are on the Trajanic road leading to Aila, and the location of Christian communities that would be natural stations for Transjordan pilgrim traffic proceeding to the Sinai. Petra. The actual inclusion of Petra on the Madaba map is now a distinct possibility after the restoration in 1965 recovered several letters in the appropriate part of the map that may reflect the phraseology related to Petra in Eusebius' Onomasticon (Donner 1984: 252254). A string of bishops are known from the Byzantine era and the Petra papyri also make it abundantly clear that the region was flourishing in the sixth century (Koenen 1996). There was also a bishop at Petra and a flourishing church into the seventh century (Schick 1995b: 427428). Its importance as a Christian center alone justifies its inclusion on the Madaba Map, but it was also a vital link and station on the Via Nova Traiana. Road stations, milestones, and well preserved stretches of this paved road have been recorded in the vicinity (Graf 1995: 242244). Sadaqa. At least three routes proceeded south from Petra, two of which were connected to Sadaqa, the ancient settlement of Zadagatta (Graf 1995: 246250), attested as Zodocatha in the Notitia Dignitatum (Or. 34.24) and in the Beersheba Edict (Alt 1921: 8). The remains of the castellum leave little doubt that this is the Castra Zodacatha mentioned in Byzantine Christian graffiti discovered in south Sinai (Negev 1977: nos. 72 and 104) and more recently in the Petra papyri (Koenen 1996: 181), which supports its inclusion on the Madaba Map. The settlement may have been embraced later in the villages of Pentapolis, a name which appears in George of Cyprus for the region. Humayma. After Sadaqa, the Trajanic road proceeded south to the edge of the escarpment, then descended at Wadi alQana into the Hisma desert, where it headed to the major settlement of Humayma, and then south to Quweira and through the Wadi Yutm to its terminus at Aila on the Gulf of 'Aqaba (Graf 1995: 252257). Humayma's multiple reservoirs and cisterns and its military station provided both sustenance and protection for travelers along this route (Oleson 1997). It is clearly the largest Roman Byzantine settlement in the Hisma desert, and is the Hauara of the Peutinger Table and the Notitia Dignitatum (Or. 34.25) and is mentioned in the Beersheba Edict (Alt 1921: 810). Christianity is well attested at the settlement. At least four Byzantine churches have been located among the ruins (Schick 1995a; cf. 1995b: 311313). Thus the available archaeological evidence and its location suggest its inclusion on the Madaba Map. Praesidium. The identification of Praesdio on the Peutinger Table with Khirbet alKhalde in the Wadi Yutm is ensured by its location some twenty (Roman) miles from Aila, (Bowersock 1983: 173). The site consists of a Roman castellum and adjacent building designated as a caravanserai (Graf 1995: 260). According to the Notitia Dignitatum (Or. 34.41) the cohors quarta Frygum was stationed at Praesidium (Or. 34.41). A Byzantine tombstone from the settlement (Oliver 1941) attests the presence of Christians at the site in perhaps the sixth or seventh century (Glueck 1939: 18, fig. 10). The Madaba Map may have included Praesidium because of its importance as a road station on the route between Auara and Aila. 'Aqaba. The importance of the Christian community at Aila/Ayla (modern 'Aqaba) in late antiquity is reflected in both literary sources and recent archaeological investigations, which have revealed a massive mudbrick structure dating to the fourth century and tentatively identified as a Christian basilica (Parker 1996: 247). Nelson Glueck discovered the remains of marble screening and two capitals from a Byzantine church at 'Aqaba (Glueck 1939: 13). A lintel with a Christian Greek inscription and symbols was found in the recent excavations of the early Islamic quarter at 'Aqaba (Zayadine 1994: 489). These facts and Eusebius' reference to the site in his Onomasticon (6.17) suggest