Landscapes of Memory and Power: The Archaeology of a Forgotten Kingdom in Ethiopia

Abstract

Unlike their northern counterparts, the kingdoms of southern Ethiopia have received little attention by archaeologists. Their relatively late emergence and absence of literacy may explain this lack of interest. However, they have much to offer to better understand the history not only of the Horn but also of the precolonial African state more generally. In this paper, the polities that developed in the southern Ethiopian highlands during the second millennium AD are briefly described and then one of them is explored in more detail: the kingdom of Anfillo. An archaeological and historical overview of the polity is provided based on two seasons of fieldwork. It is argued that in Anfillo, as in other southern Ethiopian polities, a fortified landscape materialized at the same time a persistent situation of conflict and the collective memory of the ruling classes, which used it as a mnemonic device to tell history and legitimize social divisions.

Résumé

Contrairement à leurs homologues du nord, les royaumes du sud de l’Éthiopie ont été très peu étudiés par les archéologues. Leur émergence relativement tardive et l’absence d’écriture peuvent expliquer ce manque d’intérêt. Pourtant, ils ont beaucoup à offrir, non  seulement pour mieux comprendre l’histoire de la Corne de l’Afrique, mais aussi celle de l’état Africain précolonial. Cet article décrit brièvement les organisations politiques qui se sont développés sur les hautes plateaux du sud de l’Éthiopie au cours du deuxième millénaire de notre ère, avant d'étudier plus en détail l'une d'entre elles: le royaume d’Anfillo. On offre une vue d'ensemble de l’archéologie et l’histoire de cette entité sur la base de deux saisons de travail de terrain. Il est avancé qu'à Anfillo, comme dans d’autres royaumes du sud de l’Éthiopie, un paysage fortifié matérialisait au même temps une situation conflictuelle persistante et la mémoire collective des classes dirigeantes, qui l’utilisaient comme dispositif mnémonique pour raconter l'histoire et légitimer les divisions sociales.

Of Other States

The emergence and evolution of states have been one of the most debated topics in archaeology since the 1960s. In these debates, sub-Saharan Africa has played a negligible role for different reasons, including the relatively late date of its kingdoms, their lack of written texts, and their apparently anomalous character—when compared with those of Eurasia or the Americas. As Jeffrey Herbst (2000: 36) notes, “the operations of states before the Europeans are seen as too exotic to be relevant.” In this paper, I would like to explore the polities of southern Ethiopia and more specifically the little-known kingdom of Anfillo, to do my bit in showing that African states are indeed relevant to the wider debate in political anthropology and archaeology and particularly to discussions on the role of memory and landscape in the legitimation of authority.

Since the 1990s, archaeologists have demonstrated that the continent provides many examples of political complexity that challenge our assumptions of the state, at least when defined as a centralized polity with strong vertical hierarchies of wealth, power, and status. Elements that have been noted as key for the development of the early state, such as land control, rent extraction, wealth accumulation, and private property have been often absent in precolonial sub-Saharan polities (McIntosh, 1999: 6–8), while alternative political actors, such as age sets, secret societies, and ritual specialists, have had a leading role in curtailing (or replacing) monarchic power (McIntosh, 1999: 18). Thus, many African complex polities showed some traits that are more rarely found in the states usually discussed in political anthropology and archaeology, such as strong limitations to sovereignty, power balances between different collectives, the king’s role as a guarantor of fertility, the order of the cosmos and social reproduction, and the central role of symbols for social cohesion (Fortes & Evans-Pritchard, 2015 [1940]).

Yet, while decentralized, horizontally complex formations were common in precolonial Africa (McIntosh, 1999), we should not overlook the fact that polities did exist, “characterized by the intensification of social hierarchy, territorial expansion and integration, economic specialization, control over labor, long-distance exchange, and the promulgation of state ideologies” (Monroe, 2013: 21), that is, polities that are usually identified as states elsewhere. Some defend the use of the concept even for those organizations that shared features with tribal societies, including the political weight of lineages to counterpoise the ruler’s sovereignty and a strong ritual basis—i.e., Southall’s “segmentary state” (Southall, 1988).

Graeber and Sahlins (2017: 21–22), however, argued that the term “state” is not particularly helpful, as it does not tell much about its political organization or constitution. They surmise that the state is, at best, “a fortuitous confluence of elements of entirely heterogeneous origins (sovereignty, administration, a competitive political field, etc.) that came together in certain times and places, but that, nowadays, are very much in the process of once again drifting apart.” They propose to use the term “kingdom” instead for many of the premodern polities with which anthropologists and archaeologists have to deal. This has its own problems, as a kingdom could be Egypt, ruled by an absolute, divine monarch, and the Shilluk, whose “kings” did not have any real power.

Still, I will use the term “kingdom” rather than “state” to refer to the political formations of southern Ethiopia that are the object of the present article. I define kingdom as the territorial entity inhabited by a hierarchical society under an individual ruler (often hereditary). The southern Ethiopian polities did have kings and were indeed considered kingdoms by their neighbors, although from the point of view of evolutionary anthropology, it is unclear whether they could be defined as chiefdoms or states, as the former are often centralized polities with a hereditary ruler, populations in the thousands, a strong ideology, and a class system (Earle, 1987). The key element, however, is the unequal distribution of power and resources as compared to tribal societies—which is somewhat lost in the concept of kingdom as proposed by Graeber and Sahlins (2017). The contrast between segmentary or collectivist societies and class-based societies is perhaps more evident in sub-Saharan Africa, where socio-politically egalitarian and inegalitarian communities often coexist side by side. Examples of such coexistence include the societies of the Cameroonian and Nigerian highlands, and their lowland neighbors (MacEachern, 2018), the so-called ethnic mosaics of Kenya, with communities of nomadic pastoralists, hunter-gatherers, slash-and-burn cultivators, and city-states existing in close proximity (Kusimba et al., 2005) and the Ethiopian highlands, were pockets of acephalous societies existed within kingdoms (González-Ruibal, 2022).

An element that is especially relevant in sub-Saharan African kingdoms and particularly in the Horn is the role of landscape as a vehicle for political memory (see Insoll, 2015: 329–336). Landscapes of memory are of course not exclusive of either African or state societies (Mayorgas, 2019), but the lack of literacy among most sub-Saharan polities may explain the particular prominence that material symbols—including landscape features—played in telling history—and more specifically history legitimizing the ruling classes (Bollig, 2009). In the case of southern Ethiopia, landscapes of memory and power emerged that connected natural landmarks, sacred places, cemeteries, and historical seats of sovereignty (Arthur, K.W. et al., 2019). These landscapes often materialize journeys of the first dwellers or kings, and these journeys are performed ritually every year as processions—similar pilgrimages have been documented elsewhere in Africa (Insoll, 2015: 329–330). Thus, annual processions through landscapes of memory and their mnemotopes dramatize the status of the upper and lower classes, of natives and foreigners, and inscribe them in collective consciousness and in space.

