Aksum An African Civilisation of Late Antiquity Stuart Munro-Hay

After climbing the steep valleys, such as that of the river Haddas which led from the 
coastal plain where Adulis was situated up to the highlands, the town of Koloë, 
mentioned in the Periplus (Huntingford 1980: 20; Casson 1989: 53) was reached. Koloë 
town and Maste town were also noted by Ptolemy (Stevenson 1932: 108) as among the 
towns remote from the river in the interior. Koloë derived its importance from its position 
as the first inland market where ivory could be obtained (see also 
Ch. 8: 4
). It is possibly 
to be identified with the present-day Matara in southern Eritrea. At this site the French 
archaeologist Francis Anfray, in a spectacular series of excavations (Anfray 1963, 1967, 
1974; Anfray and Annequin 1965), found numerous large and splendid mansions 
surrounded by their dependencies, together with churches, tombs, and even some 
domestic buildings in humbler residential areas. With the structures he found the material 
remains of a very sophisticated way of life. The town's history, as revealed by the 
excavations, extends back into the pre-Aksumite period, though so far this earlier 
archaeological stratum has only been accessible in extremely restricted areas of the site, 
and has not therefore been thoroughly investigated. Another town, Qohayto, further to the 
north, which also has the ruins of impressive structures (Littmann 1913: II, 148ff), but 
which has not yet been excavated, might be another candidate for identification with 
Koloë. It is most remarkable for the dam, made of seven courses of dressed stone stepped 
back in typical Aksumite fashion, which still retains water after the rains.  
From Koloë the route continued to Aksum and beyond; but although its southern 
extension led past many Aksumite communities, relatively few sites have so far been 
identified between Aksum and this main north-south route, and even fewer west of the 
capital. The eastern highlands, in contrast, contain the ruins of numerous towns and 
villages, and it is evident that this part of the Aksumite kingdom was a prosperous and 
populous region from pre-Aksumite days (Anfray 1973i: 20). Aksum itself, as is to be 
expected, lay within easy access of several villages whose produce was doubtless 
necessary to feed the capital's increasing population (Michels, in Kobishchanov 1979).  
The general homogeneity of the architecture and material goods of these ancient 
Ethiopian towns is apparent, and, despite regional differences in, for example, building 
stone and pottery types, the overall `Aksumite' nature of the civilisation is undeniable. 
The large mansions in the towns may have been the residences of sub-kings who had 
adopted the metropolitan style of living, or those of Aksumite governors and officials. No 
mansion has yet been identified as belonging to some local ruler known from the 
inscriptions, as for example the Agwezat kings who are mentioned in the time of Ezana 
and Kaleb (
Ch. 11: 5
), and information of such a specific character is likely to be very 
nearly impossible to obtain. But it is not impossible that the headquarters of the archon of 
Adulis, or that of the local controller at Matara, could be identified eventually. The 
mansion buildings at this latter place contained such symbols of wealth and authority as 
the Matara treasure (gold jewellery found in a bronze pot), an elaborate tomb, and 
apparently in one case the skeletons of prisoners still lying in their chains in basement 
oubliettes.  
Not very much is yet known about the settlement pattern of the Aksumite kingdom, but it 
has been possible to identify a number of particularly well-populated areas. The towns 

---

and villages along the main tracks south and east from Adulis towards Aksum, like 
Qohayto, Tekondo, Matara, Zala-Bet-Makeda, Ham, Etchmara, Gulo-Makeda, Haghero-
Deragweh, Yeha, Dergouah and Henzat, and those further south along the route west of 
the escarpment from Enda Maryam Tseyon Tehot and Maryam Kedih, or the branch via 
Anza, Hawzien and Degum, at least as far as Cherqos Agula and Nazret (Anfray 1970) 
must have developed along with the main trade and supply routes and at cross-routes 
leading into the interior. Such centres probably lay in areas of farming settlements, and 
acted as their market-outlet and exchange points, and some perhaps supplemented this 
activity with specific local products or, if suitably situated, could provide services 
connected with the movement of goods along the main routes.  
Aksumite settlements also appear to the west and north of Adulis, and the inscription of 
Sembrouthes from Daqqi Mahari, the buildings and coins from Arato (Piva 1907), and 
even traces as far north as Rora Laba and perhaps beyond, confirm that this region 
belonged to the Aksumite milieu. However, it is impossible to suggest what the limits of 
the Aksumite kingdom may have been at its zenith in view of the lack of archaeological 
evidence. Conti Rossini (1931) notes ruins in northern Eritrea with possible Aksumite 
affinities, particularly the thrones at Dicdic and the carved stelae at Rora Laba with lion 
and ox sculptures, and Anfray (1970) describes apparently Aksumite columns from 
Qeneda in Wollo, as well as paralleling, tentatively, the lion sculpture at Tchika-Beret in 
southern Wollo with the well-known Aksumite lion-headed water-spouts. Anfray's 
survey is the most informative we have so far, but more work is required to define the 
limits of Aksumite penetration.  
One feature often found in Aksumite town sites is the mansion-and-dependencies element 
(
Ch. 5: 4
). Such mansions were not only a feature of the towns, but also seem, at least in 
the case of Aksum itself, to have been distributed in their hinterland, perhaps representing 
local village centres surrounding landlord's houses. These élite residences are found in 
quantity, according to the survey by Joseph Michels undertaken in 1974 (Michels in 
Kobishchanov 1979; Michels 1986) in and around the capital and in the region of a 
number of villages as far as Yeha.  
The term `villa' for these mansion-and-dependencies groups is tempting (Anfray 1974: 
761), but they are evidently not all country and/or farm-houses like the Roman villas. The 
largest ones at Aksum and other towns could have had a different function from the 
smaller, more scattered country ones, although the general plan might have been common 
to both; even this is not yet properly confirmed, since a number of town mansions have 
been excavated but none of the remoter ones. Michel's published information tells us very 
little about the possible function of the larger and smaller élite residences in the Aksum 
area, and we can only guess how  to interpret them economically and socially. If we knew 
more about the chronological development of these mansions, we might be able to trace 
whether the type was developed at Aksum and moved from the city and royal context 
first to other towns and then was adopted as the model for the country mansions of a 
landowner class, a sequence which might be plausible. On the other hand, they may have 
originated elsewhere; in, for example, Adulis, with its greater exposure to foreign 
influences, and spread thence to Aksum and the countryside.  

---