INTRODUCTION
The aim of this article is to discuss the chronology and cultural periods of the Iron Age of the Gulf of Guinea. To do so, we present an important corpus of radiocarbon data collected on the island of Corisco (Muni estuary, Gulf of Guinea, Equatorial Guinea), mainly from the archaeological site of Nandá. Corisco belongs to Equatorial Guinea, but geographically speaking it is located off the Gabonese coast. Corisco is the name given by the Portuguese colonizers, meaning ‘lightning’, although the island is also known by the vernacular name ‘Mandji’, in the Benga language. It is only 15 km2 and more or less rectangular in shape, with flat topography (the maximum elevation is 35 m above sea level) (see Fig. 1).
Map of Corisco island with the location of Iron Age sites (left) and the region of Muni-Gabon estuaries (right).
At the outset, the archaeological project was designed like a rescue mission, since the construction of an international airport by the government was seriously affecting one of the most important later prehistoric archaeological sites of Equatorial Africa. Three field seasons have been carried out so far. In the first one (2009), during which we identified most of the main Iron Age sites on the island and focused our work on a general survey of the island and on cleaning the cuts in the road left by the bulldozers' tracks.Footnote 1 During the second and third seasons (2011 and 2012), we excavated some sectors of the Iron Age sites identified as Areas 7 and 8 (Nandá) that were less affected by construction, and we resumed the archaeological survey of the coast.Footnote 2 The survey led us to track down many other Iron Age sites where pottery fragments and iron artefacts were collected, as well as historical sites dated mostly between 1780 and 1970. Unfortunately, the interior of the island remained unexplored since it is still covered with dense second-growth forest. In this article we will only focus on the prehistoric part of the project, since the archaeology of colonialism is dealt with elsewhere.Footnote 3
The 22 new radiocarbon dates that we obtained from sites in Corisco encompass a period spanning from the first century bce to the twelfth century ce. This archaeological survey allows us to redefine some Iron Age archaeological groups already known in the Libreville Estuary of Gabon and to propose stylistic, technical, and chronological amendments to the island's sequence. We will also compare these data with the nearby archaeological findings of southwestern Cameroon, coastal Gabon and Congo, and the rest of Equatorial Guinea in an effort to broaden understanding of larger historical trends in the West Central African region.
It has to be pointed out that the importance of the archaeological site of Nandá does not lie on its chronology alone. This site, where most of the radiocarbon dates were collected, constitutes the largest excavated open area of any Iron Age site to date in West Central Africa (c. 900 m2 of extension, out of c. 4–5 hectares of artefact scatter). Within it, we have been able to document a broad variety of mortuary practices throughout the Iron Age and some domestic features. In this sense, this archaeological site is the first one to provide information about the organization of space both in funerary and domestic domains in this part of Africa for the first millennium ce.Footnote 4
ANALYSIS OF THE RADIOCARBON DATES
As part of our reassessment, we have compiled the currently available Iron Age radiocarbon dates from the Atlantic coast of Central Africa, including Cameroon, Equatorial Guinea, Gabon, and Congo.