In what follows, I will first briefly examine the polities of southern Ethiopia and then explore in more detail the kingdom of Anfillo, which existed as an independent political formation between the late sixteenth and late nineteenth centuries. I refer to Anfillo as a forgotten kingdom for two reasons: first, it has been forgotten, as other southern Ethiopian polities, in archaeological research, which has focused mostly on those of the north and east of the country, and on wider anthropological discussions on the nature of the precolonial African state. Second, due to a process of ethnogenesis that led to the transformation of the local Anfillo elites in Oromo as well as the interruption of oral tradition through schooling and cultural change, the memory of the kingdom is vanishing fast. Two seasons of fieldwork were conducted in the region in 2009 and 2010, which combined oral history, ethnoarchaeology, surface survey, and test pits. Archaeological work was complemented with the study of satellite imagery and GIS analyses. Regarding oral history, interviews were conducted with Tesfaye Tekaliñ in Oromo language with Oromo elders and with descendants of the old aristocratic families of the kingdom (called Busase), as well as with Boro elders, a group directly related to the Busase, now living in the region of Wenbera, some 300 km north of their original homeland in southern Ethiopia. Interviews took place in the towns of Dembidolo and Mugi (Kelam Welega Zone) and Begi (West Welega Zone), in the present-day Oromia region, and Wenbera (Metekel Zone), in Benishangul-Gumuz. As for ethnoarchaeology, we documented, among other things, pottery-making, distribution, and consumption in the Anfillo region and the bordering areas of West Welega Zones (Begi and K’ondala), which allowed us to characterize the diverse material assemblages existing today in the region and compare them with those documented archaeologically and in neighboring areas (e.g., Bula Sirika Wayessa, 2011). Further fieldwork was planned, but the conflicts that have ravaged western Oromia, where Anfillo is located, since the late 2010s, have prevented us from resuming our research.

Frontier Societies and Stranger Kings: The Kingdoms of Southern Ethiopia

Archaeological research on the early state in the Horn of Africa has focused on Eritrea and northern Ethiopia, which is where the earliest, most monumental, and most persistent polities emerged, the pre-Aksumite, Aksumite, and Solomonic kingdoms (Connah, 2015; Derat et al., 2021; Fattovich, 2019; Harrower & D’Andrea, 2014; Phillipson, 2012). These polities share many traits with classical states elsewhere (with which they were connected), such as literacy, an official religion, public monuments, and—in the case of Aksum—coinage. More recently, archaeologists have also been exploring the Islamic polities that developed in Wollo and Shewa in Ethiopia, Djibouti, Somaliland, and Puntland between the twelfth and sixteenth centuries (Chekroun et al., 2023; Fauvelle-Aymar & Hirsch, 2011; Insoll et al., 2021; Shidad Hussein, 2021; de Torres Rodríguez, 2022). These, again, fit comfortably conventional notions of the premodern state. There is, however, a third region which saw the emergence of kingdoms, but has been largely overlooked by archaeologists: southern Ethiopia.

From around the thirteenth century AD, a variety of polities appeared in the southern highlands of Ethiopia, which evince many of the traits usually associated with the early state, including a strongly stratified society, hereditary kings, a system of titles, and political offices, armies, taxes, slavery, serfdom, a religion with a priest class (Haberland, 1965, 1986), and even forms of protocurrency, such as cotton bundles, iron rods, and salt ingots (Wondu Argaw Yimam, 2020). Unlike in the northern highlands, the south was never unified. Several independent kingdoms existed until they were conquered by the armies of emperor Mïnilïk of Ethiopia during the last years of the nineteenth century. These kingdoms share several cultural traits, but one that allows us to classify them is language: they all speak North Omotic languages, which are divided into two branches: Ometo and Gonga. The Ometo languages are spoken in the area between the Rift Lakes and the upper Omo basin and include Maale, Gamo, Dawro, and Wolaytta. The Gonga speakers (Anfillo, Shekkacho, Boro, Kafa) are currently restricted to the region comprised between the headwaters of the Baro, Omo, and Alwero rivers, in southwest Ethiopia, with the exception of the Boro, living north of the Blue Nile. Originally, however, the Gonga occupied a much larger area comprising all of the present-day region of Welega (Oromia), that is, around 300 km further north than their present distribution (Fleming, 1984: 34–35). Indeed, the earliest Omotic kingdom to appear in literary records was Damot, just south of the Blue Nile, whose existence can perhaps be traced back to the tenth century AD (Huntingford, 1989: 69) and was a powerful polity by the thirteenth.

Here, the focus will be on the Gonga polities, which include Anfillo, Kafa, Hinnario, Bosha, Konch, Bizamo, and Boro (Fig. 1), although I will refer to other North Omotic traditions for comparison and contextualization. The study of the Omotic kingdoms has been undertaken by anthropologists and ethnohistorians, in a research tradition covering over a hundred years (Bieber, 1923; Haberland, 1965, 1986, 1993; Lange, 1982; Orent, 1970a, 1970b), Donham, 1994; Data De’a, 2000; Hailu Zeleke, 2007; Zegeye Woldemariam Ambo, 2021). Ethnographic and ethnohistorical work has been more recently complemented by archaeological research in southern Ethiopia. While the focus has been mostly on the emergence of agriculture and the domestication of animals (Hildebrand et al., 2010; Hildebrand & Brandt, 2010; Arthur, J.W. et al., 2019), investigations have also been conducted on the more recent history of the Gamo people, an Ometo-speaking group (Arthur, K.W. et al., 2009, 2010).

Fig. 1

Location of the main Gonga polities in Ethiopia

Nevertheless, our knowledge of the southern Ethiopian polities is still incomplete and has had little impact on discussions on the precolonial state in Africa not to say the premodern state more generally. This is due to different reasons, such as the absence of native literary records, the late emergence of the kingdoms, their secondary character (they were influenced by states in the north), and their lack of resemblance to more “classic” states, such as those of the Middle East, Egypt, and the northern Horn (Trigger, 2003).

The southern Ethiopian kingdoms have a common cultural background, but also a shared political tradition: they fit well into the African frontier society model delineated by Igor Kopytoff (1987). In frontier societies, a group (often elite) from a certain polity typically moves to the periphery of their polity for a diversity of reasons—internal rivalry, population pressure, famine, war, political aspirations—and establish an organization that replicates that of the homeland, often subduing the native population. Through time, the two polities begin to diverge, but the memory of the original plays an important role in the cultural identity of the new one, which retains symbols, traditions, rituals, and genealogies, while at the same time reworking them. Frontier societies, in fact, should not be understood as purely conservative phenomena, passive transmitters of the core culture, but rather as loci of cultural and political creativity, which hybridize elements from a diversity of traditions (Ogundiran, 2014: 5) and often influence the original society itself, as a source of institutional and material innovation (Reid, 2011: 23; Ogundiran, 2014: 21).

