INTRODUCTION
This research considers the history and archaeology of southern Ghana, an area now primarily inhabited by Akan-speaking peoples.Footnote 1 The most comprehensive model of the pre- and early Atlantic history of the region has been proposed by Ivor Wilks. According to this ‘big bang’ model of Akan origins, the history of agriculture in the Ghanaian forest and Akan sociopolitical development were the result of a series of transformations that occurred after the advent of European trade at the end of the fifteenth century. These observations and the conclusions they entail have important implications for the understanding of both the Atlantic and the pre-Atlantic past of West Africa, as well as how that past is conceptualized. Current research, including a growing amount of archaeological data from southern Ghana, suggests a substantial settlement of the forest by agricultural populations well before the beginning of European trade.
Analysis of the early history of Atlantic Africa is made particularly difficult because of the extremely fragmentary – and sometimes imagined – knowledge of the continent before the fifteenth century. The lack of early historical data for Africa has contributed to an essentialist view of Africa's involvement in the making of the Atlantic world as mainly that of a passive participant. Although archaeological and historical studies have addressed the role played by the continent in the genesis of the modern world, the relative paucity of studies of pre-Atlantic Africa makes it difficult to assess the nature of long-term historical forces, particularly those along the West African coasts and in their hinterlands – the new frontiers of the Atlantic world.
A major aspect of West African historical studies has dealt with the ‘origins’ of particular ethnic or cultural groups. In African historiography, the concept of origins often provides a nebulous foundation on which the researcher elevates the narrative walls of the pre-Atlantic past. The reliance on less demanding standards of historical criticism is obscured by the rarity of sources, by the lack of trained scholars able to assess them, and by the Western philosophical inclination to prefer weak arguments to emptiness. The concept of ‘origins’ also exists among African caretakers of various historical legacies. Here, however, it operates quite differently. Although clearly of a historical nature, traditions of origins are also constructed discourses about the past that reflect the contexts that produced them. Turning these traditions of origin into discourses that fit modern standards of scholarship is, therefore, a delicate exercise of historical criticism.
The limitations of these fragmentary sources and the challenges in their interpretation are well illustrated by the interpretation of the pre- and early Atlantic history of southern Ghana. The early political history of the region has received limited attention, with the exception of some of the coastal polities, for which our knowledge extends back into the sixteenth or even the late fifteenth century.Footnote 2 Early Portuguese sources indicate that Europeans encountered hierarchical societies, politically organized on a territorial basis. Yet the nature of these early sociopolitical organizations remains unclear.Footnote 3 The lack of written sources and the absence of oral traditions – other than potentially helpful myths – about such a remote period of time make historical evaluation difficult. It follows that the scholarly literature about state formation in southern Ghana mainly concerns political entities that arose from the seventeenth century onward.Footnote 4
All of these polities, however, are primarily known through limited European sources associated with a well-established and expanding Atlantic economy. While their emergence is well documented compared with pre-fifteenth-century political entities, they nevertheless cannot serve as pristine models of state formation and analogues for earlier, undocumented, sociopolitical systems. Interpretation is constrained by the sources, both oral and written, which notably do not antedate the 1500s. Nonetheless, the ‘origins’ of major historical events and processes such as the founding of lineages and settlement histories, as well as the genesis of sociopolitical complexity and the beginnings of agriculture are frequently placed in the fifteenth and sixteenth centuries: that is, at the time written sources first become available. Is this coincidence? We believe that such interpretations are shaped by the (documentary and oral) sources on which they depend.
HISTORY BEFORE THE ‘BIG BANG’? THE EARTHWORKS OF SOUTHERN GHANA
Archaeological evidence from earthworks, as well as from other sites in southern Ghana, establishes the existence of a sedentary agrarian society well adapted to the forest environment long before the opening of the Atlantic trade.Footnote 5 This is in contradistinction to Ivor Wilks's ‘big bang’ model of the emergence of sociopolitical complexity in southern Ghana in the fifteenth and sixteenth centuries.Footnote 6 This theory has remained largely accepted and unchallenged, with the exception of a few archaeologists who have found it difficult to reconcile such a presentation of the past with emerging archaeological data.Footnote 7
In the following discussion, the term ‘earthwork’ is used to designate a monumental trench system: an enclosure – often roughly elliptical – composed of a deep ditch and raised earthen banks.Footnote 8 They are believed to have been part of a settlement system. Earthworks are of particular note because of the implications they have for communal labor and long-term settlement.Footnote 9 Earthworks sites have been reported in southern Ghana for the past seven decades; and recent archaeological evidence on earthworks can be used to review and reinterpret earlier data.
