LIBRARY AND MUSEUM REPORTS 94. THE SETTLEMENT-MOUNDS OF CENTRAL GONJA Provisional Chronology and Historical Context by Colin Flight* - # > • • •• The last five years have seen a very rapid advance in archaeo- logical research in Ghana. Much of the momentum has come of course from the Volta Basin Research Project, set up to co-ordinate research, in archaeology as in other fields, in the area to be flooded by Lake Volta, and generously financed by the Ghana Government, there has been little time up till now for publishing or publicizing the results of this work, but a collection of interim reports was brought out a year ago which is still useful (Archaeology in the Volta Basin, 1963-66, hereafter abbreviated as AVB)i at an end. the excavation programme is now more or less ' About half of the sites Investigated recently by the V.B.R.P. are in the Gon(a district of Northern Ghana,in the area upstream and downstream from the confluence of the Black and the White Volta. The protohistoric settlement-mounds characteristic of this part of Gonja (Davies, 1964, 1967), none of which had been excavated before, have proved exceptionally rewarding (AV&j 22-41, passim). that the history of Gonja has already been studied in some detail. Two contributions should be mentioned particularly. Dr. Jack Goody has worked on Gonja for many years from an ethno-historical angle (Goody, 1954, 1967, etc.); and Professor Ivor WiIks will soon be publishing, with Nehemia Levtzion, an invaluable eighteenth-century Arabic tarikh, the Kltab Ghunja or 'Book of Gonja1. This then Is the time, before ideas on either side became settled, for discussion and sharing of information between historians and archaeologists; and it is as a contribution to that dialogue that this paper is intended. It happens * Mr. Colin Flight is a Lecturer in the Department of Archaeology. LIBRARY AND MUSEUM REPORTS 95. HISTORY Several accounts of Gonja history have been published, all of them based very largely on the corpus of oral tradition which Jones (1962) has called the 'Jakpa epic'. Jakpa, so the story goes, was a mighty warrior 'from Mande', who fought his way across Gonja from west to east, and then, before he was killed in battle, shared out the lands which were his by right of conquest among his sons. The earliest recorded version of the Jakpa epic, in substantially its modem form, Is to be found in an Arabic chronicle written in the 18901s (El-Wakkad andWMks, 1962). By contrast, the outline offered here relies primarily on the KifabGhunja, compiled In about 1751 (Wilks, 1966). This work has been known for some time In an English translation made forty years ago and published later by Goody (1954: Appendix IV). Several manuscripts have been located over the last five years, and a defini- tive edition Is in preparation. work that the Jakpa epic In its current form is a relatively recent development, summarizing something like a century and a half of early Gonja history. For the sixteenth and seventeenth century the Kit5b Is itself based on oral tradition; but it is earlier than any other recorded version by well over a hundred years, and very much more coherent and convincing. It can be shown by reference to this The arrival in Gonja of the Ngbonyo, the immigrant rulers, is described in a section of the Kit5b which has become detached from the main body of the work but survives independently (Wilks, 1966). Nabac, who was to be the first king of Gonja, had come south originally on a punitive expedition despatched by the 'Chief of Mande-Kabba1 against the trading-town of Begho. He then turned north to attack Buna, and across the Black Volta Into western Gonja. Here he built a fortified camp or stronghold called Yogbum. the KItcb GhunjS gives no dates as such for the early kings, only the lengths of their reigns; dead-reckon Ing would put the beginning of Nabac>s reign at 1549-50. Wilks (1966) believes that this date may be too early, by LIBRARY AND MUSEUM REPORTS 96. as much as fifty years, pointing out that the reigns ascribed to these early kings are on average much longer than those of the eighteenth- century rulers. made in the succession system at the beginning of the eighteenth century, however, his argument fails to convince. In view of the drastic change which seems to have been On the early history of the kingdom the Kitab has little to offer. Its author's main interest, not surprisingly, was with the conversion of the ruling dynasty to Islam early In the reign of Mawura (1580-1599?). By that time the capital of the Gbanya kingdom was at Buipe in central Gonja, on one of the trade-routes leading northwards from the margins of the forest. It may be possible to distinguish a phase of consolidation in the Buipe area followed by a phase of rapid expansion to the east and north. By around 1600 Gonja had probably reached more or less its full extent. The next phase, which the difficulties of commyn(cation .must have made almost inevitable, was one of progressive decentralization. In the Kitab this trend is reflected in the passage