Frontier societies can be considered a variant of the wider stranger king formations (Graeber & Sahlins, 2017: 5–7), which have been documented both in Africa and elsewhere. As with frontier societies, stranger kingdoms are created through the arrival of heroic outsiders, usually described as “wise men” (Triulzi, 1981) or civilizing agents, who establish a new dynasty in an alien land. This usually takes place not through violence, but through marriage with local women, a moment that is commemorated in collective memory and in the landscape. It often coincides that stranger rulers are considered sacred, as is the case with many of the Omotic kings. The new polity is dual in nature, with the stranger rulers on one side and the subdued autochthonous population on the other. The ruling class is not created in the process of polity-building, but precedes it and makes the subject class. Native people, however, usually retain a “residual sovereignty,” which includes a role as ritual specialists (Graeber & Sahlins, 2017: 7).

How do these concepts apply to southern Ethiopia? From the early second millennium AD, the region was considered an open frontier of expansion for the Christian Kingdom of Ethiopia, which saw it as that “institutional vacuum,” “open to legitimate intrusion,” which Kopytoff (1987: 6, 9) identifies as characteristic of frontier societies. Thus, several of the southern polities have stories linking their kings and elite clans to individuals and communities coming from northern Ethiopia, such as Tigray or even Aksum (Lange, 1982: 19, 59; Wondu Argaw Yimam, 2020), or simply to a “white king,” Nech’ Taro (Grottanelli, 1940: 300; Haberland, 1983). In these stories, the newcomers manage to rise to power and establish new dynasties through different strategies, which include kinship alliances with local ruling families. According to oral traditions, Amhara or Tigrayan dynasties appeared from around 1425, yet indigenous lineages can be traced back at least two more centuries (Lange, 1982: 28–29; Appendix I), which means that kingdoms in the south actually predated the arrival of the stranger kings.

The arrival of migrants brought foreign political ideas, traditions, and symbols of power to the new polities. Thus, many key political terms denote an Amharic (northern Ethiopian) origin, such as royal hall, councilor, proclamation, tax, and royal drum (Haberland, 1965). Kingly objects themselves were often imported from the north, either directly, by northern elites traveling south, or as an idea that was appropriated by the southern polities. Such objects include the drum and the umbrella, which are identical to those that are known to exist in the northern plateau since the Middle Ages. Religion was likewise influenced by the north: thus, the most important festival in the kingdom of Kafa was Mashk’aro, which is the Ethiopian celebration of the discovery of the True Cross (Mäskäl) (Huntingford, 1955: 135). In other cases, the directionality may not be so obvious, especially regarding notions of kingship, which can belong to a shared cultural background: this is the case with the importance of symbols and their materialization in specific objects, origin myths, and lineages in both northern and southern royal traditions (Haberland, 1965).

Stranger kings were not only foreigners from the north. In typical frontier society fashion, the process of creation of polities continued from those already established. Thus, disgruntled or enterprising members of the ruling clans were behind the establishment of secondary Gonga kingdoms, as they decided to try their luck in the frontier and establish their own autonomous polities (Data De’a, 2000). This was the case with the Boro polities and the kingdom of Anfillo; although in these cases, pressure from the expanding Oromo had a lot to do. During the second half of the sixteenth century, the Oromo conquered some of the southern Ethiopian kingdoms and reduced the territory of others (Mohammed Hassen, 1994: 62–81). At this point, some of the ruling families decided to migrate. Members of the elite clan known as Busase or Bushasho moved to the west and established the kingdom of Anfillo; the Boro migrated north and created a new polity to the other side of the Blue Nile, in the mountains of Wenbera. Both the Anfillo and Boro polities in turn fragmented as some of their members founded new political entities: we have documented at least four different Boro groups north of the Blue Nile, whereas Anfillo is divided into three: Anfillo proper; Sheka, which was established south of Anfillo, in today’s Sheka Zone, around 1570 (Lange, 1982: 96); and another polity, for which we have been unable to find out the name and which was created around 1750 to the north of Anfillo, around the modern town of Begi (González-Ruibal, 2014: 252–253). The creation of a new polity is usually associated with a conflict between two or more brothers, which results in one of them, usually the younger, leaving the original polity and establishing a new one elsewhere (Data De’a, 2000: 167–169).

The stranger ruler-to-be found native populations in those territories to which they arrived. In the case of the Gonga, their western periphery (between the Didessa and the Dabus rivers) was inhabited by small-scale communities of slash-and-burn agriculturalists and hunter-gatherers, speaking so-called Mao languages, which are also Omotic (Fleming, 1984), and Nilo-Saharan languages. The groups that were subdued by the Gonga were incorporated into the polities as a subaltern group and generically termed Mao, Nao, Mawo, Manno, or Manjo (Huntingford, 1955: 136; Lange, 1982: 242, 260), and they still form marginalized minorities within Gonga societies (Freeman & Pankhurst, 2001; Pankhurst, 1999). Some of these communities were incorporated as an underclass of serfs to the Gonga chiefdoms; others remained mobile in the periphery, tied to the Gonga through master-client relations: they provided (and still provide) farmers with forest products (game, fish, wood, honey) (González-Ruibal, 2014: 306–320; Worku Derara Megenassa, 2019). Once they were conquered, the Mao communities were assigned to Gonga elites and had to work for their lords, serve in the army, and perform some jobs and rituals that were (still are) considered impure or vile (such as iron-making, tanning, and circumcision).

The foreign nature of the elites that founded the Gonga kingdoms gave rise to social systems that have been described as dual (Orent, 1970a), with an elite group and a subaltern one, often speaking different languages or dialects and with their own rituals, customs, and leaders. Thus, in every Gonga polity, there was a hierarchy of chiefs (regional, clan, and family chiefs), all subjected to a king from a specific clan of the dominant group. The name of the paramount ruler (taro or tato) is similar throughout the Gonga. Typical of sacred kingship, he had important ritual functions and was believed to be endowed with supernatural powers. The ruling group, from which the king originated, was considered to have a superior mythical origin that explained its role as dominator (Bieber, 1923: 495–497; Orent, 1970b: 271–273; Lange, 1982). Although the level of political inequality varied from polity to polity, in all cases, there were corporate institutions and other political mechanisms to curtail the power of the sovereign.

This is another feature that distinguishes the Gonga polities from the Christian Kingdom of Ethiopia: we could speak, following Blanton (1998), of a corporate political system for the Gonga, opposed to the exclusionary model that ruled in the north. In the southern kingdoms, although the throne was hereditary, succession was not immediate. There was usually a senate or assembly that included representatives from all ruling clans and whose members decided whether the candidate was fit for government or which of the king’s sons was best suited to rule. This often led to conflicts between royal candidates and probably explains the continuous fission of the Gonga and the creation of new polities, as described above. One of my informants (Kebbede Wagga, from Mugi) from the Busase elite asserted that if the king was not a good leader, he could be removed from power by the assembly, which was composed by representatives of the seven clans of the kingdom of Anfillo. The existence of an assembly of clan elders, who elected a new royal candidate after the death of the king, was also recorded by Italian anthropologist Vinigi Grottanelli in the 1930s (Grottanelli, 1940: 304). A similar situation is attested in other Gonga polities, such as Hinnario, Sheka, and Kafa, where there is an institution called mikrechcho formed by seven councilors who not only appointed all political offices, but could also remove the king from power (Lange, 1982: 132–143, 215–223). They had judicial, administrative, political, and economic functions. Their capacity to make decisions and support the interest of the elites makes the Gonga system more similar to an oligarchy than a monarchy with unrestrained powers. There are also recorded instances in which the people refused to fight for the king, because they considered him unworthy of respect (Lange, 1982: 109). We have to see in this a remnant of the collectivist background of the Gonga, long lost in the north. Thus, fission, senates, and the possibility of removing the ruler from office are all limitations to kingly power that are, at the same time, typical of segmentary states. Furthermore, the sacred nature of the sovereign can be considered yet another limitation to his power, as the kings were subjected to all kinds of ritual restrictions and taboos that severely diminished their agency—what Graeber and Sahlins (2017: 461) call “adverse sacralization.”

