- 1 We write: “Kwiha” (in Tigrinya) and not “Qwiha”, an italian transcription, following the Latin one, (...)
1Kwiha1, located some 10 km east of Mekelle (Tigray), shows discontinuous human occupation starting from iiird millennium throughout Aksumite periods to Ethiopian modern time. Archaeological evidence suggests that a rock shelter in Kwiha used to be the place where obsidian lithics and later ceramics were produced (Barnett 1999). After that, Kwiha and its surroundings were occupied during the Aksumite period and possibly also during the Pre-Aksumite period. To the northeast of the site, a rectangular stone building with rows of pillars is locally believed to have been destroyed by imam Ahmed ibn Ibrahim al-Ghazi (1507-1543) during his campaigns in 1533 against the Christian highland state of Ethiopia.
- 2 Administratively, the entire area formerly belonged to Wegre Hariba, a chief place and seat of the (...)
2Because Kwiha is situated on the route leading to the Afar Depression with its salt mines, it was probably a trading centre from ancient times and throughout the medieval period (Smidt 2007, 2011). There were Muslim trading communities living side by side with the Christian community. Travelers and scholars have found a number of inscriptions in Arabic in Kwiha, more precisely in the settlement of Wegre Hariba2, dating back to the 10th-13th centuries (Smidt 2011).
- 3 Breton J.-F. (2015) – Kwiha (Mekelle Area) Preliminary Report (January 2015). Mekelle, Mekelle Univ (...)
3The aim of the study was an evaluation of Kwiha’s territory, with its natural resources and land use3 (Breton & Aytenew 2017: 46-49). Would it have been occupied during Aksumite period (50 BC-550 AD) only or later during early medieval times? Would there be a break or a discontinuity of occupation? Would there be some evidence of any pre-Aksumite occupation (450-50 BC)? These are some preliminary questions that the surface pottery collection tried to solve. This paper should be anyway considered as a preliminary study.
- 4 Belete or Bilet should be near (North-East of) Kwiha (which did not exist at that time).
4Francisco Alvares, entering Tigray in ca. 1524, first visited Wuqro (misspelled as “Agroo”), then Agula’ situated on the Eastern Tigrayan trade route. Following the track southwards, he mentions a place named Belete (or Bilet)4 on his way to Wegre Hariba where he met the governor of Inderta (Alderley 1881: 97).
5One of the earliest scholars to take an interest in the archaeological ruins of Kwiha was the British settler Nathaniel Pearce who lived in Ethiopia from 1809 to 1819. During his journeys in Tigray, he visited Kwiha (spelled “Quened” by him). “Two days after our arrival at Woger Arreva, which is situated on the top of a mountain that formed the boundary of Enderta, in the Telfain, he took me to Quened, having several men with us with instruments for digging. Quened is a small village, on each side of a swamp, full of springs, which formed themselves into a brook that runs inter the river Dola. A vast number of willows, called in the country Queha, whence it takes its name, grow in all part of the swamp” (Pearce 1938:121-122).
6N. Pearce was also taken to see the ruins where he found broken “obelisks” as well large stones curiously cut like those in Aksum. He also found stones with inscriptions in Arabic that the locals had dug out from the ground in order to show them to him (Ibid.: 125). On the northern edge of the town are ruins, including a set of monolithic pillars possibly coming from a destroyed basilica. There is also a local oral tradition that the ruins were part of a church dedicated to Saint Cyriakus (in Tigray known as Cherqos, in Geez Qirqos), thus called Inda Cherqos, and that it was destroyed by Imam Ahmad bin Ibrahim in the 16th century. The ancient remains of Kwiha have also been documented later by several scholars such as Conti Rossini (Conti Rossini 1937) and Mordini (Mordini 1944-1945: 150) in the 1930s and 1940s.
- 5 Quite evidently the place “Belete” already mentioned by Alvares.
7In 1937, C. Conti Rossini found remains of stone ruins at the locality called Bilet5 in the north-eastern part of Kwiha. Among these ruins, he discovered four steles with Arabic inscriptions. Photographs of the four steles were studied by another Italian scholar, Pansera (Pansera 1945: 3-6), who published the contents of the inscriptions.