dealing with the long reign of al-Lara (1623-1667), of whom it is said that 'he divided the country of Gonja and gave it to his brothers'. divisional chiefdoms, originally perhaps appointive, had by now become vested In certain families which were, or which chose to think of them- selves as, cadet branches of the Gbanya dynasty. In formally recognizing the divisional chiefs as his 'brothers', al-Lata was at least by implication admitting himself as king to be only first among equals. It seems that the This tendency at work within the kingdom for the divisions to Increase In power at the expense of the centre, compounded as It was with attacks from outside, led into a phase of rapid disintegration at the end of the seventeenth century. The reign of al-Lata's son Sulayman was remembered as a time of continual war. He was deposed In 1689; and a few years later, the Gbanya kingdom finally fell apart with the outbreak of civil war In 1692. The Kitab, though it is clear on the disastrous side-effects of the war, says nothing of the aims for which it was fought, or of the LIBRARY AND MUSEUM REPORTS 97. It seems apparent none the less that one of its results it achieved. results, and probably one of its objectives, was to overthrow the power of the Buipe dynasty. By 1709 Gonja was on its way to recovery, organized now on a confederate system which, with modifications in detail, survives today (Goody, 1967). The capital was transferred to Nyanga in western Gonja, where Nabac has built his war-camp, and a paramount chief was installed there whose title now is Yagbumwura, but who seems originally to have been called Gbinipewura. Seven of the divisions, Tuluwe, Kpembe and Kong among them, but for reasons which can only be guessed at not Daboya, were made 'gates' to Yagbum; that is, their chiefs were eligible for promotion to the paramountcy when it fell vacant. The number of 'gates' has varied with the course of time, but the chief of Buipe has always been rigorously excluded. Though his position even today is one of considerable prestige, his political power is negligible. SETTLEMENT-MOUNDS The group of sites discussed here, except for Jakpasere, are all settlement-mounds, analogous in structure with Near Eastern tells though on a smaller scale. Mounds of this type are very common in central Gonja, the main concentration lying .in the confluence area. In general they are irregular in plan and profile and very variable in size; the maxi- mum is probably about 250 feet across and 30 feet high.. They tend to occur in groups, either in clusters as at Butie (AVB: 31 -2), or In lines as at Silima (New Buipe) and Kjsoro. ColossalTaobab trees 30 feet or more in girth are usually found on the mounds or nearby« Over the last few years eight of these mounds have been excavated, and some cautious generalizations are possible. The larger mounds at least all seem to be of much the same origin. They consist as a rule of a series of superimposed building levels, each some 4-5 feet thick, with the collapsed or demolished superstructure filling in the rooms between the Mathewson, 1968a. Our views on mound formation are for the * most part in close agreement. A LIBRARY AND MUSEUM REPORTS 98. upstanding walls. The buildings were invariably rectangular, divided up into small compartments 8-10 feet across, with flat roofs. The volume of collapsed debris leaves little doubt that these buildings were multi-storeyed, like the 'tower-houses' at Butie of which almost nothing now survtves, but which according to one description were originally five storeys high. It is easy to imagine how the upper storeys when they collapsed would tend to fall inwards, filling the groundfloor rooms, and how the grid of surviving walls would then retain the debris and protect it from erosion, in this way a mound would form, and grow rapidly In size as the process was repeated. In mound C at Silima four structural periods were identified; two or three is probably more usual. CHRONOLOGY Accurate dating is impossible except for those sites where locally In the Accra area, Paul made tobacco-pipes occur in the later levels. Ozanne has shown how the design of tobacco-pipes underwent rapid change, especially in the configuration of the base, during the seven- teenth and early eighteenth century. Thanks to extensive documentation from European sources, it has often been possible to date the abandon- ment of individual sites in this area with certainty or fair probability, and thus to fit a precise time-scale on to the typological sequence (Ozanne, 1962). Only one large series from northern Ghana has been analysed in detail, a collection of over a hundred pipes from Yendi Dabarl (Shinnie and Ozanne, 1962). The abandonment of this site can fortunately be dated with reasonable confidence to 1713-4 (Ozanne, 1964). The latest pipes at Yendi Dabari are of so-called quatrefoll type, in which the base, or the collar on the stem-socket, or both, are In the shape of a square with rounded projections at the comers. Closely related designs are found also