NYAME AKUMA No. 79 JUNE 2013

archaeological and ethnographic map for the area, BÉNIN focusing on past craft productions. In this sense the 2013 fieldwork was very much in continuity Excavations at Tin Tin Kanza (TTK), with the 2011 and 2012 seasons (Haour et al. 2011; Northern Bénin Nixon and Haour 2012) but it also placed a much greater emphasis on regional test pitting. The over- Nadia Khalaf all aim was to achieve a wide picture of the historic Anne Haour landscape along the River between Pékinga Sainsbury Research Unit for the Arts of in the west of and Birnin Lafiya in the east, , the Americas and Oceania a distance of some 60km (Figure 1). This objective University of East Anglia was reached by conducting a series of test pit exca- Norwich, Great Britain NR47TJ vations in areas highlighted for their archaeological [email protected], potential during survey in previous seasons. [email protected] Tin Tin Kanza had been selected as a site for test-pitting for four main reasons: the presence of a past settlement site was known from ethnograph- ic interviews (Gosselain 2012); it lies at a point at which the floodplain is very narrow compared Abstract to other stretches of the river, arguably leading to more focused settlement; its location corresponded Excavations on a large settlement mound roughly with the historically-known site of Katan- nearby the small village of Tin Tin, Republic of ga, reputed to be an early focus of settlement; and Bénin, close to the , were carried out finally, it lies close to two places of current and past over two weeks in January-February 2013. Prelim- ceremonial significance (Kanza cave and ‘Nooru inary results are reported here from a 2x1m test pit, Bangu’, the cowry marsh). The site itself lies next ‘Trench II’, located on the lower half of the mound. to a track, on the edge of a small tributary that runs This excavation was extremely revealing in terms into the Niger. A mango tree sits by the location of both of structures and artefacts: it included sever- the settlement mound, marking its picturesque loca- al floor surfaces, wall stumps, and stratigraphy, as tion by the water (Figure 2). The mound evidences well as fragmented ceramics and iron artefacts. The a high density of surface pottery. An escarpment purpose of this paper is to place Tin Tin Kanza in meets the top of the mound, where Kanza cave is the published record given the threat posed to it by located. development activities in the area with the hope that work will take place in the near future at this impor- tant site. Excavation

The test pit discussed here was one of two Introduction at the site (see Champion and Haour, this volume, for a discussion of the other). It was placed within Excavations at Tin Tin Kanza were un- a modern field, about two thirds of the way up the dertaken in the context of the European Research mound, and close to a very large grinding stone sit- Council-funded ‘Crossroads of Empires’ project 1 ting on the surface. The trench was 2x1m in size led by Anne Haour . The project overall seeks to and aligned north to south, with the two metre sec- make a first inventory of archaeological sites in this tions aligned east- and west- facing. Following ini- part of the Niger Valley and, ultimately, to yield an tial clearing of surface finds, the excavation began

101 NYAME AKUMA No. 79 JUNE 2013

Figure 1: Map: location of test pit and sites in the vicinity. in spits. As is well known, given the difficulty in ent. Excavation thereafter continued by contexts. defining stratigraphic differences in deposits- dur ing excavation (these often only become apparent The two uppermost contexts had been sub- in section), excavation by spits has long been the jected to varying degrees of disturbance and were preferred option in West African [see excavated quite quickly. The top-soil (the upper- e.g., Kasana in (Nilirmi 2008); sites in Yobe, most 25cm), Context 1, loose in compaction and northern (Hambolu 2000); Gajiganna, composed of a mid to dark brown, sandy silt, was northeast Nigeria (Wendt 2007: 27) and Akumbu in heavily disturbed by cultivation and tree rooting, the Méma in (Togola 2008: 27)]. However, but included pottery, disarticulated animal bone, Tin Tin Kanza presented us with an unusual oppor- and small traces of shell. Beneath this was a con- tunity to try out context-based excavation, for after text, which had seen substantial disturbance and removal of the first spit, it became clear that distinct ashy contamination from burnt tree rooting in the stratigraphic and architectural features were pres- northern end of the trench. This Context 2, on the whole, was more compact compared to the topsoil