Cameroon has provided just 24 radiocarbon dates coming from 6 archaeological sites: 7 from Akonétye; 11 from Campo; 1 from Campo Église; 2 from Mouanko-Lobethal; 2 from Mouanko-Epolo; and 1 from Mpoengu.Footnote 5 Radiocarbon dates from Equatorial Guinea can be divided into those from sites in the mainland and those on islands. The two radiocarbon dates from mainland Equatorial Guinea come from the sites of Akom and Ayene.Footnote 6 The island of Bioko has provided 15 radiocarbon dates coming from the archaeological sites of Carboneras, Bolaopí, and Buelá.Footnote 7 Corisco has furnished 22 Accelerator Mass Spectrometry (AMS) radiocarbon dates, of which 20 come from Nandá and the 2 others are from the sites of Ulato and García. Almost half of them (ten) have already been published, while the other 12 are presented here for the first time.Footnote 8 Currently, there are 21 relevant radiocarbon dates from the Libreville estuary area of Gabon, coming from 14 archaeological sites: Oveng 1; Kango 2 and 5; Remboué 1, 3, 11, and 15; Kafelé 2; Malékou; Sablières; Angondjé; Charbonnages; Nikol Ogoum; and Okala 1.Footnote 9 Finally, 23 radiocarbon dates come from Congo, from archaeological sites of Tandou-Youmbi, BP113, Mango-Kayes, Meringue, Fignou 1 and 4, Kayes, Tchitembo, Lac Ndembo, Condé, and Loubanzi.Footnote 10
Altogether, we have compiled 107 radiocarbon dates. Unfortunately, we do not have consistent information on either the method used (conventional or AMS) or the material from which dates were obtained. Nevertheless, the archaeological contexts seem to be well documented in almost all cases. All radiocarbon dates have been submitted to an analysis of validity that allows us to evaluate them evenly and compare them with other dates, as well as to ensure that the radiocarbon dates correspond to their archaeological contexts. All dates that do not fill the requirements have been rejected and are not included in the following analysis. Our criteria fall into two categories: those associated with the radiocarbon measurement, and those related to the archaeological association of the material dated. The technical requirements include laboratory accuracy (contamination elimination and radiocarbon measurement) and measurement precision (standard deviation) whilst the archaeological requirements include the association and synchrony of the material dated with the event of interest.Footnote 11
From a technical point of view, we cannot assess the accuracy of all the laboratories that carried out the dating, with regard to removal of contamination, chemical treatment and radiocarbon content measurement, but we assume and accept that they all meet the minimum international standards, whether a research-oriented laboratory or one with a more commercial orientation. This tacit trust in the accuracy of laboratories is indispensable for this study. The dates were ascertained in 12 separate laboratories, as listed in Table 1.
List of laboratories from which radiocarbon dates from West Central Africa were obtained.
Regarding the precision (at one standard deviation) of the numerical dates obtained, there is only one date with a standard deviation greater than 100 years (Beta-20784). Even though we calibrated it, we did not use it in our comparative analysis. The precision of other dates are less than or equal to 100 years. Seventy-three out of 107 dates (68.2 per cent) have a standard deviation below fifty years. As expected, the more precise dates are those obtained by the AMS method.
Archaeologically, we must evaluate whether the dates are representative of the contexts they intend to date or not. In this sense, two Cameroonian dates, one from Campo (KIA-33074 11211 ± 45 BP) and another one from Mouanko-Epolo (KIA-8458 3860 ± 29 BP), seem not to meet the requirements of association and synchronicity of the funerary context they were intended to date, as both of them exceed the expected chronological context. In total, after the validity test, we only discarded these three dates, which suggests that both the sampling and dating processes were carried out with scientific accuracy. For the analysis and comparison we will use the 104 remaining dates.
Once the validity test was undertaken, we calibrated the 104 meaningful 14C dates using the calibration curve CalPal 2007 Hulu, included in the CalPal software,Footnote 12 almost identical to the calibration curve IntCal-04 suggested by the International Calibration Series for the last 24,000 years cal. BP.Footnote 13 After calibrating the dates, we considered the temporary lapse that corresponds with the higher probability (95 per cent), obtained using calibration at twice the standard deviation (2 σ) of the radiocarbon date. We assume that this captures, within the range of the calibrate dates, the true calendar age of the sample. We have worked with CalPal software that allows us to obtain cumulative probability curves for calibrated dates both individually and in sets of regional groups. For this reason, we have not considered it necessary to conduct a Bayesian analysis. The dates have been divided into geographical groups, depending on the countries where the sites are located, except for Equatorial Guinea, which was subdivided in three different areas corresponding with three geographical areas: the island of Bioko, mainland Equatorial Guinea, and the island of Corisco.