The History and Archaeology of Anfillo

The history of the Kingdom of Anfillo is similar to other Gonga polities. As noted above, the reason for the displacement of the ruling dynasty, the Busase, to the peripheries of the Gonga territory was probably the Oromo migration into the Gibe basin, where they used to live. The migration occurred around 1560–1570, which is when the Oromo arrived in the Gibe (Mohammed Hassen, 1994: 35–36). According to some of the Busase elders, their homeland was around the town of Jimma (which was historically part of the Kingdom of Kafa, until it was occupied by the Oromo in the late sixteenth century). Interestingly, however, a Busase informant mentioned “Damot” among the original clans that emigrated to the west. The Boro that I interviewed also assert that they originally lived in Damot south of the Blue Nile, before being expelled to the other side of the river by the Oromo migrations. Old maps (such as John Pinkerton’s Map of Abyssinia of 1814) actually situate a group of Gongas there.¹ Yet, at the same time, both Boro and Busase are mentioned as clans existing in Kafa still in the early twentieth century (Bieber, 1923: 53–55). Anfillo, Kafa, and Boro languages are strongly related (Bender, 1975: 139), meaning that their speakers have lived together in the relatively recent past. The crisis of the sixteenth century was acute, and there was probably much movement and disarray of both commoners and elites.

The historical context, in any case, offers us a post-quem date for the foundation of Anfillo. The new kingdom was in all likelihood established during the last quarter of the sixteenth century. The end is better known. After repeated conflict with the encroaching Oromo, it was finally annexed to Ethiopia by emperor Mïnilïk, during the last years of the nineteenth century (James, 1980; Negasso Gidada, 2001: 80, 180–181; 237–240).

Anfillo was created by the Gonga emigrants in an area of dense tropical forest in the westernmost edge of the southern Ethiopian highlands (Fig. 2). At the time, it was inhabited by small groups of slash-and-burn agriculturalists, the ancestors of the present Komo, Majangir, and Kwegu, all Nilo-Saharan speakers (Negasso Gidada, 2001: 86–88). These were either expelled to the lowlands, where they still live today, or were subjected and thus transformed into a serf class—Mao. The new lords established a typically Gonga political system, with a dominant class (Busase) and a subaltern class (Mao). Each group had their own clans and clan heads. Among the Mao, the paramount chief is the kedderaso under which there are several clan chiefs (nihó)—this is a good example of the “residual sovereignty” of those subjected by stranger kings (Graeber & Sahlins, 2017: 7). The equivalent among the Busase are the abeto. The king of all, Busase and Mao, is the taro (Grottanelli, 1940: 304).

Fig. 2

Approximate location of the Kingdom of Anfillo during its heyday (ca. 1600–1700 AD). For a detailed map showing the sites, see Fig. 10

While historical information on Anfillo has been available since the 1930s (Grottanelli, 1940; James, 1980; Negasso Gidada, 2001), the archaeology of the kingdom was non-existent until our first field season in the area. During 2009 and 2010, we surveyed seven hilltop sites (all called Gara, “mountain” or “hill” in Oromo) in the surroundings of Dembidolo, out of a total of 23 identified in satellite images (see “A Landscape of Conflict, Memory, and Power” section): Gara Gute, Gara K’esi, Gara Abba Saba, Gara Abba Bula, Gara Daro, Gara Dallo 1 and 2, and conducted test pits in two of them: Gara Gute and Gara Abba Bula (see Fig. 10 for locations). All sites were or had been recently plowed, and visibility was thus very good: pottery sherds and lithics were present in all seven sites and were easily spotted on the ground. Apart from the pottery, very different from the current Oromo tradition in the region, a radiocarbon sample from the central enclosure of Gara Gute (Beta 296118: 250 + / − 40 BP) corroborated its belonging to the Kingdom of Anfillo: it yielded a calibrated date of 1508–1927 cal. CE (2 σ) or 1525–1799 cal. CE (1 σ). We can discard the twentieth-century segment as we know for sure that the site was abandoned by then. We can most probably discard the sixteenth-century segment also (1525–1558), because it is a bit too early, if the Busase migrated westward as a result of the Oromo migrations. This leaves the most likely interval as 1631–1799 cal. CE.

Anfillo settlements are all very similar. They tend to be located in flat, volcanic hilltops, whose profiles are very recognizable from afar. Indeed, the most conspicuous hilltops are usually chosen, which are visible from 10 or more kilometers away. The visibility from the sites is high: they have 360° viewsheds and command views of over 20 km in many cases. Regarding their morphology, the topography of the hills has been strongly modified with stone-and-earth terraces, stone-reinforced parapets, berms, and ditches. They are disposed in a concentric manner, and there is always a combination of two or more structures (berm and ditch, ditch and terrace). The number of fortified lines varies: there are sites, such as Gara Abba Saba, where we have only been able to locate one simple enclosure and others where there are two, three, or more lines of parapets, including both terraces and proper defenses, sometimes reinforced by wide defensive ditches (Figs. 3 and 4). The size of the settlements is not always easy to calculate, because fortifications are extended by agricultural terraces, roads, and other structures, both contemporary with the fortifications and later. The size of the sites varies ostensibly, with recorded dimensions going from 1.8 to 115 ha (Fig. 5).

Fig. 3

Stone-lined parapets and terraces in the eastern slope of Gara Dallo, the main sanctuary of the kingdom

Fig. 4

The 4-m-wide ditch that surrounds the royal seat of Gara Daro, as found on the western side of the site. It blends into the Kotaa Diinaa, a 20-km-long ditch that protects the kingdom

Fig. 5

Size comparison of two fortified settlements in Anfillo

Although some sites are large, the organization of space inside was probably not very dense, if we can extrapolate what we know from the traditional organization of space in Anfillo, as recorded by Grottanelli (1940: 168–187), and in other southern Ethiopian societies (Bieber, 1923; Jensen, 1959). Inside the walls, the occupation pattern was probably quite loose, with extended family compounds (hypothetically including Busase members, serfs, and slaves) forming clusters inside the settlement. There were probably cultivations interspersed between and inside the compounds, as with contemporary Maale and Basketo (Ometo) (Jensen, 1959: Abb. 3–4). Archaeological evidence seems to corroborate this fact: in Gara Gute, the upper enclosure had a high density of finds, thus showing that there were many compounds there, but the two other terraces also yielded surface materials, and they are not too fragmented or eroded so as to be mistaken with domestic rubbish that made their way down to the limits of the site (mixed with manure or otherwise). The lower density indicates that the outer terrace had both cultivated fields, perhaps cattle yards, and some houses, and, in some cases, these extended beyond the fortified areas, based on the existence of archaeological materials.