The earthwork system of the Birim valley was first reported in the early 1930s by N. R. Junner.Footnote 10 Test excavations were undertaken at five sites and a short report published. Some of the material excavated was sent to the British Museum and examined by H. J. Braunholtz, who submitted a brief report to the Geological Survey and published a more detailed note.Footnote 11 Thurstan Shaw, who also pioneered archaeological work in Ghana, did not excavate any earthworks but made useful notes on their structure and was the first to suggest that they might be related to similar monuments in the region, particularly in Côte d'Ivoire and Nigeria.Footnote 12
In 1953, Oliver Davies excavated the Kokobin earthwork. His report provided a list of all known earthworks, with the comment added that ‘other entrenchments exist in the district, but no details are available’.Footnote 13 Davies also noted that ‘entrenchments seem confined to this district of the forest’, but prudently added that such a distribution might be a reflection of the fact that the area had been intensively prospected for diamonds.Footnote 14 From the report and the site map, it appears that Davies chose to open a trench in the western rampart and a large test unit in the interior of the earthwork.Footnote 15 Importantly, the ditch, which was part of the entrenchment system, was not fully excavated and its stratigraphy remained poorly described.Footnote 16
A more comprehensive study of the Ghanaian earthworks was carried out by David Kiyaga-Mulindwa in the 1970s as part of his doctoral research.Footnote 17 He studied a cluster of 21 earthworks in the Birim valley, excavating at two of them. He distinguished two distinct ceramic traditions, associated with different periods of occupation. These were the ‘Earthworks Ware’, the earlier of the ware types and the one that he associated with the construction of the earthworks, and the ‘Atwea Ware’, associated with the Atweafo and similar in style and manufacture to ethnographically documented ceramics.Footnote 18 The Atweafo are the current inhabitants of the Birim valley, whose oral traditions claim they have always lived there. However, they have no traditions relating to construction or occupation of the earthworks, which they describe as being made by the Tetefo, ‘the people of long, long, ago’.Footnote 19
The amount of work that Kiyaga-Mulindwa carried out at the various sites differed significantly. The two earthworks tested archaeologically were the Monsa and Batabi sites. The Monsa earthwork was the most thoroughly excavated and it provided the most significant data.Footnote 20 The stratigraphy is briefly described and the only detailed drawings available are sections of the cutting across the ditch (see Fig. 1).Footnote 21 Kiyaga-Mulindwa was able to obtain six radiocarbon dates, including four directly related to the occupation of the earthworks. Before work conducted by Gérard Chouin at Akrokrowa, these were the only available dates related to earthworks.Footnote 22 Overall, while an important pioneering study, Kiyaga-Mulindwa's research offers a preliminary assessment of a problem that appears much more complex. The limited information on the contexts of the radiocarbon dates and the uncritical interpretation of oral historical data contributed to a series of misrepresentations of the chronology, structure, and function of the earthworks. The interpretation of the Monsa radiocarbon dates is of particular importance because it is inconsistent with more recent data. For this reason, the earthworks chronology is examined in detail below.
Fig. 1. Section of the Monsa earthwork, Birim valley. Original drawing from Kiyaga-Mulindwa, ‘Earthworks’, 95–6, redrawn by Gérard Chouin.
A REVISED CHRONOLOGY: A PRE-ATLANTIC AGRARIAN SOCIETY
Archaeological data has provided increasing evidence that the Ghanaian forest was well occupied by settled agricultural communities prior to the advent of the Atlantic trade. In particular, Brian Vivian's and Peter Shinnie's data from Asante indicate the existence of stable village communities as early as 800 ce.Footnote 23 Additional data emerging from the Central Region suggests a pre-Atlantic settlement density greater than the post-1500 period. These settlements include hilltop and lowland sites of various sizes as well as entrenchments, all with substantial evidence for iron production. Sites with entrenchments are the most visible of these pre-Atlantic settlements. Their chronology is therefore of particular significance.
