1In most of the maps produced by people working either in Arabia or in Egypt, including the ones I produced myself, the side of the Red Sea in the history of which they are not directly interested is usually left empty or almost empty.1 This is explained by the fact that despite the nine Red Sea conferences which have already taken place, only a small number of scholars have, until recently, focused on the connections between its two shores. I do not pretend examining here all the aspects of these connections, which are numerous and would require the participation of specialists of many disciplines (navigation, road systems, architecture, sculpture, epigraphy, etc.), but I would like to present some reflections on the subject, which have come to my mind on three occasions. The first was in 2014, for a lecture I gave at the Collège de France, entitled “The oriental shore of the Red Sea, from Aqaba to the Farasān Islands” as part of the course given by J.‑P. Brun.2 The second was in 2017, in a paper given at the “Arabian Red Sea route” workshop organised by the Polish Centre of Mediterranean Studies in parallel with the Red Sea 8 conference in Warsaw. This paper was entitled “Oriental desert versus Arabian desert. Reflections on the connections between the Egyptian and Arabian shores of the northern Red Sea in Antiquity”. The third one was at the Red Sea 9 conference, of which this paper is the result.

2One should first bear in mind that the amount of data available is obviously not the same for the two shores of the Red Sea since extensive excavations started in the late 19th century in Egypt against the 1970s in Arabia. Besides, until the early 2000s, large scale excavations were undertaken at two sites only, one pre-Islamic, Qaryat al‑Faw*,3 and one Islamic, Rabadah* (fig. 1). Apart from that, important surveys have been undertaken in Saudi Arabia in the framework of the “Saudi Arabian Comprehensive Survey Program” the results of which were published in several issues of Atlal, the Saudi archaeological journal. On the Red Sea proper and for ancient periods, large scale excavations started at the site of ‘Aynūnah* only in 2014 under the auspices of the Polish Centre of Mediterranean Archaeology of the University of Warsaw,4 while others started the same year in the Farasān Islands under the auspices of the French Ministry for Foreign Affairs.5 To these two land excavations should be added the underwater archaeological project in the region between Rābigh* and Shu‘aybah*, respectively north and south of Jedda*, by Philipps-Universität Marburg in 2012 and 2013.6 For later periods, one should add the excavation of a 9th to 13th century site at al‑Serrayn*, 100 km south of Mecca*, by the China National Center of Underwater Cultural Heritage since 2019. One last site, an 18th century shipwreck at Umm Lajj*, is excavated by the University of Naples “L’Orientale”, since 2015.

Fig. 1 – Map of the Arabian Peninsula showing the place names mentioned in the text (A. Emery, J. Schiettecatte, with additions by L. Nehmé).

3Obviously, the most important aspect of the connections between the two shores of the northern Red Sea is the maritime one. If there were no direct and easy maritime connections, political, military, commercial and cultural links, among others, would have been difficult. The three sources used to examine this topic for Antiquity are Strabo, the Periplus Maris Erythraei, and the information gathered between 2002 and 2013 by D.A. Agius, and masterly published by him in 2019, which have the huge advantage of relying on informants from both shores.7 Because of lack of time and competence, I unfortunately did not go through the Arabic sources which deal with maritime routes. They are sometimes quoted by D.A. Agius but they would require an independent treatment. Quite apart from the medieval and modern sources, we are certain that east-west connections existed in Antiquity for at least two reasons:

• the Periplus Maris Erythraei says that it takes two or three days to sail from Myos Hormos* (Quṣayr al‑Qādim*) to Leuke Kome by sea; this is probably a reliable statement, whether Leuke Kome is to be equated with al‑Wajh* or with ‘Aynūnah, 250 km further north;8

• in his narrative of the expedition of Aelius Gallus, which took place either in 26‑25 or in 25‑24 BC,9 Strabo (XVI, 4, 24) says that the troops, on their way back from South Arabia, embarked from a place called Egra Kome, which is said to be a seaport in the territory of the Nabataean king Obodas (II), who indeed reigned in Petra at the time of Gallus’ expedition. From there, it took them eleven days to reach Myos Hormos on the other side of the sea. The Greek has Ἐγρᾶς κώμης which, as was indicated to me by P.‑L. Gatier, is usually translated “from Egra Kome” but may equally be translated “from the village named Egra”. Note here the difference between the Periplus and Strabo: according to the former, two or three days are necessary to cross the Red Sea between Leuke Kome and Myos Hormos whereas it took eleven days to cross it from Egra Kome to Myos Hormos. This difference can be explained not only by the fact that the harbours of arrival in Arabia and departure from Arabia are not the same10 but also by the fact that the navigation times in the Red Sea vary highly according to the period of the year, the winds, and the navigation techniques.11 Moreover, in Strabo’s account, the entire army is sailing whereas the Periplus mentions simple traders’ trips. All in all, there is for the moment no certainty on the location of Leuke Kome and Egra Kome.