The houses were probably similar to recent examples of Gonga architecture, which includes small round huts with thatched conical roofs for slaves, servants, and the lower classes (known as k’eto in Kafa; k’echo in Anfillo language); and larger round or oblong houses for the nobles (called kotemo and shakero, respectively, in Kafa), which were subdivided and had partition walls. The nobles in Kafa also had the herabo, a round hut with a large central space and a roof sustained by two poles that served as a reception hall (Bieber, 1923: 175–227). All Gonga vernacular architecture is made of wood, rope, and thatch and thus leaves no remains in the archaeological record. In addition, the nature of the soil in the area (dystric cambisols) and heavy rains efface any trace of post holes.

The only objects that have survived are those made of stone (grinding stones, flakes) and pottery, including cooking and serving vessels and a smoking pipe. A total of 152 diagnostic sherds were collected during the survey and in the two test pits—of which 86 were drawn—which allow for a preliminary characterization of the ceramic style of Anfillo. Fabrics are of a very high quality, well levigated, and with a very fine mineral temper (quartz and muscovite). They are either dark (brown, black, gray) or brick red in color, with dark interior. The surfaces are often burnished. Based on ethnographic parallels, a pebble or a piece of calabash was used for polishing (Arthur, 2006: 44; Bula Sirika Wayessa, 2011: 314–315). The decoration is abundant and, in the case of cooking and storing vessels, is predominately plastic (horizontal ribs, appliqués, lugs) to which simple impression, nail impression, or incision is applied. Molded necks are common—analogies can be found in other Omotic groups, such as the Aari (Haberland et al. 1959: Abb. 18) or the Gamo (Arthur, 2006: 37). In the case of fine wares, incision predominates, often very fine. There is, in fact, a marked division between cooking/storage wares and fine wares, which is revealing of the social distinctions that prevailed in Anfillo: there is no such division among the Nilo-Saharan tribal groups that lived in the surroundings of the kingdom, but the division exists, instead, in the kingdom of Ethiopia (de Torres Rodríguez, 2017: 230–232).

Regarding fine wares (Fig. 6), we have both small, thin-walled (0.5 cm or less) bowls with inverted rims, as well as globular shapes that may have been part of bottles. The few mouth diameters that could be reconstructed are between 10 and 13 cm. Fine ware was likely used to consume beer or other liquids (coffee?) and, in the case of bottles, honey wine. Fine wares (bottles, goblets, and the like) represent 26% (N = 29) of all diagnostic sherds whose function could be identified.

Fig. 6

Fine wares found during the survey (1–6, 13–18) and roulette-decorated pottery (7–12), probably acquired from Nilo-Saharan communities living in the periphery of Anfillo

As for cooking and storage vessels (Fig. 7), they constitute up to 71% (N = 81) of the assemblage of identifiable shapes, but cooking wares (bowls with inverted rims and open bowls) are only 12% (N = 14) of the total (Fig. 8). To this, we can add three lids (3%), which were probably used to cover cooking pots, as is the case today. They appeared in only one site, Gara Abba Saba.

Fig. 7

Reconstruction of the main types of storage and cooking wares of Anfillo, based on extant remains and ethnographic parallels

Fig. 8

Bowl of a smoking pipe (1), lids (2), and cooking wares (3–4)

The rest of the common pottery is made up of jars (59%, N = 67) of different sizes—their mouths being comprised between 12 and 24 cm. They have pointed rims, molded necks, ovoid bottoms, and thick walls, which were built using coils. These containers are similar in shape and technique to those still produced by potters in the southern Ethiopian Highlands, both Omotic and Cushitic (Arthur, 2006; Bula Sirika Wayessa, 2011; Cauliez et al., 2015), although the decoration differs. Jars are today used for transporting and storing water and for brewing and storing beer or honey wine, and they doubtless had similar uses in the past (Fig. 9). Pottery among the Gongas has traditionally been made by women, who belong to marginalized groups. This was probably also the case in Anfillo.

Fig. 9

Jars: A Gara Gute, B Gara Dallo, C Gara Abba Saba

We have documented several sherds (N = 8) with plaited-fiber and twisted-string roulette decoration (Fig. 6, n. 7–12). Roulette is not used by either Cushitic or Omotic peoples in southern Ethiopia. It is, instead, typical of the Nilotic peoples of South Sudan from the Iron Age (early-mid first millennium AD) to the present and some of the Koman peoples with which they are in contact (Komo, Opo, Majangir) and that are neighbors with Anfillo: it has been reported in sites just south of the kingdom, in the lowland region of Gambela, which have been radiocarbon-dated to ca. 1000–1200 cal. AD (González-Ruibal et al., 2014). In the same sites, pottery decorated with horizontal grooves has also appeared which has parallels in our area. Both types of pottery may indicate contacts between the people of Anfillo and Nilo-Saharans (Komo, Majangir?) living in their lowland periphery.

The ceramic style documented in Anfillo archaeological sites is quite different from the one existing today in the region, which is produced by female Oromo potters and is characterized by globular and hemispheric pots, open bowls, and large necked jars, all of them usually plain or with very little decoration (deep incision). Contemporary pottery-making in Anfillo is comparable to other western Oromo traditions in terms of shape, decorations, and chaine operatoires (Bula Sirika Wayessa, 2011; González-Ruibal, 2014: Fig. 5.10).

Other ceramic materials were found, including a smoking pipe bowl collected at the site of Gara Abba Saba. It is made of well-levigated fabric, with no mineral inclusions, and has a polished surface and a thickened, vertical rim. It lacks decoration, unlike most pipes known archaeologically and ethnographically in the neighboring Nilo-Saharan region. It looks like a local product. Interestingly, one of the earliest kings of Anfillo, Garo, was considered a heavy smoker (Grottanelli, 1940: 302) (Fig. 8, n.1).

Lithic tools are of flint and quartz. They may be related to hide working, an activity that was (and still is) very common in Omotic societies, which have specialized groups for these kinds of activities (Weedman, 2002). Stone (mostly obsidian) is still used for making scrapers among different south Ethiopian groups (Yonatan Sahle et al., 2012). Meaningfully, lithics appear mostly in the lower terraces of settlements: tanners (like ironsmiths and potters) have traditionally belonged to marginalized minorities in Gonga and Ometo cultures and cannot live in the same part of the village as the rest of the population (Freeman & Pankhurst, 2001; Haaland et al., 2004).

A Landscape of Conflict, Memory, and Power

Survey work on the ground was completed with the analysis of satellite photographs available through Google Earth, which has high-definition images for the area. A total of 23 fortified enclosures was recorded in an area of roughly triangular form with equal sides of around 20 km each (Fig. 10). This yields a density of one site per 7.5 km². However, this number is not necessarily very informative in demographic terms, as we surely have not identified all sites and we do not know whether the settlements were all occupied simultaneously, although some clusters certainly were, as we will see. Besides, there might be non-fortified settlements that we have not discovered yet. The area that we surveyed is the core of Anfillo and the last part of the polity to fall in Oromo hands, but originally, its territory extended to Mount Welel, as well as westward to the Mugi area, covering perhaps some 200 km², with the Gambela escarpment and the Birbir rivers serving as natural boundaries—significantly, on the right bank of the Birbir there is a Tullu Mao (Mountain of the Mao, the subaltern class of Anfillo).