on the coast, where their first appearance is dated to about 1690/95. All things considered the date cannot be very much different inland. Perhaps on general grounds It might be a little earlier. In a Gon(a context, then, it coincides conveniently, but presumably only by chance, with the end of the Gbanya kingdom. **• ' * - - • *• LIBRARY AND MUSEUM REPORTS 99. Only four sites In Gonja, all of them recently excavated, have produced tobacco-pipes In any quantity: Silima, Krunkrunmbof, Kebltokebombi, and Jakpasere. No information is available on the Kebitokebombl and Jakpasere series. The nineteen pipes from Krunkrun- mboi are all datable to the second half of the seventeenth century; with one possible exception the more elaborate designs which ought to come In at the end of the century are not represented. Probably therefore the site was abandoned around 1690. The series of over 150 pipes from Silima has not yet been fully analysed. Typologlcally It runs closely parallel with the Yendl Dabari series. Quatrefoil types occur, but in a smaller proportion here, it would seem, than at Yendi Dabari; this would suggest that Silima was abandoned around 1700. Radiocarbon dating Is frankly of little use. Several samples have been processed, from Silima and from Juni, but the significance of the results should not be over-rated. They do no more than confirm a dating for these sites around the middle of the second millennium. For proto- hlstoric Gonja carbon-14 Is probably less reliable than judicious guess- work. CROSS-DATING This section deals with each of the excavated sites in turn, It is primarily concerned with questions relying as far as possible on the interim reports in AVB, but including some unpublished information. of relative chronology. Cross-dating between one site and another is by no means easy, and for the moment depends very largely on just two criteria: has been worked through, other lines of evidence may be found to confirm or contradict the correlations suggested here; but on the whole the pottery seems to differ so much from place to place that accurate Cross- dating may never be possible except over short distances. None the fe$% tobacco-pipes and KIsoto bowls. Perhaps, when all the pottery The dating of the SI lima sequence Is discussed in more detail below. % */ * -* LIBRARY AND MUSEUM REPORTS 100. it Is already possible to work out a pattern of cross-dating, however tentative, which seems to make sense historically. 1. Krunkrunmboi (8° 4 5 ' N, 1° 30'W) The Krunkrunmboi mound was excavated by the present writer in 1966 (AVB" : 39-41). Three main structural periods were identified, with tobacco-pipes in periods II and I I I. Correlation with the Silima sequence is discussed below. 2. SJHma (8 4 6 ' N, 1° 29'W) The line of three very large mounds at New Buipe, excavated by Richard York in 1965-67, is a sife of crucial Importance. Dr. Jack Goody, in a paper shortly to appear in Ghana Notes and Queries, identifies it as the abandoned town of Silima. To avoid any chance of confusion between this site and Buipe itself, 7 miles distant to the north- west, this name is used here. All that has been published so far is an interim report on the first season's work (AVB: 33-7), by now in places out of date. A paper is forthcoming, also in Ghana Notes and Queries, which summarizes the cultural sequence as It Is now understood. Neither the earlier Iron Age levels, periods II and III, under- lying the mounds and dated by radiocarbon to the late first millennium, nor the traces of even earlier occupation, period I, are dealt with here. Period IV, thought at first (AVB; 34-5) to belong with the later levels, is now seen to be closer to periods ll-lll In cultural tradition, and probably in date as well. After period IV the site was apparently abandoned for several centuries. It is the last four structural periods, Silima V through VIIU which are relevant to the present discussion. The chronology proposed by the excavator for these periods would seem to be open to argument at two critical points. Firstly, in the excavator's view, the site continued In occupation throughout the eighteenth century; and secondly, tobacco- pipes are said to occur as early as period V. To fudge by the tobacco- * ;*-t. c LIBRARY A ND MUSEUM REPORTS 101 pipes, however, Sillma would seem to have been abandoned In about 1700, a century earlier than the excavator would allow. doubtful whether tobacco-pipes occur In mound A before the end of period V I I. In that case Silima VIII would synchronize roughly with Krunlcrunmbol l l - l l l, falling into the latter half of the seventeenth century. It also seems For cross-dating with sites further east, the occurrence of a few KIsoto bowl fragments at Silima is important. They seem to be found mostly in periods V and V I. 3. Asumpe (8° 43 *N, 1° 08'W) A low mound at Asumpe was excavated by Duncan Mathewson in 1965 (AVB: 31). No internal structures were identified, and the formation of the mound is hard to interpret. Dating depends on the occurrence of Kisoto bowls, which Imply a correlation with some part of the Sllima V-VI or Kisoto Ml sequence (see below). 