Finally, a graph that shows the six regional groups (mainland Equatorial Guinea, Corisco, Bioko, Gabon, Congo, and Cameroon) was generated (see Fig. 2). As shown, the oldest samples dated come from sites in Cameroon, Congo and Gabon, predating ce, while Corisco and mainland Equatorial Guinea begin around the bc/ce transition and the sixth century ce respectively. We have included only dates from contexts that can be unambiguously related to iron-using communities. By this, we refer to sites associated with archaeological groups in which the use of iron has been attested. Thus, a site can be associated with an iron-using community if iron or iron-related elements (slags, tuyeres) have been found, but a site can also be associated with an iron-using community if the material assemblage that appears in it coincides with the material assemblage of an archaeological group in which the use of iron has been attested. Thus, we found a tuyere associated with Middle Oveng materials in the site of Nandá and thus infer that every site with Middle Oveng pottery belongs to an iron-using community. Admittedly, the situation can be less clear-cut in the case of early iron-using communities, when some groups in the same region might be making and using iron, others only using it, and others neither using it nor making it. However, this must have been a relatively short-lived episode in general and has limited relevance for the purpose of this article.
Cumulative graph of the calibrated radiocarbon dates from West-Central Africa discussed in this article.
CORISCO ISLAND IRON AGE: AN ARCHAEOLOGICAL SEQUENCE
Before radiocarbon dates were available, pottery typology and ‘chaîne opératoires’ (Technological Choices Framework) made possible the identification of three archaeological groups on the island.Footnote 14 Since most of them have been documented previously in the Libreville Estuary (Gabon) by archaeologist Bernard Clist we will use the same terminology he suggested with some adjustments.Footnote 15 According to that schema, the Oveng group developed during the Early Iron Age, and the Nandá – former Group II – and Angondjé groups during the Late Iron Age.Footnote 16
EARLY IRON AGE: OVENG
The Oveng group was first documented at eight different sites of the Libreville Estuary of Gabon, and nine radiocarbon dates are available.Footnote 17 Radiocarbon dating spanned the period between the first century bce and the sixth century ce. One of the dates (Gif-8151) traces the period back to the fourth century bce. However, we do not consider that this date contradicts the majority of dates that set the beginning of Oveng in the second and first centuries bce (Beta-14832, Lv-1519, and Arc-343), since accurate contextual data of its collection is not available. The possibility that it refers to an earlier, pre-ceramic phase cannot be ruled out. The Oveng group has also been documented in Areas 7 and 8 of the Grasslands of Nandá (Corisco), as well as at other places on the island (Area 1, Cruz, etc.). Thirteen radiocarbon dates are available for this period (see Table 2). The differences noticed in style and chronology of the materials throughout the Early Iron Age enable us to outline three different phases that we have called Early, Middle, and Late Oveng.
Radiocarbon dates from Corisco. Notes: All dates are AMS and were obtained from charcoal samples. a SU = Stratigraphic unit.
Seven radiocarbon dates set Early Oveng in Corisco approximately from the first to fifth centuries CE. All those radiocarbon dates come from a burial site where pottery fragments and iron artefacts have been recovered from a total of twenty burials. The kind of ritual performed during this period was the secondary interment. This means that the corpse was first exposed, perhaps with the grave goods, until it was entirely defleshed, and then some bones and the grave goods were buried in rounded or elongated simple shallow pits.Footnote 18 The majority of grave goods consisted of iron objects: axes, spoons, sickle-knives, bracelets, anklets, necklaces, special-purpose money (ekuele), and spears (Fig. 3). Some of the burials yielded pots similar to the Gabonese Oveng style, particularly to Clist's Oveng types A and D, vessels with simple flared rims and vessels with convex rims respectively.Footnote 19 They were invariably decorated with comb impressions and incisions (see Fig. 3). Both the form of burial and the artifacts associated with them are of great relevance to understand early Bantu social organization and cultural practices. Thus, the manipulation of ancestral bones is a ritual custom widespread in the Bantu area, which continues to the present day. Special-purpose money (which included axe hoards) was probably used in marriage transactions; this represents another typical feature of western Bantu populations, which archaeology is now proving to be two millennia old.Footnote 20
Diagnostic materials from the Early Iron Age in Corisco: Early, Middle, and Late Oveng.
Middle Oveng is defined by four radiocarbon dates that come from an ordinary settlement. No burials have yet been found for this phase at Nandá. The archaeological features of this period comprise occupation floors with postholes, refuse pits and a possible iron furnace reused as a rubbish pit. Radiocarbon dates set this phase from the fourth to sixth and seventh centuries ce, slightly overlapping with both the previous and the following periods. Fragmented pieces of pottery that fit the Oveng tradition have been found inside the refuse pits. These were from vessels with pointed, convex rims, and semiclosed shapes and bowls decorated with impressions (see Fig. 4).