Fig. 10

Sites identified during field survey or analysis of satellite photographs. Numbered sites indicate places mentioned in the text: (1) Mount Konki, (2) Gara Dallo, (3) Gara Daro, (4) Gara Yingi, (5) Gara Abba Bula, (6) Gara K’esi, (7) Gara Abba Saba, (8) Gara Gute

The fortification of the Anfillo landscape is explained by the elderly Oromo of the area as a defense against the constant Sayyoo Oromo menace to the east. In fact, the Sayyoo kept expanding between the seventeenth and eighteenth centuries and reducing the extension of Anfillo (Negasso Gidada, 2001). The mountainous region to the north was soon lost to the Sayyoo, probably already during the seventeenth century. The next great loss occurred during the mid-eighteenth century, when the Busase and the Sayyoo fought over the rights over a salt lake (hora) that was used by the cattle of both groups. The Busase lost and had to withdraw into the Anfillo forest (Negasso Gidada, 2001: 82–83). By the late nineteenth century, most of the present Dembidolo wereda, which is where most fortifications are located, was already in the hands of Oromo or Oromized Busase. The Kotaa Diinaa has to be put in relation with the Oromo pressure. This linear structure is said to encircle all the settlements in the Gara Yingi ridge (see below), and it is believed to stretch over 20 km. Its function, according to our informants, was to stop the Oromo cavalry. They translated to us Kotaa Diinaa as the “Enemy’s Ditch” and “Diinaa” is indeed “enemy” in Oromo, but Kote, Kot, or Kotano is the name of a mythical Gonga ruler whose kingdom extended from the Omo to Anfillo (Lange, 1982: 59–61) and Gonga people assign all kinds of infrastructures to this king—including roads, bridges, and ditches. Linear works similar to the Kotaa Diinaa, with analogous function, are known in other Omotic kingdoms, as we will see. During our survey, we could locate only some parts of it. In all likelihood, it has been reused by the dense network of paths and roads between farms that cover the area. An intensive survey work coupled with ethnohistorical research would be needed to trace the entire fortification.

The Busase elders that we interviewed insisted, in turn, in the internecine problems of their group. Kebbede Wagga mentioned the troubles that ensued the proclamation of the Tuma Gimbi Garo, the Laws of King Gimbi Garo. According to him, while some accepted the laws, others did not, and there was “conflict between father and son, one clan, and another clan.” A consequence of these fights was the split of the Busase community: one group remained in Anfillo and the other emigrated to the north, to the present Begi area, some 150 km north of Dembidolo (see González-Ruibal, 2014: 251–253). In general terms, the situation depicted by oral tradition is one of instability and conflict that tallies well with the fortified landscape documented archaeologically.

We analyzed the data from the survey and satellite images with GIS software. If we look at the total view shed of all the fortified sites (Fig. 11), we see what can be considered a plausible image of the territory of Anfillo as it was around the eighteenth century, prior to the second Oromo advance. Within the area of fortified sites, however, we can identify at least two clear clusters: a NW group and a SE one. The SE group is formed by five sites, including Gara Gute and Gara Abba Saba, which have already been mentioned and which have yielded abundant materials. These settlements have no oral traditions associated with Anfillo, as proved by the fact that they are linked to the Italian occupation of 1936–1941. This is consistent with an older moment of abandonment, in turn coherent with the occupation of this area by the Oromo during the second half of the eighteenth century or early nineteenth. The 14C date from Gara Gute also suggests an early abandonment.

Fig. 11

Cumulative viewshed of the Anfillo sites

The NW group is more interesting. This cluster, which is where all the sacred and royal sites are located, and the SE are not mutually visible (Fig. 12). The NW cluster is located along a ridge comprised between two elevations: Gara Yingi to the west and Gara Dallo and Gara Daro to the east.

Fig. 12

Viewsheds from the sacred lake of Dällo (A), Mount Konki (B), and the royal seat of Daro (C). Their visibilities cover the NW cluster. They do not see or are seen by any of the sites in the SE cluster

There is a rich oral tradition surrounding these sites: Gara Yingi (Fig. 13), one of the largest settlements of Anfillo, is considered to be the home of the first Mao, named Tokko, and of his clan; Gara Daro (Fig. 14) is the royal seat of the Busase, the settlement of the paramount king or taro; Dallo is a lake from which the first Busase (Goddi) and the first Mao (Tokko and Goshero) are said to have emerged (Fig. 15).

Fig. 13

Map of Gara Yingi, the seat of the Tokko clan

Fig. 14

Gara Daro, the royal seat of Anfillo, as seen from Gara Dallo, the main sanctuary

Fig. 15

The sacred lake of Dallo during the dry season. Oromo farmers have felled virtually all autochthonous trees that surrounded the lake when Grottanelli visited the site in 1939. Those in the background are recently planted eucalyptus

This cluster is surrounded by the aforementioned Kotaa Diinaa, the 20-km ditch. Besides, these sites are among the largest and most protected: Gara Daro, for instance, has a 10-m deep ditch that surrounds the entire settlement, which occupies 7.5 ha (see Fig. 4). The fortified ridge was, on the one hand, a scenario of power that displayed the strength of Anfillo against both the Oromo in the east and the “wild” tribes of the lowlands, to the west: the ridge marks the limit between sedentary agriculturists, on the one hand, and hunter-gatherers and swidden cultivators, on the other, and between the highlands (baddaa in Oromo) and the lowlands (gamoojii), between the settled Anfillo and the mobile Komo. East of the earthen ramparts lay a land open for exploitation: slaves, gold, and ivory.

Yet the monumental scenario is also history told through the landscape: a materialization of an origin myth. The story did not only include human-made settlements (fortifications) and modified natural places (the sacred lake of Dallo) but also unmodified natural landmarks, most notably Mount Konki. If Gara Dallo was the place from where the Busase and the Mao emerged, Gara Konki is where Tapi, a blind old man from the local indigenous group (the Gumnao for the Busase), had a prophetic vision about the arrival of the foreigners (Grottanelli, 1940: 294). The Gumnao (most likely Komo people) would be later expelled by the foreigners to the lowlands and those that remained in the highlands subdued by the Busase and made into the subaltern Mao class. History becomes natural history: the myth explains the inequalities between the groups, anchors them into the landscape, and constructs a political geography.

It was not enough, however, to have history inscribed in the land: it was necessary also to bring it back to life to make it more effective. Therefore, there was a pilgrimage every year from Mount Konki to Gara Dallo that reenacted the history of Anfillo, and a ceremony was held in the latter place—attended by all Busase and Mao clans—in honor of the paramount god Sanchi Gai, creator of the peoples of Anfillo (Grottanelli, 1940: 333) (Fig. 16), an epiphany of power to show everybody her or his place in the social order.