- 6 Breton J.-F. (2015) – Kwiha (Mekelle Area) Preliminary Report (January 2015). Mekelle, Mekelle Univ (...)
- 7 A protocol of Amendment to the Memorandum of Understanding regarding the cooperation within the Kwi (...)
8Mekelle University has started a long-term program of surveys at Kwiha, as part of an Ethiopian-led interdisciplinary research project in archaeology, epigraphy and ethnohistory. From 2014 to 2017, scholars from the Department of History and Heritage Management (MU) surveyed the site and collected a large amount of pottery, with Jean-François Breton as the archaeological field director appointed by the responsible Department of History and Heritage Management6 (Breton & Aytenew 2017; Breton 2019). The project serves as a field school within the newly founded PhD Programme in History and Cultural Studies. This paper deals with the preliminary assessment of the surveys and the pottery7.
9In a rock shelter excavated in the 1940s by F. Moysey, “the assemblage comprised ceramics, flaked and ground lithics and faunal remains” (Barnett 1999: 128). The rock shelter at Kwiha contains an interesting series which can be dated to the Late Stone Age, based on an assessment of the pottery and the typology of the stone industry. Following Moysey’s excavation, the material was taken to Nairobi, where it is currently stored in the foreign collections of the National Museum of Kenya. It was recently studied by Barnett (Barnett 1999). She could classify a Level 4: spanning from the 6th millennium to the 5rd millennium BP; a Level 3: spanning from the late 5th millennium to the early 3rd millennium BP; and finally a Level 2 and top of Level 3: datable to the Pre-Aksumite period, ca. 3rd century AD (Ibid.: 137).
- 8 There is however a doubt on the evidence of an occupation of the site during the 3rd/2nd millennium (...)
10The assemblage consists predominantly of microliths, but contains large utilized blades, burins and scrapers reminiscent of the Hargeisan culture of the northern Somali plateau (Ibid.: 137-138). The ceramics found at the site consist of reddish non-decorated pots which were in use in the prehistoric period (Ibid.: 128-137)8.
11The ancient site of Kwiha occupies a rocky spur, at an altitude of 2,200 m, dominating wide areas to the North, and May Bandéra River (refering to Tgn. bandéra, derived from Italian: Flag River) to the East. This permanent water resource strengthens emphasizes the central position of the site and gardens are still cultivated in its narrow valley (fig. 1, 2).
Figure 1 – Map of the site of Kwiha with areas of surveys
Drawing: J.-F. Breton and Y. A. Aytenew ; graphic computer: R. Douaud
Figure 2 – Map of the center of Kwiha with areas of surveys
Drawing: J.-F. Breton and Y. A. Aytenew ; graphic computer: R. Douaud
- 9 Oral tradition mentions that a church was constructed here by atsé Made Tsiyon (1314-1344), while o (...)
12In the northern corner there stands the compound of the church of Inda Cherqos (or Inda Qirqos) with its cemetery9. The modern church under construction, built directly over the ancient mud-brick one, is a landmark visible from afar. All around the church there are remains of the ancient settlement of Kwiha.
13Some 300 m east of the Inda Cherqos Church lie the ruins of a monumental building (Building 1: fig. 3) some 13.70 m long by 7.5 m wide (13°29’198 N, 39°32’929 E; altitude: 2,192 m) (Godet 1977: 44; Smidt 2004: 268). The plan was drawn in October 2014 with Yohannes Aytenew after cleaning of the whole eastern part (Breton sous presse). These remains consist of large limestone blocks lying on the ground, which can be readily identified as stone bases, pillars with or without capitals and separate capitals. From this plan, one can hypothetically reconstruct a three-aisles building, orientated east to west.
Figure 3 – Building 1: seen from the east
© J.-F. Breton 2017
- 10 We were told that some pillars had been taken to Mekelle in the 1960s.
14Some 100 m west of the new church of Inda Cherqos, currently under construction, three monolithic pillars are lying (Building 3). Two of them are the same size: 2.40 m (by 0.50 m by 0.50 m), while the third one is 1.88 m (by 0.50 m by 0.50 m). Because of their similarities, these pillars may stem from the same building10.