Diagnostic materials from the Late Iron Age in Corisco: Angondjé and Nandá traditions.
Late Oveng corresponds with a tradition not previously documented. We named the phase ‘Late Oveng’ because it overlaps slightly with Middle Oveng (see Fig. 2) and its material culture bears some resemblance to that of earlier periods. Two burials with grave goods and perhaps three more without goods (Area 7), one rubbish pit (Area 8), and some pottery remains recovered from surveys in the southern part of the island (Cruz II) make up the archaeological features identified for this phase. Two radiocarbon dates (one coming from one of the tombs and the other from the pit) set this phase in the seventh to eighth centuries ce. The only material recovered from the burials was two pairs of simple bracelets (two in each tomb), which are stylistically similar to others found in Early Oveng burials, and a simple collar, all of iron. Nevertheless, unlike in previous phases, burials were primary inhumations, apparently without excarnation. Three other elongated shallow pits were documented, with a similar size, shape and orientation, but no grave goods were found inside them. They may have been humble burials of this period. The pottery recovered from the refuse pit recalls in shape the pottery of Middle Oveng, especially the pointed rims, but it lacks surface decoration (see Fig. 3). Late Oveng might be the beginning of a period of social and demographic decline that lasts until the late first millennium ce. More archaeological data may clarify whether this phase is actually part of the Oveng group or something different altogether.
After Late Oveng, radiocarbon dating and archaeological survey suggest a time of depopulation: Area 1 and Cruz I disappeared and the large Areas 7 and 8 were reduced to a small site situated between these two areas (probably less than half a hectare wide). Neither funerary nor domestic contexts have been found on the island dating to the three centuries between the eighth and the tenth centuries ce, when Corisco seems to be uninhabited.
LATE IRON AGE: NANDÁ
After several centuries of depopulation, a new archaeological group appeared in Corisco: the Nandá tradition. This tradition was named Group II when it was first identified at one burial site and three settlements in the Libreville Estuary area of Gabon. Only three radiocarbon dates were previously available for these contexts, and one of them (Beta-20784) has a 140-years deviation. Even though this sample is hardly representative, the dates nonetheless imply that this tradition developed between the tenth and eleventh centuries ce.Footnote 21
In Areas 7 and 8 of Corisco, a Nandá settlement and a burial site have now been documented. We first identified them as Group II because the pottery types are identical to those of coastal Gabon. A total of seven radiocarbon dates, of which six come from burials, place this tradition between the eleventh and twelfth centuries ce. The burial site has yielded 16 graves so far. Both the shape of the graves and the disposition of the grave goods suggest that the funerary ritual performed was primary inhumation. The funerary goods include standardized iron items and pottery vessels (see Fig. 4). The types of iron objects differ from the ones buried during the Early Iron Age. In this phase, necklaces, bracelets, anklets, knives, and adzes are found. The pottery vessels include gourd-shaped and globular pots with everted rims, both of them with incised decoration. A peculiarity of this pottery, as we pointed out elsewhere, is that the pots were barely fired and crumbled easily.Footnote 22 The Nandá pottery assemblages were made purely for burial during the funerals. Another Nandá burial ground has been identified nearby (Area 3) and, although it was completely or almost completely destroyed during the construction of the airport, it yielded numerous vessels, mostly gourd-shaped pots, which could be photographed and drawn. The Late Iron Age funerary context also provides important social information. Belonging to this period, we have found the burial of a prominent individual or chief, whose rank was signalled by 4 heavy iron collars and 18 pots.Footnote 23 Still, there do not seem to be remarkable social differences during this period, since most burials seem to be very similar in size and funerary offerings. Among the latter, it is worth noting the appearance of knives and razors that can be associated with scarification and circumcision; both practices, which are associated with rites of passage, still exist in the region or have existed until very recently.Footnote 24
LATE IRON AGE: ANGONDJÉ
The Angondjé tradition was first identified on numerous domestic sites in the Libreville Estuary area.Footnote 25 Nine radiocarbon dates situate the tradition between the eleventh and fifteenth centuries CE.Footnote 26
The Angondjé tradition has not been found in Areas 7 and 8 at Nandá, but many Angondjé-style pottery scatters were found when surveys were conducted all over the coast of Corisco in places where bulldozers had cleared the jungle (Fig. 1). As a result of such clearing, several Angondjé sites were located, making this tradition the best represented in the archaeological record of the island. However, none were excavated. Despite the absence of an undisturbed archaeological context we were able to document two profiles (in Ulato and García) where we obtained charcoal samples associated with typical Angondjé materials for radiocarbon dating. The results date the tradition to between the tenth and the twelfth centuries ce.
