Introduction

1One of the most interesting aspects of the history of the northern Horn of Africa is the intense interaction between people from the coast and people from the inland, which has been ongoing since Antiquity. The coastal areas of present-day Eritrea have always been a gateway of goods as well as of people, ideas, and knowledge. These, over time, strongly influenced the economic, social, political and religious dynamics of the internal regions, and vice versa.

2“From Adulis it is a journey of three days to Koloê, an inland city that is the first trading post for ivory, and from there another five days to the metropolis itself, which is called Axômitês [Axum]; into it is brought all the ivory from beyond the Nile through what is called Kyêneion, and from there down to Adulis”.1

3This well-known paragraph reported in the Periplus Maris Erythraei represents a milestone contribution to the process of reconstructing the routes linking the southern Red Sea coast with the adjoining highlands and inner desertic regions, as it describes in detail the logistics of the itinerary from the sea port of Adulis (in Eritrea) to the metropolis of Aksum (located in the mountainous regions of the central Tigrean plateau, in present-day northern Ethiopia). As can be inferred from the text itself, the Adulis-Koloê-Aksum itinerary illustrated in the Periplus (fig. 1) was only a segment of a larger and more complex network of medium- to long-distance sea and land routes which extended “from beyond the Nile through what is called Kyêneion”. Moreover, this itinerary originates from the mid 1st century CE, a period wherein Aksum was already progressively emerging as a powerful polity and Adulis – with the establishment of the Indo-Roman maritime trade – was gaining an increasingly prominent economic role, leading to its inclusion within Aksum’s political sphere.2 It is therefore evident that both Aksum and Adulis constituted major nodes within this network at that time (as well as for the following six centuries, at least).

Fig. 1 – Sketch of the Adulis-Koloê-Aksum itinerary and the hydrography of Ethiopia, Eritrea and the Sudan relevant to this paper (L. Sernicola/University of Naples “L’Orientale”).

4Before entering in the detail of the paper, a discussion about the possible location of Kyêneion desserves some space. While there is general agreement among scholars in accepting Bent’s first identification3 of ancient Koloê with the huge Aksumite archaeological remains at Qohaito4 for its situation and the fitting with the Periplus’ travel time5 (see fig. 2 for the location of Qohaito and the situation of all place names), Kyêneion is a toponym that has not yet been identified with any archaeological site, nor its definite geographic location has been ascertained. According to Schoff,6 who assumed that the anonymous author of the Periplus intended to refer to the river known by that name, Kyêneion has to be identified with the region of Sinnar, indeed just across the Blue Nile, some 560 km south-west of Aksum. His hypothesis is followed by several scholars including Huntingford,7 Hable‑Selassie,8 Munro‑Hay.9 On the other hand, it is widely accepted that the ancients were indisputably vague and sometimes confused when describing the uppermost course and waters of the Nile, entitling various rivers in Ethiopia to be part of it. One of them is the Takkaze, the most important affluent of the Nile-tributary Atbara. Running westwards for about 608 km from the central Ethiopian highlands until it flows into the Atbara, the Takkaze is already attested in Aksumite inscriptions.10 In Ge‘ez (ancient Ethiopic), the word “takkaze” means “river”. Thence, since in the Bible the Nile is referred to as “the river”, in Ge‘ez this was rendered with “Takkaze”. This reflects its original identification with the Nile, of which it is the easternmost affluent, closest to the coast and thus best known to the ancient geographers. With this same meaning it is used in Ethiopian medieval texts.11 In the second and chronologically later Greek inscription of the Monumentum Adulitanum, the anonymous king vaunts his conquest of the territories “beyond the Nile” where snow-capped mountains rise ‒ most likely referring to the Semen mountains, which suggests that this river was the Takkaze.12 The same “confusion” could have been happened in the Periplus; if this is true, then Kyêneion could be identified with the Walqayt region, ca 120 km south-west of Aksum. As we will see in the 16th to 20th century historical sources, the old and important trade route from the Red Sea to Gondar, the Lake Tana region and the south (via Aksum) had to cross the Takkaze.