it appeared on the Madaba Map. Aila also formed an important station and nexus for Christian pilgrim routes in the Gulf of 'Aqaba region (Mayerson 1982: 4849). Across the Sinai, the Peutinger Table lists Clisma, Medeia, Phara, Haila. The Ravenna Cosmography [7, 817] follows the same pattern, listing Satanna [Mt. Sinai], Olelum [Madelam], Pnarana [Phara], and Minea [Hailanea], which appears to be the same route preserved on the Peutinger Table (Schnetz, 1990; 20.817). The route probably represents the pilgrim route to Mount Sinai (St. Catherine's) in the Byzantine period (Mayerson 1981; cf. Finkelstein 1979: 3031), and its inclusion on the Madaba Mosaic Map seems assured. By 451, there also was a bishopric on the island of Iotabe in the Red Sea -- 1,000 stadia from Aila, but its location remains a matter of dispute. Exploration of the island of Tiran at the head of the Gulf yielded no evidence for the Byzantine period and the island of Jazirat Fir'awn contains some Byzantine sherds (Glueck 1939: 11), but fails to fit the description of the site in ancient sources (Mayerson 1992: 3). In addition, the region of Midian in the northern Hijaz appears to have been a monastic center in the preIslamic era. During the Umayyad period, the poet Kuthayyir 'Azza mentions the "monks of Madyan," perhaps originating from the nearby bishopric of Aila (Shahid 1989: 294). The site is identified with Madyan Shu'ayb, known as Madiana in Ptolemy (6.7.27) and mentioned in Eusebius' Onomasticon (124.8). The region's biblical associations with Moses, Jethro, and Sepporah obviously provided the attraction for Christian monks, and became part of the Islamic tradition (Bosworth 1984: 54). Its inclusion on the Madaba Map therefore seems likely, but the minimal evidence for Christianity elsewhere in the Hijaz (Beaucamp and Robin 1981: 5051, cf. 5861) suggests that Midian formed the limits for the southwest section of the Madaba Mosaic Map. There were also monasteries in the desert east of Aila at Kilwa and Dayr Hisma (Schick 1995b: 308, 387), but such monastic sites alone are not normally included in the Madaba Map and these communities remain obscure (Shahid 1989: 294295). What seems clear is that the region of southern Transjordan and beyond was well represented on the Madaba Map, particularly because of Christian pilgrimages from Transjordan to the Sinai, for which the traditional Roman roads were extremely important.

Bibliography Agusta Boularot, S., Mujjali, A., and Seigne, J. 1997 Un "nouveau" gouverneur d'Arabie sur un milliaire inédit de la voie Gerasa/Adraa ADAJ : ? Alt, A. 1921 Die Griechischen Inschriften der Palaestina Tertia Westlich der 'Araba. Berlin and Leipzig: De Gruyter. AviYonah, M. 1954 The Madaba Mosaic Map. Jerusalem. Beaucamp, J. and Robin, C. 1981 Le christianisme dans la péninsule arabique d'après l'épigraphie et l'archéologie. Hommages à Paul Lemerle: Travaux et Mémoires 8: 4561. Berchem, D. Van 1937 Annone Militaire dans ['empire Romain ai III siècle, Mémoires de la société rationale des antiquaires de France 80: 117-202. Bosworth, C. E. 1984 Madyan Shu'ayb in Prelslamic and early Islamic Lore and History. Journal of Semitic Studies 29: 5364. Bowersock, G. 1983 Roman Arabia. Cambridge: Harvard University. Bulliet, R. 1975 The Camel and the Wheel. Cambridge: Harvard University. Cuntz, 0. 1929 Itineraria Romana, Vol. 1: Itineraria Antonini Augusti et Burdigalense. Leipzig: Teubner. Dilke, 0. A. W. 1987 "Itineraries and Geographical Maps in the Early and Late Roman Empires," and "Cartography in the Byzantine Empire. Pp. 234-275 in The History of Cartography, Vol. 1, J. B. Harley and D. Woodward, eds. Chicago: University of Chicago. Dillemann, L. 1976 La Carte Routière de la Cosmographie de Ravenne, Bonner Jahrbücher 175: 165170. Donner, H. 1984 Transjordan and Egypt on the Mosaic Map of Madaba. ADAJ 28: 249257. 