The typical archaeological findings from this period are sherds of pottery that differ in shape, decoration and quality from those of the Nandá tradition. Angondjé pottery is well-fired; it presents carinated shapes and straight or slightly open rims. Decorative patterns usually leave the strip closest to the lip undecorated and then fill the rest of the vessel with a baroque style that combines different incised and impressed carved wooden roulette designs. Lids, lid knobs, and annular bases are also common (see Fig. 4). Iron implements from Angondjé sites include adzes, gouges and small knives, which were similar to, but smaller than, those found in Nandá burials.
THE IRON AGE IN SOUTHERN CAMEROON, EQUATORIAL GUINEA, AND COASTAL CONGO
In order to provide a broader picture of the Iron Age of West Central Africa we have extended the scope of our analysis to include some representative sites of southwestern Cameroon, coastal Congo, the coast of mainland Equatorial Guinea and, further north, material from Bioko island. We have chosen these sites for several reasons: their geographical and cultural proximity to the area where we are working, the type of archaeological features found in them and, finally, the fact that they have been radiocarbon dated. We compare information from these sites with the data coming from Corisco in order to develop a general archaeological sequence for the area.
From Cameroon we have taken into account six archaeological sites, namely Akonétye, Campo Center, Campo Église, Mouanko-Lobethal, Mouanko-Epolo, and Kribi-Mpoengu.Footnote 27 Except for Akonétye, which is located slightly inland, the rest are distributed along the coast from the south of Douala (opposite Bioko island) to Campo river.Footnote 28
The excavators describe the features found in those sites as ‘shaft-like pits ranging in depth from 2–3 m’ that ‘usually contain ceramics, stones, macro-botanical remains and charcoal, but also slag, tuyère fragments and in some cases iron objects such as rings, axes and knives’.Footnote 29 Although human remains were virtually absent, the authors interpreted these pits as burials for the same reasons we did on Corisco island.Footnote 30 The 24 radiocarbon dates obtained place this burial custom between the first century bce and the fifth century ce, which completely overlaps with the Early Oveng period on the island of Corisco.
Radiocarbon dating suggests that southwestern Cameroon and Early Oveng Corisco burials are contemporary. They share many traits, although some differences can also be noticed. First of all, the burials from Cameroon seem to be primary inhumations (at least those from Akonetyé and Campo; the layout of the pits at Mpoengu have not been described),Footnote 31 while the Early Oveng burials are secondary, as we discussed above. However, the iron artefacts deposited inside the graves are quite similar: the axes, bracelets, anklets, and special-purpose money from Akonétye are almost identical to the ones from Corisco. Special-purpose money is also found at Mouanko-Epolo.Footnote 32 In contrast, pottery vessels, which sometimes were placed upside down, differ stylistically from Corisco pottery, although the flat bases and some lip decoration remain similar.Footnote 33 This is not surprising, since metallic implements tend to have greater stylistic similarities across large areas than pottery, which often has more localized stylistic variation and distribution.
For Congo we have taken into account 11 archaeological sites, namely Tandou-Youmbi, BP 113, Mandigo-Kayes, Meningue, Fignou 4, Kayes, Fignou 1, Tchitembo, Lac Ndembo, Condé, and Loubanzi. All of them are located in coastal Congo, north of Pointe-Noire.Footnote 34
All findings come from pits that yielded mostly pottery and carbonized oil palm nuts and, to a lesser degree, iron slag and tuyère fragments. According to James Denbow these sites can be divided chronologically into Early Iron Age (Tandou-Youmbi, Kayes, Fignou 4, BP 113, Mandigo-Kayes, Lac Ndembo, Meningue, Fignou 1, and Tchitembo) and Late Iron Age (Condé and Loubanzi). We have only considered the Later Iron Age sites that do not contain European trade goods. Radiocarbon dates place all Iron Age sites in this area between the second century bce and the seventeenth century ce. A gap in the sequence from the ninth to twelfth centuries ce separates the Early and Late Iron Age.