Fig. 2 – Map showing sites and localities mentioned in the paper: early 1st millennium BCE to late 1st millennium CE (L. Sernicola/University of Naples “L’Orientale”).

5In both cases, Kyêneion has to be considered the southern limit of an axis linking the African side of the southern Red Sea coast with the Nile catchment area to the south-west of Aksum. Another possibility is that the term “Nile” of the Periplus refers to the Mareb, the main border river now separating the central Eritrean highlands from Ethiopia. On this basis, Conti Rossini13 suggests identifying Kyêneion with Cohain, the area within the loop of the Mareb. This interpretation, if correct, would bring Kyêneion even closer to Aksum (ca 40 km) and to the north rather than the south-west. This latter hypothesis seems to be less probable as the Mareb river is not strictly related with the Nile. Born in the Hamasen highlands of central Eritrea, it runs westwards in the Kunama lowlands flowing underground and then reappearing before ending its course in the Sudanese desert; only in exceptional cases an arm of this river joins the Nile-tributary Atbara.

6Anyhow, whatever the location of Kyêneion itself, the archaeological and historical evidence presented in this paper demonstrates of the existence of a network of routes linking Aksum to the regions south-west of it.

7This network did not remain consistently the same. Most importantly, it certainly did not begin, nor end, with Aksum or with the Indo-Roman trade: the network started significantly before Aksum’s ascent and continued long after Aksum’s decline. It was part of a longer process of exchange and interactions, and, like any process, was fluid and mutable over time. The network followed a relatively stable pattern of natural corridors while simultaneously readapting to the changing ecological, social, economic and political conditions that affected the nature and distribution of required goods and natural resources, as well as the importance and location of major nodes and hubs. The result was a dynamic, long-term network of local, regional and interregional circuits forming major axes and ultimately overcoming the disparate nature of this territory by connecting the sea with the innermost lands of the northern Horn.

8A segment of this network linked the territories of the southern Red Sea with the regions of present-day central western Ethiopia and eastern Sudan via the Tigrean plateau, and might have constituted the south-western prolongation of the Adulis-Koloê-Aksum itinerary reported in the Periplus.

9Later sources confirm the significance of this route. Sections 97 and 416 of Antoine d’Abbadie’s 1890 Géographie d’Éthiopie (translation in Huntingford 1989) state that the route of the great caravan from Adwa to Gondar arrived at Mai Abeqat (in the south-western suburb of Aksum) on the second day, at the Takkaze river on the ninth day, and at Gondar on the twenty-fifth.14 The Géographie d’Éthiopie also includes a second traditional itinerary, which describes the distance between Adwa to Gobedra (Aksum) as requiring a day to travel across, and notes that the western edge of the Aksum conurbation was a stopping place for salt merchants. This itinerary lists the destination for the third day as Addi Gidad, which was a toll post (as at Aksum). Dambaguina (Indabaguna) is listed for the fifth day of the itinerary, and is described as a place where the traders separate, some going to Walqayt and the rest to Gondar.15 Although recorded in the 19th century, these routes – which were largely determined by the topographical features of hill slopes, river crossings and the availability of drinking water – were likely to have been in use for many centuries previously.

10The primary aim of this paper is to expand the existing knowledge of this north-east/south-west exchange axis in a chronological and spatial manner, in light of recently acquired archaeological evidence. The first part of this paper will discuss excavations and surveys conducted along the Mai Negus river valley (to the south-west of Aksum) which enabled further reconstruction of this ancient trade route. The second part of this paper will explore the identification of an archaeological port-site to the south of the Gulf of Zula (at Irafali) and its implications for our understanding of the changes in this coastal-inland connection route during the 16th-17th centuries.16

The north-east/south-west circuit in ancient times

Evidence from the early 1st millennium BCE: the site of Seglamen

11The valley along the Mai Negus/Haselo river crosses the plain to the south-west of Aksum’s conurbation and flows south-westwards: following the valley, it is a journey of six days on foot to the Takkaze river, the regions north of Lake Tana (where the Blue Nile originates), and the upper course of the Atbara (see fig. 2).