The main challenge is to determine whether earthworks belong to the pre- or post-Atlantic period; if they were built before or after the opening of the Atlantic trade. The amount of work involved in their construction suggests a social organization capable of mobilizing, controlling, and co-ordinating labor, and relatively sedentary settlement over long periods of time.
Unfortunately, sites excavated before the 1970s were poorly dated, mainly because radiometric dating was not yet available. Chronologies were posited on the basis of the relative depth of the archaeological deposits, the weathered nature of some of the ceramics, and the presence/absence of European trade materials. Prior to Kiyaga-Mulindwa's excavations in the Birim valley, most authors agreed – not without some doubt – that earthworks were ancient but belonged to the Atlantic period or to the immediate pre-Atlantic period.Footnote 24 Kiyaga-Mulindwa did not dispute this assessment and he provided a more substantive chronology by obtaining radiocarbon dates. His interpretation was that the period of earthwork construction coincided with an ‘ethnocide’ and period of population replacement, the latter being the direct explanation for the shift in pottery traditions observable archaeologically.Footnote 25 Kiyaga-Mulindwa published six radiocarbon dates; four of them came from the Monsa Earthworks, and more precisely from two units: MN4 (date N-2208) and MN5 (date SI-2716, SI-2717, SI-2718).Footnote 26Table 1 summarizes Kiyaga-Mulindwa's interpretation of the occupational history of the Monsa earthwork site, while Table 2 presents the details available for the four dates.Footnote 27 These results are very heterogeneous: two mid-second millennium dates, one from the first millennium ce, and one with an unknown, and therefore invalid, standard deviation.Footnote 28
Table 1. Summary of Kiyaga-Mulindwa's chronology for the Monsa earthwork. The 1550–1740 range is given in Kiyaga-Mulindwa, ‘Earthworks’, 191
Table 2. Recalibrated radiocarbon dates from the Monsa earthwork, Birim valley. The most probable age ranges (i.e. age ranges with the largest areas under distribution) are shown in bold italics
SI-2717 is an intriguing date because it is very early, and because we do not know at what depth it was found. There is some confusion over the origin of the sample.Footnote 29 From Kiyaga-Mulindwa's brief description, it is clear that this sample was found in the ditch. If the date is not contaminated and incorrect, which is always a possibility, two interpretations of SI-2717 are possible: either the sample was found in the ditch, in deposits associated with the occupation of the earthworks, and is evidence that the construction of earthworks started at the beginning of the first millennium ce; or the sample was found just below the bottom of the ditch, in which case it is unrelated to the earthworks. Further evaluation can only be assessed on the basis of additional work at the site.
Sample SI-2718 was also taken from the ditch. Unfortunately, the fact that the standard deviation is unknown makes its interpretation difficult. Recalibrating the date using standard deviation ranging from 50 to 90 yielded 2σ age ranges clustered between 782 and 1151 ce, strongly suggesting that this sample dates between the eighth and the twelfth centuries ce. Kiyaga-Mulindwa, however, suggests that this sample has ‘nothing to do with the construction of the earthworks having been carried into the ditch by erosion and subsequently buried in the silt’.Footnote 30 Yet this conclusion merits further consideration. The content of a pit or ditch is usually considered to be more recent than the making of the depression, unless specific depositional features suggest a different conclusion. In the case of the Monsa earthworks such indices are missing. On the contrary, Kiyaga-Mulindwa notes that ‘most of the bank soil appears to have been washed away from the ditch and not the reverse’.Footnote 31 In addition, he found a later date (SI-2716) for the bank. Evaluating the dates, John Sutton rightly noted: ‘since the silting of the ditch could not have begun until after the bank (with its sixteenth-century date [sic]) was built, there is something wrong here’.Footnote 32 Kiyaga-Mulindwa's assessment of the date appears to have been derived from his own conviction that although ‘this valley has been inhabited continuously from the beginning of the first millennium A.D.’, the ditch was a ‘post-settlement phenomenon’ post-dating the fifteenth century.Footnote 33 On the basis of the information provided, SI-2718 suggests, in fact, that the Monsa ditch is a first-millennium or early second-millennium feature, an age consistent with the new earthwork dates presented below.