The Early Iron Age findings, characterized by pits, suggest a sedentary occupation. Even though the pottery found inside the pits is partially coeval with the Oveng material, it bears little stylistic resemblance to it, which is not surprising, given the distances involved.Footnote 35 Among the features that recall the Oveng style, we found comb-stamped herringbone surface decoration, which predominates in the Loango coast during the first half of the first millennium ce, and grooved decoration found on pottery from BP 113 site, overlying the Early Iron Age horizon.Footnote 36 Mandigo-Kayes and Lac Ndembo have yielded a different pottery style named by Denbow as ‘spaced curvilinear ware’.Footnote 37 Some of the pottery from these sites, dated to between the seventh and eighth centuries ce, seems to resemble the Oveng style.Footnote 38
The Later Iron Age sites of Condé and Loubanzi yielded a quantity of carved wooden roulette decorated sherds quite similar to those found in Angondjé tradition pottery both in Corisco and Libreville Estuary, with which they are contemporary.Footnote 39 They are the ‘woven motifs’, also found on cloth, ivory, and woodcarvings in Loango, northern Angola, and the Democratic Republic of Congo. Those motifs were even tattooed on human bodies on the nineteenth-century Loango coast, as Debow emphasizes.Footnote 40
Finally, some dates are available for Bioko island and the coast of mainland Equatorial Guinea. The date samples from Bioko were gathered during the colonial period by missionaries and put together and published by Bernard Clist.Footnote 41 The two radiocarbon dates from coastal mainland Equatorial Guinea come from the archaeological reconnaissance undertaken by Clist in 1985.Footnote 42
Perhaps the most outstanding cultural characteristic of the population of the island of Bioko was the absence of metallurgy, while it was widespread among the continental groups. This original feature made missionaries and colonizers consider the Bubi as people living in a ‘Neolithic stage’. It is to missionaries that we owe the archaeological investigation of remains related to the Bubi people. Amador Martín del Molino outlined a five-stage cultural sequence for the Neolithic of Bioko: Carboneras, Bolaopí I, Bolaopí II, Buelá, and Balombe.Footnote 43 The Carboneras phase, dated to the second half of the first millennium ce, was the more accurately studied period.Footnote 44 Little resemblance exists between Carboneras ware and the pottery of Corisco, except for a globular vessel with open rim decorated with wavy comb impressions and gourd-shaped pots that are reminiscent of those from the Nandá tradition.Footnote 45
The survey undertaken by Clist along the coast of continental Equatorial Guinea uncovered several Iron Age archaeological sites.Footnote 46 Among them, Akom and Ayene are the most interesting ones: Akom consists of several refuse pits, and Ayene is a shell midden. Both sites have been dated to the first half of the second millennium CE. Akom has yielded four iron bracelets similar to the simple ones found in Corisco. The pottery sherds found at Ayene fit the Angondjé tradition.Footnote 47
DISCUSSION
The information provided by the excavation of the site of Nandá and the survey around the coast of the island of Corisco allows us to suggest a sequence for the island's Iron Age.
Compared with the available data from the continent mainland, the sequence of Corisco shows chronological and cultural particularities. To begin with, the Early Iron Age, associated with the Oveng tradition, seems to start at the end of the first millennium bce. In the Libreville Estuary area, it starts around the same date, or maybe some time before. The more likely hypothesis is that Corisco began to be inhabited at this moment by populations coming from the coast, thus marking the beginning of the Iron Age on the island. However, earlier Late Stone Age visits have been attested by lithic artefacts located during survey (both in Corisco and in the nearby island of Elobey Grande).Footnote 48
Calibrated dates from Corisco.