12Since 2010, the Italian Archaeological Expedition at Aksum of the University of Naples “L’Orientale” and ISMEO (“Associazione Internazionale di Studisul Mediterraneo e l’Oriente”) has been conducting archaeological investigations in a 100 km² transect along the Mai Negus/Haselo river valley, from the south-western periphery of Aksum to Adet. The major goals of the project are to provide a long-term reconstruction of the population history, cultural interaction dynamics, and environmental dynamics in this region; and to provide a detailed archaeological map of this area for the cultural heritage management of Central Tigray, to complement the assessment of the archaeological area of Aksum implemented between 2000 and 2008.17 The area was selected as an area of study because the Mai Negus/Haselo river valley represented an important traditional, and perhaps ancient, exchange route linking Aksum (and the northern Ethiopian highlands generally) to the Takkaze river in the south-west – and thus connecting Aksum to the internal regions of the northern Horn of Africa.

13A systematic survey of the study-area and archaeological excavations at site SG 1 in the modern village of Seglamen have been conducted so far, both providing stimulating insights into the ancient cultural and economic interactions along this corridor, in addition to providing information on the emergence of the Aksumite culture.18

14The modern village of Seglamen is located on a remarkably flat terrace, about 12 km to the south-west of Aksum (fig. 3). SG 1 is a site dated to the first half of the 1st millennium BCE located in the easternmost sector of the modern village, at the confluence of the Mai Negus with a minor tributary. Awareness of the possibility of an archaeological site in this area dating back to this period (the so-called “pre-Aksumite” period) came about in the early 1970s. A royal inscription in monumental South-Arabic, which commemorated the re-erection or restoration of a temple dedicated to the god “HBS” (“Hawbas”), was found by a local farmer at Amda Tsion, in the south-eastern sector of modern Seglamen.19

Fig. 3 – The location of Seglamen and Medogwe along the Mai Negus river valley (L. Sernicola/University of Naples “L’Orientale”).

15Preliminary excavations aimed at detecting archaeological remains chronologically and culturally related to the inscription were carried out in 1974 by the University of Rome “La Sapienza”, under the direction of Lanfranco Ricci. These brought to light a large post-Aksumite (ca 800/850‑1300 CE) rural house, apparently built on earlier foundations.20 The actual location of the pre-Aksumite settlement was determined in 2006, during the systematic survey of Aksum and its vicinities conducted in the framework of the World Bank Ethiopian Cultural Heritage Project.21 This project occurred after a hiatus of more than 30 years, during which research in the area was completely abandoned. The location of the settlement was definitively confirmed in 2009.

16The site extends over an area of about 7 hectares in the easternmost sector of the modern village, overlooking the river gorge. It encompasses the present-day areas of Amda Tsion, where the ancient settlement was located, and Mogareb, a gently raised terrace to the north-west of the settlement, where the remains of the ancient cemetery have been detected and investigated (fig. 4).

Fig. 4 – The location of site SG 1 at Seglamen with, encircled, the approximate extension of the ancient settlement and cemetery (L. Sernicola/University of Naples “L’Orientale”).