The interpretation of SI-2716 date also requires careful assessment. The sample is located in layer 4 in the drawing of the south-east section of the ditch and banks.Footnote 34 Based on its shape, that layer appears to be the original outer bank of the entrenchment, although the author describes it as ‘the layer representing the original ground level before the construction of the Monsa earthworks’.Footnote 35 Based on the available drawing, however, layer 3 – located statigraphically beneath layer 4 – seems to be the original ground level (see the plan redrawn in Fig. 1). It is also notable that the SI-2716 sample appears to have been located beneath the post-holes found ‘at almost equal intervals along the top of the outer bank at Monsa’ and interpreted as the remains of a ‘barricade made of posts of about 15 centimeters in diameter spaced out at intervals of about 2 to 3 meters’ (see Fig. 1).Footnote 36 The presence of intrusive features such as post-holes may suggest that this sample was from a disturbed context. It is even possible that the organic sample from which the date was obtained relates to the inferred palisade. The palisade, however, was probably not part of the original earthwork. In other contexts, palisades have only been noted in earthworks assemblages that include Akan ware, indicating that these features were added during more recent occupations. The possibility that SI-2716 is related to the palisade could explain its closeness in age to N-2208, another radiocarbon date clearly associated with a later occupation of the site.Footnote 37 It follows that SI-2716 is not associated with the original construction of the banks and ditch system.
An alternative chronology emerges from this critical assessment of the Monsa data (summarized in Table 3). This chronology enables us to explain the seemingly brutal transition from the early ware to the Akan ware observed in the archaeological record. In fact, according to our interpretation, this transition does not indicate a replacement of population but could rather hide a gap in occupation. Some of the sites were abandoned and reoccupied after a century or two. Some were never resettled. During such a gap, the society experienced drastic changes that have yet to be understood but which had a deep impact on its material culture.Footnote 38
Table 3. Revised chronology for the Monsa earthwork
NEW CHRONOLOGICAL EVIDENCE FROM THE AKROKROWA EARTHWORKS
Archaeological research undertaken as part of the Central Region Project was specifically aimed at evaluating the age of the earthworks. Although some sites, such as the Monsa earthwork excavated by Kiyaga-Mulindwa, were reoccupied during historic times, others are only associated with early ceramics and, therefore, appear not to have been reoccupied after their abandonment. To avoid archaeological disturbances associated with settlement reoccupations, entrenchment sites only showing evidence of a single occupation were considered to be the best sites from which to obtain radiocarbon chronologies. The Akrokrowa earthworks site, identified near Abrem Berase by Gérard Chouin in 2002, produced only early ceramics and no evidence of reoccupation after its abandonment.Footnote 39 It was therefore an ideal site from which to collect samples for radiocarbon dating and to compare results with Kiyaga-Mulindwa's.
The Akrokrowa site was subsequently excavated by Chouin and six samples sent for radiocarbon dating (see Figs. 2 and 3). The ditch is the only place within the earthwork to allow archaeological access to the 260 cm deep, well-stratified layers that testify to the very long occupation of these settlements. To compare the Akrokrowa data with those of Kiyaga-Mulindwa's, samples were collected from the inner bank (Unit K), the ditch (Unit L), and the outer bank (Unit M). They included four samples collected at different depths in the ditch as shown in Figure 4 (Unit L: Beta-184397, Beta-184398, Beta-184399, Beta-216471). Two samples were collected from the banks (Unit M: Beta-184396 and Unit K: Beta-216826). The dates provide a very homogeneous grouping, with the exception of Beta-184399. These results are summarized in Table 4.
Fig. 2. Earthworks and other known settlement sites in the area around Abrem Berase. Drawn by Gérard Chouin.
Fig. 3. Location of units A to M at Akrokrowa, Abrem Berase. Drawn by Gérard Chouin, 2009.
Fig. 4. Stratigraphy of unit L at Akrokrowa. Drawn by Gérard Chouin, 2009.