Fig. 16

Ideal paths connecting the sites of the NW cluster and the sacred places

Discussion: Fortified Kingdoms and Cosmic Landscapes

The monumental earthworks of Anfillo, while impressive, are not unique. If there is something that characterizes the landscape of the Omotic kingdoms, both Gonga and Ometo, it is large fortifications. Each and every hierarchical polity that developed in the area devoted enormous efforts to modify the topography where it did not offer enough natural protection. The extent to which the landscape was transformed has probably no parallel in the Horn of Africa. It is not just that shrines, palaces, or even villages were fortified. Ramparts of rammed earth, berms, and ditches did encircle royal residences and settlements, but they often linked settlements and compounds and, in several cases, enclosed entire kingdoms. The problem is that the continuous use of the land for farming and the thick forests have in many cases camouflaged these extraordinary works.

All known Gonga states were fortified in one way or the other. The fortifications of Hinnario or Ennarya were legendarily large and intricate: they played a paramount role in preventing the Oromo cavalry destroying the kingdom until the eighteenth century—the last of the Gonga polities in the Gibe region to fall (Mohammed Hassen, 1994: 47). Hassen (1994: 52) speaks of the “impregnable fortress of Gibe Ennarya,” composed of an “elaborate network of trenches and fortifications,” a real maze that thwarted both the Oromo attacks and those of Ethiopian King Susïnyos. The labyrinthine effect was probably enhanced here and in other Gonga kingdoms by the myriad of compounds surrounded by hedges, which are today characteristic of the spatial organization of Gonga (and Oromo) villages and that in the past surely helped to slow down any advance—what Insoll (2015: 312–313) calls “botanical architecture.” The powerful Kingdom of Kafa also benefitted from “an impressive defense system. The entire state was protected with moats, fortifications, gates, guards, and traps, and an elaborate road network” (Asmarom Legesse, 2000: 65), which discouraged attacks and helped preserve the polity’s independence. The inheritors of the Gonga kingdoms, the Oromo polities of the Gibe basin, also had their kingdoms surrounded by impregnable fortifications and could only be accessed through specific gates, which were still in use in the 1880s (Traversi, 1888: 903–904).

The king’s palace and those of the provincial chiefs were protected by fortified enclosures: to enter the royal compounds, it was necessary to pass through gates that were defended by members of specific clans (Lange, 1982: 226). Smaller kingdoms were equally concerned with physical protection: the Kingdom of Yem² “had built a vast complex of fortified entrenchments against the Oromo since about 1700” (Caulk, 2002: 290–291), and Sheka was likewise protected by ditches and parapets. It was only possible to enter the state though heavily guarded gates, which were controlled by the hunting caste, the Manjo, who were also in charge of protecting the king’s “soul” by guarding the fortifications of his royal residence (Lange, 1982: 161). A similar concern for the king’s integrity existed in Bosha, where the royal compound had four gates, with a gatekeeper in each one. In addition, the Bosha constructed ditches along their western border similar to those of other kingdoms in the region (Lange, 1982: 53). In the case of Yem, this fixation with protection was extended to the tomb of the sovereign: a complex of stele palisades surrounding a burial site on top of an isolated hill has been recently documented (Kinahan, 2013).

The Ometo also invested huge efforts in fencing the land (Arthur, K.W. et al. 2019). Fortified settlements with berms, ditches, and walls have been documented through field survey and interviews with elders. Interestingly, radiocarbon dates have confirmed the antiquity of these sites, in line with oral tradition, the earliest of the fortified settlements going back to the thirteenth century AD (Arthur, K.W. et al. 2019). While the Gonga usually resorted to earth and turf, the Ometo seems to have made use of drystone walls more frequently (see Hailu Zeleke, 2007; Arthur, K.W. et al., 2009; Admasu Abebe, 2014). Through centuries of struggles with the Oromo, the Wälayta became experts in digging ditches against their cavalry and they managed to repel the attacks until the late nineteenth century. Fortifications were built especially on the eastern frontier. A wall seen by Italian explorer Vittorio Bottego in the late nineteenth century extended for 1 km and was 2 m tall. The most impressive Ometo fortifications were probably those of the kingdom of Dawro (Hailu Zeleke, 2007). They were built during the reign of Kati Halala (1757–1782), thus their name Halala Kella—the Wall of Halala (cf. Admasu Abebe, 2014). The aim of the walls, in this case, seems to have been the protection against their powerful neighbors, the Wälayta. Oral tradition tells about three enclosures with high walls, and recent archaeological surveys have documented a wall in basalt masonry 2 to 3 m high and that is said to be 170 km long. The idea of building large fortifications was not new. There is a reference to the constructions of ditches in Dawro in the Futuh al-Habasha, the book about the campaigns of Ahmed Grañ between 1529 and 1537. The chronicler writes that the ditch was “long enough as to be only passable by one place” (‘Arab Faqih, 2003: 127–128). The people of Zeqala had also dug a long trench against Ahmed Grañ’s cavalry, which proved useless in this case (ibid: 298).

Some non-Omotic peoples neighboring the Ometo also built large defensive systems: during the nineteenth century, the Kambata built a wall-and-ditch system (Braukämper, 1983: 62)—in the territories occupied during their expansion. Large part of the Kambata country that was not naturally defended was surrounded by this wall-ditch system, called bōho (Braukämper, 1983: 62), which recalls in its purpose Anfillo’s Kootaa Diinaa and Dawro’s Kella.

Similar processes were going on in neighboring regions further to the south and in roughly similar dates. In western Uganda, complex polities developed during the fifteenth century which undertook large construction works: the main settlements had colossal earthworks which, in some cases, such as Bigo, covered several square kilometers. The excavators of Bigo and similar sites coincide in that the fortifications were devised to protect cultivated fields and cattle, as well as to fulfill symbolic needs, including the demonstration of the power of their inhabitants (Robertshaw, 2002). Fortified sites also emerged in the Lake Victoria Basin, in eastern Kenya, with radiocarbon dates that overlap with those of Anfillo and western Uganda (Odede, 2008: 47).

Why this fixation with defensive infrastructures? There are several reasons for it, not all of them of practical nature. From the thirteenth century, southern Ethiopia became a shatter zone: a point of contact between four different kinds of polities: the Christian kingdoms in the north, the Islamic sultanates in the east, the Oromo pastoralists following indigenous religions in the southeast, and the Omotic polities. The first two saw the Omotic area as an open frontier of predation. Wealth (slaves, gold, and ivory) could be extracted through trade or through violence, and both conflict and cooperation are historically attested between the Omotic polities and their neighbors. The rise of the slave trade in northeast Africa and the Middle East from the thirteenth century onwards also boosted conflict—southern Ethiopia was one of the main sources of slaves in the Horn until the early twentieth century (Fernyhough, 2013)—and this conflict was not only between the Omotic kingdoms and the rest, but also internecine. The latter is well attested in oral tradition (Lange, 1982).