102 NYAME AKUMA No. 79 JUNE 2013

Figure 2: Photo taken from the top of Tin Tin Kanza mound, overlooking the Niger River. and comprised of a mid to light yellowy brown san- partial remains of a potsherd pavement (Pavement dy silt. Once again, the deposit featured many pot- 1), and what appeared to be a complete, but broken, tery and bone inclusions. vessel in the west section, were notable occurrences here. Pavement 1, centred in the middle of the exca- Beneath this was a mixed deposit (Contexts vation unit, was composed of one complete vessel 3-5) composed of grey clay building material, red- that had a vertical stratigraphy spread in a circular pigmented soil, and charcoal inclusions. A single fashion (c. 0.27m in diameter) integrated with other cowry shell was recovered here. Although the red potsherds; thus this pavement included a variety stratum (context 6) was very thin and fragile during of decorated and undecorated sherds, incorporated excavation it, and another that appeared subsequent- with pebbles (Figure 3). The complete vessel was ly, showed up very clearly in the trench sections; we decorated with folded strip roulette. Charcoal from interpret these as floor layers. Indeed, underneath Context 8, which included this fragmented pot and were the remains of another more substantial floor, pottery pavement, subsequently yielded a date of which included large fragments of pottery. The grey 920 ± 30 bp (Beta 348774), after calibration at 2 clay reminiscent of building material or rubble (Con- sigma AD 1030-1190, AD 1200-1210. text 7) was again present, perhaps associated with the destruction and levelling of earlier features. A The pavement was carefully removed, along hearth associated with a large smashed pot, the very with the layers of large pottery fragments. The de-

103 NYAME AKUMA No. 79 JUNE 2013

Figure 3: Potsherd pavement 1 with associated pot.

104 NYAME AKUMA No. 79 JUNE 2013

Figure 4: Chalk floor with associated postholes, pottery and clay-hardened surface.The rectangular shape in plan close to the large posthole was a result of over-digging. North is at left. posit beneath was reddish brown in colour and had Judging by their shallowness it may be that these a moist, friable texture. This context contained frag- postholes were cut from a higher level but had been ments of animal bone, some of which had degraded missed in excavation due to the nature of the depos- onto the pottery. This deposit sealed a semi-circular its. chalk floor, 1.26m x 0.94m in size; the eastern edge of the floor remains unexcavated, but it appears the Context 11, sitting directly on top of this floor could be a complete circle. The chalk floor, chalk floor, was a mixed deposit of reddish brown, although only 0.05m thick, was very compact. It friable silty sand, which had animal bone and pot- had five clear post-holes cut into it and was over- tery inclusions. There was also charcoal present lain at the northern edge by hardened, cracked clay within this context, which produced a date of 1120 (Figure 4). The postholes were very generic in their ± 30 bp (Beta 348775), after calibration at 2 sigma form and shape, all circular in plan, with flat bases AD 880 - 990. and steep sides. The post in each case had not sur- In the layer (Context 13) directly above the vived. The postholes were excavated independently, chalk floor were several refitting sherds while at the and assigned individual context numbers (14-23). north end of the excavation unit and associated with

105 NYAME AKUMA No. 79 JUNE 2013

Figure 5: Hearth with associated iron object and (left) potsherd pavement 2.

Figure 6: Excavation of chalk floor and associated Pavement 2 and wall. Hearth seen in Figure 5, now removed, was at bottom left.

106 NYAME AKUMA No. 79 JUNE 2013

Figure 7: Pavement 3 and associated wall stump, left in-situ for next field season.