17Excavation in the settlement area uncovered the remains of overlapping stone buildings belonging to three major architectural phases. These remains included small rectangular and circular houses and large multi-room and multi-storied palaces, all ascribable to the first half of the 1st millennium BCE on the basis of associated artifacts and radiocarbon dating. More precisely, radiocarbon dating suggests an occupation of the settlement from ca 980 to 450 BCE.22

18Excavations in the cemetery have so far exposed the remains of 37 tombs consisting of circular and roughly rectangular shafts, which can be grouped into two major types. The first type of tomb is those associated with stela – a sandstone monolith (entirely or partly carved) is erected inside the shaft. The second type is tombs without stela. Both types can be either circular or rectangular and, in both cases, human skeletal remains and grave goods are found. Two of the tombs associated with stela are pits with multiple burials, all the others, with or without stela, are single burials. Small votive deposits characterized by shallow pits are located around the graves, in the proximity of outcropping natural boulders. Associated artifacts include ornaments, zoomorphic clay figurines, and miniatures of cups and beakers.23

19Analysis of the material culture and first data from faunal remains and archaeobotany suggest that the local economy was mainly agro-pastoral. Specifically, the economy was likely based on the cultivation and processing of cereal and legumes (Hordeum vulgare, Eleusine sp., Sorghum sp., Eragrostis sp., Linum usitatissimum, Vicia fava) and the breeding and butchering of cattle, sheep, and goats. A variety of craft, and possibly industrial activities were also performed at the site, including stone knapping, pottery making, cleaning and refining animal skins, and wood working.24

20Alongside the manufacturing and use of local products, a significant corpus of imported materials has been collected at both the cemetery and the settlement. Among these are ceramics from the neighboring regions,25 blue glass beads from the Nile Valley, the typical South Arabian “Torpedo” amphorae or “Type 4100” jars,26 and several fragments of pearly oyster (Pinctada margaritifera), a marine species living in the Red Sea and Indo-Pacific Ocean.27 One such fragment of pearly oyster is a worked marginal fragment of an upper valve, found in a context dated between the 7th and 5th centuries BCE. The fragment measures 2.6×2.7×0.3 cm and shows an abraded edge and a polished surface, suggesting it was part of a pendant or inlay (fig. 5).28

Fig. 5 – Fragment of Pinctada margaritifera with abraded edge and polished surface from excavated context at site SG 1, Seglamen (L. Phillipson).

21The site was therefore likely to have been included in a medium- to long-scale interchange circuit involving the Nile Valley, the Tigrean highland, and both the African and Arabian coastal regions of the Red Sea. Its location in a prominent position overlooking the river valley leading to the south-west corroborates this evidence. The location is also reflective of a deliberate occupation pattern recurrent in this period, characterized by the situation of relatively small but affluent centers, often associated with temples, along the major river valleys linking the Tigrean plateau with the regions south of it.29

22In this period, the settlement pattern of the area of Aksum was characterized by sparse rural dwellings and few small villages. No traces of ceremonial complexes nor evidence of prominent elite-related structures have been found, although the existence of a certain degree of economic (and potentially social) inequality appears to be proven by the differences in architecture and in the composition of artifact assemblages resulted from the test excavations conducted at the D‑site in the area of Kidane Mehret.30 This pattern suggests that Aksum was a marginal centre during the first half of the 1st millennium BCE – one of several rural settlements in the Tigrean plateau with no particular relevance, while other sites (primarily Yeha, and perhaps Matara) were more significant political centres and commercial hubs.31

From the mid 1st millennium BCE to the late 1st millennium CE: the site of Medogwe and a general overview of the Aksum region

23Three km north-east of Seglamen and 8 km south-west of Aksum is the site of Medogwe, situated along the uppermost course of the Mai Negus/Haselo (fig. 2‑3). First described by de Contenson32 and Hailemariam33 in the 1950s, the site has recently been systematically mapped and surveyed by our expedition and dated, on the basis of the exposed materials, from the mid 1st millennium BCE to the mid 1st millennium CE.34 The site presents elements clearly ascribable to the proto-Aksumite culture and to the early, classic and middle phases of the Aksumite culture; scanty evidence of pre-Aksumite materials has been also noticed.