Table 4. Radiocarbon dating of samples collected at Akrokrowa. The most probable age ranges (i.e. age ranges with the largest areas under distribution) are indicated in bold italics
Beta-184397 and Beta-216471 were obtained from charred palm kernels found in Unit L, respectively in Level 6 (50–60 cm below the surface) and Level 19 (180–190 cm below the surface). Beta-184398 was obtained from a piece of charred wood found in the same unit, in Level 10, at 100 cm below the surface. From their relative stratigraphic positions, we expected Beta-184397 to be more recent than Beta-184398, and Beta-216471 to be the oldest of the three. This relative chronology was confirmed by the radiocarbon results. The fact that the radiocarbon dates are consistent with the stratigraphic position of the three samples demonstrates that the deposits are relatively undisturbed.
Beta-184396 was obtained from a palm kernel found within the outer bank, in Unit M, Level 4, between 30 and 40 cm below the top of the bank. The bank is a by-product of the ditch and, logically, by dating the bank one dates the construction of the earthworks. Interpretively, this is somewhat simplistic, as it is likely that the banks – the portions of the earthworks most susceptible to erosion – were reworked regularly over the centuries and may have included older material. Nevertheless, comparison of Beta-184396 and Beta-216471 seems to strengthen such reasoning.
Beta-216826 was derived from a fragment of palm kernel found in the inner bank, in Unit K, Level 11, between 107 and 110 cm below the surface. This fragment was associated with an occupation layer that seems to have been cut by the ditch. We might expect the date to be anterior to the construction of the ditch, roughly contemporary with Beta-184396, and older than the deposits found in the ditch. In fact, the results seem to show that the level of occupation found in Unit K is contemporaneous with the construction of the ditch that cuts it. They also help definitively place the construction of the Akrokrowa earthworks in the eighth or ninth century ce, which is consistent with the radiocarbon dates obtained for the deposits found in the ditch.
The single outlier in the series is Beta-184399, a sample collected from Unit L at the bottom of the ditch, 198 cm below the surface. It yielded a 2-sigma calibrated age of 5310–5060 bce, which clearly lies outside the earthwork's occupation. The sample probably represents an old surface or organic material buried in subsoil.Footnote 40
The Akrokrowa radiocarbon dates point to the construction of the entrenchment system in the second half of the first millennium ce, most likely during the eighth century, long before the arrival of the first Europeans on the Costa da Mina.Footnote 41 Notably, these results coincide nicely with SI-2718 collected by Kiyaga-Mulindwa in the ditch at Monsa. The radiocarbon results from the Akrokrowa earthworks provide the first series of radiocarbon dates undoubtedly entirely related to the construction and occupation of earthworks by the makers of the Atetefo pottery. The fact that the site is located far away from the Birim valley, but seems to belong to the same historical period, radically changes our perception of sociopolitical and economic life in the forests of southern Ghana before the opening of the Atlantic trade. In the forest area that Wilks envisioned as being sparsely populated by bands of hunter-gatherers,Footnote 42 a network of fully formed agrarian settlements now comes to light.
STRUCTURE AND FUNCTIONS OF EARTHWORKS
At this stage, it is too early to make broad generalizations about the nature of the earthwork societies. A large quantity of fragments of palm-nut shells was found during excavations, suggesting large-scale cultivation of palm trees. Fragments of oil lamps were found. The inhabitants of the earthworks also made iron tools. Fragments of slag as well as pieces of iron objects were found at the Akrokrowa site, together with possible quartz tools, polished axes, and stone beads. No bone, shells, or macrobotanical remains were found on the site to provide more information about the earthwork builders. There is therefore a need for further archaeological investigation, incorporating specific strategies to gather botanical evidence from archaeological deposits.
In the same vein, we do not have the proper data or epistemological tools to infer from the excavations the nature of the political system of which the earthworks were a part. We can only notice that their construction suggests a type of social organization capable of mobilizing material and human resources to build the huge embankments. On the basis of available data, however, we cannot be more precise about the character of the political organization or forms of authority.