107 NYAME AKUMA No. 79 JUNE 2013 the chalk floor were the remains of another potsherd two areas; the fact that this wall appeared only as a pavement, ‘Pavement 2’, constructed using an inte- stump suggested its deconstruction at some point in gration of stones and pottery sherds. Interestingly, the past to make room for another living arrange- stone predominated and, unlike the case of the other ment, perhaps contemporary with the creation of pavements excavated, the pottery used in the sur- Pavement 3. The surface design of Pavement 3 is face was taken from the same pot and not pieced striking: potsherds were used to create leaf-shaped together from different individual sherds. patterns pressed into the stones making up the rest of the floor. It was at this stage that, faced with the Unfortunately this pavement was in the cor- end of the current field season, we resolved that we ner of the trench and its full extent could not be re- should continue excavation in 2014. vealed, but it appeared to be sitting on a wall. The wall was made up of very compact whitish grey building material, clearly a different deposit from the surrounding more friable material; it resembled The Pottery Assemblage an interior wall. This was the area where the hearth The ceramic assemblage from this trench, had been recovered, and it may be that the hearth consisting of 829 sherds, has been subjected to a cut the (earlier) pavement. Placed at the base of the preliminary study. Of these, 229 sherds can be dis- hearth (Context 28) were an elongated iron blade counted as they issue from entire vessels, which (Figure 5) and 11 pieces making up a small vessel, will be refitted and treated separately. These arise around 7cm in diameter, with a thickness of 1cm, in Context 8 (192 sherds decorated with a folded and undecorated. The co-occurrence of the vessel strip roulette), Context 26 (25 sherds, undecorated), and blade may suggest a possible ritualistic aspect. and Context 28 (11 sherds, undecorated). There was Charcoal associated with the hearth has returned also a single, burnished, worked sherd from Con- the date of 1010 ± 30 bp (Beta 345503), after cali- text 8, which we similarly discount here. bration AD 990-1040, AD 1110-1120. The hearth fill was sampled for archaeobotanical remains. Of the remainder 600 sherds, (Table 1), we can observe the following. Decorated sherds The potsherd pavement and the wall were account for between 50 and 80% of the material maintained in the trench until the circular floor had within levels, with no clear pattern from layer to been entirely excavated, in order to check that the layer. Globally, folded strip rouletting, which re- pavement had been fully exposed and did not con- sults from the use of a roulette made of one or more tinue beneath the chalk floor. Figure 6 shows the de- strips of flat-sectioned materials folded into an ac- posit underneath the chalk floor: a mottled deposit cordion shape, is the most common decoration and of red, white and grey, perhaps a packing material is consistent throughout contexts. Pierced potsherds to seal and level previous floors. At this stage the (couscoussière) are quite common in the top of the trench was levelled to this mottled surface through- stratigraphy, less so after context 9. We plan to con- out. tinue the analysis of this material in parallel with Following the removal of the circular that from the other test pits excavated in the course floor, potsherd pavement and hearth, a -grey de of the Crossroads of Empires project. posit with some large stone inclusions and very few archaeological materials was revealed. It was thought that we were approaching the sterile level, Conclusion but this proved not to be the case: underneath this deposit was uncovered a very fine potsherd pave- Security permitting we intend to return to ment, Pavement 3 (Figure 7). It covered the entire Tin Tin Kanza in January 2014, which should al- excavation unit, and the stump of a wall separated low us to reach a much better understanding of this

108 NYAME AKUMA No. 79 JUNE 2013

-

1 4 1 7 1 4 1 3 10 19 10 54 34 22 11 14 261 186 414 600 142 Totals

0 0 0 0 0 0 6 1 0 0 0 0 0 0 0 0 0 11 48 48 30 29 Cont. Cont.

0 0 0 2 0 2 4 0 0 0 0 0 9 0 0 1 0 0 0 18 18 26 Cont. Cont.

0 0 2 0 0 0 0 2 0 0 0 0 2 0 0 0 0 0 15 17 13 25 Cont. Cont.

0 0 0 1 0 9 0 0 2 4 4 0 0 8 2 1 0 2 27 42 51 24 Cont. Cont.

0 0 0 0 1 0 0 0 0 2 0 0 0 0 0 0 13 11 15 26 10 13 Cont. Cont.

0 0 0 1 0 0 0 0 0 2 2 0 1 0 0 0 0 0 0 0 0 12 Cont. Cont.

0 1 0 0 1 3 4 0 0 1 6 3 3 0 0 0 14 42 56 24 10 11 Cont. Cont.

0 0 0 7 2 0 7 0 0 0 0 0 0 1 4 0 2 0 1 9 10 17 Cont. Cont.

0 0 0 7 0 0 8 0 0 0 1 7 0 0 8 0 1 0 0 8 16 24 Cont. Cont.

3 0 9 0 0 0 4 0 0 0 0 0 0 6 3 0 0 4 7 21 30 10 Cont. Cont.

0 0 0 8 0 0 6 0 0 0 0 1 1 0 6 1 0 0 0 5 11 17 Cont. Cont.

0 0 0 3 0 5 0 0 0 2 0 0 2 0 0 0 0 0 4 16 18 23 Cont. Cont.

0 1 0 3 0 4 0 5 0 2 0 1 0 0 3 0 27 13 11 51 64 10 Cont. Cont.