24At Medogwe, like at Seglamen, local artifacts are paralleled by imported ones. Among these, fragments of black Adulitan ware dateable to a period prior to the 3rd century BCE on the basis of the stratigraphic sequence at Adulis (fig. 6).35 These fragments suggest that the site was included in this complex trade network with the Red Sea regions from at least the mid/late 1st millennium BCE to the late 1st millennium CE.

Fig. 6 – Fragments of Adulitan ceramics from surface collection at Medogwe (L. Sernicola/University of Naples “L’Orientale”).

25Other literary sources possibly refer to an ancient exchange circuit between Aksum and the regions south-west of it during the kingdom of Kaleb in the 6th century CE. In particular, in his Topographia Christiana, Cosmas Indicopleustès mentions that the king of Aksum used to send men to trade gold with the governor of Agaw (in the region of Sasu) every two years:36 this region was located to the south-west of Aksum, in the Blue Nile valley.37 In this region, the Aksumites exchanged gold for salt, ox and iron. Cosmas Indicopleustès lists the goods that the kingdom of Aksum exported abroad from its sea port, including gold, ivory (from elephants and hippos), emeralds, live animals, aromatic plants, and slaves.38

26The Mai Negus/Haselo river valley, which was first being crossed during the 1st millennium BCE, might have still been used during the Aksumite period (1st millennium CE) to access the regions to the south-west of the Tigrean plateau, where the upper courses of the Atbara and Blue Nile are located. From Aksum, goods and merchandises reached the sea following the well-known Aksum-Koloê-Adulis route. Although still scanty, archaeological evidence of this is starting to emerge from systematic surveys and settlement pattern analysis.

27A large Aksumite satellite village, dating from the early to the mid 1st millennium CE has been detected at Mai Abeqat, immediately south-west of the Aksum conurbation.39 The site surface presents a great amount of imported items and clay tokens suggesting of commercial and administrative activities; it was abandoned when Aksum started its process of decline. Interestingly, this is the same area where, according to the previously-mentioned 19th century sources, merchant caravans used to stop on their journey from Adwa to Gondar, that is, from the Tigrean plateau to the Nile Valley region.40

28Moreover, recent surveys conducted along the upper course of the Mai Negus demonstrate the existence of a peculiar settlement pattern, completely different from the one found in the area of Aksum or around other major Aksumite centres. Here, remains have been found of small, isolated Aksumite buildings situated at a regular distance of about 15 km that overlook the river valley; fine wares and administrative devices, including counters and coins, form a significant component of the surface materials.41 These few, isolated buildings are situated out of the salubrious and agriculturally productive northern areas, at a lower altitude which continues to decrease as one moves southwards, and are placed on a narrow plateau flanked on both sides by steep escarpments: they seem to be outposts, placed to mark Aksum’s expansion into the south-west, and to control and somehow secure this long-established natural corridor. They are part of a landscape characterized almost exclusively by large lithic sites with no traces of constructed structures nor ceramics. These were probably seasonal camps, reused over time by pastoral groups. No absolute dating evidence has been collected from any of these “lithic” sites yet. However, at least some traits of the lithic industry preliminarily dated to the 1st millennium BCE and the 1st millennium CE, on the basis of similarities with the lithic technology attested at other Aksumite sites in this region.42 Of course, future investigations are necessary to enable more precise interpretation, to refine the chronological attribution of these sites, and to better understand the dynamics of economic and cultural interactions that occurred at different spatial scales in ancient times.

29As Aksum started to decline in the late 7th century CE, so too did the satellite villages around Aksum’s conurbation, including Mai Abeqat,43 the sites along the Mai Negus,44 and most of the largest Aksumite cities, like Matara (located along the Adulis-Koloê-Aksum route) and Adulis itself.45 After having reached the peak of its hegemony in the southern Red Sea during the 6th century, the port city of Adulis was abandoned, as both written sources and archaeological evidence testify to.