4A few years ago, respectively in 2012 and 2016, P. Arnaud and F. De Romanis12 drew attention to the fact that the sailing speed between Myos Hormos and Leuke Kome is expressed in the Periplus in dromoi, as opposed to stadioi which were used to express distances between some other ports. Dromoi and stadioi are said to be the translation of navigation days into spatial terms but dromoi are reserved to express straight-line navigation.13 The variety of measuring systems reflects the existence of different sorts of informants. According to F. De Romanis, the distances in dromoi would have come from Arab sailors who sailed these routes in very small boats or knew about them from local seamen. He also suggests that, quote, “the sea between Leuke Kome and Myos Hormos was probably sailed only occasionally by those South Arabian large ships that visited Leuke Kome”,14 and that seems correct on the basis of the text of the Periplus. Indeed, most often, the people sailing between Myos Hormos and Leuke Kome must have been either Egyptians or Nabataean Arabs, whose sailing abilities are mentioned by Diodorus in a passage related to Nabataean piracy in the Gulf of Aqaba, the source of which goes back to Agatharchides of Cnidus, between 130 and 110 BC.15

5The location of Egra Kome is not really better known than that of Leuke Kome. Note that in the Loeb edition of Strabo, Egra is written with a smooth breathing, but this does not necessarily exclude a Semitic equivalent starting with h or ḥ such as ḥegrā, the ancient name of Madā’in Ṣāliḥ.16 This “Egra” is said to be a kome, i.e. a small locality. In the early 1990s, this site was identified by A.I.H. al‑Ghabban with a cape located about 40 km south of the city of al‑Wajh. This cape is called Rās Kurkuma or, on a Canadian map of Arabia, Rās Karīkumā.17 One also finds the toponym “Ekre”, on a map which draws the itinerary from Cairo to Mecca according to the account of a man named Dom Philippo of Tunis (Prince Mohamed Chelebi’s name after he converted to Christianism), who undertook the travel in 1657.18 It is likely, however, that this “Ekre” corresponds to present-day Bi’r Akra, ca 30 km east from the shore, where there is a water source and which was a stop on the hajj road.19

6A site identified as Akra Komi is well documented in the National Museum of Riyad, where a few showcases present objects which come from a building where a few soundings were undertaken in 1992, among which architectural blocks and fragments of bronze statues (fig. 2). More precisely, the site where the objects come from is located at al‑Quṣayr*, which is 16 km north-east of Rās Kurkuma. Al‑Quṣayr lies 6 km away from the sea, on the edge of a desertic plateau located at the outlet of Wadi al‑Ḥamḍ*. It overlooks a palm grove which is still exploited and it is therefore an oasis. Many blocks which come from the building were reused as tombstones in the neighbouring Islamic period cemetery.

Fig. 2 – Nabataean capital from al‑Quṣayr, on display in the Riyad National Museum (L. Nehmé).

7In 2016, in the course of the second field season of the al‑‘Ulā*‑al‑Wajh Survey Project (UWSP), Z.T. Fiema and his team explored more carefully not only the site of al‑Quṣayr but also the cape itself. After a close examination of the best preserved main building, it became clear that the structure is not a temple but a Nabataean monumental triclinium, i.e. according to Z.T. Fiema’s own words, “a gathering place for ritual banquets and official meetings”.20 As does the Dūmat al‑Jandal triclinium excavated by G. Charloux in the so-called western settlement and dated to the interval between the last third of the 1st century BC and the first third of the 1st century AD,21 the al‑Quṣayr one was relatively isolated and overlooked both the wadi and the associated settlement, the latter lying ca 280 m south-west of the triclinium. The identification of this new triclinium and the parallel of the Dūmat one may, as suggested in Fiema et al. 2020, “indicate the importance of this type of communal structures in the ‘Nabataeization’ of the peripheral areas of the kingdom”. The role of the ritual banquets and the fact that they acted as a cultural marker of Nabataean identity was examined and pointed at recently.22 The surface pottery assemblage collected in the area south of the cemetery, an area which corresponds probably to the ancient settlement associated with the monumental building, was predominantly dated to the 1st centuries BC and AD, i.e., the core period of the Nabataean presence in Northwest Arabia.