Despite this evident connection and reciprocity with the Gabonese coastline, we must emphasize the particularities of the historical process of Corisco. The archaeological groups documented on the island have some resemblance to those from Gabon, but they show technological and stylistic specificities that lead us to define them as technological variations of those archaeological groups. Likewise, we perceive a cultural connection with the Iron Age of southwestern Cameroon, as indicated by the same iron items buried in funerary contexts, although ceramics are different. In general, the area that extends from south Cameroon to Congo went through a similar process around the middle of the first millennium ce (fifth to seventh centuries depending on the area) resulting in a reduction in the number of sites and/or an impoverishment of material culture, as documented in Corisco. This shows that, despite its insular character, Corisco reflects wider patterns well and that data from the island can be used as a proxy to better understand historical processes in the Gulf of Guinea as a whole.
At the same time, similarly to Western Europe, we could speak about different iron ages.Footnote 49 Despite sharing a common socioeconomic and technological background and the existence of contacts and migrations, every region has its chronological and technological particularities. In our case, the Early Iron Age would have started some time later than on the continental mainland (around the turn of the era) and would have finished around 100–200 years later, with strong signs of cultural and demographic decline. Although the population came from the Gabonese coast and was therefore part of the same archaeological group, the inhabitants of Corisco developed cultural peculiarities, especially in the case of ceramics, although not so in the case of metals. This seems to suggest a common intragroup function and meaning for the former, and a common coastal intergroup symbolic meaning for the latter. In this respect, the special-purpose money represents the most significant class of object, since its use as bridewealth was documented at least until the beginning of the twentieth century among local groups like the Fang.Footnote 50
In Corisco, the end of the Early Iron Age and the beginning of the Late Iron Age is separated by three centuries (seventh and eighth to tenth centuries ce) during which the island was probably abandoned. We lack information to clarify the archaeological sequence of the Gabonese coast during this period. The limited information available points to a possible demographic collapse.
As we have seen above, the Nandá group appears on the island around the eleventh to twelfth centuries ce. We could consider Corisco a peripheral and marginal place only occupied during climatic periods and socioeconomic contexts characterized by coastal population growth. Actually, the decay of Oveng and the depopulation of the island coincide with the period 400–700 ce when a general population drop occurred across all West Central Africa.Footnote 51 This has been associated with an increase in humidity followed by a regression of the savanna and an expansion of rainforest ecosystems.Footnote 52 From 1000 ce a new population growth is noticed.Footnote 53 This phase coincides fully with the emergence of the Nandá/Angondjé traditions in Corisco and, consequently, with the beginning of the Late Iron Age it lasts until the thirteenth century ce, when the island was definitely abandoned until the eighteenth century. Once again, chronological mismatches between the island and the continental coast are documented: in Gabon the Angondjé tradition overlaps with the era of growing European contact.
The Late Iron Age runs for a period of three centuries in Corisco and its remains challenge the current archaeological sequence that has been proposed for the area. To address this issue we will make some preliminary clarifications. Throughout this article we have used the term archaeological group when referring to material culture with a certain degree of homogeneity in their technological, geographical, and chronological features. A point that we would like to clarify is that we understand these archaeological groups in a different way from the classic model of ‘archaeological cultures’.Footnote 54 Our starting point is a critique of the culture-history paradigm and its direct association of the archaeological record with ethnicity, language, or race.Footnote 55 This is the case with the traditional association of some pottery decoration with population movements, Bantu languages, iron metallurgy and food production, because in most cases the archaeological and linguistic data present much more complex scenarios.Footnote 56 The case of Corisco is one such case of a complex scenario. An analogous opinion is defended by Denbow, when he declares that
whereas it had been thought that the linguistic and cultural history of western Bantu south of the tropical forest was a relatively straightforward one of gradual expansion of a single cultural unit southward, the archaeological data suggest much more complex processes of ethnogenesis in the period prior to AD 800.Footnote 57
In Gabon, Nandá and Angondjé pottery styles were identified as two different traditions belonging to different human groups, as the scarce radiocarbon dates for the Nandá (Group II) period seemed to suggest. However, the dates associated with Nandá and Angondjé traditions in Corisco overlap (see Table 2). Considering Nandá and Angondjé traditions as two different human groups has been the traditional hypothesis since this is what the dates from Gabon seemed to imply. The great difference between pottery wares make this point believable, as we, as archaeologists, are accustomed to infer the existence of different human groups from different wares. However, the dating from Corisco suggests otherwise: the fact that Angondjé dates share the same timespan with Nandá on such a small island suggests that they could be the same human group. However, we must bear in mind that the traditions do not share physical space in Corisco as far as we know, but the fact is that the distance between them is very small (around 100–200 m).