Reassessing hegemonies and hubs: from the decline of Aksum to the 15th century

30Very few historical and archaeological evidence illuminate the period immediately after the progressive decline and abandonment of Aksum as a political centre during the 7th century CE. Apparently, the focus of the Christian polity shifted in eastern Tigray, although knowledge about the late 1st millennium in this area is currently limited to a small number of ancient churches.46

31It seems that, in this period, the maritime hegemony in the southern Red Sea shifted to the Persians. Their supremacy in the Red Sea trade may have taken place from the second half of the 6th century to the first half of the 7th century, until the beginning of the Arab expansion in the Red Sea during the 9th‑10th century CE.47 A persistence of the Persian merchants’ presence in the southern Red Sea (or coexistence with Arab traders) can be seen in a series of inscribed tomb stones found at Dahlak Kebir, in Kufic characters, dating from between 1040 and 1100 CE.48 In this period, the centre of the maritime trade in the northern Horn of Africa became the Dahlak Islands, concurrent with the emergence of a local sultanate. Archaeological evidence of this may be seen by the presence of Islamic mausoleums called “qubba”at Dahlak Kebir and on the Eritrean coast, near Massawa (see fig. 7 for the location of all place names in this section).49

Fig. 7 – Map showing sites and localities mentioned in the paper: 16th to 20th century (L. Sernicola/University of Naples “L’Orientale”).

32Regardless of the causes catalyzing the Aksum polity’s decline, this shift led to a reorganisation of the coastal/inland exchange network (including the north-east/south-west axis), as is apparent from Arabic sources dating to the 9th-10th centuries. In the 9th century, the Arab historian, geographer and traveler Abu al‑Hasan ‘Ali al‑Mas’udi stated that the Ethiopian kingdom had many towns and extended until the sea, where there were the towns of Zula, Basi (perhaps Massawa) and Dahlak.50 He also attests that these towns were inhabited by Muslims paying tribute to the Abyssinians (supposedly, the people of the inland). At the beginning of the 10th century, the Dahlak sultanate was drawn into the Yemenite/Persian orbit due to the strength of their trade contacts: the sultanate stopped paying tribute to the Abyssinian king until it became independent.51

33The route connecting the coast to the inland in this period can be determined from information about the presence of Muslim traders’ settlements distributed among the coast and the inland. In fact, it has been found that the Shewa region housed a sultanate (founded in 896‑897 CE) during this period.52 In the Enderta region, Islamic stelae dating before the 10th century have been found, attesting to the presence of Islamic people in the area.53

34Major changes took place in about the 11th century, when a new centralized authority based further to the south in the mountainous region of Lasta was established. Its political centre was located at Adefa (a toponym mentioned both in Arabic and Ethiopic sources), near the ecclesiastical establishment called Roha – subsequently renamed Lalibela, after one of the prominent kings of this period. Archaeological evidence is still scanty and very few written Ethiopian historical records survive from this period. Even the circumstances of the coming to power of this centralized authority are still unclear and the extant information about this event is mostly fabulous and incomplete; what appears clear from the propaganda successive to its decline is that an ethnic group – presumably of Agaw descent – distinct from the one that had ruled previously, had come to power.