1992 The Mosaic Map of Madaba. Palaestina antique 7. Kampen, Netherlands: Kok Pharos. Finkelstein, I. 1979 The Holy Land in the Tabula Peutingeriana: A Historical-Geographical Approach. PEQ 111: 2734. Funaioli, G. 1920 Ravennas Geographus. Cols. 305310 in PaulyWissowa's RealEncyclopädie, 2 Reihe 1. Stuttgart: Metzler. Glueck, N. 1939 Exploration in Eastern Palestine III. Annual of the American Schools of Oriental Research, vol. 1819. New Haven: ASOR. Graf, D. 1995 The Via Nova Traiana in Arabia Petraea. Pp. 241-267 in The Roman and Byzantine Near East: Recent Archaeological Research. Ann Arbor. Journal of Roman Archaeology Supplementary Series 14. 1997 The Via Militaris and Arabia. Dumbarton Oaks Papers 51. Kennedy, D. L. 1982 Archaeological Explorations on the Roman Frontier in North East Jordan. Oxford: BAR, International Series 134. 1997 Roman Roads and Routes in NorthEast Jordan. Levant 29: * Kennedy, D. L and AlHusan, A. G. 1996 New Milestones from Northern Jordan: 19921995. Zeitschrift für Papyrologie und Epigraphik 113: 257262. Koenen, L. 1996 The Carbonized archive from Petra. Joumal of Roman Archaeology 9: 177188. Mayerson, P. 1981 The ClysmaPhara Haila Road on the Peutinger Table. Pp. 167176 in Coins, Culture and History of the Ancient World: Numismatic and Other Studies in Honor of Bluma L. Trell. ed.L. Casson and M. Price. Detroit: Wayne State University Press. 1982 The Pilgrim Routes to Mount Sinai and the Armenians. IEJ 32: 4457. 1992 The Island of lotabê in the Byzantine Sources: A Reprise. BASOR 287: 14. McNicoll, A. W 1992 The Milestones of Mugharat alHales. Pp. 122124 in Pella in Jordan 2. Sydney: Mediterranean Archaeology Supplement 2. Miller, K. 1916 Itineraria Romana. Stuttgart. Mittmann, S. 1966 The Roman Road from Gerasa to Adraa. ADAJ 11: 6587. 1970 Beiträge zur Siedlungs und Territorialgeschichte des nördlichen OstJordanlandes. Abgamndlungen des Deutschen Palästinavereins. Wiesbaden: Harrassowitz. Negev, A. 1977 The Inscriptions of Wadi Haggag, Sinai. Qedem 6. Jerusalem: Institute of Archaeology, Hebrew University. Oleson, J. 1997 Humeima. Pp. 121122 in The Oxford Encyclopaedia of Archaeology in the Near East, vol. 3. E. M. Meyers, ed. New York/Oxford: Oxford University. Oliver, J. 1941 A Christian Epitaph from Eastern Palestine. American Journal of Archaeology 45: 542-543. Parker, S. T. 1986 Romans and Saracens: A History of the Arabian Frontier. Philadelphia: American Schools of Oriental Research. 1996 The Roman 'Aqaba Project: The 1994 Campaign. ADAJ 40: 231257. Piccirillo, M. 1987 Le iscrizioni di Um ErRasasKastron Mefaa in Giordania, Liber Annuus 37: 196204. 1993 The Mosaics of Jordan. Amman: American Center of Oriental Research. RassonSeigne, A.M. and J. Seigne 1995 Note préliminaires a l'étude de la vole Romaine Gerasa/ Philadelphia. ADAJ 39: 193210. Reed, N. 1978 Pattern and Purpose in the Antonine Itinerary. American Journal of Philology 99: 228251. Schick, R. 1995a Christianity at Humayma, Jordan. Liber Annuus 45: 319342. 1995b The Christian Communities of Palestine from Byzantine to Islamic Rule: A Historical and Archaeological Study. Studies in Late Antiquity and Early Islam 2. Princeton: Darwin Press. SchillingerHäfele, U. 1963 Beobachtungen zum Quellenproblem der Kosmographie von Ravenna. Bonner Jahrbücher 163: 238251. Schnetz, J. 1990 Itineraria Romana, 2 vols. Stuttgart: Teubner. Shahid, I. 1989 Byzantium and the Arabs in the Fifth Century. Washington: Dumbarton Oaks. Staab, F. 1976 Ostrogothic Geographers at the Court of Theodoric the Great: A Study of Some Sources of the Anonymous Cosmographer of Ravenna. Viator 7: 27 64. Tsafrir, Y. 1986 The Maps used by Theodosius: On the Pilgrim Maps of the Holy Land and Jerusalem in the Sixth Century C.E. Dumbarton Oaks Papers 40: 129145. Wilkinson, J. 1977 Jerusalem Pilgrims. Warminster: Aris & Phillips. Zayadine, F. 1992 L'espace Urbain du grand Pétra, le routes et les stations caravanières. ADAJ 36: 217239. 1994 Ayla'Aqaba in the Light of Recent Excavations. ADAJ 38: 485505.

This contribution was first published in: The Madaba Map Centenary, Jerusalem 1999, 227-229.