15Between 2014 and 2015, different walls were exposed on the eastern side of the Kwiha site as result of terrace building by famers and of road building. South-east of the Inda Cherqos church, a deep hole was recently dug in order to build a burial vault; this showed evidence of stratified layers of occupation and walls built in rubble stones.
- 11 All the inscribed stones were found at Bilet, the plain directly below the Inda Cherqos church, and (...)
16A large number of inscribed tomb stones testify to the occupation of Kwiha and its surroundings during the early Islamic Period11. In addition to the inscriptions discovered by Conti Rossini, twelve more were found in 1962 and taken to the National Museum in Addis Ababa. In the 1960s renewed research on the inscriptions was carried out by M. Schneider (1967). She tried to compare the Arabic writing styles in the inscriptions with steles from the Dahlak Islands (Schneider 1983a).
- 12 Muluberhan Adane (2016) – About Kwiha cemetery. Mekelle, Department of History and Heritage Managme (...)
17Later, more Islamic inscriptions were discovered (Tekle Hagos et al. 2013)12. These inscriptions should be integrated into the medieval context of an Islamic community trading along commercial routes. The most important route, the eastern Tigrayan trade route running south to north, links Kwiha north to Wuqro and Negash. Another inscription, probably from one of the earliest mosques of the region, published by W. Smidt and approximately dated to the 10th century based on paleography (Smidt 2009, 2011), was kept in the church of Wuqro for a long time. In turn, Negash is well known for its former Muslim community, dating back to the 7th century according to Muslim tradition, and also displays more recent Arabic inscriptions (Smidt & Haggag Rashidy 2012).
18Kwiha village faces north, crowning a group of hills at a general altitude of 2,250 m. The crest is totally occupied by Kwiha town with its market, municipality, churches, hospital and schools initially developed during the Italian occupation. The old western villages on the top of the limestone outcrops are now connected to Kwiha center by modern roads forming a continuous line of dwellings.
19All these hills dominate, to the north, the gently sloping plain irrigated by three rivers, joining at different points (fig. 4).
Figure 4 – The gardens by river Gembela
© J.-F Breton 2017
20Stretching from east to west, there are about five main hills (fig. 1).
21- The most easterly hill (n° 1; fig. 1) dominate the new city of Kwiha at an altitude of 2,275-2,285 m. As some cliffs could have been strongholds, such as Imba Feqadu, a survey was conducted in 2015, but no fortifications were recorded and no significant pottery was collected (Area 18).
22- Below Imba Feqadu, to the north, is hill n° 2 (fig. 1), reaching 2,230-2,240 m, and ending on its western side in a steep cliff over the river May Bandéra. The slopes and the top of this hill are covered with terraced fields and small rocky spurs but as a large military camp occupies most of its south-western part, survey is impossible.
23East of the old asphalt road Mekelle-May Mekden-Wuqro, hill n° 2, ends by a steep cliff over this road. Pottery is abundant just on the crest (near the TV antenna) with a high density of obsidian (Area 11). On its south side, there is an ancient building (probably Aksumite), rectangular and made of flat stone, ca. 6 m by 7 m and preserved up to 2 m high (13°29’118 N, 39°32’990 E; altitude: 2,218 m). Many potsherds have been recorded nearby (Area 12). So, it seems that the crest east of the road was a densely occupied area during Aksumite period.
- 13 Slopes were probably covered by trees in Antiquity as on many sites in Aksumite territories.
24To the north, the slopes becoming more and more gentle, are now occupied by agricultural terraces (Areas 16 and 22)13. In the plain, the lowest part being called “Bilet” (Schneider 1983b: 113), partly occupied by new plants, there is an Islamic cemetery. Preliminary excavations in the cemetery were conducted by J. Loiseau in 2018. All around, potsherds are few, but bricks were found here and there.