35Traditionally known as the “Zagwe dynasty” – although others prefer to refer to it as the “Lasta period” –54 the rulers of this period, which lasted until the third quarter of the 13th century, administered a centralized Christian realm, but apparently exerted non-direct control east of the Shewan plateau, in a territory increasingly occupied by such Muslim states as Ifat and Hadyya.55 Their main outlet to the sea seems to have been Zula, from where they got engaged in long-distance trade, although to a much lesser extent than Aksum. Unlike Aksum, the Zagwe kingdom was virtually unknown to the coeval Mediterranean powers; the few available archaeological and historical information about this period suggest relations with Egypt and the Holy Land.56 As documented in the repeated correspondence with the patriarchs of Alexandria as well as the sultans of Egypt,57 Ethiopian delegations were recurrently sent to Egypt to ask for a Metropolitan with gifts from the Ethiopian kings to the sultans. Ethiopian/Egyptian relations in this period are attested by the presence of Ethiopian monks in Coptic monasteries and are also observable in the monuments:58 the painting and decorations of medieval Lasta churches provide many examples of encounters between Ethiopians and other Oriental Christians – mostly Coptic and perhaps Byzantines.59 Recently acquired evidence of imported cedar woods and painting minerals such as lapis lazuli – presumably via Egypt – at the Church of Yimrhana Kristos gives insights on the nature of some of the goods brought from abroad.60 Finally, Ethiopian pilgrims and communities are attested in the Holy Land,61 also during the Crusades period, when Ethiopian kings, mainly for economic and political reasons, preferred to maintain their relations with the Islam-dominated Middle East; monasteries, plots of land and other important places in Jerusalem were donated to Ethiopians.62 In 1270 the Zagwe dynasty came to an end, and the political power was transferred back to “semitic-speaking” people. This represented a further stage in the continuing southward shift of the power centre.

36Apart for some of the tomb stones found at Dahlak Kebir, archaeological evidence on the maritime networks for the period from the 10th‑11th centuries to the 16th century seem to be very limited in the maritime space of the northern Horn of Africa. Despite that, textual evidence suggests that there was an active maritime trade network and relationships among Egypt and the southern Red Sea which involved both Egyptian, Arab and Jews merchants. Some aspects of this period commercial interactions in the Red Sea, and the role of the Dahlak Archipelago, have been outlined by Bramoullé63 and Margariti64 on the basis of textual evidence.

37A phase of network fluidity, among the coast, lowland and highland, in the northern Horn of Africa, is suggested for this period, with areas subject to “semitic-speaking” authorities being in a state of constant flux and no evidence of a permanent capital city until the 16th century, when Gondar progressively emerged as a commercial hub and political centre.

Changing network routes: the 16th and 17th centuries

38After a long period of hiatus in the archaeological evidence, it seems that the connections among the coast and the inland through the south-western route were revived in the 16th and 17th centuries, probably in part as a consequence of the Portuguese and Ottoman presence in the Red Sea and the related revival of the trading routes towards the eastern Indian Ocean.

39Starting from 1506‑1507, in order to close off the Mameluke sultanate from the trade to and from the Indian Ocean, a Portuguese fleet under the command of Tristão da Cunha and Afonso de Albuquerque aimed at blocking the entrance to the Red Sea. The armada set sail to Cape Guardafui, went to Socotra, and then proceeded to Aden. After an unsuccessful attempt to seize the town, Albuquerque sailed towards the Red Sea, the fleet set sail and went to some islands near Kamaran, and then to Dahlak Kebir and Massawa. The aim of the Portuguese to control the Red Sea entrance through Dahlak and Massawa was accomplished in 1520 but it ended in 1557 with the Ottoman conquest of Massawa. Towards the end of the 16th century, the Portuguese opened the maritime route around the Cape of Good Hope, thus enabling the Portuguese empire’s expansion into South and Southeast Asia. This event led to the contraction of the trade routes along the Red Sea route.65

Gondar

40In this period, Gondar became the new capital and trade centre for a long period of time (and with a strong Portuguese influence) after many centuries of Ethiopia lacking a permanent capital city. Geographically, Gondar was located at the center of commercial activity: situated about 35 km north of Lake Tana, Gondar was on, or close to, the point at which the main caravan track going from Metemma to Tana joined the route from Ennarya to the Red Sea (fig. 7).