8Even more interesting is the fact that among the pottery collected and identified by C. Durand, several sherds belong to Dressel 2‑4 wine amphorae from Campania and one sherd is probably an AE3 type biconical Egyptian amphora.23 These two types are precisely, as noted by C. Durand, the two main groups attested in the amphorae jetty of Myos Hormos24 and it is interesting to find them in both Myos Hormos and al‑Quṣayr. Conversely, it is now well known that Nabataean pottery was put to light at both Myos Hormos and Berenike*.25

9The existence of an important Nabataean site with Egyptian connections through the sea at al‑Quṣayr suggests the existence, in parallel, of inland connections between al‑Quṣayr and the largest site on the Arabian side in the late 1st century BC and the 1st century AD, ancient Hegra.26 The hypothesis of such a connection is precisely what drove Z.T. Fiema to launch the al‑‘Ulā‑al‑Wajh Survey Project (UWSP) in 2013. Three possible routes were explored during the first season (fig. 3), one mostly along the Wādī Fuḍalā*, one mostly along the Wādī Tharī* and one mostly through Wādī al‑Jizl* and then through al‑Kurr* and as‑Sudayd*. Neither of these routes, however, was convincing because no indisputable sign of the presence of an ancient caravan track such as ancient wells, caravanserai or groups of inscriptions left by passers-by was discovered. Following this, Z.T. Fiema judiciously decided, three years later, to explore the Wadi al‑Ḥamḍ, which is the longest but most convenient natural communication route to connect al-‘Ulā (ancient Dadan, 20 km south of Hegra) to al‑Quṣayr and the Red Sea. Unfortunately, in the kind of landscape crossed, the only types of structures which could be found are stone circles, enclosures, cairns, standing stones, etc., none of which could be dated precisely. Besides, the rock which borders the wadi is, in most cases, not suitable to carve inscriptions.27 It should be noted in passing that even when rock-cut inscriptions are discovered in a survey, they are not necessarily evidence for the circulation of people involved in the caravan trade. Some may have been written by simple travellers or by pastoralists.28

Fig. 3 – Google Earth image of the al‑‘Ulā region (Google Earth, with additions by L. Nehmé).

10After meeting with the Wadi al‑Jizl, the Wadi al‑Ḥamḍ turns south in the direction of Medina. Coming from al‑‘Ulā, these two wadis, al‑Jizl followed by al‑Ḥamḍ, were used during the Islamic periods to link Syria and Medina, with variants according to the periods from Umayyad to Ottoman. According to both A.I.H. al‑Ghabban and A. de Maigret, the ancient routes, in Antiquity and in the Early Islamic period (7th‑12th century AD), passed through Wadi al‑Jizl and Wadi al‑Ḥamḍ, following a list of steps which differ slightly.29 In 2019, during the first season of the al‑‘Ulā‑Medina survey undertaken in the framework of the Madā’in Ṣāliḥ Archaeological Project by the author, the team followed the track as suggested by the two above mentioned scholars. Among the sites visited30 where surface pottery was collected, two have yielded pottery identified by C. Durand31 as being earlier than the Umayyad period. One, called Umm Hidim, is located 30 km south of al‑‘Ulā, at the point of junction between the Wadi al‑Jizl and the Wadi Hidim (fig. 4).32 Based on the analysis of the pottery, and especially the presence of both a pithos sherd in the so-called “Hegra 9” fabric,33 characteristic of the older periods at the site, i.e. at the latest 1st century BC, most possibly earlier (fig. 5) and the existence of many sherds the fabric of which is very similar to that of the Umm Daraj pottery near al‑‘Ulā, C. Durand suggested that this site may have been occupied from the Lihyanite to the Early Nabataean period. The second site, a very large fort like structure, is locally called al‑Muṣanna‘ah, i.e. “the built”. It stands exactly at the junction between the Wadi al‑Jizl and the Wadi al‑Ḥamḍ, in an extremely strategic position. It is interesting to note here that this fort lies exactly where the “minimum energy expenditure” only path between al‑‘Ulā and al‑Wajh, as calculated by the GIS software, turns west.34 If this path was used rather than the other possible routes such as the “minimum time and energy expenditure” and other routes explored by the UWSP project, it means that al‑Muṣanna‘ah was more or less in the middle.

Fig. 4 ‑ The Wadi Umm Hidim in 2019 (L. Nehmé).

Fig. 5 – Sherd of the Hegra fabric no. 9 (Tourtet, Müller 2011, fig. 1).