4. Kisoto-East (8 4 9 ' N, 1° 05'W) The mound at Kisoto excavated by Mathewson In 1966 was the largest of a string of six (AVB; 28-31). The upper levels, periods I and I I, were formed in the normal way, by the collapse or demolition of storeyed buildings. same origin, but not unlikely. It is uncertain whether the basal levels are of the This was the first site from which the distinctive grey-ware bowls now known as Kisoto bowls were recovered In any number (AVB: 30). They are recorded from fifteen sites so far, concentrated in an area on both sides of the White Volta upstream from the confluence. Silima and Kebi- tokebombi, where they are rare, are marginal to the main distribution, but * Duncan Mathewson and the present writer have a paper in preparation dealing with Kisoto bowls. LIBRARY AND MUSEUM REPORTS 102. Important as the only sites where they can be shown on direct evidence to belong, In part at least, to an earlier period than tobacco-pipes. At KIsoto they are found throughout period In the basal levels. This would suggest that KIsoto HI are broadly contemporary with Slllma V-VI (Mathewson, 1968a). For the time being there Is no other evidence for cross-dating the two sequences, but the complete absence of tobacco-pipes at KIsoto does seem to Imply that nothing occurs here parallel with the latest stages of the Sifima sequence. I and II, and possibly also 5. Junl (9° 06'N, 1° 03'W) The excavation by Mathewson In 1967 of a mound near Kusawgu is as yet unpublished. Three structural periods were recognized, all of which are taken to represent collapsed tower-houses. KIsoto bowls occur In the two upper periods, and possibly In period I. A straightforward correlation of Junl II with KIsoto I, and of Junl III with KIsoto I I, seems permissible. 6. Kebitokebombi (8° 32'N, 0° Little information is available on the large mound near the Sheribong river excavated by York in 1967 (the site Is referred to briefly In AVB: 22). Tobacco-pipes and a few KIsoto bowl fragments are reported, and cross-dating with the Silima sequence should there- fore be possible in due course. 7. Jakpasere (8° 29'N, 0° 31 'W) Jakpasere In eastern Gonja Is not a mound, but a large ruined building claimed to have been one of Jakpa's palaces. Trial excavation by Mathewson in 1966 (AVB; 25) has recently been followed up by further work. The tobacco-pipes recovered are all of seventeenth- century types (Mathewson, 1968b). V •* A- « LIBRARY AND MUSEUM REPORTS 103. TOWARDS SYNTHESIS On any interpretation the period from 1550 to 1700 which saw the rise and fall of the Gbanya kingdom coincides with part at least of the settlement-mound sequence. Historically we can recognize some of the main trends affecting the Gbanya kingdom in the course of time, a period of aggressive expansion followed by decentralization and in the end by disintegration. Archaeologically we have a series of sites in central Gon[a, all of the same type, showing a pattern of settlement and abandonment which can be ordered chronologically on a relative basis, but only the last part of which can be dated in terms of years. The problem is to bring these two process-models together. Settlement-mounds were not abandoned everywhere at the same time. The latest stage in the sequence is represented at only two of the excavated sites, both of them in the Buipe area. On the evidence of tobacco-pipes Silima and Krunkrunmboi went out of occupation at about the same time, around 1700; and the obvious inference is that they were abandoned during the civil war which began in 1692, according to the Kitab Ghunja, and lasted for twelve years. It would in fact be very surprising if the fighting had not left some mark archaeologically on thfe area. The war was a painful episode which the author of the Kit5b, writing only forty or fifty years later, preferred to gloss over; but he does make special mention of the depopulation which it caused. 'Then a war broke out1, he says, 'in which many people were killed. Brother killed brother, and those who were spared had no strength left in them; their villages disappeared as well as their power.....' On this point, then, historical and archaeological evidence fit together so well as to leave little doubt that settlement-mounds fell out of use in the Buipe area at the very end of the seventeenth century. Further east, beyond the White Volta, Juni and Kisoto were apparently both abandoned at about the same time, possibly during the second quarter of the century, possibly earlier. This seems to have been a period of peace and prosperity, to judge by the KItSb, during which the population might b© expected to have Increased rather than declined. LIBRARY AND MUSEUM REPORTS 104. (Krunkrunmboi was apparently founded in about 1620, and may perhaps reflect this trend). No wars or Invasions are on record which would explain why these sites were abandoned, and never resettled. Perhaps a political explanation is the most likely. It may