The function of earthworks has been debated by both archaeologists and historians, most of the times in relation to their structural features, but the relatively poor description of earthworks in Ghana did not allow for a detailed understanding of the nature and functions of these sites.Footnote 43 The interpretation of earthworks as defensive sites is based on common-sense observations as well as on comparative data from other parts of the world. As early as 1932, comparisons were made with Iron Age entrenchments found in Great Britain.Footnote 44 In his paper on the Kokobin earthwork, Davies did not discuss the issue of the function of the entrenchments, but it is clear from his use of words such as ‘rampart’ or ‘fosse’ that he did not doubt that they were defensive features.Footnote 45 In 1967, he called these sites ‘entrenched forts’, although noting that ‘the fortifications do not follow strategic contours’ and even spread ‘across small valleys’.Footnote 46 Clearly, he doubted the military value of these sites.Footnote 47 In his 1957 study on the rise of the Akwamu polity in the seventeen century, Ivor Wilks suggested that they were fortifications that may have represented ‘Akwamu attempts to strengthen its western frontier against Akim aggression’.Footnote 48
Kiyaga-Mulindwa also suggested that earthworks were fortifications that evolved as the result of the military pressure of the ‘Atwea’ against the ‘Tetefo’: ‘earthworks were constructed so as to be used as community defenses. … They were a response to small, intermittent attacks. … they seem to be responses to intermittent, low-level pressure from the outside rather than to large-scale invasion’.Footnote 49 The long-term occupation of the earthworks and the extensive spread of earthworks throughout the forest zone of West Africa do not seem to support such a hypothesis.
The efficiency of entrenchment as defensive works is difficult to evaluate because we are far from understanding what these sites would have originally looked like and what the usages were in warfare. Once again, there is need to strengthen our understanding of the archaeological record. It could be that entrenchments were part of a more complex system of fortification combining fosse, banks, and vegetal defensive rings not unlike those documented by European explorers in nineteenth-century Sierra Leone.Footnote 50
Earthworks may also have been used as barriers to keep elephants away from settlements. The protection of communities against the threat posed by elephants should probably be studied very seriously, especially considering the fact that pre-Atlantic earthworks do not seem to have had any permanent causeways. The presence of defensive forest belts added to the earthworks could have represented a form of fortification strong enough for the type of warfare practiced at the time or to keep away dangerous animals. At the same time, forest belts would have protected the village from bushfires, a strategy well known in Guinea.Footnote 51
As early as 1964, in an unpublished paper, another theory emerged that suggested earthwork systems were made ‘to prevent exit and not entry’.Footnote 52 This theory, which held that earthworks might have been used as camps for gold miners, was part of Ozanne's 1971 review of the state of archaeology in Ghana and captivated for some time the attention of Marxist historians.Footnote 53 Ozanne never excavated an earthwork. His argument was based on the mistaken impression that earthworks would have served as a rampart to forcibly retain laborers inside the camp. He noted that they were ‘defined by a deep ditch with a bank on either side – the outer one usually being the higher’.Footnote 54 The argument did not stand up to new evidence brought forward by Kiyaga-Mulindwa, who demonstrated that the distribution of earthworks did not coincide with that of ancient placers.Footnote 55 After 1978, to our knowledge, only Wilks continued to regard them as possible work camps.Footnote 56
This theory still subtly persists in the historiography, however, although all archaeologists with a practical knowledge of earthworks agree on the relative profile of the embankments. In 1951, Shaw could write that ‘it is usually higher on the inner side than on the outer’.Footnote 57 His view is supported by the profile of the Kokobin entrenchment published by Davies.Footnote 58 The profile of the Manso earthwork published by Kiyaga-Mulindwa also shows that the inner part is higher than the outer bank.Footnote 59
Yet the profile of any entrenchment varies from one segment of an earthwork to another. At Akrokrowa, we took measurements at three different places along the entrenchment and observed important variations depending on the position of the entrenchment vis-à-vis the existing slopes. Where the fosse was dug halfway down a slope, outer banks are naturally positioned below the inner bank, but where the trench is sunk on leveled ground, the outer banks are usually of greater height than their inner corollary. Looking at the general profile and commenting upon the relative height of banks is not enough. It is also necessary to study the structure of the banks and their stratigraphy, to understand when and how they were constructed.
In fact, we may have stratigraphic evidence from Akrokrowa that what we consider today as an inner bank might actually be the melted remains of a clay wall that defended access to the settlement behind the ditch. Such a wall, if confirmed, would reinforce the military function of the early earthworks.