The development of Omotic defenses has of course much to do with an environment of continuous conflict, yet they also have to do with a concern with symbolic separation and boundary-making, which was mirrored in a fragmented landscape of large, often impassable rivers, thick tropical forests, and inaccessible mountains. Thus, Braukämper (1983: 69) considers that linear works were not only defensive in nature, but reflect a desire to make a boundary real, an expression of property over occupied lands. But the many ramparts and gates that separated the king from his subjects in places like Kafa were also a materialization of the symbolic separation of the sacred ruler from the people. They are also a materialization of a highly segmented and stratified society. All Omotic polities have several classes, from just two to five or more. The class bar (usually with racial connotations) was most often impermeable and impossible to overcome: no Manjo (a member of the subaltern class) could ever become a Mingo (the ruling class) in Kafa.

Structurally related to this need for separation was the need for protective devices, both physical and symbolic, to avoid contact with members from inferior groups: hence the fact that the term “caste” has been employed to define intrasocial communities among the Omotic. Although the term is incorrect (Pankhurst, 1999), it captures the stark and impregnable divisions between groups and that at times includes even the use of different languages by the stranger rulers and the autochthonous people (Graeber & Sahlins, 2017: 393), as in the kingdom of Yem (Huntingford, 1955: 137). People believed (and still do in many places) that violating group boundaries can cause illness and death. The fear of pollution was particularly acute in relation to the king, who, like in other cases of sacred kingship (Graeber & Sahlins, 2017), stood for the healthy reproduction of the entire society and the order of the cosmos: thence the enormous effort at sparing the sovereign physical contact with the rest of the population. All rulers had to eat in seclusion and concealed from the rest of the population, and in some cases, they could not be seen or touched at all, the trespassers being condemned to death (Bieber, 1923: 133; Orent, 1970b: 287; Lange, 1982: 44). In the kingdom of Yem, the separation was expressed in a very material way: in public audiences, the monarch sat alone on a 5-m tall platform, while the nobles stood below at the foot. The rest of the time, he spent inside his palace, as it was believed that the country could not “be illuminated by two stars at the same time” (Huntingford, 1955: 140). This concern for separation extended fractally from the royal compound to the city to the entire region, because, as Graeber and Sahlins (2017: 66) remind, “Royal palaces, royal cities, or royal courts almost invariably become microcosms, images of totality.”

Sub-Saharan Africa has been usually described as a region where the idea of political borders as well-defined boundaries has been largely absent (Herbst, 2000: 45–52). Borderlands and open frontiers have been crucial in the continent’s history (Kopytoff 1987; Reid, 2011; González-Ruibal, 2014; MacEachern, 2018), but the limits of kingdoms have been seldom defined with precision, and uniform territorial control has been the exception rather than the norm. In that, they were not dissimilar to early complex polities elsewhere (Smith, 2005). The lack of concern for limits is true for many parts of Africa (including much of the Horn and Sudan (Pankhurst, 1997)), but it is not universal. There are examples of polities that went to great pains to enclose their territory with linear infrastructures. The most famous example is that of Benin (Darling, 2016), whose earthen walls and ditches extended for thousands of kilometers. More modest in scale, the kingdom of Mankon, in Cameroon, was also completely surrounded by a ditch during the nineteenth century, which was some 16 km long and 3–6 m wide (Warnier, 2007: 131–158). In turn, Feti, in Angola, had a 12 km-long ditch during the early second millennium AD, one of several linear fortifications delimiting the settlement (Vansina, 2004: 171). Limits were not only political but also spiritual: they divided the world of the spirits from the human world, and a diversity of rituals was performed in them (Insoll, 2015: 296–298). It is interesting that a concern with territorial boundaries often comes hand in hand with a concern for the body of the sovereign, as a provider of life and stability—this was certainly the case in Mankon and many Gonga polities.

The symbolic nature of the fortified landscape had not only to do with a concern for separation (of social classes, of different polities, of inside and outside) but also with memory. All throughout southern Ethiopia landscapes were suffused with mnemotopes: sacred mountains, springs, lakes, and forests and fortified sites that were associated with the history of a kingdom (Arthur, K.W. et al. 2019). Landscape became a way of narrating the past effectively in an illiterate social environment: a material pedagogy. The history that it told was one of origins, ethnicity, and social division—the existence of (alien) lords and (local) vassals. That is, the main pillars of Gonga societies. The legitimizing history of social division was doubly fixed: first, through the landscape and then through the rituals performed in that landscape. Connerton (1989: 70) argues that rituals—unlike myths—do not permit variation or reinterpretation, at least in theory, because they do change, albeit more slowly than myths. Commemorative ceremonies are thus the best way of creating a fixed, master narrative that tells people who they are. The daily use and perception of the landscape and the ceremonies that took place in them worked as incorporating practices through which people assumed their status. Considering the persistence of social marginalization in southern Ethiopia (Freeman & Pankhurst, 2001), it can be said that they worked remarkably well.

Conclusion

The archaeology of complex polities in northeast Africa has focused so far primarily on those formations that most closely resemble the model of the early state usually discussed in archaeology (Trigger, 2003). This has left out of the picture other polities that can add much to the debate. In this article, I have examined the kingdoms of southern Ethiopia, whose history can be traced back a thousand years. These kingdoms, which were recognized as such by their neighbors, developed strong forms of social inequalities and elaborate rituals and materialities of power. With their foreign rulers endowed with divine powers, they shared features with stranger-king and sacred-king formations (Graeber & Sahlins, 2017), as well as with other African frontier societies (Kopytoff, 1987). Yet at the same time, they also possessed some original traits, the most striking of which—though not unique—was their obsession with fixed boundaries, which they monumentalized in different ways. Thus, they created elaborate landscapes that speak not only of endemic conflict and social division but also of an enduring collective memory. To date, our knowledge of these kingdoms comes basically from oral tradition and passing references in texts produced by neighboring polities, as little archaeological work has been undertaken. In this article, I have tried to show that archaeology has much to offer to our understanding of these polities. I have presented the case of Anfillo, one of the least known of the so-called Gonga kingdoms. The polity existed between the late sixteenth and late nineteenth centuries as an independent organization. Like other Gonga kingdoms, Anfillo is a perfect embodiment of a frontier society: an elite group from a core polity that moved to the periphery so as to establish a new political entity, while retaining strong cultural and symbolic links with the homeland. One such link was the landscape: they replicated the Gonga cultural landscape, which meant a heavy physical intervention in the territory. The people of Anfillo built permanent settlements, agricultural systems, and linear fortifications that thoroughly modified the environment. This behavior was at odds with conceptualizations and uses of the land by the mobile indigenous groups, whom they either expelled or subjected. Land modification was one of the ways in which the Anfillo elites appropriated the new territory. The inscription of the rulers’ collective memory into it was another. This they did through oral history, long genealogies, the setting of mythical events in key points of the landscape and through the rituals and processions that took place in those same places. In that, Anfillo (and other southern Ethiopian polities) behaved not unlike many empires and expansive polities in the past. Rather than seeing Anfillo and other similar polities as bizarre political experiments, we should better take them as what they are: complex political formations that have much to offer to our understanding of states and political culture in Africa and elsewhere.