be that the centrifugal tendencies at work during the reign of a I-La ta (1623-67) were marked by a progressive abandonment of tower-houses in the outlying areas. If this is true, Ff the decline and disintegration of the Gbanya kingdom find archaeological expression In the excavated settlement- mounds, two conclusions would seem to be in order: first, that tower- houses were somehow part of the political superstructure imposed on this area by the Ngbanya; and second, that the settlement-mound sequence begins not earlier than about 1550. Archaeologically there is some reason to think that tower-houses are an Intrusive feature. The three major sites, Sillma, Kisotoand Juni, all seem to have been founded (or In the case of Silima refounded) at least approximately at the same time. AH belong to a uniform and highly specialized architectural style which appears to have no local ancestry. What Is especially significant, the distribution of tower-house mounds does not correspond with any underlying cultural continuum. The pottery is very far from uniform. The Silima pottery, for instance, compares closely only with that from Krunkrunmboi and from an unexcavated mound across the river at Kadelso (Davies, 1964), Were It not for the fortunate chance that a few KIsoto bowls found their way as far as Silima, cross-dating with sites further east would be very difficult. The pronounced local differentia- tion obvious in the pottery may be supposed to reflect the ethnic diversity of the Indigenous population, fust as tower-houses reflect the political structure Imposed on this heterogeneous substratum. There is no direct evidence for dating the beginning of the mound sequence accurately enough for the purposes of historical correlation, but a date of 1550 would not be unacceptable archaeologically. A later date, as suggested by Professor Wilks, seems out of the question. Even to fit the whole sequence Into a space of 150 years is probably to compress it as tight as it will go. V , - * *• , *• * LIBRARY AND MUSEUM REPORTS 105. The sJmpfej't and most likely interpretation Is that the Gbanya kingdom and the settlement-mound sequence are correlated, not fust in part, but completely. The Inception of tower-houses would coincide with the arrival of the Ngbanya, and their distribution would define the area of initial consolidation in central Gonja. The absence of mounds outside central Gon[a would suggest that the outlying provinces were from the beginning administered differently; but even in central Gonfa the trend towards decentralization during the seventeenth century Is marked by the abandonment of tower-houses beyond the White Vofta. Around Buipe itself tower-houses only fell out of use during the civil war which tore the Gbanya kingdom apart at the end of the century. It only remains to emphasize in conclusion the provisional I am very grateful to Richard character of this attempted synthesis. York and Duncan Mathewson, who did most of the hard work, for Informa- tion on the sites they have excavated, and for discussing and disagreeing with me ever the last two years. They do not necessarily share my views. A consensus Is neither possible ribr'deslrable at this time. I would certainly not want to see the Interpretations set out here accepted without argument; but of this, I dare say, there Is little danger. LIBRARY AND MUSEUM REPORTS 106. References AVB: Archaeology In the Volta Basin, 1963-66, by R. York, D. Mathewson, D. Calvocoressl, and G. Flight. (Legon: Volta Basin Research Project and Department of Archaeology.) Davles, O ., 1964. 'Gon|a painted pottery'. Trans. Hist. Soc. Ghana 7: 4 - 11. j 1967. West Africa before the Europeans. (London: Methuen.) El-Wakkad, M ., and I. Wilks, 1962. 'Qfisatu Salaga Tarikhu Gon|a*. Ghana Notes and Queries 3 : 8- 31 and 4t 6 - 25. Goody, Jack, 1954. The Ethnography of the Northern Territories of the Gold Coast west of the White Volta. (London: Colonial Office.) , 1967. (O.U.P.) 'The Over-Kingdom of Gonja1. In (eds.) D. Forde and P.M. Kaberry, West African Kingdoms In the Nineteenth Century: 179-2051 Jones, D . H ., 1962. 'Jakpa and the foundation of Gonja'. Trans. Htst. Soc. Ghana 6: 1 - 2 9. Mathewson, R.D., 1968a. 'SomeNotes on the Settlement-mounds of Central Gonla'. Research Review 4(2): 1 0 8 - 1 4. , 1968b. 'Excavations at Jakpasere, 1968'. Forthcoming. Oranne, Paul, 1962. 'Notes on the Early Historical Archaeology of Accra*. Trans. Hist. Soc. Ghana 6t 51 - 7 0. LIBRARY AND MUSEUM REPORTS 107. - , 1964. 'The Archaeologfcol Contribution to theAshantl ReMarch Project1. A.R.P., First Conference. Reportt 5 0 - 9. (Lefloni I . X XJ Shlnnl«# P . L, and Paul Ozanns, 1962. 'Excavations at Y*ndl Dabarl1. Trans. HI»t. Soc. Ghatw 6t 8 7 - 1 1 8. . Tomllnson, H . H ., 1954. Th« Custorm/ Constitution and History of th» Gonja People. Wllks, Ivor, 1966. 'A note on the chronolog/, and origins, of H»e Gonfa kings'. Ghana Notes and Queries 8i 2 6 - 2 8. York, R.N., 1968. 'Excavations at New Buipe: a preliminary report1. Ghana Notes and Queries, forthcoming. <•+ • -C