0 1 0 5 0 0 0 0 4 6 0 0 0 2 1 0 2 24 33 36 69 26 Cont. Cont.

2 1 0 0 1 0 0 2 0 4 5 1 0 0 1 69 46 10 69 54 12 138 Cont. Cont.

ated

Total Undec Total Dec Detail of pottery recovered from Tin Tin Kanza Trench II. All contexts that included material large enough for analysis (sherds All contexts that included material large II. Trench Kanza Tin Tin Detail of pottery recovered from Couscous Carin Rims Body sherds BOT Rim, FSR Rim, channel cord Rim, braided Rim, burnished Rim, incised Body, FSR Body, incised Body, burnished Body, channel Body, channel FSR+ Body, incision FSR+ Body, ONE Body, slip TOTAL

UNDEC DECORATED

Table 1: Table of a diameter above 25cfa coin) are included here. Detail contexts which did not feature any material: Context 6 – red floor; Con text 10 – pottery pavement; Contexts 14, 16, 18, 20, 22 postholes; 15, 17, 19, 21, 23 posthole fills (of contexts 18, 20 and 22 respectively); Context 27 – cut of floor; 28 hearth fill. Complete vessels are not included in the totals above. They comprise fragmented vessels in contexts 8, 26 and 28, as discussed the text. FSR = folded strip roulette. ONE one-fibre strip. BOT=bottle neck. Decoration terminologies follow Livingstone Smith et al. 2010.

109 NYAME AKUMA No. 79 JUNE 2013 complex site. Future excavation, involving section- Akuma 76: 23-32. ing the wall stump and Pavement 3, will for ex- ample confirm the relationship between wall and the pavement. One obvious question is how far the sequence will extend, and what was the initial date Livingstone Smith, A., O. Gosselain, A. Mayor, A. of settlement of the site; as a reminder, our three and S. Ndèye Guèye current radiocarbon dates, reported above, span the 2010 Roulettes in sub-Saharan Africa today. In very late ninth to the very early 13th centuries. At a Anne Haour, K. Manning, N. Arazi and O. first assessment, Tin Tin Kanza seems to share simi- Gosselain, editors, African Pottery larities with sites of the Kainji area some 200km Roulettes Past and Present: Techniques, downstream (Nzewunwa 1983: 42-43) and thus Identification and Distribution. Oxford: may hold a unique promise for an archaeological Oxbow, pp. 1-90. understanding of West African pavement cultures. There is also a question of archaeological mitiga- tion: we learnt in April 2013 – just six weeks after our excavations there – from Mardjoua Barpougou- Nilirmi, N.S. ni, an MA student associated with our project and carrying out follow-up investigations in the region, 2008 The archaeology of : A study of that widening of the Birnin Lafiya-Pékinga track di- Kassana, Upper West region, Ghana. In T. rectly threatened the site. Insoll, editor, Current Archaeological Research in Ghana. Oxford: BAR International Series 1847, pp. 103-115. Bibliography

Nixon, S., and A. Haour Gosselain, O. 2012 Potsherd Pavements and Associated 2012 Mission au Dendi 2012. Unpublished field Structures From First and Second report. Millennium AD Sites in Northern Benin. Presentation to African Archaeological Research Days, University of Southampton UK, 3rd – 4th November 2012. Hambolu, M.O.

1996 Recent excavations along the Yobe Valley. In Vorträge, Internationales Symposium Nzewunwa, N. SFB 268, Frankfurt am Main, 13.12-16.12. Volume 2: 215-229. 1983 A Sourcebook for Nigerian Archaeology. Lagos: National Commission for Museums and Monuments.

Haour, A., O. Banni Guené, O. Gosselain, A. Liv- ingstone Smith and D. N’Dah Togola, T. 2011 Survey along the Niger River Valley at the Bénin-Niger Border, Winter 2011. Nyame 2008 Archaeological Investigations of Iron Age

110 NYAME AKUMA No. 79 JUNE 2013

Sites in the Mema Region, Mali (West Footnotes Africa), Cambridge Monographs in African 1 Archaeology 73. Oxford: BAR (ERC Independent Starter Grant # 263747) International Series 1736.

Wendt, P.K.

2007 Gajiganna: Analysis of Stratigraphies and Pottery of a Final Stone Age Culture of Northeast Nigeria, Journal of African Archaeology Monograph Series, Vol. 1. Frankfurt a. M.: Magna Verlag.

111