11Al‑Muṣanna‘ah is a very impressive fort, built in basalt, almost exactly 50x50 m (fig. 6). Its plan and dimensions show similarities with two Roman forts of the Eastern Desert, Maximianon and Krokodilô, which are dated to the last quarter of the 1st century AD.35 It can also be compared with Islamic forts such as the 8th century AD fort built north of the site of ancient Dadan, 120 km north-west of al‑Muṣanna‘ah, discovered in 2004 and at which excavations were undertaken a few years ago and resumed in 2020.36 It is difficult, on the basis of the plan as it appears on the surface and on Google Earth, to establish whether al‑Muṣanna‘ah is Roman or Islamic. The only sherds picked up during the survey (almost no pottery was visible on the surface) are nine small and eroded body sherds none of which shows the typically Islamic creamy/greenish fabric. One of them corresponds to the same “Hegra 9” fabric mentioned above. The date of this fort, thought to be Islamic, is worth a reassessment. It would certainly be very interesting that it is either Nabataean or Roman but only soundings may answer this question.37 If it turned to be Roman, the historical implications would be of importance. It would mean that there was a Roman presence in the area, almost 150 km south of Hegra, as early as the late 1st century AD and this might, in turn, be put in perspective with the issue of the centurion known by the Periplus to have stationed at Leuke Kome, the political affiliation of which, Nabataean or Roman, is debated. P. Arnaud and M.A. Speidel have both put forward several convincing arguments in favour of him being Roman.38 Considering that the Periplus is dated to the mid 1st century AD, the building of a Roman fort in al‑Muṣanna‘ah before AD 106 would renew the available data on the Roman presence in the Arabian Peninsula. As for the hypothesis first put forward by M.A. Speidel according which the Romans may have initiated a domination of South Arabia after the expedition of Aelius Gallus, scholars do not agree on its validity.39 It is rather assumed that the South Arabian kingdoms maintained with the Romans an alliance based relationship, an amicitia.40 The two Latin inscriptions from the Farasān Islands are less relevant here because one of them is dated to between AD 114 and 132 and the other to AD 144,41 and they are therefore later than the creation of the Roman province of Arabia in AD 106.

Fig. 6 – Google and ground view of al‑Muṣanna‘ah (Google Earth; L. Nehmé).

12In the present state of the research, one has to admit that there is no absolute certainty that the ancient track from al‑‘Ulā to Medina followed the itinerary as it is described by A. de Maigret and A.I.H. al-Ghabban. During the 2020 season of the Madâ’in Ṣāliḥ Archaeological Project, another itinerary was explored, starting from two interesting sites with many Nabataean inscriptions, as‑Sīj* and aṣ‑Ṣuwaydrah*, respectively south-east of Madā’in Ṣāliḥ and ca 60 km east-north-east of Medina. The final results of this survey are not available yet. Note that the southernmost Nabataeo-Arabic text, yet unpublished, was discovered at aṣ‑Ṣuwaydrah.

13Coming back to the connection between al‑Quṣayr and ancient Hegra, the existence of a track was obviously useful only if there was also a caravan track from Hegra to the north, i.e. up to Petra and Gaza, two important commercial hubs. The author had the chance, in 2004, to explore this route, called Darb al‑Bakrah, “the Bakrah road”, with a team directed by A.I.H. al‑Ghabban, from the Saudi Department of Antiquities. This route had already been identified by ancient travellers as a camel path, in particular by G.A. Wallin in 1848, C. Doughty in 1877 and A. Musil in 1910.42

14During the 2004 exploration of this route, thousands of inscriptions were recorded from various sites, including 912 Nabataean and Nabataeo-Arabic inscriptions, along with inscriptions in Imperial Aramaic, Greek, Ancient North and Ancient South Arabian and even one in Palmyrene. All these were published in 2018.43 The evidence from all the inscriptions suggests that this route was used from the second half of the first millennium BC (the Imperial Aramaic text) down to the 5th century AD, which is the date of the latest Nabataeo-Arabic text, UJadhNab 109.44 Up to the turn of the Christian era, there is evidence of people coming from South Arabia. After that, the evidence points to an integration of the Darb al‑Bakrah into a regional communication network which extends from Yathrib, modern Medina, to the site called Ṭal‘at al‑Midrāt*, with an important hub at Umm Jadhāyidh*, where 572 texts were recorded. It is absolutely certain that the Darb al‑Bakrah was a caravan track in Antiquity not only because several sites have been interpreted as caravan stations but also because several examples of the same individual writing his name and father’s name in different places were identified. The examples even include a man named Šullay son of ’Awšū who left his name at eight years interval in the same area, UJadhNab 309 in AD 295 and UJadhNab 538 in AD 303 (fig. 7).45 It is interesting to note that this Šullay, despite bearing an Arabian name, was probably Jewish because he wrote the second inscription “the day of the feast of the unleavened bread”, i.e. Passover. On the basis of the material collected, it can now be suggested that the Darb al‑Bakrah started to be used regularly and intensively only after the Nabataeans had settled in Hegra, in the second half of the 1st century BC, and this would explain why the earliest Nabataean inscription on the Darb, UJadhNab 110, is dated AD 41‑42. The road continued to be used in the Roman period, as we know from the few inscriptions dated to years of the Roman province of Arabia.