There are more arguments supporting the hypothesis that considers Angondjé and Nandá the same group. First of all, the small number of radiocarbon dates from Gabon associated with the Nandá tradition and the larger assemblage of dates from Corisco suggest that the traditions coexisted, at least in Corisco. The great differences between pottery wares could be explained as follows: all Nandá pottery has been recovered from a funerary context, while Angondjé pottery comes from domestic areas. Although it is not very common, there exist archaeological examples of human groups that have two different wares for two different life domains, as happened with the Moche of ancient Peru, in the Meroitic Kingdom, and in the Argaric Bronze Age Culture of southeast Iberia.Footnote 58
In this particular case, we have to refer to the concept of archaeological group or tradition exclusively in relation to material culture, with the aim of ordering it in space and time, combining association-dissociation and synchrony-diachrony. We must emphasize as well that our study of material culture takes into account another critique of culture-history classic typologies, too focused on the final form of objects and too concerned with decorated items. Our analysis of material culture stems from the concept of ‘chaîne opératoire’, in which every technological decision involved in the manufacture of any object is considered culturally significant.Footnote 59
As noted above, the pottery of the Late Iron Age burials (Nandá tradition) is characterized not so much by its shape or decoration, but for having been fired at a very low temperature, thus producing ceramics that crumble easily. Rather than indicating a lack of technical proficiency, those technological decisions speak about the funerary rituality during that prehistoric period on the island. It seems to us that the rituals performed during the burial of the corpse involved the production of a high quantity of pots, expressly manufactured for the dead. Those pots could have contained food and drink (palm wine) for the dead, and none of those pots would be reused. However, Angondjé pottery differs from Nandá not only in shape and decoration, but in its manufacture, which is much more suitable for cooking and storing – indeed many Angondjé pots show traces of soot. All in all, we conclude that the ‘chaînes opératoires’ differ, as do the contexts in which they appear. Based on the available data, we suggest that the pottery traditions Nandá and Angondjé do not represent two different human groups; rather, they belong to the same tradition but relate to two different communities of practice – everyday life and burial.
At the moment, however, it is perhaps prudent to allow space for other interpretations given the limited data at our disposal, data entirely sourced from a 15 km2 island. Thus, while the chronological sequence Nandá–Angondjé has to be abandoned with our radiocarbon dates, it is not totally implausible that two different ethnic groups cohabited in the region during the Late Iron Age, as cases of multiethnic communities or different ethnic groups living side-by-side is certainly not unknown in the area.Footnote 60 Against this theory, however, runs the fact that no clear Nandá-style pottery appears in Angondjé assemblages or vice versa. We know of multiethnic contexts in many areas of Africa, but in the cases that have been studied ethnoarchaeologically, there is always some mixture of different traditions in the assemblages of each ethnic group.Footnote 61 Therefore, although we remain cautious with regard to the interpretation proposed here (same group, different pottery styles for different social contexts) and are open to other possibilities as the archaeological record of Equatorial Guinea and Gabon becomes better known, we consider that the presence of two different ceramic traditions fulfilling different social functions is the most viable hypothesis at the moment.
This has some consequences for our knowledge of the Late Iron Age communities of the region. Given that the same ceramic assemblages appear in coastal Gabon and southern Equatorial Guinea, we can conclude that this was definitely not an isolated occurrence. The sites of Corisco suggest that we have to be open to unexpected cultural phenomena in the still scarcely-known record of Central Africa, and that we must not take for granted the pottery-archaeological culture equation. Other, more complex scenarios are possible. The existence of a domestic/funerary duality in the material culture of Late Iron Age Corisco, in fact, has important cultural and social implications. It underscores the autonomy of the funerary world as a realm of social life with its own rules, both material and immaterial, a world that does not necessarily replicate with exactitude the realm of the living.