Earthworks were settlements, and trenches marked the limit of these settlements. This was clear at Akrokrowa, where we found no evidence of archaeological deposits outside the area delimited by the ditch. This latter, we believe, did not fulfill a single, unique function but played a multifunctional role in the life of the people who lived there. For instance, the fact that entrenchments were used for waste disposal is clearly demonstrated by the nature of the material found in the ditches, as well as their stratigraphy, which illustrates the slow accumulation of refuse throughout centuries. There is a parallel to be drawn between the ditches used as rubbish dumps and midden deposits that form a belt of small mounds around historic and contemporary settlements. This area forms the kurotia: that is, the ‘border, outskirt … of a town’.Footnote 60 Pescheux suggested that the kurotia was ‘the borderline located at the edge of a … settlement (cultural space) and, more exactly, the area where settlement (the cultural space) and forest (the natural space or world) interact’.Footnote 61 Entrenchments might have been a form of kurotia, an area where everything broken, useless, or considered as dirty was deposited.
Pescheux also reminds us that the kurotia was mainly a liminal area, where rituals aimed at protecting the community against social disorder, diseases, and attacks were conducted. Indeed, the use of the trenches as sites for mundane waste disposal could have been extended to include ‘spiritual’ waste disposal. By strongly marking the border between the settlement and the forest, the ditch formed a physical separation between the mundane and spiritual worlds. As a liminal space, it must also have been the ideal place to carry on rituals and sacrifices. Traces of such ritual activities may be found in the archaeological record in the form of objects or features that would not be expected in an ordinary waste disposal site. In the case of Akrokrowa, the test unit was far too narrow to enable us to examine this hypothesis. It is clear that further excavations of earthworks should be planned, on a larger scale, with the trenches as a particular focus.
At this stage of the research, many other potential functions could be suggested. We can only point out the fact that available data are not sufficient to allow us to state what the actual functions of earthworks were, and if they evolved through time. Consequently, a multifunctional approach to these structures is recommended, to embrace a larger perspective and to fuel and orient further research.
Polet and Saison noted that the earthwork they excavated in Côte d'Ivoire was located away from main water bodies and that there was no ground water available up to 20 meters below the surface.Footnote 62 Getting water would have been a tedious task for a community living in such an environment. The situation might have been different for each earthwork, but the fact remains that, being most often constructed on hills and raised grounds, they tend to be away from main river courses. Large quantities of shards may testify to the necessity to store water in the entrenched settlements. Trenches might also have been used to retain water during the dry season, in the same way that artificial ponds are dug in contemporary villages to cope with the chronic lack of water. If this is the case, we could better understand why trenches are often built half-way down slopes, since there would be no better way to collect runoff. Possibly the shade provided by the surrounding forest cover would have prevented quick evaporation. Water would probably not have filled the trench but would form a series of small ponds all around the earthwork, in depressions. Such reservoirs of stagnant water would have been fairly similar to those observed near Nsadwer, along the course of the Hua river, during the rainy season. They would also have resembled the small ponds of standing water observed by Kelly and Norman at Savi, and Connah at Benin City, albeit only during the rainy season.Footnote 63
Iron and iron technology were crucial factors in the process of digging earthworks, the latter being tangible evidence of the working power of this metal and, at the same time, a display of prestige for the elite who could demonstrate their ability to accumulate and control kin and followers. Indeed, to build earthworks, two major things were needed: iron excavation tools and human resources. In fact, evidence of iron smelting and smithing were found on all excavated earthworks in the form of slag and fragments of iron. Akrokrowa was no exception, although we did not find any fragments of furnace. The presence of slag suggests that iron was smelted locally. The dialectic relationship between iron production, earthworks, and prestige/demography is yet to be explored in Ghana. One of the functions of earthworks could have been to raise a community and/or elite groups to a higher status position in a larger sociopolitical system. To achieve such a ‘rite of passage’, communities would have had to acquire the requisite technological knowledge and the adequate number of tools. In addition, they would have been demographically strong enough to carry on large-scale excavation works. In this case, the ditch would also play the role of a social boundary. Such a process of social differentiation could explain the presence, during the same period, of smaller settlements without earthworks.
ABANDONMENT OF THE EARTHWORKS
Excavations at Akrokrowa and associated sites afford a glimpse of a well-settled landscape predating initial European contact in the fifteenth century. The chronology of Akrokrowa is in agreement with an abandonment of the site by the mid-fourteenth century. This in turn is amplified by data coming from other earthworks in southern Ghana suggesting a major change in settlement pattern before the opening of the Atlantic trade, characterized by the abandonment of earthworks. This remains a working hypothesis that needs to be tested on a larger number of entrenchment sites. If the hypothesis is confirmed, why were they abandoned? Whatever happened must have been quite sudden, able to affect a vast area, and traumatic enough to dramatically alter people's way of life within a generation, wiping out the structures of a centuries-old agrarian order.