25- The central hill (n° 3; fig. 2) is separated by a rather deep gorge in which flows the Bandéra River whose source is located near the Elementary school. The old asphalt road Mekelle-May Mekden-Wuqro, probably constructed in the 1930s, runs through this gorge. A large part of the hill is occupied by houses, small fields and eucalyptus. On its north corner stands the compound of the church of Inda Cherqos with its cemetery; this modern church, still under construction and built on top of the ancient one, is a landmark (altitude: 2,197 m) visible from afar. All around the church are remains of the antique settlement of Kwiha (see above) where pottery was collected (Areas 1 to 6).
26The deforested northern flanks of this hill show a terrace, roughly following the contour lines, used nowadays for dry-farming agriculture and animal grazing (Areas 7-9). One building of about 6 by 7 m made of massive granite boulders (13°29’385 N, 39°32’656 E, altitude: 2,170 m) may belong to 2nd millennium, but no pottery is associated (Area 10). To the east, along May Bandéra river, there is one highly concentrated zone of pottery (13°29’532 N, 39°32’779 E) but no trench pit was possible (Area 10).
27- Hill n° 4 stands above a spring whose scarce water is used for gardens below. All this area is covered with an eucalyptus forest and gardens are irrigated by a small watercourse whose source, May Ayni, is located at an altitude of 2,218 m (Area 15). South of Area 15, another river flows, whose water is also used for vegetable crops. Close to this river, abundant ceramics exist in some places (13°48’81° N, 39°54’258 E) but no ancient building is visible.
28In the west, a ravine, where a small seasonal river flows, separates hills n° 4 and 5. There again, gardens are laid out around water canals, from 2,210 m down to 2,180 m. This river joins May Bandéra River some 500 meters to the north.
29- Hill n° 5 (fig. 1) has a flat rocky top with abrupt slopes all around but is now fenced and survey is not allowed.
30The May Bandéra River flows north-west in a small gorge cut into alluvium and, at some places, into limestone. Small-sized pebble and stone walls, probably recently built, perpendicular to the river course, are intended to reduce the effects of floods. At an altitude of 2,125 m, May Bandéra joins Dollo River, flowing from east to west. Before reaching the junction, the Dollo River flows through gardens and this is probably the largest cultivated zone of all the Kwiha neighborhoods. These are certainly the “swamps full of springs” recorded (Area 13) by N. Pearce (1938).
31Further down the river (from now on named Gembela or Dora’), small gardens are also to be seen along the eastern side and limestone is gradually replaced by schist and granite. Small waterfalls alternate with pools. A long canal supplies water to a large agricultural area (altitude: 2,115 m) and some meters below, another canal on the west side irrigates more gardens (Area 20; fig. 4). Collected pottery – even in small quantities – demonstrate that these gardens were used in late Aksumite period, and maybe later.
32In the valley of the Gembela, the altitude of 2,110 m seems to have been the ancient limit of Kwiha’s irrigated territory. Further west is the Inda Gabir village where no pottery was ever collected, so the eastern limit of the ancient Kwiha territory was probably located there. Gardens continue down the valley to Elala (Area 21), now one of the northern suburbs of the urban area of Mekelle. Both sides of the river are dominated by rocky spurs deprived of trees, and irrigation canals are still used for watering the gardens.
- 14 Inventory, drawings and photographs were carried out at the Department of History (Mekelle Universi (...)
33This preliminary report is based upon the substantial pottery finds from the 2015 to 2017 surface collections. In each area pottery was systematically collected. The collection comprises more than 250 potsherds with specific forms and respective decorations14.
34As there are no complete vessels, the corpus is arranged according to four main locally made wares, each named based on its diameters or its shapes (cf. annex). Some of them seem to fall into reasonably discrete geographical and chronological groupings in Tigray. The material is clearly derivated of the shape of the Aksumite ceramic tradition.
35- Open and closed shapes have been classified only according to their diameters because no complete form is available: between 30 and 50 cm large (fig. 5), between 25 and 30 cm large (fig. 6). Small bowls are between 15 and 20 cm in diameter (fig. 7).