41Since at least the 14th century, Gondar acted as a trading center along this route, and was subsequently designated as a small market town inhabited by Muslim traders. In the 17th century, Muslim traders made Gondar the major emporium of northern Ethiopia, importing foreign wares in exchange for civet, coffee, gold, hides and slaves. Salt bars served as the medium of exchange. Trade and commerce were the economic mainstays of the city, underpinning the caravan trade going north to the Red Sea coast, west to Metemma, and south to Ennarya. Later on, in the 17th century, King Fasilidas selected Gondar as his capital city, with the indubitable intention of directly supervising customs collection on the Red Sea, Tana, and Ennarya trade routes.

Irafali

42In 2014, a team from the Massawa Museum identified a coastal site in the south of the Gulf of Zula (4 km to the north of the present-day village of Irafali and 14 km to the south of Adulis).66 The site extends ca 1,000×500 m west-east from the Massawa-Assab road to the coastline, in the Bay of Irafali, in the southern part of the Gulf of Zula (fig. 7). This is an open bay enclosed by coral reef, and is marked by the presence of the Dola Vulcan. The area is rich in hot water sources and there are many nearby wells (fig. 8).67 On the opposite side of the road, a long and extended stone wall runs perpendicular to the modern road. Several large rectangular stone structures are visible on the site surface; a large circular structure is close to the road, has white pebbles or quartz on top of it, and is likely a burial site (fig. 9a, 9b).

Fig. 8 – The site of Irafali in the Gulf of Zula, extending ca 1,000×500 m (C. Zazzaro/University of Naples “L’Orientale”).

Fig. 9a – The site of Irafali, a circular structure with quartz (C. Zazzaro/University of Naples “L’Orientale”).

Fig. 9b – The site of Irafali, squared stone structures (C. Zazzaro/University of Naples “L’Orientale”).

43Several potsherds were observed on the surface, as well as colored opaque glass bracelets and semi-precious gems. The pottery assemblage included a large variety of local and imported wares, including green underglaze potsherds from Iran and Chinese porcelain. Preliminary pottery analysis dates to the 16th century, suggesting the site may correspond to the ancient settlement of Irafali (fig. 10a, 10b).

Fig. 10a – The surface pottery collection from Irafali, Chinese porcelain (C. Zazzaro/University of Naples “L’Orientale”).

Fig. 10b – The surface pottery collection from Irafali, Iranian glassware (C. Zazzaro/University of Naples “L’Orientale”).

44The name of this port was recorded in maps since the 15th century and in literary sources since the 17th century. According to historical sources, Irafali was an important trade center that was linked to the inland by two main caravan routes, one going to the Afar lands and the other to the Akkele Guzai and to Tigray. The main trade goods were salt, grain, cattle, and luxurious goods (such as tobacco, sugar and textiles). Between 1635 and 1638, Jesuit missionaries hid at and around Irafali.68 It seems that the port lost its importance in the 19th century, and was replaced by other commercial centers like Zula, Massawa, and other Afar ports along the Red Sea coast.

Conclusions

45Investigations along the Mai Negus/Haselo valley, south-west of Aksum, demonstrate that the involvement of various sites in the trade network linking the coastal regions of present-day Eritrea with the internal areas of the northern Horn of Africa predates both the ascent of Adulis as a commercial hub and the rise of the Indo-Roman trade. Similarly, recent researches and observations indicate that a coastal, lowland and highland network resumed long after Aksum’s decline, reorganizing itself in response to the fluctuating importance of the involved centres, e.g. Gondar and Irafali in the 17th century.

46Although archaeological evidence is still scanty, for the period after Aksum’s decline until the modern period, the combination – when possible – of material evidence and literary sources is providing new and stimulating insights in the reconstruction and analysis of this dynamic network. Based on a pattern of natural corridors through the coast, lowland and highland, this network continuously readapted to changing ecological, social, economic and political conditions. These affected the nature and distribution of required goods and natural resources as well as the weight and location of major nodes and hubs. The archaeological record, for its part, helps illuminating those periods for which historical sources are not available, making it possible to investigate the inherent fluidity of this network in a “longue durée” perspective. Future research in this direction will contribute to a better understanding of this complex historical dynamic.