Looking at world history during this period, it seems that only one event can possibly explain such a large-scale phenomenon: the occurrence of the Black Death or Great Plague.Footnote 64 Well documented in Europe,Footnote 65 as well as in North Africa, where it may have killed a third or more of the total population between 1346 and 1352, for lack of direct evidence, the Great Plague is often believed to have spared sub-Saharan Africa.Footnote 66 Looking at the highly infectious nature of this disease and the dense network of links between West Africa and North Africa, as well as between eastern Africa and the Arab world and Asia during this period, this seems highly improbable. The Plague reached the Maghrib in 1348Footnote 67 and it is plausible that it reached the forests of West Africa in the mid-fourteenth century. In affecting societies living in forest areas of southern Ghana, the Plague would have had exactly the same effect as in other parts of the world: it would have destroyed a large portion of the population living in densely populated settlements, resulting in their abandonment – a well-documented phenomenon in many other contexts.
Here it is useful to quote Merrick Posnansky, one of the few researchers to have considered the Plague as an unknown component of West African history: ‘We have no idea if the plague spread across the Sahara in the 14th century but we should not discount its possibility’.Footnote 68 We believe the widespread abandonment of earthworks in the fourteenth century is an indirect indication that Posnansky might have been right. While the only consideration here is the abandonment of the earthworks of southern Ghana, it may be more than coincidental that other abandonments and transformations in settlement organization appear to have occurred in other parts of sub-Saharan Africa during the same period.Footnote 69
CONCLUSION
Compared with America and Europe we know very little about the forest zone of West Africa at the opening of the Atlantic trade. Past and present scholarship has been insufficient to evaluate the pre-Atlantic past and previous reconstructions require critical reassessment. Limitations in our knowledge have had a direct impact on the perception of Africa's role in the Atlantic revolution. We should now conclusively admit that the Atlantic trade did not bring agrarian order to the peoples of the West African forest. In the first millennium ce, a sociopolitical order that was well adapted to the forest environment probably extended across the whole of the Gulf of Guinea. Iron tools were used to modify the forest environment, to farm, and to dig entrenched settlements. Palm oil and yam may have formed an important part of the subsistence base. This pre-Atlantic agrarian order is coming to light as a growing number of sites and earthworks are discovered in the forest zone of West Africa.Footnote 70
There is a growing body of evidence suggesting that this pre-Atlantic settlement pattern collapsed rapidly, possibly following a deadly plague. People abandoned their settlements and dispersed across the farmlands, before reorganizing themselves according to new patterns. This hypothesis cannot be ignored, and we need to revisit the available data pertaining to the archaeology of Africa along these new lines. The advent of people from the Atlantic world may have coincided with the aftermath of this epidemic. The first Europeans arriving on the West African coast in the late fifteenth century may have viewed a region still emerging from the demographic and social impacts of the Plague. As in other world areas, forests may have re-conquered previously farmed lands and dramatically altered the landscape. It was this landscape – largely reforested and evocative of wilderness – that early Europeans met as they arrived on the West African coast. The Atlantic market may have become central to a restructuring of society, a means of empowering a new generation of West Africans seizing new opportunities.
The Atlantic factor in African history, as well as Africa's role in the Atlantic past, are essential themes in the study of the modern world. Yet examination of the myriad of intersections represented have often been conducted from the perspective of a history of impact – sometimes coupled with an archaeology of impact – that looks at change in the later part of the second millennium ce as a process primarily resulting from the irresistible spread of capitalism outside its European cradle. The history of the Atlantic world was transformative in terms of the ways in which people (particularly aspiring elites from different continents) manipulated new opportunities in exchanging their goods, genes, germs, and values on an unprecedented scale. Yet such a Eurocentric view of Atlantic history is not consistent with historical realities.Footnote 71 We prefer to conceptualize Atlantic history as one aspect of a complex past that encompasses interactions both within and without Africa. This is a provocative way of rethinking Africa's integration into the Atlantic space.