Figure 5 – Large basins, jars and bowls (30-50 cm in diameter)
Graphic computer: R. Douaud
Figure 6 – Large jars and bowls (25-30 cm in diameter)
Graphic computer: R. Douaud
Figure 7 – Small bowls and basins (15-20 cm in diameter)
Graphic computer: R. Douaud
36- Ledge-rim basins: There is a small range of decorations on the wide-ledge-rims basins. The rounded-edge everted rim with a reaching lip may be among the later rims (late Aksumite). (Munro-Hay 1989: 252). Variations of the basic idea of incised lines joining to a central impressed hole (fig. 8). Chevrons or incised zigzags do not occur.
Figure 8 – Ledge-rim basins
Graphic computer: Rozeen Douaud
37- Hole mouth: rare (QW-2-5; fig. 5).
38- Small pots: few (QW-1-3; fig. 10).
39- Handles or grips are considered for their incised lines only (fig. 9).
Figure 9 – Handles and grips
Graphic computer: R. Douaud
Figure 10 – Bases
Graphic computer: R. Douaud
40As it was not possible to employ microscope or thin sections on site, and also no detailed physical analysis has yet been undertaken, the fabric could not be considered as a criterion. The coarse-wares of uniform type, being hand-made with brown and orange slip on both faces, the body being red clay tempered with crushed limestone or decayed basalt. There were very few fine wares. Fine orange (or orange reddish) clay appears only in Area 2 (QW-2-5) and 13, lithic black clay in Area 5 QW-5-1; fig. 11), grey clay ware (QW-1-4). Burnishing is generally brown or red (Area 16). Decoration comprises mainly crossed lines and finger prints. There was one image of a possible animal on a potsherd (QW-7-1; fig.11).
Figure 11 – Decors and specimens
Graphic computer: R. Douaud
41Without any excavation having been undertaken so far, the surface collection of pottery only allows some preliminary conclusions. It seems that most of the potsherds belong to the Middle and Late Aksumite periods.
42Fine and common red wares, almost exclusively found in the oldest levels (Proto and Early Aksumite) of numerous sites, such as Wakarida, were not found during the survey. Therefore, it is probably during the Middle and Late Aksumite period that Kwiha experienced its greatest prosperity and the maximum extension of its territory.
43Obviously, pottery has been collected in every plot of Kwiha territory. As all surveyed areas (n° 1 to 22) display pottery in various percentages: on hill n° 2, the Areas 11 and 12 display one of the highest surface pottery densities (Middle/Classical Aksumite periods), it is probable that all of these areas were occupied during the Middle and/or Late Aksumite periods.
- 15 See comparisons with Wakarida landscape and land use (Gajda et al. 2015: 211-216) and Aksum landsca (...)
- 16 But the extent of forest on these hills during Aksumite period remains largely unknown.
44Therefore two types of agriculture may have probably coexisted: dry-farming and gardens15. Plough cultivation may have been practiced on the slopes of hills 3, 4 and 516, down to the junction of the two rivers May Bandéra and Dollo. Irrigated perimeters appear in two main places: firstly, on the high slopes between Hills 3 and 4, using permanent or seasonal waters from the springs and, secondly, in the lower course of Dollo River, after it joins May Bandéra River, and then further down in the floor of the Gembela valley.
45The relatively small amount of Late Aksumite pottery would suggest difficult times. A decline of demographic pressure, as a result of environmental degradation, can, however, not be confirmed due to the lack of geo-environmental studies. Late Aksumite pottery is mainly attested on the top of hill n° 3 (Areas 1-6), suggesting that the peripheral zones were less densely occupied, and that this hill remained the core of the settlement.
46To summarise, most of the pottery from Kwiha can be broadly placed in a time span from the end of the Middle Aksumite period up to the post-Aksumite period. As Mifsas Bahri can be mainly placed into the late Aksumite Period (550-700 AD; Gaudiello & Yule 2017: 272-273), Kwiha may represent a link between Wakarida, Wuqro and Lake Hashenge during this period and possibly during the Middle Aksumite period. Therefore it should be considered that the late sphere of Aksum extended far south of the capital.
Figure 12 – Potsherds decorated (a) with wavy lines (QW-11-4) and (b) with incised triangles (QW-11-3)
© J.-F. Breton
Flux RSS