Fig. 7 – UJadhNab 309 (AD 295) and UJadhNab 538 (AD 303), written by the same individual (L. Nehmé 2004, Farīq aṣ‑Ṣaḥrā’ 2017).

15In Hegra proper, the Roman period is now well documented thanks to the excavations undertaken since 2008 in various areas of the site, including in the so-called residential district, which forms a 52-ha settlement (never completely urbanized) surrounded by a 3 km-long city wall along which thirty-eight bastions out of ca eighty-one (based on the average interval between the preserved ones) were identified on the field. In the sectors with a domestic occupation excavated so far, the occupation extends until the 4th, possibly the 5th century AD.46 Apart from these, two structures are particularly relevant to the Roman period at Hegra and have yielded artefacts which are relevant to the present contribution. These are the so-called south-east gate on the rampart, excavated between 2011 and 2017 by F. Villeneuve, and the Roman fort, excavated by Z.T. Fiema since 2015.47

16The fort was built in the first half of the 2nd century AD along the southern section of the rampart and it was reinforced at the end of the same century (fig. 8). A military occupation is attested in it until the end of the 3rd century. After that, civilians started living inside the camp and continued doing so until the end of the 4th century, not later. This fort is the southernmost Roman camp known so far, 430 km south-east of that of Ḥumayma, its closest parallel. In both the south-east gate and the camp, several Greek and Latin inscriptions were put to light during the excavations, either on movable stones or reused in the walls of the towers flanking the south-east gate.48 Four of them mention individuals who are said to be stationarii and are referred to either as such or with the abbreviation “S”.49 Two, no. 1 (35004_i09b) and 2 (35004_i09a) are probably dated to AD 161‑169. After others, Z.T. Fiema describes the milites stationarii as being under the command of a centurion or a decurion and being in charge of controlling the travellers’ traffic in rural zones. They maintained security on the roads, chasing criminals and fugitives. They also checked travel documents either at the city gate or during the road patrols. The stationarii are also well known in the Egyptian Desert where they appear in early 2nd century documents.50 The inscriptions from Hegra suggest that they were sent to outposts two by two. Thanks to these inscriptions, we know for sure that Hegra was a statio where pairs of soldiers were stationed for a certain time.

Fig. 8 – Plan of the Roman fort of Hegra after the 2019 season (Madā’in Ṣāliḥ Archaeological Project).

Conclusion

17After mentioning the harbour of Egra Kome, Strabo says that the army of Aelius Gallus “marched by land over to Coptus*, and, with all who had been fortunate enough to survive, landed at Alexandria”. He also says that “at the present time they [the loads of aromatics] are for the most part transported by the Nile to Alexandria; and they are landed from Arabia and India at Myus Harbour; and then they are conveyed by camels over to Coptus in Thebaïs, which is situated on a canal of the Nile, and then to Alexandria” (Geography, XVI, 4, 24). This is what happened on the Egyptian side around 20 BC. On the Arabian side, a maritime route, up to Leuke Kome, combined with a land route going further north, was added in the 1st century BC to the north-south land route along which the South Arabian and Minaean merchants were active. See also now Nehmé 2021. At that time, it seems that the Nabataeans had completely taken over the land part of the route, founding the city of Hegra in the second half of the century probably for this reason. The Romans just followed their path, and continued to control the trade using the same recipe, installing stationarii to control the road network. “Nabataean” individuals continued to be active along the roads after the creation of the Roman province of Arabia in AD 106. They were also still active on the Egyptian side, as shown by the Nabataean inscriptions dated to the era of the Roman province of Arabia discovered along the east-west roads joining Myos Hormos to Coptos (and not only there), for which I refer most recently to the publications by C. Durand and R.Z. Mohamed.51 Both regions, the Eastern Desert and the Arabian Desert, were probably more connected than we usually think of and functioned in a parallel way, despite a probably lower investment in Arabia than in Egypt. Future work at al‑Quṣayr and al‑Bad‘*, added to the survey of the Wadi al‑Ḥamḍ and of the al‑‘Ulā‑Medina ancient road, combined with soundings at the fort of al‑Muṣanna‘ah, will certainly continue to shed new light on the region.